Anda di halaman 1dari 698

US.

Department of Transportation
Federal Highway Administration

Publication No. FHWA-HI-94-021 February 1994

NHI Course No. 13127

Pavement Deflection Analysis


Participant Workbook

National Highway Institute

Contents
Section 1 - Course Introduction 1.1 Deflection Measurements and Their Uses 1.1.1 Surface deflections 1.1.2 Deflection Uses 1.2 Course Objectives Section 2 - Materials Characterization 2.1 Introduction 2.1.1 General 2.1.2 Need for Elastic Moduli 2.2 Elastic Moduli in Pavement Systems 2.2.1 Elastic Modulus 2.2.2 Modulus of Elasticity Not a Measure of Strength 2.2.3 Determination of Elastic Modulus 2.2.4 Laboratory vs. Field Moduli 2.2.5 Different Types of Moduli 2.2.6 Difference Between Modulus of Elasticity and Resilient Modulus 2.2.7 Poisson's Ratio 2.2.8 Nomenclature and Symbols 2.2.9 Stress Sensitivity of Moduli Laboratory Determination of Elastic Moduli 2.3.1 Introduction 2.3.2 Diametral Resilient Modulus 2.3.3 Triaxial Resilient Modulus 2.3.4 Flexural Modulus SHRP Resilient Modulus Laboratory Tests [2.3] 2.4.1 Asphalt Concrete SHRP Protocol P07 2.4.2 Asphalt Treated Base and Subbase 2.4.3 Unstabilized Materials SHRP Protocol P46

Page
1-1 1-1 1-1 1-3 2-1 2-1 2-2 2-4 2-4 2-5 2-5 2-7 2-8 2-8 2-10 2-13 2-13 2-13 2-13 2-13 2-17 2-21 2-21 2-21 2-21 2-23

2.3

2.4

Page
2.5 Typical Values of Elastic Moduli 2.5.1 Typical values of modulus of elasticity for various materials 2.5.2 Typical Pavement Materials Estimating 2.6.1 2.6.2 2.6.3 2.6.4 2.7 Elastic Moduli of Pavement Materials Asphalt Concrete Moduli Portland Cement Concrete Moduli Stabilized Materials Moduli (includes base, subbase and subgrade) Unstabilized 2-24 2-24 2-24 2-25 2-25 2-26 2-27 2-27 2-41 2-41 2-41 2-42 2-43 2-44 2-45 2-46 2-47

2.6

Variations in Modulus 2.7.1 General 2.7.2 Temperature 2.7.3 Moisture 2.7.4 Freeze-Thaw Conditions 2.7.5 Time of Loading 2.7.6 Stress Level 2.7.7 Material Density Poisson's Ratio

2.8

Section 3 - Fundamentals of Mechanistic-Empirical Design 3.1 Introduction 3.1.1 Overview of Historical Development 3.1.2 Reasons for Using Mechanistic-Empirical Procedures (rather than empirical) Layered Elastic Systems 3.2.1 Assumptions and Input Requirements 3.2.2 One-layer System With Point Loading (Boussinesq) 3.2.3 Odemark's Method i3.181 3.2.4 Two-layer System (Burmister) 3.2.5 Multi-layer System 3.2.6 Elastic Layer Computer Programs 3-1 3-1 3-1

3.2

3-7 3-7 3-10 3-16 3-20 3-24 3-24

Page
3.3 Analysis of Rigid Pavements 3.3.1 Introduction 3.3.2 Continuously Supported Slab Models 3.3.3 Elastic Layer Model 3.3.4 Finite Element Models 3.3.5 Coupled Models Design Process 3.4.1 Flexible Pavements 3.4.2 Rigid Pavements Existing Overlay and Mechanistic-Empirical Design Procedures 3.5.1 Introduction 3.5.2 New design Procedures 3.5.3 Overlay design Procedures Example 3.6.1 3.6.2 3.6.3 3.6.4 3.6.5 3.7 3-36 3-36 3-36 3-38 3-39 3-39 3-40 3-40 3-58

3.4

3.5

3-63 3-63 3-63 3-72 3-88 3-88 3-89 3-91 3-92 3-94 3-94 3-94 3-96 3-102 3-102

3.6

Introduction Asphalt Institute Effective Thickness Procedure Asphalt Institute Deflection Procedure WSDOT Mechanistic-Empirical Summary

Use of Elastic Analysis Software 3.7.1 Introduction 3.7.2 Software Demonstration 3.7.3 Description of "Standard" Sections 3.7.4 Classroom Exercise

Section 4 - Nondestructive Testing Devices 4.1 Introduction 4.1.1 Types of Data Collected 4.1.2 Benefits Surface Deflection Measurements (NDT for Structural Evaluation) 4.2.1 Deflection Measurement Uses 4.2.2 Categories of Nondestructive Testing equipment 4.2.3 Typical NDT Patterns 4-1 4-1 4-3

4.2

4-3 4-3 4-4 4-5

Page
4.3 Static or Slow Moving Deflection Equipment 4.3.1 Benkelman Beam 4.3.2 Plate Bearing Test 4.3.3 Automated Beams [4.31 4.3.4 Curvature Meters [4.3] 4.3.5 Typical Applications 4.3.6 Advantages/Disadvantages of Static or Slow Moving Load Deflection Equipment Dynamic Vibratory Load (Steady State Deflections) 4.4.1 General 4.4.2 Dynaflect 4.4.3 Road Rater 4.4.4 WES Heavy Vibrator [4.21 4.4.5 FHWA Cox Van (Thumper) f4.21 4.4.6 Typical Uses of Steady State Pavement Surface Deflections Impact (Impulse) Load Response Devices 4.5.1 General 4.5.2 Dynatest Falling Weight Deflectometer (FWD) 4.5.3 KUAB Falling Weight Deflectometer f4.5. 4J5, 4 J] 4.5.4 Foundation Mechanics Falling Weight Deflectometer [4.12] 4.5.5 Phonix FWD [M , 4J] 4.5.6 SASW Approaches 4.5.7 Typical Uses of Impulse Pavement Surface Deflection 4.5.8 Advantages and Disadvantages of Impulse Load Equipment Comparisons and Correlations Between FWD and Other Devices 4.6.1 Introduction 4.6.2 Comparisons Between Devices 4.6.3 Correlations Between Deflection Measuring Equipment Calibration of Load Cell and Deflection Sensors 4-6 4-6 4-8 4-9 4-9 4-9 4-9 4-10 4-10 4-11 4-14 4-17 4-17 4-18 4-18 4-18 4-21 4-25 4-26 4-27 4-27 4-28 4-28

4.4

4.5

4.6

4-29 4-29 4-30 4-37 4-45

4.7

Page
Section - Deflection Analysis Techniques Introduction 5.1.1 General Deflection Basin Parameters (Including 5.1.2 Maximum Deflections) Regression Equations for Predicting 5.1.3 Moduli 5.1.4 Surface Moduli Backcalculation 5.1.5 Combining Indices for Project Analysis 5.1.6 Joint Evaluation in Rigid Pavements 5.1.7 Void Detection in Rigid Pavements 5.1.8 Class Exercise A - Deflection Basin 5.1.9 Parameters Class Exercise B - Load Transfer 5.1.10 Efficiency Manual Backcalculation Initial Estimates 5.2.1 Class Exercise 5.2.2 Problem Summary 5.2.3 Automated 5.3.1 5.3.2 5.3.3 5.3.4 5.3.5 5.3.6 5.3.7 5.3.8 Backcalculation Introduction Typical Flowchart Measures of Deflection Basin Convergence Class Exercise Convergence Error Measure of Modulus Convergence Convergence Techniques Summary of Backcalculation Programs Verification of Backcalculation Results 5-1 5-1 5-1 5-4 5-12 5-15 5-15 5-26 5-42 5-51 5-56 5-58 5-63 5-66 5-73 5-74 5-74 5-74 5-76 5-83 5-84 5-85 5-90 5-95 5-101 5-101 5-103 5-109

Backcalculation of Rigid Pavements Introduction 5.4.1 Backcalculation Methods for Rigid 5.4.2 Pavements Backcalculation of Composite Pavements 5.4.3

5.5

Critical Sensitivity Issues in Backcalculation 5.5.1 Input Data 5.5.2 Compensating Layer and Non-Linearity Effects 5.5.3 Subgrade "Stiff' Layers 5.5.4 Pavement Layer Thickness Effects 5.5.5 Relative Layer Stiffness Effects 5.5.6 Seasonal Effects 5.5.7 Fixing Layer Moduli 5.5.8 Rules of Thumb Reliability and Errors in Deflection Analysis 5.6.1 Introduction 5.6.2 Types of Measurement Errors 5.6.3 Sources of Errors in Backcalculation 5.6.4 Effects of Errors on Backcalculated Moduli 5.6.5 Procedures to Minimize Errors Expert System

Page 5-112 5-114 5-114 5-115 5-135 5-140 5-142 5-144 5-145 5-151 5-151 5-151 5-155 5-156 5-159 5-159

5.6

5.7

Section 6 - Backcalculation Programs 6.1 Overview 6.1.1 6.1.2 Computer Programs for Backcalculation Selection of a Backcalculation Computer Program 6-1 6-1 6-6 6-13 6-14 6-22 6-35 6-48 6-49 6-49 6-65 6-66 6-66 6-67

6.2

Specific Programs 6.2.1 BOUSDEF 6.2.2 EVERCALC 6.2.3 MODULUS 6.2.4 MODCOMP Class Project Description 6.3.1 Perform Backcalculation 6.3.2 Perform a Basic Mechanistic-Empirical Analysis Additional Project Data 6.4.1 Data File Format 6.4.2 Project Data

6.3

6.4

Page
Section 7 - Course Wrap-Up 7.1 7.2 7.3 7.4 Summary and Review Future Trends in Pavement Deflection Analysis Questions and Answers Course Evaluation 7-1 7-2 7-3 7-3

SECTION 1.0 COURSE INTRODUCTION


1.1 D EFLEC TIO N M EA SU REM EN TS AND TH EIR USES 1.1.1 Surface Deflections A simple and convenient method to assess the structural integrity o f pavements is to apply a load to the pavement surface and measure the resulting deflections. In this course, a variety o f m ethods for utilizing pavement deflections to obtain information regarding the structural condition and load carrying capacity o f pavement systems will be presented. Pavem ent deflection measurement techniques are numer ous and can be categorized according to the characteris tics o f load applied to the pavement surface. Static or slow moving load deflection measurements represent the first generation approach which basically originated with the development o f the Benkelman Beam at the W ASHO Road Test in the early 50's. The next generation involved application o f a dynamic vibratory load, exemplified by the Dynaflect and Road Rater. These pieces o f equipment are more mobile and productive than the static equipment, and led to deflection measurements becoming a routine pavement condition survey task. Falling weight deflectometers can be considered third generation deflection equipment and measure deflections resulting from a dy namic impulse load which attem pts to simulate the effect o f a moving wheel load. Future equipment will likely measure deflections caused by an actual wheel load mov ing at highway speeds. 1.1.2 Deflection Uses Early use o f deflection data typically involved considera tion o f maximum deflection directly under the load, rela tive to empirical standards. Usually some statistical m easure o f deflections on a pavement section is compared w ith a "tolerable" deflection level for that section under the expected traffic. I f the measured value exceeds the

1-1

tolerable deflection then an empirical procedure deter mines the corrective measure required, usually an overlay, to reduce the measured deflections to the tolerable level. Examples o f this approach include The Asphalt Institute's M S-17 and CalTrans' Test M ethod 356. In some states maximum deflections are monitored during spring thaw and load restrictions are placed when the thawing pave ment's deflection reaches a certain level. Empirical use o f deflection basin data usually involves one o f the "basin pa rameters" which combine some or all o f the measured basin deflections into a single number. W ith a trend tow ards mechanistic pavement analysis and design, which is based on fundamental engineering prin ciples, the use o f deflection data has become more sophis ticated. Complete deflection basins are used, in a proce dure known as backcalculation, to estimate in-situ elastic moduli for each pavement layer. Knowledge o f the exist ing layer thicknesses are typically necessary for this pro cedure. A typical deflection basin is shown in Figure 1.1. The backcalculated moduli themselves provide an indica tion o f layer condition. They are also used in an elastic layer or finite element program to calculate stresses and strains resulting from applied loads. These stresses and strains are used with fatigue or distress relationships to evaluate damage accumulation under traffic and predict pavement failure. They can also be used to evaluate cor rective measures such as overlays, rehabilitation or reconstruction. It is these mechanistic analyses o f pave ment deflection that this course is intended to address. The backcalculation procedure is covered in detail in sub sequent sections. Briefly, however, it involves calculation o f theoretical deflections under the applied load using assumed pavement layer moduli. These theoretical deflections are com pared with measured deflections such as those shown in Figure 1.1. The assumed moduli are then adjusted in an iterative procedure until theoretical and measured deflection basins match acceptably well. The moduli derived in this way are considered represen tative o f the pavement response to load, and can be used to calculate stresses or strains in the pavement structure for analysis purposes. 1-2

TYPICAL DEFLECTION BASIN

DEFLECTION

(microns

or m ils)

Figure 1.1 - A Typical Deflection Basin

1.2 CO URSE OBJECTIVES

The specific course objectives are to familiarize participants with:

(a)

Empirical and mechanistic-empirical pavement deflec tion based design procedures, with emphasis on the latter. M aterials characterization for these procedures, with emphasis on elastic modulus. Selection o f deflection test equipment - strengths & weaknesses 1-3

(b)

(c)

(d)

Backcalculation procedures for flexible and rigid pavements theory and application. Deflection measurements and factors affecting them, including unusual field conditions. Errors in deflection data and how they affect backcal culation procedures. Practical applications o f backcalculation results.

(e)

(f)

(g)

1-4

SECTION 2.0 MATERIALS CHARACTERIZATION


2.1 IN TRO D U CTIO N General Backcalculation is an iterative process that uses a theo retical model, pavement layer thickness, Poisson's ratio, and estimated moduli, adjusted during the backcalcula tion procedure, to produce theoretical deflections that match field measured deflections within a specified tol erance. The end result o f the backcalculation process is a modulus value for each pavement layer. SECTIO N 2 concentrates on materials characterization, with the intent o f ensuring that course participants are familiar with typical moduli for common pavement materials. In particular, a participant should be able to deduce w hether any particular modulus produced by back calculation is reasonable, and, if outside typical expected values, the possible reasons for the deviation. Typical pavement materials will range from poor qual ity, unstabilized, natural, in-situ subgrades to high qual ity m anufactured materials such as Portland Cement C oncrete (PCC) and asphalt concrete (AC). It should be kept in mind that, although one o f the primary objectives o f pavement construction specifications is to ensure material consistency, significant material vari ability is common and will be reflected in the surface deflection measurements. In particular, subgrade response is likely to show the greatest variability, and subgrade response typically accounts for up to 90% o f the measured deflection for some pavements. This will be dem onstrated through the use o f layered elastic pro grams to be introduced later in this course.

2.1.1

2-1

2.1.2

N eed for Elastic Moduli M uch o f the structural deterioration o f a pavement structure is caused by the stresses or strains in the indi vidual materials o f the pavement. Strains at the bottom o f an asphalt layer are related to cracking o f the asphalt, while stresses or strains on top o f the subgrade may cause rutting or roughness. These stresses or strains are not related in any simple way to the overall deflection o f the pavement. Stresses, strains and deflections are all pavement response param eters and it may be tempting to assume that sim ple relationships exist between different types o f response. The relationships are. simple in a semi-infinite half space, but they are not in a layered system. That deflection is a poor substitute for strain, may be illustrated by an example. In Figure 2.1 the pavements are loaded by a 50 kN (11.3 kip) dual wheel load. The tire pressure is 0.7 M Pa (102 psi) and the distance between the tire centers is 350 mm (13.75 in.). In Case I, at the center point between tires, the applied load results in a deflection o f 0.464 mm (18.3 mil) and a maximum compressive strain on top o f the subgrade o f 476 (^strain (10-6 mm/mm), These deflections and strains, as well as those described subsequently, w ere calculated with the elastic layer program ELSYM 5. Case II is identical to Case I, except that the subgrade modulus is only half the value used for Case I. In Case II, at the same locations considered for Case I, the same load results in a deflection o f 0.705 mm (27.8 mil) and a subgrade strain o f 659 (^strain. Clearly the pave ment in case II has a poorer "bearing capacity" than Case I, exhibiting higher stresses and strains under the same load level.

2-2

50 kN (Pressure = .7 MPa)

AC 150 mm, 2000 MPa Base 300 mm, 300 MPa OVERLAYS :

Deflection Com pressive strain AC 185 mm 65 mm

V///

Base

ubgra^"
40 MPa Case 1 Defl. (mm) 0.464 Strain (*10A6)
4 7 6

Case II 0.705 659

Case III 0.594 476

Case IV 0.463 285

Figure 2.1(a)

AC

in., 290,000 psi


1 2

11250 lb. (Pressure = 102 psi) Base

4 3

50Q psj

OVERLAYS :

Deflection

ComDressive

7.3 in.

u b g ra c ^ )
'11,600 psi Case I Defl. (in.) Strain (*10A6) 0.018
4 7 3

iubgrac^'')
5,800 psi Case IV 0.018 285

5,800 psi Case II 0.028 659

5,800 psi Case III 0.023 476

Figure 2.1(b)

2-3

To reduce deflections and strains in Case II, an addi tional layer o f asphalt concrete (AC) (modulus 2000 M Pa or 290 K SI) could be added. To reduce the sub grade strain to the same level as Case I, 65 mm (2.6 in.) o f AC should be added. H ow ever to reduce the deflection. to the level for Case I, 185 mm (7.3 in.), almost three times as much AC is required. The use o f deflections as a direct measure o f the struc tural capacity o f a pavement should, therefore, be avoided. Instead the deflections should be used to determine the pavement layer moduli, and the moduli then used to com pute stresses or strains which can be used to evaluate structural capacity or remaining life.

2.2

ELA STIC M O D U LI IN PAV EM EN T SYSTEM S Elastic M odulus Elasticity refers to the ability o f a substance or object to return to its original state after undergoing deformation due to the application o f force. Elastic modulus is simply the stiffness o f a material within its elastic range. Elastic modulus has been adopted in the 1986 A ASHTO Guide for the Design o f Pavem ent Structures for characterizing paving materials [2.61 Elastic modulus is sometimes called Young's modulus since Thom as Y oung published the concept o f elastic modulus in 1807. Essentially, elastic modulus can be determined for any solid material and represents a con stant ratio o f stress (a ) and strain (e). E = stress/strain Thus, the "flexibility" o f any object (be it pavement or airplane or bridge or w hatever) depends on its elastic m odulus and geometrical shape. In fact, the product o f m odulus (E) and moment o f inertia (I) is a common measure o f structural stiffness.

2.2.1

(Eq. 2.1)

2-4

A material is elastic if it is able to return to its original shape or size immediately after being elongated or com pressed. Almost all materials are elastic to some degree as long as the load placed on a material does not cause it to deform permanently. H owever, in the case o f highway materials this often is not the case. 2.2.2 M odulus o f Elasticity N ot a M easure o f Strength It is im portant to remember that a measure o f a mate rial's modulus o f elasticity or the resilient modulus is not a measure o f strength. Strength is the stress needed to break or rupture a material (as illustrated in Figure 2.2), w hereas elasticity means that the material returns to its original shape and size. However, modulus o f elasticity is a measure o f material stiffness, and may provide an indication o f material condition or quality. 2.2.3 Determ ination o f Elastic M odulus Elastic moduli are generally determined by the follow ing tw o methods: 1. Lab Procedures D irect lab measurement o f resilient modulus can be per formed using AASHTO M ethod T292 and T294 for non-plastic subgrade and unbound materials and A STM D4123 for asphalt concrete and other stabilized m ate rials. These tests are fairly sophisticated and costly, and require experienced lab personnel to run them reliably. 2. N on D estructive Testing (NDT) N D T techniques are being used more than ever to assess the structural condition o f existing pavement systems. This assessment requires calculation o f pave ment layer moduli.

2-5

strengrth

Stress

>

Strain

Figure 2 .2 Sketch of Stress vs. Strain o f a Materia] in Compression

2-6

In general, tw o types o f N D T procedures may be used. These include: deflection based m ethods that utilize devices such as the falling weight deflectom eter (FWD), the dynaflect, road rater, etc., and w ave propagation techniques such as spectral analysis o f surface waves (SASW ) method (briefly covered in Appendix E).

In the deflection based methods the deflection basin data is analyzed, usually using elastic theory, to backcalculate moduli o f pavement layers and an average modulus o f underlying soils. W ave propagation techniques employ high frequency w aves o f extremely low magnitude to determine elastic properties o f the pavement layers. SASW methods are not yet autom ated and require significant effort, and as a result, are generally not utilized for production work. 2.2.4 Laboratory vs. Field Moduli Comparison o f moduli obtained from standardized laboratory tests with those backcalculated from field deflection measurements often produce varying results. This is not surprising, since it is fairly unlikely that conditions o f temperature, stress, moisture, loading rate, load duration, material volume and density, am ongst others, are likely to be the same for laboratory and field tests. M ost pavement materials are sensitive to one or more o f these factors in term s o f apparent modulus response, so that test results may need signifi cant adjustment prior to making any comparisons. As an example, a typical impulse load from a falling weight deflectom eter has a duration o f 25 - 30 ms, while dy namic load pulses o f 100 ms or more are fairly common for laboratory tests, with many tests using 100 ms as a standard. Asphalt concrete would show a modulus o f up to about 50% or more higher for the FW D load duration than the laboratory test at 25 C (77 F), all other factors being equal. This varies with tem pera ture, and the range is about 25% at 5 C (40 F); based on the Asphalt Institute equation [2.121. A. few degrees difference in tem perature can have the same effect as this difference in load duration.

2-7

The intent o f this brief discussion is to emphasize that com parisons should be made only if conditions are essentially the same for field and laboratory tests. I f this is not the case, adjustments should be made to "normalize" the tests (information provided in Section 2.6 should assist in doing this). 2.2 .5 Different Types o f Moduli Discussions about moduli can be complicated by the num erous kinds o f moduli such as: M odulus o f elasticity Diametral resilient modulus Triaxial resilient modulus Bulk modulus M odulus o f resilience M odulus o f rigidity M odulus o f rupture M odulus o f roughness Secant modulus Tangent modulus Young's modulus Shear modulus Fineness modulus ...etc. W e are only interested in the modulus o f elasticity and resilient modulus and throughout these notes, w e will use the term "modulus" to mean the same. 2.2.6 Difference Between M odulus o f Elasticity and Resilient M odulus W hat is the difference between modulus o f elasticity and resilient modulus? The modulus o f elasticity for a material is basically the slope o f its stress-strain plot within the elastic range (as shown in Figure 2.2). Fig ure 2.3 shows a stress versus strain curve for steel. The initial straight-line portion o f the curve is the elastic range for the steel. I f the material is loaded to any value o f stress in this part o f the curve, it will return to its original shape. Thus, the modulus o f elasticity is the slope o f this part o f the curve and is equal to about 207,000 M Pa (30,000 ksi) for steel. On the other hand, resilient modulus is usually based on stress and strain measurements from rapidly applied loads more like those that pavement materials experience from wheel loads.

2-8

Stress

Strain

Figure 2.3 - Stress-Strain Diagram for Steel

2-9

M any pavement materials exhibit a significant amount o f plastic or permanent deformation under applied loads, as well as an elastic or recoverable deformation. Resilient modulus is the ratio o f the applied stress to the recoverable (elastic) strain, i.e. resilient modulus relates to the elastic com ponent o f the response only. It is an estimate o f the modulus o f elasticity. This is illustrated in Figure 2.4. 2.2.7 Poisson's Ratio The other material param eter used in elastic analysis o f pavement systems is Poisson's ratio. This is defined as the ratio o f transverse to longitudinal strains o f a loaded specimen. This concept is illustrated in Figure 2.5. In realistic terms, Poisson's ratio can vary from 0 to 0.5 (assuming no specimen volume increase occurs after loading). Generally, "stiffer" materials will have lower Poisson's ratios than "softer" materials. You might see Poisson's ratios larger than 0.5 reported in the literature; however, this implies that the material was stressed to cracking, experimental error, etc. This can also occur in granular materials if applied stresses cause particle re-orientation which results in a volume increase. Poisson's ratio varies from .15 for Portland cement concrete to .45 for subgrade soils. Typical val ues are shown in Section 2.8. Poisson's ratio is tem perature sensitive but for backcalculation purposes it is always assumed to be constant.

2-10

Mr= a d/er

F igure 2.4 - R esilien t M odulus (M r) fo r a Plastic M aterial

2-11

. U2
L

Where

HeD 1

Poisson's ratio - strain along the diametrical (horizontal) axis

AD

strain along the longitudinal (vertical) axis

Figure 2 . 5 Illustration of Poissons Rato

2-12

2.2.8

N om enclature and Symbols The nomenclature and symbols from the 1986 A ASHTO Guide [2.6] will be used in referring to pavement moduli. For example: (a) E aC asphalt concrete elastic modulus base course resilient modulus subbase course resilient modulus roadbed soil (subgrade) resilient modulus

(b) (c)

EbS E sb
M r (or E sg )=

(d)

The only exception is that M r and E sg will be used interchangeably. 2.2.9 Stress Sensitivity o f Moduli Changes in stress can have a large impact on resilient modulus for certain types o f pavement construction materials. "Typical" relationships are shown in Figures 2.6 and 2.7, and are discussed in more detail later in this section. As shown in Figures 2.6 and 2.7, coarse grained materials tend to show stress stiffening behav ior and fine grained materials are likely to be stress softening.

2.3 LA BO RA TO RY D ETER M IN A TIO N OF ELASTIC M O D U LI 2.3.1 Introduction M oduli can be measured in the laboratory using the diametral or split tensile tests for bound materials such as AC or PCC, and the triaxial test for unbound mate rials. M oduli can also be measured using a flexural test. PCC moduli are often correlated to compressive or split tensile strength test results. 2.3.2 Diametral Resilient M odulus Diametral resilient modulus is the stiffness o f a material subjected to a repeated, dynamic pulse-type loading.

2-13

(6= 0 ,

+2 3 ) a

Figure 2 . 6Resilient Modulus vs. BuUc Stress for Unstabilized Coarse Grained Materials

(od = o 1 - C 3 )

Figure 2 . 7 Resilient Modulus vs. Deviator Stress for Unstabilized Fine Grained Materials

2-14

Diametral deformation is measured along the horizontal diameter (in fact, the term "diametral" simply means "diameter" or measured across a diameter). This test is most commonly used for AC materials. One standard method for this test is A STM D4123 Indirect Tension Test for Resilient M odulus o f Bitumi nous Mixtures. It generally takes about 10 minutes to test one sample. A compressive load (to produce tensile stress) is applied to an AC core or laboratory com pacted sample, typically 100 mm (4 in.) in diameter and 63.5 mm (2.5 in.) thick or 150 mm (6 in.) in diameter and 75 mm (3 in.) thick. The AC sample is loaded ver tically in compression (Figure 2.8) which produces a relatively uniform tensile stress across the vertical diameter {Figure 2.9). The horizontal deformation is measured with LV D Ts across the diameter o f the sam ple as shown in Figure 2.10. The formula below can be used to calculate the resilient modulus:

EAr = P ^ + 27)
w here (t)(A H ) E a c = asphalt concrete resilient modulus, psi, P H t = repeated load, lb., = Poisson's ratio (usually assumed), = thickness o f the sample, in.,

(Eq. 2.2)

AH = recoverable horizontal deformation, in. [To convert to M Pa use M Pa = psi/145]

2-15

Load

Figure 2 . 8Vertical Loading of ^ AC Core or Laboratory Prepared Specim en for Determining Diametral Resilient Modulus

Figure 2 . 9 Vertical Loading Produces a Relatively Uniform T ensile Stress Across the Vertical Diameter

Figure 2 . 1 0 Measurement of Horizontal Deformation in the Diametral Resilient Modulus Test

2-16

To conduct this type o f test, the needed test equipment includes (after A STM D4123): Testing machine capable o f applying a load pulse over a range o f frequencies, load durations, and load levels (typical load duration is 0. Is at 1 H z with load ranges 4 to 35 N/mm (20 to 200 lb./in.) o f specimen thickness (10 to 50 percent o f the AC tensile strength). Tem perature control system capable o f controlling tem peratures from 5 to 40C (41 to 104F). Typi cally, moduli are determined at 5, 25, 40C (41, 77, and 104F). M easurem ent and recording system. The horizontal m easurements are made with linear variable differ ential transformers (LVDTs) capable o f measuring deformations o f 0.00025 mm (0.00001 in.). Loads are measured with an electronic load cell. D ue to possible creep effects at the higher tem pera tures, caution is warranted for such resilient moduli results. 2.3.3 Triaxial Resilient M odulus One commonly used triaxial standard test m ethod is A ASHTO T292 and T294 (currently under revision). The specimen consists o f a cylindrical sample normally 4 in. (100 mm) in diameter by 8 in. (200 mm) high (Figure 2.11). The sample is generally com pacted in the laboratory; however, undisturbed samples are pre ferred if available (which is rare). The specimen is enclosed vertically by a thin "rubber" membrane and on both ends by rigid surfaces (platens) as sketched in Figure 2.12. The sample is placed in a pressure cham ber and a confining pressure is applied ( a 3) as sketched in Figure 2.13. The sample then undergoes repeated pulses o f an axial stress referred to as "deviator stress." This deviator stress is designated and it equals the total vertical stress applied by the testing apparatus (a^)

2-17

(100 mm)

F ig u re 2 . 1 1 Basic TriaxiaJ Specimen Configuration

-Platen
f c- *

-M

X - Sample

L *

Membrane

V V

4;
.

'

* ___ ^

'Pialen

F ig u re2 12 Enclosure of Triaxial Specimen

o3 confining stress
3 . F r .; -

Chamber

r ' >

Figure 2 . 13Triaxial Specimen in Pressure Chamber

2-18

minus the confining stress ( 0 3 ). In other words, the deviator stress is the repeated stress applied to the sample. These stresses are further illustrated in Figure 2.14. The resulting strains are calculated over a gauge length, which is designated by "L" (refer to Figure 2.15). As illustrated in Figure 2.15, the initial condition o f the sample is unloaded (no induced stress). W hen the deviator stress is applied, the sample deforms, changing in length as shown in Figure 2.16. This change in sam ple length is directly proportional to the stiffness. The following equation can be used to calculate the resilient modulus: M R ( o r E R) = ^ e.

(Eq. 2.3)

Mr (or Er) = resilient modulus,


w here a j = = P A er = = = = L = deviator stress, P/A repeated load, cross sectional area o f the sample, recoverable axial strain, AL/L gauge length over which the sample deformation is measured, change in sample length over the gauge length due to applied load.

AL

I f the material is relatively strongly bound then the confining pressure is not necessary and the modulus can be measured in uniaxial compression.

2-19

c, > total axial stress


o ^ deviator stress

c 3 - confining stress

*=0<j+ o 3

or

od *=o,-C 3

Figure 2 . 1 4 Stresses Acting on Triaxia] Specim en

No Load

L length over which repeated deformation Is measured

Figure 2 . 1 5 Gage Length for Measurement o f Strain o n Triaxia] Specimen

Figure 2 . 1 6 Deformation of Triaxia! Specimen Under Load

2-20

2.3.4

Flexural M odulus Flexural tests are most commonly used to determine the modulus o f rupture o f PCC. They are also used with cyclic loading to determine the fatigue characteristics o f bound materials, particularly AC. D ata from flexural tests can provide a flexural modulus. M odulus o f rup ture (M R) is defined as the maximum tensile stress in the beam sample at failure. Typically third-point load ing is used as in AASHTO T97-86 Flexural Strength o f Concrete (using simple Beam with Third-Point Load ing). PCC beams are usually 150 mm x 150 mm (6 in. x 6 in.) in cross section, with length more than three times the depth. AC fatigue tests have been performed on samples with cross-sections varying from 38 mm x 38 mm (1 X in. x VA in.) through 150 mm x 150 mm (6 A in. x 6 in.). The sample is loaded at the third point as shown in Figure 2.17. N ote t hat modulus o f rupture (M R) is the tensile strength in bending for PCC and is different from modulus o f elasticity (E) or resilient modulus (M r).

2.4 SHRP R ESILIEN T M ODULUS LABORATORY TESTS

[11]
2.4.1 Asphalt Concrete SHRP Protocol P07 (currently under revision): U ses the repetitive indirect tensile test (similar to A STM D4123). The preconditioning, applied stress and seating loads are shown in T able 2.1. 2.4.2 Asphalt Treated Base and Subbase SHRP Protocol P33 has been eliminated. For materials w ith sufficient cohesion SHRP Protocol P07 should be used. Otherwise use SHRP Protocol P46.

2-21

P/2

P/2

b = Thickness A = Deflection

77

23 PU 1296/A
b h 3

12
Figure 2.17 - Flexural Testing in Third Point Loading

2-22

2.4.3

Unstabilized M aterials SHRP Protocol P46: U ses a triaxial compression test conceptually similar to AASHTO T294 (which is under revision). The test sequence for granular materials is shown in Table 2.2 and fine-grained (cohesive) materials in Table 2.3,

Table 2.1.

SHRP Resilient M odulus Test Requirements Asphalt Concrete, Protocol P07.

Test Temp. (F) 41 77 104

Applied Stress* 30 15 5

Seating Stress* 3.0 1.5 0.5

Min. Load Applications 30 30 30

* as a percentage o f tensile strength at 77 F

Table 2.2.

SHRP Resilient M odulus Test Requirements N on-Cohesive (Unbound) Granular, (Base or Subbase) Protocol P46 Type 1 or Type 2 Repet. Appi. Dev. Stress, psi 15 3, 6 ,9 5, 10, 15 10, 20, 30 10, 15, 30 15, 20, 40 Min. Load Applications 500 100 100 100 100 100

Confining Press., psi 15 3 5 10 15 20

Comment Pre. Test Test Test Test Test

2-23

Table 2.3.

SHRP Resilient M odulus Test Requirements Subgrade Soils (Fine Grained), Protocol P46 Type 1 or Type 2

Confining Press., psi 6 6, 4 ,2

Repet. Appi. Dev. Stress, psi 4 2, 4, 6, 8, 10

Min. Load Applications 500 100

Comment Pre. Test

2.5 TY PICAL VALUES OF ELASTIC M ODULI 2.5.1 Typical values o f modulus o f elasticity for various materials include E o o o o o M aterial Rubber W ood Aluminum Steel Diamond
(psi)

(M Pa) 7 7,000-14,000 70,000 200,000 1,200,000

1,000 1,000,0002,000,000 10,000,000 30,000,000 170,000,000

Typical pavement materials E o o o o o o o o o M aterial Asphalt Concrete (32F (0C)) Asphalt Concrete (70F (21C)) Asphalt Concrete (120F (4 9 Q ) Crushed Stone Sandy Soils Silty Soils Clayey Soils Stabilized Soils Portland Cement Concrete (psi) 3,000,000 500,000 20,000 20,000-100,000 5,000-30,000 5,000-20,000 5,000-15,000 5,000-3,000,000 3,000,000-8,000,000 (M Pa) 21,000 3,500 150 150-750 35-210 35-150 35-100 35-21,000 20,000-56,000

2-24

2.6 ESTIMATING ELASTIC MODULI OF PAVEMENT MATERIALS Introduction In most cases it will be fairly obvious whether backcalculated moduli are reasonable for the type o f material in a given pavement layer. However, in some cases, a backcalculated modulus may be significantly different than expected, yet be feasible and consistent with the conditions at the time the deflection data was obtained. This section attempts to identify factors most likely to affect material response, and provides some guidelines in estimating moduli from information other than that used in the backcalculation procedure. 2.6.1 Asphalt Concrete Moduli (Range Approx. 345 MPa to 20,700 MPa (20,000 to 3,000,000 psi.)) i. ii. Laboratory tests for resilient modulus Shell method i.e. Van der Poel's nomograph (or McLeod's modification) The Asphalt Institute regression equation [2.12] (which is easily programmed into a spreadsheet): log |E*| = 5.553833 +

iii.

0.028829 (P 2 0 0 / f 0 17033) - 0.03476 Vv + 0.070377 ^ 7 0 ^ 106 + 0.000005 X (Eq. 2.4)

0.00189 (X /f u ) + 0.931757 ( 1 /f 0.02774) where |E *| = dynamic modulus (psi)


X

- tp (1-3 + 0.49825 l o g f ) p ac0.5

(Eq. 2.5)

2-25

and P 2 0 0 = % f aggregate passing f Vy = load frequency (Hz) = % air voids in mix


# 2 0 0

sieve

T 170F, 10 = absolute viscosity o f asphalt cement at 70F, in poises (.3 x 106 to 5 x 106) Pac tp = % asphalt (by weight o f mix) = temperature (F)

This equation is "highly satisfactory" for densegraded crushed stone and gravel mixes, but needs to be corrected for different mixes such as sandasphalt or slag asphalt [2 . 1 1 ]. iv. Emulsion Mixes a. b. 2.6.2 Uncured - treat as unbound material Cured - use (i), (ii) or (iii)

Portland Cement Concrete Moduli (Range Approx. 14,000 to 56,000 MPa ( 2 to 8 x 106 psi.)) i. ii. Laboratory tests E = 33 p 1 5 fc
0 5

(psi)

(Eq. 2.6)

p = unit weight o f concrete (pcf) fc = compressive strength at 28 days (psi) or E = 57,000 fc 0 -5 for normal weight PCC (psi)
(Eq. 2.7)

2-26

2.6.3

Stabilized Materials Moduli (includes base, subbase and subgrade) (Range Approx. 35 MPa to 14,000 MPa (5 x 103 to 2 x 106 psi.)) i. ii. Laboratory tests (all) Lime stabilized.

Compression E c (in ksi) = 1 0 + 0.124 UC, [2 J ] UC Flexure E f = 4.6 MR - 139 (ksi) , [2 J \ M R = modulus o f rupture (psi) iii. Cement Stabilized. Some general information is shown in Table 2.4. E = fc + 500 (ksi) [2.131 fc = compressive strength (psi) iv. 2.6.4 Asphalt stabilized. Similar to AC.
(Eq. 2.10) (Eq. 2.9) (Eq. 2.8)

= unconfmed compressive strength (psi)

Unstabilized (Range Approx. 35 to 690+ MPa (5,000 to 1 0 0 , 0 0 0 + psi.)) i. ii. or where Laboratory tests. Mr=k (psi)
(Eq. 2.11) (Eq. 2.12)

M r = k 1 0 ^ 2 (psi) ad
0

= deviator stress = bulk stress = a ! + + C3 J

2-27

Table 2.4 - Summary of the properties of Cement-Stabilized Soil. f2.81

(UC = unconfined compressive strength;

C = cement content, percent by weight)

Property
Density

Granular Soils
1.6-2.2 t/m2

Fine-Grained Soils
1.4-2.0 t/m3

Notes
May be higher or lower than untreated soil. Delay between mixing and compaction causes density reduction. UC in psi UC in MN/m2 d * age (days) (d > d0) (UC)d0 = UC strength at age of d0 days

Unconfined Compressive Strength

UC = (90 to 150) C UC = (0.5 to 1.0) C (UC) = (UC)d0 k = 7 0 C (p s i) k * 0.5 C (MN/m2)

UC = (40 to 80) C UC = (0.3 to 0.6) C + k log (k/do0) k = 10 C (psi) k = 0.7 C (MN/m2) To a few hundred psi To a few MN/m2 30 - 40

Cohesion

To a few hundred psi To a few MN/m2 40 - 45

Depends on C, d

c = 7.0 +0.225 (UC) psi c = 0.05 + 0.225 (UC MN/m2) Friction Angle Flexural and Tensile Strength Strength under combined stress states CBR May decrease at high confining pressures. Need 1 - 3% cement to develop. Relationships developed using Griffith crack theory UC in psi

Tensile Strength = (1/5 to 1/3) compressive strength (c , - c 3)2 = UC(a1+a3) for a3/UC < 0.1 t t a, = UC + 5ac for a3/UC > 0.1 (compression positives) CBR = 0.55 (UC)1431

2-28

Table 2.4 (continued) [2.81


Property
Modulus-Compression

Granular Soils
1 x 10s - 5 x 10s psi 7 - 35 GN/m2 r E ,_ [

Fine-Grained Soils
10* -1 0 s psi 0.7 - 7 GN/m2

Notes
Depends on stress level E, = initial tangent modules E* = tangent modulus
c3 = j

0 . 7 5 ( 1 - s in ) ( , - * , n 2 E
2 c cos < + 2CT3 sin<t> |> Ei = Kpa(a3/pa)n

confining pressure

J 1

pa = atmospheric pressure n =0.1 -0.5 k = 1 ,0 0 0 -1 0 ,0 0 0 4 = internal friction angle

Modulus - tension and Flexure Resilient Modulus Compression Resilient Modulus - Flexure

Same order of magnitude as in compression Ec > E* (usually) MR = Kc(a1 - a 3)^l(a3)''2(UC)" c k, = 0.2 to 0.6 k2 = 0.25 to 0.7 n = 1.0 + 0.18C m = 0.04(10)-186C Effect of confining pressure not known

Mrf = Kf(10)m- UC

Fatigue Behavior

No fatigue for F/T( < 0.50 Ti= initial tensile strength F= for a . + 3 o j > 0

8 ( , + s)
F = <r3 Poissons Ratio 0 . 1- 0. 2
c1 t

+ 3c 3 < 0 t

0 .15 -0 .3 5

2-29

Table 2.4 - (Concluded) [2jy


Property
Shrinkage

Granular Soils
A few tenths of 1%

Fine-Grained Soils
Upto1%

Notes
Shrinkage cracks generally inevitable

Thermal Properties (a) conductivity k = 0.6 k = 1.0 (b) Heat Capacity C = 0.82 C = 8.40 (c) Thermal Expansion k = 0.3 k = 0.55 BTU - ft/hr * ft2 * #F w/m * K BTU/lb * #F J/kg * 0 F-1

e = 5 x 10-*

s = 9 x 10-6
1 psi = 6.89 1 1<H PA = 6.89 x 10-J MN/mm2

c-1

iii.

See Table 2.5 Hicks & McHattie [2.10]. Table 2.6 and Figure 2.18 Rada & Witczak [2.7]. CBR estimates (subgrade). Generally these relationships were developed for CBR values o f less than 15-20% (see Figure 2.19).

iv.

Shell: E(MPa) = 10 CBR [E (psi) = 1500 CBR] W ES:E(M Pa)= 37.3 CBR0- [E(psi)=5409CBR-711]

(Eq. 2.13) (Eq. 2.14)

TRRL: E(MPa)=17.6 CBR 0 -6 4 [E(psi)=2550 CBR0-64]

(Eq. 2.15)

2-30

Table 2.5 - Selected Measured Dynamic Moduli for Pavement M aterial [2.101
Frequency & Unbound C r a n u l a r Base C o lo r a d o ; s t a n d a r d B a s e , V ' max and 8.7% < No. 2 0 0 ; s t a n d a r d s ubbase, 2 V and 7 . 9 \ < No. 200 C a l i f o r n i a ; w e l l - g r a d e d and subrounded g r a v e l 3 / 4 " max; c l a s s 2 a g g r e g a t e base 30 cpm, 0.1 sec 100 D ry 1 0 .0 0 0 t o 1 3 ,0 0 0 p s i , 8 . 0 0 0 t o 9 ,0 0 0 p s i , P a rtia lly satu ra te d 3% < No. 20 0, 0 3 * 55 7 . 0 0 0 t o 1 0 ,000 p s i , C a l i f o r n i a ; w e l l - g r a d e d and a n g u l a r c ru s h e d s t o n e 3 / 4 " max; c l a s s 2 a g g r e g a t e ba se 30 cpm, 0.1 sec 100 Ory 1 1 .0 0 0 t o 1 2 ,0 0 0 p s i , 1 4 .0 0 0 t o 1 5 ,0 0 0 p s i , P a rtia lly s a tu ra te d 3% < No. 20 0, o 3 * 57 10% < No. 200, o 3 * 57 9 , 0 0 0 t o 1 0 ,0 0 0 p i i , 7 ,5 0 0 t o 9 , 5 0 0 p s i , S o i 1- a g g r e g a t e o f 17% s i l t y sand and 83% c r u s h e d g r a n i t e ; 100% T - 1 8 0 ; w - c 5.1% C a l i f o r n i a ; w e l l - g r a d e d and subrounded g r a v e l 3 / 4 " max; c l a s s 2 a g g r e g a t e ba se ( d r y ) S ilty fin e sand 100% 33 cpm. 0.1 sec 10,000 1 . 8 5 6 0 * 81 3 . 1 2 6 0 * 37 AASHO T - 9 9 ; <0% < No. 200 w -c * 1 3 . *% Subgrade AASH0 c l a s s A -6 s i l t y c la y ; 120 cpm, 0 . 2 sec 33 cpm, 0.1 sec S i l t y c la y O v d (AASHO T e s t ) ; 20 cpm. 0 . 2 5 sec 5 t o 10 p s i ; o 3 < 3 p s i ; * 110 t o 115 p c f 10 ,000 1 0 ,000 3 , 0 0 0 t o 4 , 0 0 0 p s i , 18% w -c 7 ,0 0 0 t o 8 , 0 0 0 p s i , 16% w -c 1 5 ,0 0 0 t o 2 0 ,0 0 0 p s i , 14% w -c 3 , 0 0 0 t o 4 , 0 0 0 p s i . w e t season 1 ,5 0 0 t o 2 ,0 0 0 p s i . d r y season 1 3 ,0 0 0 p s i , 1 0 ,0 0 0 p s i , 8,000 p s i , 13% w -c 14% w - c 15% w - c 17% w - c 30 cpm, 0 . 2 sec 10 ,000 7 , 0 0 0 o 3 * 55 33 cpm, 0.1 sec 10,000 3 ,8 3 6 0 * 3% < No. 20 0, 0 3 * 57 10% < No. 20 0, O3 * S0 3% < No. 20 0, C3* 53 200, o 3 * ^ 9 8% < .N o .
m x

Load R e p e titio n 10,000

Dynamic Modulus l O ^ I S a ^ 1 *7 , 2.4% w -c *1 10 .019O3*1 5 , 6 . 3 \ **-c *6 8 , 6 8 7 a 3 * 496, 8.2% w - c

D u r a t io n 120 cpm, 0 . 2 sec

5 . 0 0 0 t o 7 ,0 0 0 p s i , 8% < No. 200, a 3 * 60

3 ,u s e * 55

w-c * 14 t o 18 p e r c e n t ;

* 110 t o 114 p c f sand s u b g r a d e

M icaceous s i l t y

7 . 0 0 0 p s i , 16% w - c 2 .0 0 0 t o 5 ,0 0 0 p s i . 2 , 0 0 0 p s i . 18% w -c

H ig h ly p l a s t i c c la y and s i l t y c la y

(PI 3 6 .5 ) 2 5 .5 ) c la y ;

30 cpm, 0 .1 sec 120 cpm, 0 . 2 sec 30 cpm. 0.1 sec

10,000 1 0 ,0 0 0

4,1 50o 3 ,2 0 0 o 5 * 2 7 ,0 0 0 t o 1 0 ,0 0 0 p s i . 20% w -c 1 5 ,0 0 0 t o 1 6 ,0 0 0 p s i , V8% w-c 1 4 ,0 0 0 t o 1 5 ,0 0 0 p s i , 16% w -c

(PI -

AASHO c l a s s A - 7 - 6 s i l t y w - c = 11 t o 20% y * 102 t o c la y 105 p c f AA^HO c l a s s A -6 t o A - 7 - 6 s ilty

100

10,000 p s i , te s t)

1 atm ( s o i l m o i s t u r e s u c t i o n d u r i n g

1 0 0,00 0 p s i . 10 atm

2-31

Table 2.6 - From Rada & Witczak 2.71

(a)
Typical values for k1 and k2 for unbound base and subbase materials (MR= k.,0 k ). 2 (a) Base Moisture Condition Dry Damp Wet k/ 6,000 - 10,000 4,000 - 6,000 2,000 - 4,000 (b) Subbase Dry Damp Wet 6,000 - 8,000 4,000 - 6,000 1,500 -4,000 0.4 - 0.6 0.4 - 0.6 0.4 - 0.6 k2 * 0.5 - 0.7 0.5 - 0.7 0.5 - 0.7

2-32

Table 2.6 (cont'd) - From Rada & Witczak [2.71

(b) Dry, Sr < 60 Percent Aggregate


DGA - limestone - 1a DGA Limestone - 2a CR-6-crushed stone3 CR-6-slaga Sand-aggregate blendsb Bank-run gravelb

Standard CE
ki 8,500 11,500 6,000 12,500 3,800 5,000

Modified CE
ki 10,500 15,000 9,000 20,000 6,000 8,000

k2
0.5 0.3 0.5 0.35 0.5 0.4

k2
0.5 0.3 0.5 0.35 0.5 0.4

ak1-k2 values are typical for fines percentage (no. 200) of less than 15-18 percent. bk1 value should be decreased and and k2 value increased if fines percentage is greater than 10 percent. (CE = Compactive Effort)

2-33

Table 2.6 (cont'd) - From Rada & Witczak [2.71

(b), (contd) Wet, Sr > 85 Percent Aggregate


DGA - limestone - 1a DGA Limestone - 2a CR-6-crushed stone3 CR-6-slag3 Sand-aggregate blendsb Bank-run gravelb

Standard CE
ki 7,000 6,000 3,500 5,600 1,900 1,250

Modified CE
ki 9,000 7,500 5,000 9,000 3,000 2,000

k2
0.4 0.5 0.7 0.35 0.7 0.7

k2
0.4 0.5 0.7 0.35 0.7 0.7

ak1-k2 values are typical for fines percentage (no. 200) of less than 15-18 percent. bk1 value should be decreased and and k2 value increased if fines percentage is greater than 10 percent. (CE = Compactive Effort)

2-34

Value

(psi)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

K2 V a l u e Figure 2.18 (Rada & W itczak 2.7])

2-35

Danish Road Laboratory: E(MPa) = 10 CBR 0 where CBR = California Bearing Ratio (%) N ote: Method also uses Eg= modulus o f granular layer on subgrade where Eg = K E sg K = factor varying between 1.5 and 4.8 depending on who is using method (Shell, TAI, Kentucky, etc.) Shell also suggests that K is a function o f the thickness o f the granular layer so that Eg = 0.2hg - 4 5 E sg where hg = granular layer thickness in (mm)
(Eq. 2.19) (Eq. 2.18) (Eq. 2.17)
73

[E(psi) =1500 CBR 0 -7 3 ]

(Eq. 2.16)

[Eg = 0.86hg 0 -4 5 E Sg for hg in inches] v. A relationship between R-value and modulus is: E Sg =1155 + 555R where R = Hveem stabilometer R-value. vi. General correlation relationships, e.g.,
Table 2.7 Yoder & Witczak [2,9] Table 2.8 Hicks & McHattie i2.101

(Eq. 2.20)

2-36

FIG. 2-19 : CBR-MODULUS RELATIONSHIPS

MODULUS

(M Pa)

CBR

----- SHELL

----------WES

------ ----- TRRL

----- ----- DK

2-37

Table 2.7 - Characteristics Pertinent to Road and Runway Foundation [2.91


Value as Foundation When Not Subject to Frost Action (5) Excellent Value as Base Directly Under W earing Surface (6) Good

Major Divisions (1) (2)

L etter (3) GW

Name (4) Gravel or sandy gravel, well gTaded Gravel or sandy gravel, poorly graded Gravel or sandy gravel, uniformly graded Silty gravel or silty sandy gravel Clayey gravel or clayey sandy gravel Sand or gravelly sand, well graded Sand or gravelly sand, poorly graded Sand or gravelly sand, uniformly graded Silty sand or silty gravelly sand Clayey sand or clayey gravelly sand Silts, sandy silts, gravelly silts, or diatomaceous soils Lean clays, sandy clays, or gravelly clays Organic silts or lean organic clays Micaceous clays or diatomaceous soils Fat clays Fat organic clays

Potential Frost Action (7) None to very slight None to very slight None to very slight Slight to medium Slight to medium None to very slight None to very slight None to very slight Slight to high

GP Gravel and gravelly soils

Good to excellent Good

Poor to fair

GU

Poor

GM

Good to excellent Good

Fair to good

GC Coarse grained soils

Poor

sw
SP

Good

Poor

Fair to good

Poor to not suitable Not suitable

Sand and sandy soils

su
SM

Fair to good

Good

Poor

SC

Fair to good

Not suitable

Slight to high

ML Low compressi bility LL < 50 Fine grained soils High compressi bility LL > 50

Fair to poor

Not suitable

Medium to very high Medium to high Medium to high Medium to very high Medium Medium

CL

Fair to poor

Not suitable

OL

Poor

Not suitable

MH CH OH

Poor Poor to very poor Poor to very poor Not suitable

Not suitable Not suitable Not suitable

Peat and other Pt fibrous organic soils

Peat, humus, and other

Not suitable

Slight

2-38

Table 2.7 (cont'd.) - Characteristics Pertinent to Road and Runway Foundation f2.91
Compressi bility and
Expansion

Drainage
Characteristics Compaction Equipment (10) (9)

Unit Dry Weight


(pcf) (11)

(8)

Field CBR (12)

Subgrade Modulus, k (pci) (13)

GW

Almost none

Excellent

GP Gravel and gravelly soils

Almost none

Excellent

GU GM

Almost none Very slight

Excellent Fair to Poor

GC

Slight

Coarse grained soils


SW Almost none Almost none Almost none Very slight

Poor to practically impervious

Crawler-type tractor, 125-140 rubber tired equipment, steel-wheeled roller Crawler-type tractor, 120-130 rubber-tired equipment, steel-wheeled roller Crawler-type tractor, 115-125 rubber-tired equipment Rubber-tired equipment, 130-145 sheepsfoot roller, close control o f moisture Rubber-tired equipment, 120-140 sheepsfoot roller

60-80

300 or more

35-60

300 or more

25-50 40-80

300 or more 300 or more

20-40

200-300

Excellent Excellent Excellent Fair to poor

Sand and sandy soils

SP

su
SM

SC

Slight to medium

Poor to practically impervious Fair to poor

Crawler-type tractor, rubber-tired equipment Crawler-type tractor, rubber-tire equipment Crawler-type tractor, rubber-tired equipment Rubber-tired equipment, sheepsfoot roller, close control of moisture Rubber-tired equipment, sheepsfoot roller

110-130 105-120 100-115 120-135

20-40 15-25 10-20 20-40

200-300 200-300 200-300 200-300

105-130

10-20

200-300

ML Low

Slight to medium Medium Medium to high

compressi bility
LL < 50

CL OL

Practically impervious Poor

Rubber-tired equipment, 100-125 sheepsfoot roller, close control of moisture Rubber-tired equipment 100-125 sheepsfoot roller Rubber-tired equipment 90-105 sheepsfoot roller

5-15

100-200

5-15 4-8

100-200 100-200

Fine grained 1 soils High compressi bility


LL > 50

MH CH OH

High High High

Fair to poor Practically impervious Practically impervious Fair to poor

Rubber-tired equipment, 80-100 sheepsfoot roller Rubber-tired equipment, 90-110 sheepsfoot roller Rubber-tired equipment, 80-105 sheepsfoot roller Compaction not practical

4-8 3-5

100-200 50-100

3-5

50-100

Peat and other

Pt

Very high

fibrous organic soils

2-39

Table 2.8 - Crude Empirical Relationships Between Resilient M odulus and O ther Test Data [2.101

p a i 5K 10K 20K . K g /c m 2 200 I CBR 2 I I 3 4 1 5 1 6 8 . 10 1 . 20 1 30 1 40 1 50 1 , . 80 1 , 100 1 D Y N A M IC M O D U L U S 500 . 1 IK 1 . 2K 1 5K 1 . 10K 1 50 K 1 , 10 OK |

1 .1

Bearing Value, pai (12" sia. plate, 0.2" deflection, 10 repetitions)


20 25 30 40 80 100 200 300 400

_ _ J___ L J___ I _ u_J__ ,_ |___ i___ 1 _i_


General Soil Rating as Subgrade, Sub-Base or Base Very Poor Subgrade Poor Fair Med. Good Med. Good Med. Good Sub Sub Sub Sub Sub Sub Base Base grade grade grade grade base base Excellent Base

A.A.S.H.O. SOIL CLASSIFICATION I A-2-7 A4 A5 A6 A-7-6 A-7-5 I I_____ I_____ I___ A-2-6

A-l-b

A-2-5 A3

A-1-a

A-2-4

I
T

T^~

I
I I

I___ I

I I

UNIFIED SOIL CLASSIFICATION

2-40

2.7 VARIATIONS IN MODULUS 2.7.1 General To some extent, the preceding approaches for estimat ing moduli from material and loading characteristics have explicitly included specified ranges or values o f the temperature, loading frequency, etc. However, it is o f interest to summarize these effects below, with some indication o f the relative importance o f each parameter or condition in affecting the modulus. 2.7.2 Temperature i. Asphalt bound materials Temperature is the most significant factor affect ing the modulus o f asphalt treated materials as illustrated by The Asphalt Institute (TAI) modu lus equation. At low temperatures (below about 0-5 C [30-40 F]) the modulus tends towards a value o f 14,000 to 20,000 MPa (2 to 3 million psi). At high temperatures (above about 45-50 C [110 to 120 F]) the modulus tends to a rela tively low value, usually less than about 700 MPa ( 1 0 0 , 0 0 0 psi), with the actual value related to the modulus o f the untreated aggregate. ii. Cementitious bound material These comments relate to materials stabilized to the point o f exhibiting an unconfined compres sive strength, such as soil cement or cement stabilized subbases. Other cement- or limetreated materials would exhibit behavior similar to unbound granular materials with a similar gra dation. Cement bound materials show little tem perature effect in terms o f modulus at normal pavement temperature ranges. However, deflec tion data may be affected by temperature due to expansion or contraction movement at joints or cracks, or due to shape effects resulting from temperature gradients. 2-41

iii.

Unbound materials Temperature has little effect on the modulus of unbound materials, except in the way that mois ture in the material is affected by temperature. Below freezing (0 C or 32 F), modulus is sig nificantly increased due to cementing action o f ice. Otherwise, no effect is expected except if temperature causes moisture content changes, which is not a direct load-related response effect during the deflection test, but a longer term material condition effect.

2.7.3

Moisture i. Asphalt bound materials Moisture has little or no direct response effects on deflection data for asphalt bound materials. If the asphaltic material is moisture sensitive and subject to stripping, the reduction in modulus will reflect the extent to which stripping has occurred at any given time, but, again, the moisture does not show a direct response effect. ii. Cement bound materials The presence or absence o f moisture will have no effect on the direct response o f cement bound materials during deflection testing. Moisture will have a long term effect on modulus similar to the effect it has on strength. iii. Unbound materials At a given density and stress level, moisture content is probably the most significant factor affecting the modulus o f unbound materials, as illustrated by the relationships shown in Table 2.6. Modulus can increase by a factor o f five or more for a material as it dries out. This is true for coarse grained and fine grained materials, although the effect o f drainage conditions may be

2-42

more significant for coarse grained materials. Reduction in modulus with increased moisture content is usually related to decreased inter particle friction due to increased lubrication for cohesionless materials. In cohesive soils, modulus changes are usually related to the soil fabric effects associated with clay-waterelectrolyte systems, which are fairly complex. Under saturated conditions applied loading may result in excess pore water pressures which can dramatically affect apparent structural response. 2.7.4 Freeze-Thaw Conditions This usually refers to the spring thaw period in areas where significant seasonal frost penetration o f the pavement occurs during winter. Thawing occurs from the surface down, resulting in thawed material under lain by frozen material until the structure is completely thawed. There is usually an excess o f surface water present, so that the thawed zone is often saturated and unable to drain due to underlying ice. This combination o f low temperatures and saturated conditions results in material modulus combinations that are particularly susceptible to load associated damage. i. Asphalt bound materials The direct response effect on asphaltic materials during spring thaw is governed by temperature conditions at this time. Typically temperatures are low, resulting in high asphalt moduli and relatively brittle behavior for these materials. Damaging conditions under load are com pounded by the lack o f support from unbound, saturated materials in the thaw zone. ii. Cement bound materials Deflection o f cementitious materials is independ ent o f temperature or moisture conditions. Cement bound materials may be damaged by

2-43

freeze-thaw or wet-dry cycles, resulting in low ered moduli and strengths. This damage may be reflected by increased deflections. Freeze-thaw effects are more o f a factor for stabilized bases or subgrades than the pavement surface. iii. Unbound materials The modulus o f unbound materials in the thawed zone are generally at their lowest level, particu larly in the area close to the thawing front, where they are likely to be saturated and subject to excess pore pressures under a dynamic load due to lack o f drainage. Moduli may be as low as 2 0 % or less o f the modulus that would be exhib ited by the material in a dry condition. For frost susceptible materials the response may be perma nently affected if ice lenses form causing frost heave during the freezing process, which results in a loss in density on thawing. 2.7.5 Time o f Loading i. Asphalt bound materials At any given temperature, the modulus o f asphaltic materials is strongly influenced by time o f loading due to the visco-elastic nature o f the material. This effect is reduced at low tempera tures. The Asphalt Institute equation in Section 2.6.1 can be used to illustrate the effect. The dif ference in time o f loading between laboratory tests (typically 0.1 sec.) and FWD tests (typically .025 to .035 sec) can result in a 50% difference in moduli, with the higher modulus exhibited for the shorter load pulse on the FWD. ii. Cement bound materials Dynamic moduli for cementitious materials can be approximately twice the static moduli. (Neville, 2.4) However, this effect reduces as the modulus increases, with good quality PCC 2-44

moduli in the 28,000-35,000 MPa (4 to 5 million psi) range showing similar moduli for dynamic or static loading conditions, according to Neville. Some practitioners suggest that dynamic PCC moduli determined from FWD tests will show significantly higher values than laboratory (static) measured moduli.

iii.

Unbound materials The effect o f time o f loading on the modulus of unbound materials is generally small compared with stress level and moisture content effects.

2.7.6

Stress Level i. Asphalt bound materials Within typical ranges o f stress level encountered in pavement structures, the effect on asphaltic material modulus is insignificant. ii. Cement bound materials Generally speaking, stress level does not signifi cantly affect modulus for strongly bound ce mented materials, although Table 2.4 (Ref. 2.8) shows a relationship between stress state and compressive modulus for cement stabilized soil. For cement or lime modified materials showing little or no cohesion, stress level effects would be similar to those exhibited by unbound materials. iii. Unbound materials Many unbound materials are extremely stresssensitive with moduli very significantly affected by stress level. Granular materials will often show stress-stiffening behavior, as described in Section 2.2.7 and 2.5.4, with the apparent modulus increasing as the applied stress level

2-45

increases. Fine-grained materials often exhibit stress-softening behavior, showing a decreasing modulus as the stress level increases. This has a significant bearing on deflection measurements and the associated analyses. If deflections are measured with an FWD, unbound material stress levels are very different for the center sensor compared with outer sensors which may be 1.8m (6 ft.) or more away from the load plate. This means that the material response would be differ ent at these locations which needs to be recog nized in the analysis. Stress level effects should also be considered if the deflection test load is different from the expected design wheel load. 2.7.7 Material Density Density is considered to be an indicator o f the quality o f constructed pavement layers and is typically reflected by the material modulus. i. Asphalt bound materials For a typical adequate quality asphalt concrete material, modulus will vary somewhat with den sity, and this variation can be illustrated with the Asphalt Institute modulus equation in Section 2.6.1. The variation is minor compared with temperature and time o f loading effects. How ever, significant variations will occur if densities are significantly lowered due to poor compac tion, for instance. ii. Cement bound materials Cementitious bound materials show some modulus variation with density as illustrated by the PCA modulus equation in Section 2.6.2. The modulus o f light weight concrete [1600kg/m3 (100 lb./ft3)] may be 50% to 60% that o f normal weight PCC.

2-46

iii.

Unbound materials Density effects are relatively minor for unbound materials that are compacted to typical pavement structural layer levels. The modulus does in crease with increased density, as illustrated by Rada & Witczak's data in Table 2.6 and Figure 2.19, or by the relationship between CBR and modulus (Fig 2.20). Stress level and moisture effects are more significant than density effects for unbound materials.

2.8 POISSON'S RATIO Poisson's ratio was defined and described in Section 2.2.5. Typical values o f Poisson's ratio (n) include: Material Steel Aluminum PCC Flexible Pavement o Asphalt Concrete o Crushed Stone o Soils (fine-grained) Poisson's Ratio 0 .2 5 -0 .3 0 0.33 0.1 5 -0 .2 0 *

o o o o

0.35 () 0.40 () 0.45 ()

*Dynamic determination o f n could approach 0.25 for PCC [Neville (1.4)]

The particular significance of Poisson's ratio lies in the fact that it essentially defines the three dimensional state of stress or strain in the material, as illustrated by the general ized Hooke's law for multiaxial loading i.e., (see Figure 2 . 21 ),

8x

= Y

'

E ( y + a z ) = i [ * - ^(a y + a z)]

(Eq. 2.21)

2-47

= ^

- K

+ z) = ^ [ a y - ^ ( a x + a 2)]

(Eq. 2.22) (Eq. 2.23)

where

e = strain a = stress E = elastic modulus = Poisson's ratio

As a simple example, consider the uniaxial load case with Oy = a z = 0 which results in:

(Eq. 2.24) (Eq. 2.25)


i.e. response along the y and z axes are directly related to Poisson's ratio. It is obvious from this that the magnitude o f Poisson's ratio is important to mechanistic pavement analyses. Fortunately large variations in Poisson's ratio for a given pavement material are unlikely. Since it is very difficult to measure Poisson's ratio the value is usually assumed. It is believed that the analysis is fairly insensitive to this value.

2-48

a.

a,

Figure 2.20 - Three Dim ensional Stress Diagram

2-49

2.9

Class E x ercise

Objective: T o faniliarize the student with the typical range of values of oduli that are comaon to highway pavecent saterials. fart A Directions: U s i n g the information contained in this anual, along with y o ur personal experience, enter a "reasonable" value of the n odulus of elas t i c i t y and Poisson's ratio for aach of the saterials described. 1. A new a sphalt c o n c r e t e surface at 25 degrees Celsius {77 degrees F a h r e n h e i t ) . Poisson'a R atio fHPa) of Elasticity lEill

2.

A ne w p o n t l a n d c e n e n t concrete slab. P oiss o n ' a Ratio Modulus of Elasticity fMPa) Itill

3.

A clean, c r u s h e d s t one base layer, well drained. P ois s o n ' a Eaii-e Hc*3ulus e l Elasticity m m lEill

4. A c e m e n t - s t a b i l i z e d base, with 40 a (1 1/2 in) axisua particle size, and a 2 8 -day conpressive strength of approxisately 7000 kPa (>1000 p s i ) . Pois s o n 'a
Eg tig

Modulus pf Elasticity
fMPa) *

5. A s i lty sand subgrade, subject to seasonal frost penetration, in late spring of t h e year. Poisson's Rfitifl Hadiilili-fil JklAfilASLilX IKEA1

2-50

6. The sane s i l t y and sub-grade in problen 5, but in let* fall of the year. Poisson' Ratio Modulus of Elasticity (MPa) lESl)

7. A recycled b i t u m i n o u s surface, vell-co*pacted, et 40 degrees Celsius (104 d e g r e e s Fahrenheit). Poisson' Eat I p Modulus of Elasticity 1Ma 1 l ii i

I. A 1 iae-stabilized subgrade with an unconfined coapressivs strength of 1000 kPa (-150 psi). Poisson's Patio Modulus of E lastici ty IMPAl

A poorly d r a i n e d d a y subgrade having a California Bearing Ratio of 3.


9.

Poisson'. La tig

(MPa)

Modulus of Elasticity Ids!)

10. A clean, c r u s h e d 9 ravel base, separated fro the ^ r a d e in problen 9 by e l ayer of filter fabric, and covered by e thin bituminous surface (eg., chip seal). Poisson's Modulus of Elastic ity

Ratio

1KP&1

2-51

Fart I Directions: Based u p o n the description of each pavement systea, write down an expected aodulus of elasticity and Poisson'* ratio for each of the mate r i a l s in the pavement. Also indicate a ainimum probable v a l u e and a aaxiaum probable value for the acxJulus of each layer. Pavercent Ko. 1 - New Concrete P avement A newly constructed jointed, reinforced concrete pavement has a 150 Km (6 in) surface over a 200 am (8 in) ceDent treated base (CTB) . It is located o n a county road which serves 5500 ADT. The subgrade soil is a plastic clay, and underdrains have been placed along both s i d e s of the pavement at a depth of 1.1 aetera (42 in) . Layer Wuftfrer 1 2 3 Material PC Concrete CTB Plastic Clay Poisson' R atio __________ Modulus of Elasticity iK Fa I I b s II ____ __________ ______________

Paverent No. 2 - A g i n g Asphalt Pavesent A flexible p a vement w a s constructed 20 years ago on s state primary road which c u r r e n t l y serves 12,000 ADT. The pavement is composed of a 110 u s (4.5 in) asphalt concrete surface over a 400 an (16 in) c r u s h e d stone base course. The subgrade is a silty sand. The road is located in a northern clinate where the frost penetration is t y pically about 1 aeter (40 in) each winter. Due to weather and traffic, the asphalt surface is severely alligator c racked and slightly rutted in the wheelpaths. Layer
VliEbSI

Material Aging H M A C Cr. Stone Silty Sand* Silty Sand

poisson's Ratio __________ __________

Modulus of Elasticity Xii lSil ____ ___________ _______________ _______________ _______________

1 2 3 4

___________________ __________ ________

the zone o f frost penetration in

2-52

SECTION 2.0 REFERENCES Bu-bushait, A. A., "Development o f a Flexible Pavement Fatigue Model for Washington State," Ph.D. Dissertation, Department of Civil Engineering, University o f Washington, Seattle, Washington, 1985. Southgate, H. F., and Deen, R. C., "Temperature Distributions in Asphalt Concrete Pave ments," Highway Research Record No. 549, Highway Research Board, Washington, D.C., 1975, pp. 39-46. Hadley, W. O., Laboratory Techniques for Resilient Modulus Testing, Proceedings, Strategic Highway Research Program Products Spe cialty Conference, American Society o f Civil Engineers, Denver, Colorado, April 8-10, 1991. Neville, A.M., Properties o f Concrete, John Wiley and Sons, New York, 1975, p. 320. Houston, W.N., Mamlouck, M.S., Perera, R.W.S., "Laboratory versus Nondestructive Testing Pavement Design", ASCE Journal o f Trans portation Engineering Vol. 118 No. 2, Mar/Apr 1992, New York, pp 207-222. AASHTO, "Guide for Design of Flexible Pavement Structures", 1986. Rada, G. and Witczak, M.W. "Comprehensive Evaluation o f Laboratory Resilient Moduli Results for Granular Material", TRB, TRR 810, 1981. Terrel, R.L., Epps, J.A., Barenberg, E.J., Mitchell, J.K. and Barenberg, E.J., "Soil Stabilization in Pavement Structures: A User's Manual" Volumes I and II. US FHWA-IP-80-2, 1979. Yoder, E.J. and Witczak, M.W., "Principles o f Pavement Design", John Wiley & Sons, 1975.

2-53

2.10

Hicks, R.G. and McHattie, R.L., "Use o f Layered Theory in the Design and Evaluation o f Pavement Systems", Alaska DOT Report UFHWA-AD-RD-83-8, 1982. Miller, J.S., Uzan, J. and Witczak, M.W., "Modification o f the Asphalt Institute Bitu minous Mix Modulus Predictive Equation", TRB TRR 911, Washington D.C.

2.11

2.12

The Asphalt Insititute, "Research and Development the Asphalt Institute's Thickness Design Manual (MS-1) Ninth Edition, Research Report No. 82-2, 1982 Thompson, M R., "A Proposed Thickness Design Procedure for High Strength Stabilized Base (HSSB) Pavements", Transportation Engineering Series No. 48, Illinois Cooperative Highway and Transportation Series No. 216, University o f Illinois, Champaign, Illinois, 1988.

2.13

2-54

SECTION 3.0 FUNDAMENTALS OF MECHANISTIC-EMPIRICAL DESIGN 3.1 INTRODUCTION 3.1.1 Overview o f Historical Development a. b. c. d. e. f. gh. i. jk. 1. m. 1848 Kelvin - elastic half space 1885 Boussinesq - elastic half space (point load) 1926 Westergaard - two layers 1928 Love - elastic half space (circular load) 1943 Burmister - two layers 1948 Fox - solutions two layers 1949 Odemark - transformed section (equivalent thickness) 1951 Acum and Fox - solutions three layers Early 50's - finite element method 1961 Jones and Peattie - three layers 1963 Commercial programs >five layers 1970's - Widespread use o f layered theory (main frame) 1980's - Microcomputers used as tools

3.1.2 Reasons for Using Mechanistic-Empirical Procedures (rather than empirical) Mechanics is the science o f motion and the action of forces on bodies. When we refer to a mechanistic ap proach in engineering, we are talking about the applica tion o f elementary physics to determine the reaction o f structures to loading. The primary concern in pavements is how the structure distributes vehicle loads to the under lying soil layers. Weak pavements concentrate the load over a smaller area o f the subgrade than strong pavements resulting in higher stresses as shown in Figure 3.1. In order to quantify how the load is being distributed, certain fundamental properties o f the materials must be known along with the thicknesses o f the pavement layers and the load characteristics. These will be discussed later. An empirical approach is one which is based on the results o f experiments or experience. Generally, it requires a number o f observations to be made in order to ascertain the relationships between the variables and outcomes o f trials. It is not necessary to firmly establish the scientific basis for the relationships as long as the limitations are recognized. In some cases, it is much more expedient to rely on experience than to try to quantify the exact cause and effect o f certain phenomena.

3-1

Strong Pavement Load

Weak Pavement Load

____________________I

I_____________________

__________________

Surface Base

rrrTTTTr> _
Subgrade

Figure 3.1 - Load Distribution Characteristics of Strong versus W eak Pavement

M ost o f the pavement design procedures used in the past have been empirical in that their failure criteria were based on a set o f given set o f conditions, i.e., traffic, materials, layer configurations, and environment. The equation for the thickness o f cover (total pavement structure) for asphalt pavements developed by Hveem and Carmany [3.1] for California highways is an example o f empirical pavement design: T = K'(TI)(90 - R)/(c) 2 where T = thickness o f cover, (ft.) 0.095 (coefficient depending on design wheel load and tire pressure with a factor of safety), traffic index, resistance value, and cohesiometer value.

(Eq. 3.1)

K1 =

TI = R = c =

Although the above equation encompasses parameters for the bound materials (c-value) and the underlying unbound materials (R-value) as well as the traffic volume (TI), it is based on a 22 kN (5000 lb.) wheel load with a tire pres sure o f 480 kPa (70 psi). This approach is still used by CalTrans, but the equation has been adjusted to the cur rent form which no longer involves c, for instance. Yo der [3.2] noted that it is unlikely that this design proce dure could be successfully adapted to a region with severe frost problems or different rainfall characteristics. Another illustration o f an empirical design procedure is the AASHTO process. The fundamental information for developing the design procedure came from the AASHO Road Test which was constructed and tested during the late 1950s and early 1960s. The most recent version of this flexible pavement design process is illustrated in Fig ure 3.2. This figure shows the basic design nomograph which was developed from the following empirically de rived performance equation: [3.16]

3-3

logio W is = (Z r ) (S o) + (9.36)(log (SN + 1)) - 0.20 f


10

APSI 4 2 - 15
1094

J - + (2.32) (logi 0 MR) - 8.07

(Eq. 3.2)

0.40 +

(SN + 1)5.19

where W is = 18,000 lb. (80 kN) equivalent single axle loads predicted to pt,
Zr = Z-statistic associated with the selected level o f design reliability,

So

= overall standard deviation o f normal dis


tribution o f errors associated with traffic prediction and pavement performance,

SN

= Structural Number (essentially a "Thickness Index"),

APSI = overall serviceability loss = po - Pt, p0 = initial serviceability index following con struction, = terminal serviceability index, and = resilient modulus o f the roadbed soil(s). (psi)

Pt Mr

The various constants in the above equation were ob tained from regression analysis o f the AASHO Road Test data hence this too is an empirical design procedure as opposed to a mechanistic-empirical approach.

3-4

SM
Figure 3 . 2 . Elustration o f AASHTO Design Nomograph for Flexible Pavements

3-5

A mechanistic-empirical approach to pavement design in corporates elements o f both approaches. The mechanistic component is the determination o f pavement structural responses such as stresses, strains, and deflections within the pavement layers through the use o f mathematical models. The empirical portion relates these responses to the performance o f the pavement structure. For instance, it is possible to calculate the deflection at the surface of the pavement using some o f the tools discussed later. If these deflections are related to the life o f the pavement, then an empirical relationship has been established between the mechanistic response o f the pavement and its expected performance. There are currently no pure mechanistic approaches to pavement design. The basic advantages o f a mechanistic-empirical pavement design procedure are: (a) The ability to accommodate changing load types and quantify their impact on pavement performance. The ability to utilize available materials in a more ef ficient manner. The ability to accommodate new materials. M ore reliable performance predictions. A better evaluation o f the role o f construction. Use o f material properties in the design process which relate better to actual pavement behavior and performance. An improved definition o f existing pavement layer properties. The ability to accommodate environmental and aging effects on materials.

(b)

(c) (d) (e) (f)

(g)

(h)

Currently, the primary means o f mathematically modeling a pavement is through the use o f layered elastic analysis. Although more complicated techniques are available (e.g., dynamic, visco-elastic models), we will restrict the dis cussion to basic linear elastic models subjected to static loading. Layered elastic analysis computer programs can

3-6

easily be run on personal computers and do not require data which may not be realistically obtained. Finite ele ment approaches are likely to become more common as computing power increases. These approaches can be used for both flexible and rigid pavements. Several other methods specific to rigid pavement analysis will be described in a separate section.

3.2 LAYERED ELASTIC SYSTEMS 3.2.1 Assumptions and Input Requirements The modulus o f elasticity and Poisson's ratio o f each layer define the material properties required for computing the stresses, strains, and deflections in a pavement structure using layered elastic or finite element models. Typical values for the moduli and Poisson's ratios o f pavement materials were given in Sections 2.5 and 2.8 o f these notes. In addition to the material properties o f the layers, the thickness o f each pavement layer must also be described. For computation purposes, the layers are assumed to ex tend infinitely in the horizontal direction, and the bottom layer (usually the subgrade) is assumed to extend infinitely downward. Given the typical geometry o f pavements, these assumptions are considered to be fairly representa tive o f actual conditions, except when analyzing jointed PCC pavements in the vicinity o f the joints or edges, as well as edge loadings on asphalt pavements. It is assumed that material behavior is perfectly linearly elastic, homogeneous and isotropic. Homogenous refers to pavement layers which are composed o f the same ma terials throughout. Isotropic means that the material will possess the same properties along all axes (as opposed to wood which possesses differing material properties with respect to the direction o f the grain). If non-linear or stress-sensitive behavior is modeled, iterative procedures are usually involved. As will be pointed out in Section 3.2.3, most pavement materials do not exhibit these ideal ized characteristics. This can result in problems during the evaluation o f deflection basins leading to unrealistic values o f moduli. 3-7

The loading conditions must be specified in terms o f the magnitude o f the load, the geometry o f the load, and the number o f loads to be applied to the structure. The magnitude o f the load is simply the total force (P) applied to the pavement surface. In pavement analysis, the load geometry is usually specified as being a circle o f a given radius (r or a), or the radius computed knowing the con tact pressure o f the load (p) and the magnitude o f the load (P). Although most actual loads more closely represent an ellipse, the effect o f the differences in geometry be come negligible at a very shallow depth in the pavement. Effects o f multiple loads on a pavement surface can be approximated by summing the effects o f individual loads. This is referred to as the Law o f Superposition and is considered valid as long as the materials are not stressed beyond their elastic ranges (subject to plastic or perma nent deformation). To summarize, the following information must be avail able to compute the response o f a pavement to loading: (a) Material properties of each layer. (i) (ii) (b) (c) Modulus o f elasticity (E). Poisson's ratio (fa).

Thickness o f each pavement layer. Loading conditions. (i) Magnitude o f load. (ii) Geometry o f load. (iii) Number o f loads.

Figure 3.3 shows how these inputs relate to a layered elastic model o f a pavement system. The outcome o f a layered elastic analysis is the computa tion of stresses, strains, and deflections in the pavement.

3-8

Total Load

Radius r or a

--------

Surface Base

^1

>m

!
k

hi

^2 ^2
\r

h2
ra

Subgrade

E 3

, (J-3

Figure 3.3 - Layered Elastic Pavement Model

As will be mentioned in Section 3.2.5, the use o f a layered elastic analysis computer program will allow one to calcu late the theoretical stresses, strains, and deflections any where in a pavement structure. However, there are only a few locations in which we are generally interested for the calculation o f critical responses. These are. Location Pavement Surface Bottom o f AC or ATB Bottom o f PCC or CTB Top o f PCC Slab (corner) Top o f Intermediate Layer (Base or Subbase) Top o f Subgrade Response Deflection Horizontal Tensile Strain Horizontal Tensile Stress Horizontal Tensile Stress Vertical Compressive Strain Vertical Compressive Strain

The locations o f these responses relative to a pavement structure and load are illustrated in Figure 3.4. The hori zontal tensile strain at the bottom o f the asphalt concrete layer is considered indicative o f potential cracking o f the surfacing (fatigue failure). Fatigue o f PCC is related to tensile stress. Rutting failure in the subgrade can be pre dicted using the vertical compressive strain at the top of the subgrade. Deflections under load at the pavement sur face are used in imposing load restrictions during spring thaw and overlay design (for example). In the next three sections, we will discuss the evolution o f layered elastic analysis and try to get a qualitative under standing o f the process. 3.2.2 One-layer system With Point Loading (Boussinesq) The origin o f layered elastic theory is credited to V J. Boussinesq [3.3] who published his classical work in 1885. He developed solutions for computing stresses and deflections in a halfspace (soil) composed o f homogene ous, isotropic, and linearly elastic material. Boussinesq influence charts are still widely used in soil mechanics and foundation design. The governing differential equations for a point load on an elastic half space were postulated earlier by Kelvin. Boussinesq developed the closed form mathematical solution in 1885 for point load, while Love extended this work to a circular load in 1928.

3-10

1.Pavement surface deflection 2. Horizontal tensile strain at bottom of bituminous layer 3. Vertical compressive strain at top of base 4 .Vertical compressive strain at lop of subgrade

Figure

3 .4

Pavement Response Locations U sed in Evaluating Load Effects

3-11

In this approach, the stresses and deflections are calcu lated for a point load applied to the surface o f a deep soil mass. Distance variables are expressed in terms o f cylin drical coordinates, in which the distance from a point on the surface may be expressed as: R 2 = r 2 + z 2 = x2 + y2 + z 2 as shown in Figure 3.5. The vertical stress, o z, radial stress, c r, and the vertical deformation, u, can be calculated using the following for mulae for a point load: Vertical stress: c?z -3Pz 2;zR5 (Eq. 3.4) (Eq. 3.3)

Radial stress: _P(1+A0 Gr,z ~ 2 jS J -3r2z ( l - 2 |i)R R 3 + R+z (Eq. 3.5)

Vertical deformation below the surface:


_ P 0 + /4

uz,r =

2 tE

g - n) R

z2' R3

(Eq. 3.6)

Surface deflection at a distance, r, away from the load (i.e., Eq. 3.6 with z = 0 and R = r):

_ o _ V )p

(Eq. 3.7)

Uf

(* )(E )(r)

For all practical purposes, the equations for a point load can be used for a distributed load at points more than about two radii from the load. [~3.18] The deflection beneath the center of a rigid, circular load o f radius, a, can be estimated by the equation: _ (1 - U2)P (2)(E)(a) (Eq. 3.8)

3-12

Figure 3.5 Cylindrical Coordinates in Onc-Laycr System

3-13

Example 1; A load o f 40 kN (9,000 lb.) is placed on a 300 mm. (11.8 in.) diameter plate. The plate is resting on a subgrade which has an elastic modulus o f 51.7 MPa (7,500 psi) and a Poisson's ratio o f 0.4. What is the deflection at the center o f the plate? [i P a E uo = = = = _ 0.40 40,000 N 150 mm 51.7 MPa (1 - (0.4)2)40000 _ --------------------------- z .l / mm (.Uo5 in.) 2(51.7)(150) V 7

Exam ple 2: Given the loading conditions above, what is the modulus o f elasticity o f the subgrade if the deflection at the center o f the plate is 0.72 mm (0.028 in)?

f )(a) ( 2 )(u 0e ^

( 1 ~ ( 04)2) 40000 MPa 2(0.72)(150)

(22,500 psi) Conclusions: As you can see from examples 1 and 2, pavement model ling is a simple process for one-layered pavement systems. Note from the examples that modulus and deflections are inversely, linearly related so that if the modulus increases by a factor o f three the deflections will decrease by a fac tor o f three. The Boussinesq equations were modified through mathe matical integration to approximate the effects of a circular distributed load on the pavement surface. The equations for stress, strain, and displacement below and along the centerline o f a circular load are as follows: Vertical Stress at depth z:

Oz Go 1 -------------- XJT

(Eq. 3.9)

[\ + { a! z f j

3-14

Radial and Tangential Stress at depth z:

/i

+ a/ (
1

O r O? O o

1+

+) K

/\2

M i

(Eq. 3.10)

J)

Vertical Strain at Depth z:

( i + / / ) Ob
z

1+

(% )

2^2

(Eq. 3.11)

Deflection at depth z: / \

Where: Co = stress on surface (MPa), T E = elastic modulus (MPa) a= z= plate radius (mm) depth below pavement surface (mm)

= Poisson's ratio

3-15

3.2.3 Odemark's M ethod [3.18] In 1949 Odemark developed an approximate solution to the calculation o f stresses, strains and displacements in a layered system. Since then a number o f exact solutions to the same problem have been devised (ELSYM5, BISAR, ALIZE III, CIRCLY, etc.). So why use time on an approximate solution? First, it should be recalled that the "exact" solutions are only close to "exact" in a mathematical sense related to the numerical integration procedures. The assumptions made with respect to equilibrium, compatibility and con stitutive equations (Hook's law) are not correct for pave ment structures. Loads are dynamic, materials are not continuous, some are even particulate (granular) and deformations are not only elastic, but also plastic, viscous and visco-elastic, and they are mostly non-linear and ani sotropic. In a physical sense, therefore, all solutions are approximate. There are two advantages to using Odemark's method: (a) It is simple and very fast, it may be included in a spread sheet or used in a Pavement Management System, where millions o f computations must be performed. a non-linear elastic subgrade (or a subgrade where the modulus, or apparent modulus, varies with the distance from the load) may easily be included. This may be extremely important for the interpretation o f deflection data.

(b)

Odemark's method is based on the assumption that the response below a given layer, will depend on the stiffness o f that layer only. The stiffness o f a layer is: E *I (1 where:
-

(Eq. 3.13)

E is the modulus, I is the moment o f interia, and is Poisson's ratio


3-16

This assumption is used to change a layered system into a semi-infinite halfspace, for which Boussinesq's equations may be used. Consider a two-layer system as shown below: hi> ^1, |ij
E 2> ^ 2

A layer with thickness hj, modulus E^ and Poisson's ratio Hi, rests on a material with modulus E 2 and Poisson's ratio (j-2 The stiffness o f the upper layer is: I * Ei
i - m 2 =

X2 * b * h ,3 *Ei

(Eq. 3.14)

where:

b is the width under consideration

If the system is transformed to the following: he, E2, [x2

e 2 >^ 2 The stiffness will be:

I *E, Xa'b he ; ________________ e E? i - n 22 "


For the new stiffness to be identical to the original stiff ness: Hi ^ 1 ~ H22 E2
1 -H i2

(Eq. 3.15)

h = h. * 3 e

(Eq. 3.16)

The new system is a semi-infinite halfspace where Boussi nesq's equation can be used. With a multi layer system the method is used successively. First layer one is changed to the elastic parameters of 3-17

layer two and the equivalent thickness, he j, is calculated. Any materials below layer two are assumed to have the same elastic parameters as layer two. Then layer one and two are changed to the elastic parameter o f layer three, etc. It has been found that the best agreement with the "exact" solutions normally is obtained when Poisson's ratio is assumed to be the same for all layers. The equation for the equivalent thickness may then be written as:
n-1

(Eq. 3.17)

To get a better agreement with the exact solutions, the equivalent thickness is normally m ultiplied by a factor, f. For the first structural interface (e.g., between surfac ing and base) f is 0.9 for a two-layer system and 1.0 for a multi layer system. For all other interfaces f is 0.8. Exam ple 1 A 300 mm (11.8 inch) diameter plate is loaded to 40 kN (9,000 lbs.) on an asphalt concrete pavement over a sub grade. The asphalt concrete is 15 cm ( 6 inches) thick and has a modulus o f elasticity o f 3450 MPa (500,000 psi); the subgrade modulus is 69 MPa (10,000 psi). What is the deflection at the center o f the loaded area? What is the vertical stress at the top o f the subgrade? Use Poisson's ratio = 0.35. H int: The total deflection o f the pavement surface is the sum o f the subgrade deflection and the compression o f the asphalt layer. First, calculate the deflections at the surface and at a depth o f 150 mm for an asphaltic half space. Compute the compression o f the asphalt layer by subtracting the deflection at the bottom o f the layer from the deflection at the surface. Convert the asphalt layer to an equivalent subgrade thickness using Odemark's method and calculate the subgrade deflection at this depth. The sum o f the asphalt compression and the deflection o f the subgrade at this point can then be summed to obtain the overall deflection. The stress at the top o f the subgrade can also be calculated at a depth equal to the equivalent asphalt thickness. 3-18

1)

Calculate the plate pressure, then the deflection at the surface o f the asphalt half-space using Equation 3.12 (Note that since z=0, most terms drop out:

load 40,000N ... .... .. <jc = ------ = -------- = . 5 6MPa (81.2 psi) area tt<150)2 _ 2(1 - S ) c jo a _ 2(1-.352)(.56)(150) E ~ <.=.043 mm (.002 in.) (ELSYM5 gives .0017) N ote: The above equation is an algebraic simplification o f Equation 3.12 fo r the case where z= 0 (give it a try!). 2) Calculate deflection at 150 mm (Eq. 3.12):
(1+.35)(.56)(150) 3450

3450

a iso = -

(i-2 (.3 5 )O u (l % < , ) - '%

150

150

=. 027 mm (.001 in.)

(ELSYM5 gives .0011) 3) Asphalt compression = do-dz = 0.043 mm 0.027 mm = 0.016 mm (.0006 in.) Calculate equivalent thickness of asphalt

4)

h e = f * h , * 3 ST = . 9 ( 1 5 0 ) ^ p = 497 mm (19.57 in.) VEh 5) Calculate deflection at he


(l+.35)(.56)(150) 69

dh. =

d h .- 0.547 mm (.022 in.)

(ELSYM5 gives .0218) 3-19

6)

Total Deflection = 0.016 mm + 0.547 mm = 0.563 mm (.022 in.)>/ (ELSYM5 nan as a 2-layer system give .0231)

7)

Stress at top o f subgrade:

-----------------

3 /2

= 069 MPa = 69 KPa (10 p s i ) /

(l + (l50/497) ] It can be seen that these equations are quite cumbersome to utilize on a calculator, but on a computer they execute quite fast and give reasonable results, as can be seen when the problem is repeated in the following section. 3.2.4 Two-layer system (Burmister) Burmister [3.4] extended the one-layer solutions to two and three layers. We will restrict the discussion in this section to two layers, analogous to a full-depth asphalt layer over subgrade. In his work, Burmister assumed that the layers have full frictional contact (no slip) at the inter face, and that there are no shear or normal forces on the surface outside o f the loaded area. He also assumed that Poisson's ratio for each of the two layers is 0.5 in order to simplify the mathematics. The exact equations are rather long and complicated even with this simplification, but the important parameters in the solution are listed below: p a h load distributed over a circular plate (contact pressure) radius o f the flexible plate thickness o f the surface layer

E 1 modulus o f elasticity for the surface layer E 2 modulus o f elasticity for the subgrade In the equations, the geometry o f the load and surface layer are specified by the ratio a/h, and the moduli o f the layers by E 2 /Ej. Graphs based on these ratios were de veloped to aid in determining the surface deflections and stresses in pavements. Figure 3.6 is a graph which can be used to find the displacement coefficient Iaz, for 3-20

calculating the deflection on the surface at the center of the loaded area using the following equation:

(Eq. 3.18)
Figure 3.7 can be used to find the ratio o f the vertical stress in the pavement structure to the applied stress. Please note that Figure 3.7 is only valid when the radius o f the load is equal to the surface thickness. Other graphs would be needed for different values of a/h. Exam ple 1 (again): A 300 mm (11.8 inch) diameter plate is loaded to 40 kN (9,000 lbs.) on an asphalt concrete pavement over a sub grade. The asphalt concrete is 15 cm ( 6 inches) thick and has a modulus o f elasticity o f 3450 MPa (500,000 psi); the subgrade modulus is 69 MPa (10,000 psi). What is the deflection at the center o f the loaded area? What is the vertical stress at the top o f the subgrade? E2 _ 69 M P a __ 1_
50

E I " 3450 MPa

Area o f Plate = na2 = 0.0707 m 2 (1 1 0 in.2) a h 150 150

40 kN p = P/A = ----------- - = 566 kPa (82 psi) .0707 m

3-21

1.0

3-22
Ratio of the Radius of fhe Flexible Bearing Area to the Thickness of the Surface Layer (a/h)

Figure 3 . 6 Graph for Determining Displacement Coeffecient for Tw o-Layer System

Values of Parameters, z/a

Vertical Stress Influence Coefficient

Figure 3 .7 Vertical Stress as a Function o f Depth in a Two-Layered System, for a = h

Deflection at center o f load: From Figure 3.6: Iaz = 0.29 Az = 1.5 ( 566) ( 15 ) (0.29) = 0.54 mm (.02 in.) 69,000 V 7 V ' Vertical stress at top o f subgrade: From Figure 3 .7: o z/p = 0.12
gz =

0.12(566) = 68 kPa (9.9 psi)

Class Exercise Use the information in Example 1 to calculate center deflection and subgrade stress for a 450 mm (18 in.) plate with an 80 kN (18,000 lb.) load on a 225 mm (9 in.) AC layer. Deflection = .769 mm (.030") Stress = .062 MPa (8.9 psi) 3.2.5 Multi-layer system The majority o f pavement structures are more compli cated than the two-layer system discussed above. Usually a base layer with or without an underlying subbase layer are placed between the asphalt concrete and subgrade. These layers may be either unbound or stabilized granular materials which have distinctly different properties from the surface and subgrade layers. One can imagine that the solutions for systems with three or more layers become increasingly more complex. For a three-layer pavement, several charts and tables have been developed by Peattie, Jones and Fox [3.5] to deter mine the stresses, strains, and deflections. Peattie [3.6] developed graphical solutions for vertical stress in threelayer systems. Jones [3.7] presented solutions for hori zontal stresses in a tabular form. Both o f these solutions were based upon a Poisson's ratio o f 0.5 for all layers. 3.2.6 Elastic Layer Computer Programs The logical extension o f these solutions was the develop ment o f computer programs in order to expedite analysis and allow greater flexibility in accommodating material properties and multiple loads. Even the most elementary o f these programs allow for materials with Poisson's ratios 3-24

other than 0.5. Some are capable o f ascertaining the effects o f multiple wheel configurations and/or nonlinear material behavior. During the 1980s, as these mainframe computer programs were converted to run on microcom puters, a greater potential was developed to use elastic analysis tools in pavement design and rehabilitation deci sion making. A list o f layered elastic computer programs is given in Table 3.1. Examples o f these programs include BISAR, ELSYM5, and WESLEA. BISAR, developed by Shell Oil Co. in the early 1970's is generally considered the benchmark to which all other layered elastic modelling programs are compared. Its complex mathematical models yield the most rigorous analysis o f stresses and strains within the pavement structure. BISAR has two important features: 1) It can solve for the influence o f both normal and tan gential loading on the pavement surface, and 2) it allows the user to specify the degree o f friction between the pavement layers. BISAR is a proprietary program avail able from Shell Oil Co. ELSYM5 is similar to BISAR, except it cannot model vari able friction between the pavement layers. ELSYM5 was developed at the University o f California at Berkeley from code developed by Chevron Research. ELSYM5 was recently modified by the FHWA to include a user friendly interface. Modifications were also made to the program by Dr. Lynne Irwin o f Cornell University to correct prob lems in the mathematical integration routines which pro duced irregular answers under certain conditions. ELSYM5 is available from the FHWA. WESLEA, developed by the U.S. Army Corps o f Engi neers, Vicksburg Mississippi, represents the latest attempt at improving the state-of-the-art pavement modelling pro grams. WESLEA executes much faster than similar pro grams allowing it to be used more efficiently on micro computers. WESLEA is available from the U.S. Army Corps o f Engineers. This is not an exhaustive list o f layered elastic computer programs currently in existence, they are simply a list o f the most widely used. Typical input required for using these computer programs include: (a) Material properties o f each layer. (i) (ii) Modulus o f elasticity. Poisson's ratio.

3-25

Table 3.1 - Layered Elastic Computer Programs


Number Programs Number o f Loads Continuity Conditions at Interface Full Continuity to Frictionless Full Continuity to Frictionless Full Continuity Full Continuity to Frictionless Full Continuity

of
Layers (max)

Probabilistic Considerations

Program Source

Remarks

WESLEA

10

10

No

U. S. Army CE Waterways Experiment Station Shell International Petroleum Co., Ltd London, England Chevron Research Company University o f California, Berkeley National Cooperative Highway Research Program (Project 110B) FHWA-US DOT U.S. Army C.E.

Short running times Considers horizontal as well as vertical loads. Comparatively long running times (on 286 computers) Considers horizontal as well as vertical loads Nonlinear response o f granular materials accounted for in DAM A program o f the Asphalt Institute which makes use o f the CHEV program. Short running time for particular point. Latest version dated 1993 - integration procedures enhanced bv Cornell University Running time is long for degrees o f reliability other than 50% (deterministic mode) Interative process used to arrive at moduli for untreated granular materials. Running time is long in probabilistic mode. Program considers materials both as time independent and time dependent (elastic and viscoelastic). Modification o f CHEV program. Includes provision for stress sensitivity o f granular layers. Permits consideration o f horizontal and vertical loads; in particular permits consideration o f radially directed horzontal forces. Can consider orthotropic material behavior. Permits consideration o f strain energy.

B1SAR

10

10

No

3-26

CHEV

No

ELSYM5

10

No

PDMAP (PSAD)

Yes

VESYS

Full Continuity Full Continuity

Yes

CHE VIT

12

Yes

Waterways Experiment Station MINCAB Systems Canterbury, Australia (For Australia Research Board

CIRCLY

5+

10+

Full Continuity to Frictionless

No

(b) (c)

Thickness o f each pavement layer. Loading conditions (2 o f 3 listed below). (i) (ii) (iii) Magnitude o f load. Radius o f load. Contact pressure.

(d) (e) (f)

Number o f loads. Location o f load(s) on the surface (x,y coordinates). Location o f analysis points for output (x,y,z coordinates).

For the purposes o f illustration, three typical pavement cross-sections have been selected as shown in Figure 3.8. The initial conditions we will examine are: (a) Material properties o f each layer. Elastic Modulus Layer Asphalt Concrete Crushed Stone Base Fine-grained Subgrade MPa 3450 172 52 psi 500,000 25,000 7,500 Poisson's Ratio 0.35 0.40 0.45

b)

Layer Thicknesses AC Thickness mm in 50 125 230 2 5 9 Base Thickness mm in 150 200 150 6 8 6

Section A B C

3-27

2 (50 mm) ACP

6 (150 mm) Base

Fine-grained subgrade

Seciion A (Thin Thickness Section)

5 (125 mm) ACP

8* (200 mm) Base

Fine-grained subgrade

Section B (Medium Thickness Seciion)

9* (230 mm) ACP

6" (150 mm) Base

Fine-grained subgrade S e d b n C (Thick Sedion)

Figure 3 .8 Typical" Pavement Sections

3-28

(c)

Loading Conditions Magnitude o f Tire Load = 40 lcN (9,000 lbs.) Tire Pressure = 552 kPa (80 psi) Number o f Loads = 1 Location o f Load: x= 0 y= 0 Location o f Analysis Points (all at x=0, y=0) (i) (ii) Surface Deflection (all at z=0) Horizontal Tensile Strain at Bottom o f ACP Thin Pavement: z = 50 mm (2 in) Medium Pavement: z = 125 mm (5 in) Thick Pavement: z = 230 mm (9 in) Vertical Compressive Strain at Top o f Subgrade Thin Pavement: Medium Pavement: Thick Pavement: z = 200 mm (8 in) z = 330 mm (13 in) z = 380 mm (15 in)

(d) (e)

(f)

(iii)

A layered elastic analysis was performed using the inputs as defined above, and the results are listed in Table 3.2 under the heading "Standard Pavement". As would be expected, the surface deflection, the horizontal strain at the bottom o f the AC, and the vertical compressive strain at the top o f the subgrade all decrease with increasing pavement thickness. A comparison o f deflection basins for the three standard sections is given in Figure 3.9. In order to demonstrate the sensitivity o f the analysis to load and material changes, several other cases were run and compared to the Standard Pavement with the results shown in Table 3.2. These included: L ow Tire Load: Tire load decreased from 40 lcN(9,000 lb.) at 552 kPa (80 psi) to 4 kN (900 lb.) at 207 kPa (30 psi). This is a comparison o f the effects o f a truck load versus a passenger vehicle. High Tire Pressure: Increase from 552 kPa (80 psi) to 965 kPa (140 psi). This demonstrates the difference in pavement response from a standard truck tire pres sure to one which is an extreme. 3-29

Table 3.2 - Sensitivity Analysis of Various Input Parameters

Pavement Response Parameter


Surface Deflection Top of AC Section A - Thin Standard Pavement Low Tire Load High Tire Pressure Stabilized Subgrade Asphalt Treated Base Moisture Sensitive 1.219 mm (0.048") Section B - Med. 0.686 mm (0.027) Section C - Thick 0.457 mm (0.018)

0.152 mm (0.006)

0.076 mm (0.003")

0.051 mm (0.002")

1.321 mm (0.052)

0.711 mm (0.028)

0.483 mm (0.019)

0.914 mm (0.036)

0.584 mm (0.023")

0.406 mm (0.016)

0.533 mm (0.021)

0.356 mm (0.014)

0.305 mm (0.012)

1.346 mm (0.053)

0.838 mm (0.033)

0.610 mm (0.024)

Pavement Response Parameter


Horizontal Tensile Strain - Bottom of AC or ATB (x 10") Section A - Thin Standard Pavement Low Tire Load High Tire Pressure Stabilized Subgrade Asphalt T reated Base Moisture Sensitive 467 Section B - Med. 279 Section C - Thick 145

121

44

18

735

352

163

368

246

128

196

88

71

482

433

257

3-30

Table 3.2 (con t'd ) - Sensitivity Analysis o f Various Input Parameters (Subgrade Strain Calculated 2.5mm (0 .1 ") Below interface) Pavement Response Parameter
Vertical Compressive Strain - Top of Subgrade (x 10-*) Section A - Thin Standard Pavement Low Tire Load High Tire Pressure Stabilized Subgrade Asphalt Treated Base Moisture Sensitive -2,220 Section B - Med. -747 Section C - Thick -370

-280

-81

-40

-2,520

-786

-384

-957

-437

-253

-512

-229

-177

-2,580

-1,030

-608

Location of FWD Sensors

D0 (0mm or 0 ft.)
0

D,

d2

d3

(305mm or 1 ft.)

(610mm or 2 ft.)

(914mm or 3 ft.)
0.00

FWD Deflec tion 250


(nm )

FWD Deflec0.01 tion (in.)


MEDIUIV

500-

0.02

750THIN

0.03

0.05 Figure 3.9

3-31

Stabilized Subgrade: Increase subgrade modulus from 52 MPa (7,500 psi) to 345 MPa (50,000 psi) in the top 150 mm (6 in) o f soil. This case shows the effect o f subgrade improvement.

Asphalt Treated Base: Increase base course modulus from 172 MPa (25,000 psi) to 3450 MPa (500,000 psi). This represents an increase in the structural capacity o f the pavement. Moisture Sensitive: Decrease AC modulus from 3450 MPa (500,000 psi) to 1380 MPa (200,000 psi). This illustrates a weakening o f the AC layer due to strip ping in the mixture. Comparisons between the "Standard Pavement" case and the modified input just described are given in Table 3.3. As would be expected, there is a dramatic decrease (on the order o f 75 to 90 percent) in all o f the pavement re sponse parameters when a passenger vehicle loading is compared to that o f a truck load. These decreases be come even more meaningful when they are discussed in the context o f pavement life later on in these notes. The increase in tire contact pressure from 552 to 965 kPa (80 to 140 psi) results primarily in an increase in the ten sile strain in the asphalt concrete. This parameter increased 11 percent for the thick pavement and 57 per cent in the thin pavement. It is interesting to note that very little change in the surface deflection or compressive strain in the subgrade resulted from the higher tire pres sure. In general, the surface layer o f the pavement is sen sitive to changes in tire pressure, particularly "thin" sur face courses. The lower layers are most sensitive to changes in load. The incorporation o f a stabilized subgrade layer affected the vertical compressive strain in that layer most effec tively, although significant improvement in the surface deflection and tensile strain in the asphalt surface can be noted for the thin pavement. The use o f an asphalt treated base greatly influenced the response parameters in all three pavement sections.

3-32

Table 3.3 - Comparisons of Changes in Pavement Responses from Standard Pavement

Low Tire Load


P e r c e n t Change from standard P a v e m e n t Section A Surface Deflection -88 Tensile Strain in AC -74 Compr. Strain in Subgrade 87

-89

-84

-89

-89

-88

-89

High Tire Pressure


P e r c e n t Change from standard P a v e m e n t Section A Surface Deflection +8 Tensile Strain in AC +57 Compr. Strain in Subgrade +14

+4

+26

+5

+6

+12

+4

Stabilized Subgrade
P e r c e n t Change from standard P a v e m e n t Section A Surface Deflection -25 Tensile Strain in AC -21 Compr. Strain in Subgrade -57

-15

-12

-41

-11

-12

-32

3-33

Table 3.3 - Comparisons of Changes in Pavement Responses from Standard Pavement

Asphalt Treated Base


P e r c e n t Change from standard P a v e m e n t Section A Surface Deflection -56 Tensile Strain in AC -58 Compr. Strain in Subgrade -77

-48

-68

-69

-33

-51

-52

Moisture Sensitive
P e r c e n t Change from standard P a v e m e n t Section A Surface Deflection +10 Tensile Strain in AC +3 Compr. Strain in Subgrade +16

+22

+55

+38

+33

+77

+64

The thick pavement, Section C, demonstrated the greatest increase in pavement responses to the presence o f a moisture susceptible asphalt mixture. Section A was the least affected o f the three. Further illustration o f the effects o f a weaker surface layer is presented for all three sections in the FWD basins shown in Figure 3. JO .

3-34

Location ol FWD Sensors

Figure 3 '10 Differences in Deflection Basins for the "Standard" vs. "Moisrure Sensitive" Pavement Sections

3-35

FWD Deflection (jim)

FWD

Deflection (in.)

FW Deflection (^m) D

FW Deflection (in.) D

FW D Deflection (tim)

FWD

Deflection (in.)

3.3 3.3.1

ANALYSIS OF RIGID PAVEMENTS Introduction Several theories have been developed over the past years to analyze rigid pavements. These can be divided into four major groups: a. b. c. d. Continuously Supported Slab Models Elastic Layer Models Finite Element Models Coupled Models

A description o f each o f these models follows. 3.3.2 Continuously Supported Slab Models Equations for analysis o f concrete slabs on grade were first published by Westergaard in Denmark in 1923. Westergaard developed closed-form solutions to compute critical stresses and deflections under a single load for three loading cases. These loading cases were: (a) interior load (b) edge load and (c) corner load. Westergaard defined interior loading as the case when the load is at a considerable distance from the edge, while edge loading was described as the case when the load is at the edge, but at a considerable distance from the corner. Westergaard developed these equations by considering the concrete slab as a medium-thick plate supported by the subgrade. The plate was assumed to be o f constant thick ness and to be infinite in both horizontal directions. The subgrade was represented by a set o f uniformly distributed springs, which is referred to as the Winkler foundation. The stiffness o f these springs was referred to as the modulus o f subgrade reaction (k). In this model, the sub grade cannot transfer shear stresses. Therefore, the slab is discontinuous at the edges as the soil beyond the edge does not provide any support. The slab was also assumed to be fully supported with no discontinuities in the slab. The disadvantages o f this model are: (a) Slabs which have cracks cannot be analyzed 3-36

(b) (c)

Load transfer between slabs cannot be considered Effect o f voids or partial subgrade support cannot be considered The assumption that the subgrade or subbase beyond the edge does not provide support is incorrect.

(d)

In spite o f these limitations, Westergaard's equations are still widely used. Westergaard's original equations have been modified many times by different authors partly to bring them to better agreement with elastic theory and to get a closer fit to the experimental data [3.181. Ioannides et al [3.221 carried out a detailed analysis o f Wester gaard's solutions to determine their form, theoretical background, limitations, and applicability. They studied Westergaard's original equations as well as his "new for mulas" from 1948 and compared the results to the ILLISLAB finite element program. Based on this comparison they developed new expressions for corner loading. They also developed slab size requirements for using Wester gaard's equations. The Westergaard's equations for inte rior loading and edge loading, as well as the new solution for comer loading which were developed by Ioannides et al are given in Table 3.4. Table 3-4 Westergaard's Equations

Interior Loading Interior Loading Edge Loading

Maximum Bending Stress Maximum Deflection Maximum Bending Stress

{ 3P(1+ji)/27ih2} {ln(2Ub) + 0.5 -y} + [3P(1+n)]/64h2 (bit)2 ] (P/8k/2) {1+(1/2t0 [tn(al2l) +y -5/4] (all)2 [3(1+n) P/7i(3+n)h2 {/n(Eh3/100ka4) + ] 1.84 - 4p/3 +[(1-n)/2] +1.18(1+2|a)(a//)} {P(2+1.2^)1 }/{Eh3k}1/2{1-(0.76 /2 +0.4|i) (all)}

Edge Loading Corner Loading Corner Loading

Maximum Deflection Maximum Bending Stress Maximum Deflection

(3P/h2) [1.0- (clI)07 2]

(PlkP)tl.20S - 0.69(c/0]

3-37

For the interior and edge loading cases, the load is applied through a circular area with a radius 'a'. For the comer loading case the load is applied through a square loaded area with a side length o f'c ' Maximum Bending Stress (psi) Maximum Deflection (in) P= E= H= h= k= a= c= ft = b= b= y= in = total applied load (lb.) slab Young's modulus (psi) slab Poisson's ratio slab thickness (in) modulus o f subgrade reaction (psi/in) radius o f circular load (in) side length o f square load (in) (Eh3/[12(1 - n2)k ]} {I is the radius o f relative stiffness). [(1.6a2 + h2) 1 ] - 0.675h if a < 1.724h / a if a > 1.724h Euler's constant (=0.57721566490). natural logarithm

While Westergaard considered the subgrade support as a set o f springs, H ogg and Hall (5.30') considered the sub grade as a semi-infinite elastic solid to develop a mathe matical model for determining maximum stress and deflection in a concrete slab under a single load applied at the interior o f the slab. 3.3.3 Elastic Layer Model The elastic layered model for rigid pavements is similar to the elastic layered model for flexible pavements. In this model each layer is defined by its elastic modulus, Pois son's ratio and thickness. A limitation o f using the layered elastic model is that only the interior loading case, where the load is at a considerable distance from the edge can be analyzed. Edge and corner loading cases cannot be evaluated by models based on elastic layer theory. In addition, elastic layer theory cannot be used if cracks exist in the concrete slab or if voids or loss o f support exist below the slab.

3-38

Computer programs which use layered elastic theory can be used to analyze rigid pavements. Some elastic layer programs such as microcomputer version o f ELSYM5, assume that the interfaces between the layers are rough, although it is possible to define a no friction condition. Some programs such as BISAR and WESLEA have the ability to model the interface between the layers as having full adhesion, complete slip or at any condition intermedi ate between these two extremes. It is incorrect, however, to assume that full friction exists between the interface o f the concrete slab and the unbound aggregate base or the subgrade. An elastic layer program which can handle vari able interface condition gives the user the ability to model any interface condition. Therefore, such a model will give a better representation o f the existing conditions in the field, assuming that these conditions are known. 3.3.4 Finite Element Models Numerous finite-element models for analyzing rigid pave ments are available. These finite element programs are capable o f incorporating some or all o f the following fac tors into analysis: partial slab support, load transfer, dowel looseness, aggregate interlock, voids and cracks. ILLI-SLAB is a finite element program which has been widely used to analyze rigid pavements (3 .23). This pro gram was developed in the late 1970s at the University o f Illinois and models a medium-thick plate on a Winkler foundation. This program can evaluate the structural response o f a concrete pavement with joints, cracks or both. Ioannides et al (3.24) expanded ILLI-SLAB for a variety o f support conditions including stress dependent resilient subgrade and Boussinesq elastic half space. 3.3.5 Coupled Models These are structural models that have been obtained by coupling two or more o f the following models: (a) (b) (c) finite element multiple-layers systems analytical (closed-form)

3-39

A description o f coupled models that have been developed is summarized by Majidzadeh (3.25). An example o f a coupled model is the stress analysis model (RISC) devel oped by Majidzadeh (3.25). which couples finite element plate theory with multi-layer elastic theory. This model consists o f a two layer rigid slab resting on a semi-infinite three-layer elastic solid foundation.

3.4 DESIGN PROCESS 3.4.1 Flexible Pavements 3.4.1.1 Introduction The design process described in Section 3.3 and detailed under Section 3.3.4 covers mechanisticempirical concepts applied to the technical per formance o f the pavement alternative. It does not cover the complete decision process, where the choice o f a pavement rehabilitation alternative may be controlled by factors other than technical considerations. 3.4.1.2 Fatigue and Damage Concepts The useful lives o f highway and airport pave ments are finite, and like everything else, tend to "wear out". The rate at which a pavement wears is a function o f its structure, the materials from which it is constructed, traffic levels and charac teristics, and the environment in which it exists. Pavement wear due to traffic is referred to as "damage". Damage can be defined as an altera tion o f the physical properties o f the pavement structure due to the application o f wheel loads. Generally the alterations take the form o f defor mation or cracking o f the pavement layers. Dam age increases exponentially with increasing wheel loads. When the damage reaches a specified level, the pavement is considered to have failed. For example, a single wheel load application may permanently compress or deform the asphalt layer by some minute amount. Over a period o f time as thousands o f wheel loads are applied, the 3-40

permanent deformation becomes significant leading to measurable rutting. When the rutting reaches a pre-specified level (such as 12.5 mm (.5 in.) for example) the pavement is said to have failed. The number o f wheel loads required to reach failure is typically referred to as "number o f wheel load applications to failure", or Nf. The damage at any point in time, D, is defined as the ratio o f the accumulated wheel load applications N (since the pavement was new) to the total number expected to cause failure (Nf) or D --S -.

N,

Figure 3.11 illustrates this concept. At N = 0, or 0 wheel load applications, D = 0. At failure or N = Nf, D = 1. Each wheel load application "consumes" 1/Nf o f the pavement's life. Up to this point, the discussion has concerned only one load level for the wheel in question. H ow are many levels o f wheel loads taken into account? Consider the simple case o f two load levels. As stated earlier, higher wheel loads impart greater damage per application. Figure 3.12 shows two curves for two load levels A and B. Curve A represents the heavier wheel load. Let nt represent a given number o f applications o f wheel load A followed by n2 applications o f wheel load B followed by n3 applications o f wheel load A. The total damage, D, is given by: D = ADj + AD 2 + ADj nl n2 n3

N f* + NfB + NfA This is an illustration o f Miner's Law which states that damage is linearly cumulative, or N D= L where i represents various wheel
1

loads or standard wheel load applied during a particular season. Note that for this approach to be successful, Nfi must be known for any magni tude o f wheel load. 3-41

Number of Load Applications, N

Figure 3.11 - Damage vs. Number o f Wheel Load Applications

3-42

D = AD., + AD2 + AD3


Damage, D = 1.0 Wheel Load A

N um ber of A pplications, N

Figure 3.12 - Procedure to Summarize Damage due to M ultiple Wheel Loads

3-43

As shown in Figure 3.4 and mentioned in Section 3.2.1, pavement distress relationships for flexible pavements typically consider horizontal tensile strain at the bottom o f the surfacing layer for sur face cracking, and vertical compressive strain on top o f the subgrade for rutting. Current rutting problems are often associated with plastic defor mation o f the AC surfacing and is not addressed in this approach. This is for cracking and rutting due to applied wheel loads, with any given load resulting in a specific strain level. As with many materials, the distress relationships relate to strains resulting from stress levels well below the ultimate material strength i.e. a fatigue approach. The typical form o f these distress relationships is: Nfi = A EtjBEc for cracking N ri = A' Svj B' for rutting where Nfj = number o f repetitions o f a given wheel load (i), causing strain level 8tj, which will result in pavement failure by fatigue cracking. = horizontal tensile strain at the bottom o f the surfacing layer caused by wheel load i = vertical compressive strain on top o f subgrade caused by wheel load i elastic modulus o f surfacing (Eq. 3.19) (Eq. 3.20)

8ti

evi

E
A B ,C ,A ',B '

= experimentally determined constants = number o f repetitions o f a given wheel load, causing a strain level 8vi, which will result in pavement failure by rutting

N ri

Failure in cracking would occur when the number o f actual wheel loads o f magnitude i that have passed over the pavement equals Nfj, and simi larly for rutting. The basis o f the design process lies in the following assumptions:

3-44

(i)

if a wheel load i passes over the pavement a certain amount o f the permanent damage occurs so that UA o f the cracking life o f
-*V ri

the pavement is "consumed"; and the rutting life is "consumed" (ii)

of

Miner's hypothesis applies, i.e., that damage (D) is linearly cumulative, so that total damage caused by Nj actual repetitions o f load level i is Dy, = (Eq. 3.21)

and Dr/ = for cracking and rutting,


-Ar

(Eq. 3.22)

ri

respectively. Further, for the spectrum o f loads i, total damage for all loads is

I Df = , i

Nj_ N fi

(Eq. 3.23)

and

D' = ?

f -,

(Eq. 3.24)

and failure occurs in cracking when D f = 1.0 or rutting when D r = 1.0. Whichever occurs first controls the design. The process is often extended to include seasonal effects by summing damage caused by load i during season j, with Nij repre senting actual repetitions o f load level i during season j, etc. It should be noted that the equations pre sented above characterize rutting as a func tion o f the subgrade strain. 3.4.1.3 Distress Criteria The main empirical portions o f the design process are the equations used to compute the number o f loading cycles to failure. These are derived by observing the performance o f pavements and re lating the type and extent o f observed failure to an initial strain under various loads. Currently, 3-45

two types o f failure criteria are widely recog nized, one relating to fatigue cracking and the other to rutting initiating in the subgrade. Other relationships have been developed but are not commonly used, although they are gaining popu larity. As an example, compressive stress in unbound base course materials has been used to predict onset o f pavement roughness [3.18]. Similarly, compressive stresses on top o f cement stabilized subbase layers have been related to crushing damage under wheel loads [3.19]. Fur ther developments should be expected, since it is not reasonable to focus only on cracking o f the surfacing or rutting o f the subgrade while essen tially ignoring other structural layers and modes o f distress. Many equations have been developed to estimate the number o f repetitions to failure in the fatigue mode for asphalt concrete. Some rely on the horizontal tensile strain at the bottom o f the asphalt layer, e t , and rely on the modulus o f the asphalt mix, E, as well. One commonly accepted criterion was suggested by Finn et al. [3.81: log Nf = 15.947 - 3.291 log The above equation assumes that failure is defined as fatigue cracking over 10 percent o f the wheel path area. Figure 3.13 shows the relation ship between tensile strain in the asphalt concrete and the number o f cycles to failure for two levels o f asphalt concrete modulus. It should be kept in mind that other relationships may be based on dif ferent definitions o f failure. For PCC analyses the distress indicator is typically horizontal tensile stress at the bottom o f the PCC layer, and in most cases this is normalized by considering the ratio o f this stress to the modulus o f rupture. Rutting can initiate in any layer o f the structure, making it more difficult to predict than fatigue cracking. Current failure criteria pertain to rut ting which can be attributed mostly to an over stressed subgrade. This is typically expressed in terms o f the vertical compressive strain, ev, at the top o f the subgrade layer:

(Eq. 3.26)
[This equation is rearranged from Chevron, as referenced in 2.121

3-46

Failure in this case is defined as 12.7 mm (0.5 in) depressions in the wheel paths o f the pavement. Figure 3.14 illustrates how the vertical compres sive strain relates to the number o f cycles to fail ure.

Load Applications (N f )

Figure 3.13 - Lim iting Horizontal Strain Criterion for Asphalt Concrete Fatigue Cracking.

3-47

SUBGRADE RUTTING

COMPRESSIVE STRAIN ( * 1 0 6)

LOAD REPETITIONS IN)

Figure 3.14 - Limiting Subgrade Strain Criterion for Rutting

3-48

The number o f cycles to failure for our example pavement sections are given in Table 3.5 for both fatigue and rutting. The lower number o f the two criteria indicates the expected mode o f failure. The "Standard Pavement" with the thin crosssection for example is expected to fail after about 1.000 cycles in rutting. The medium section, B, o f the same case would show rutting after 130.000 load repetitions. Section C o f the "Standard Pavement" would be expected to fail after about 3.3 million cycles in either fatigue or rutting. The "Low Tire Load" results indicate that structural failure o f the pavements is unlikely for any o f the pavement sections. Increased tire pressure primarily affects the fatigue life o f all three sections, although sections A and B will still fail in rutting first. Subgrade stabilization tends to reduce rutting while having less effect on cracking. The use o f an asphalt treated base dra matically benefits all three sections from a struc tural viewpoint. The weakening o f the pavement due to the presence o f a moisture sensitive asphalt mixture has its most severe effect on the rutting failure. The basic conclusion that can be drawn from Table 3.5 is that structural failure for thin (Section A) pavements and weak ("Moisture Sensitive") pavements will be in rutting according to these criteria. Table 3.5. Failure Criteria Applied To Typical Pavements

Millions o f Cycles to Failure Section Criterion Standard Pavement 0.07 0.001 0.39 0.13 3.36 3.24 Low Tire Load 6.10 10.73 170.37 3,331.86 3,895.76 88,714.00 High Tire Pressure 0.02 0.001 0.18 0.11 2.38 3.09 Stabilized Subgrade 0.15 0.05 0.59 1.65 5.07 20.82 Asphalt Treated Base 1.31 0.79 17.41 32.39 42.70 110.08 Moisture Sensitive 0.14 0.001 0.20 0.03 1.10 0.35

A B C

Fatigue Rutting Fatigue Rutting Fatigue Rutting

3-49

3.4.1.4

Typical Approach Designing a pavement using a mechanistic-empirical approach is an iterative process which can include the several steps shown in Figure 3.15. These include: (a) Determining the appropriate number o f sea sons. Analyzing traffic for the design period to determine the total number o f traffic loads in each season (n). Computation o f strains at critical points in the pavement for each season under consid eration. Calculating the number o f cycles to failure (Nf) for each season. Calculating the damage ratio (n/Nf) for each season. Summing the damage ratios for all seasons (D). Increasing or reducing layer thicknesses if D is not close to 1. Determining final cross-section design.

(b)

(c)

(d)

(e)

(f)

(g)

(h)

3-50

Figure 3 . 1 5 flo w Diagram o f a McchanisticEmpirical Design Procedure

3-51

At the outset, one should establish the lengths o f the seasons to be considered in the design proc ess. This selection should reflect time periods during which the material properties do not change substantially. For instance, if an area has three months o f freezing weather, say from De cember through February, the assumption could be that the base and subgrade materials are frozen (high moduli) and that the asphalt concrete is very stiff (high modulus). During the spring, March and April, the moduli o f the underlying layers may be very low, and the asphalt concrete modulus will be an intermediate value. The as phalt concrete modulus will be at its lowest dur ing the summer (May through August) and the base and subgrade will have intermediate moduli. In the fall (September through November), the lower layers will probably maintain intermediate stiffnesses, and the asphalt concrete will again have an intermediate modulus. This type o f analysis would have to be customized to your particular materials and climate. An example o f this kind o f information is shown as follows:

Season Season Winter Spring Summer Fall Length 3 months 2 months 4 months 3 months

Relative Modulus o f Elasticity AC Base Subgrade high inter low inter high low inter inter high low inter inter

The traffic analysis is normally accomplished by using the initial traffic volume, an assumed growth rate, and the design period to calculate the number o f traffic loads on the pavement over its life. For a highway pavement, this normally results in a number o f 80 kN (18,000 lb.) equiva lent single axle loads (ESALs). The 80 kN (18,000 lb.) ESAL is a standard pavement design axle load with the standard 40 kN wheel design 3-52

load representing one side o f the axle. However, it should be noted that since any realistic, mixed traffic loading condition can be used in layered elastic analysis, one is not restricted solely to an ESAL configuration. For the purposes o f illustration, we will use the loading condition in the previous example for the "Standard Pavement". Recall that this was a 40 kN (9,000 lb.) wheel with 552 kPa (80 psi) con tact pressure. I f nT represents the total number o f this type o f load expected over the design life, then the seasonal distribution can be estimated: Winter: Spring: Summer: Fall: nw = (3/12) n j nsp = (2/12)nT ns = (4/12) n j nf = (3/12) nT

In this example, we are assuming that the ex pected traffic loads will be distributed evenly throughout the year. If there are obvious sea sonal patterns in the traffic, spring load restric tions for instance, then an adjustment would have to be made in the calculation to account for this. A layered elastic analysis would need to be per formed for the assumed loading condition for each season being considered. The loading condition as it would appear in the layered elastic analysis is illustrated in Figure 3.4. The initial thicknesses could be selected as the minimum allowable thicknesses for given classes o f pavement. The critical strains in the pavement structure are determined at the locations noted in Figure 3.4. These strains are calculated for each seasonal change in modulus values for the various layers. Next, the number o f load cycles to failure (Nf) for fatigue cracking and rutting are computed using failure criteria. These will be subsequently

3-53

discussed in detail. For the fatigue criterion, the number o f cycles to failure is a function o f the horizontal tensile strain at the bottom o f the asphalt concrete and the modulus o f the asphalt concrete. For rutting, it is a function o f the verti cal compressive strain at the top o f the subgrade. The number o f cycles to failure is computed for each season. The damage for each season is defined as the ratio o f the number o f loads expected for that time period, n, to the number o f cycles to failure, Nf, for that condition in the pavement. The total damage, D, over the design life is then computed by summing the damage in each season: D = V N f w + "sp^fsp + "s ^ fs + nf/Nff I f the total damage is close to 1, it means that close to 100 percent o f the pavement life has been expended. If D is greater than 1, then the pavement has been underdesigned, and the layer thicknesses should be increased. If the damage is much less than one, then the opposite is true. 3.4.1.5 Sensitivity o f design to failure criteria To illustrate the impact o f failure criteria on design, Figures 3.16 through 3.18 were devel oped using the material parameters and loading conditions in the previous example for the "Standard Pavement". Various thicknesses o f bituminous surface were plotted against the com puted number o f repetitions to failure for 150 mm (6 in.), 250 mm (10 in.), and 350 mm (14 in.) thick granular bases. The controlling criterion in the design is the one requiring a greater thickness o f surface material for a given traffic level. Keep in mind that these charts were developed holding the material properties constant; they are not to be used in actual design.

(Eq. 3.27)

3-54

)
E i - 500,000 psl (3,450 MPa) m 25,000 psi (172 MPa) e 2 E3 7,500 psl (52 MPa)

3-55

AC Thickness (In.)

AC Thickness (mm)

Figure 3 . 1 6 Design Chart o f Example Pavement (6 in (150 mm) Base)

E i - 500,000 psl (3.450 MPa) E2 E3 25.000 psl (172 MPa) 7,500 psl (52 MPa)

3-56

AC Thickness

(In.)

AC Thickness (mm)

Figure 3 . 1 7 Design Chan o f Example Pavement (10 in (250 mm) Base)

E i - 500,000 psi (3.450 MPa) E2 E3 25.000 psi (172 MPa) 7,500 psi (52 MPa)

3-57

AC Thickness (in.)

AC Thickness (mm)

Figure 3 . 18Dcsign Chart o f Example Pavement (14 in (350 mm) Base)

The pavement designs resulting from three traffic levels are given in Table 3.6. In this table, the resulting thickness o f asphalt concrete was rounded to the nearest 12.5 mm (0.5 in), and the assumed minimum thickness o f the bituminous layer was 50 mm (2 in.). It is interesting to note that the rutting criterion controlled the design in all cases except those involving the thicker two pavement sections at 2,000,000 load repetitions. The choice o f which o f the three equivalent pavement designs to use at any o f the traffic lev els would depend on a life-cycle cost analysis. 3 .4.2 Rigid Pavements 3.4.2.1 Fatigue and Damage Concepts In a concrete pavement, a crack will result when the tensile stress at the bottom o f the slab exceeds the tensile strength o f the concrete. Such stress can develop due to traffic loading and/or environmental conditions. Furthermore, a number o f laboratory tests have shown that the cracking o f concrete beams can occur with repeated application o f stress, which are smaller than the tensile strength o f the concrete. This type o f cracking is referred to as fatigue cracking. Fatigue tests on concrete beams show that the number o f repeated loads the concrete can sustain before fracture is a function o f the stress ratio, which is the applied stress divided by the modulus o f rupture o f the concrete. The modulus o f rup ture o f the concrete is the maximum tensile stress that the concrete can sustain.

3-58

Table 3.6 - Comparison of Example Designs

40 kN (9,000 lb) Wheel Loads

Base Thickness mm
150

AC Thickness mm
100 60 50 150 130 90 220 200 190

Controlling Criterion

in.
6 10 14 6 10 14 6 10 14

in.
4.0 2.5 2.0* 6.0 5.0 3.5 8.5 8.0 7.5 Rutting Rutting Rutting Rutting Rutting Rutting Rutting Fatigue Fatigue

2X104

250 360 150

2X105

250 360 150

2X106

250 360

* Assumed Minimum Thickness


Typically a concrete pavement is subjected to a variety o f loads. The tensile stress in the concrete caused by each load can be computed. The most critical pavement stress occurs when the wheels loads are applied at or near the pavement edge and midway between the joints. For each load, the ratio between the applied load and the modulus o f rupture (stress ratio) can be com puted. Relationships have been developed between the stress ratio and the allowable load repetitions. Therefore, for each load the fatigue damage during the design period can be com puted from:

D i= ~

(Eq. 3.28)

3-59

where, Dj = Fatigue damage for load i nj = Actual load repetitions o f load i during the design period applied

Nj = Allowable load repetitions for load i ob tained from a fatigue relationship (based on the stress ratio) Applying Miner's hypotheses, when, -ZDj =1 the pavement fails due to traffic load associated dam age. Generally, when the stress ratio caused by a load is less than 0.5 it is assumed that no fatigue damage results from that load. Many failures in rigid pavements occur due to factors that are unrelated to fatigue. Pavement failures may result from relatively few repetitions o f heavy axle loads at slab corners and edges which cause: pumping; erosion o f subgrade, subbase and shoulder materials;

voids under and adjacent to the slab; and faulting o f joints.

Failure due to these other causes should also be considered in the design process. 3.4.2.2 Distress Criteria The failure criteria most commonly used in rigid pavements is that relating to fatigue cracking. Models for fatigue have been developed based on analysis o f laboratory fatigue studies as well as analysis o f performance observations. Concrete fatigue relationships are generally plotted in terms o f a stress ratio and load applications to failure. The stress ratio is the stress induced due to the applied load divided by the modulus o f rupture.

3-60

The fatigue equation based on the relationship for concrete that is used by the Portland Concrete Association is [3.28]: lo g N = 11.78- 12.11 (t7t/M R )fo r 0 .5 < o -t/M R < 1 log N = infinity for o^/Mr < 0 .5 where, M R = Modulus o f rupture, psi crt = Induced tensile stress, psi N = Allowable repetitions As with flexible pavements the load repetitions to failure for a given load level can be obtained from this equation. The maximum stress that is induced in the slab, which corresponds to that for edge loading should be used in these equations. The Portland Cement Association has developed deflection based criteria to account for erosion o f material beneath slab corners. This criteria was developed from pavement performance and faulting data from the AASHO road test. The erosion criteria is used together with the fatigue criteria in the thickness design procedure o f the Portland Cement Association. An alternative fatigue relationship is [3.27]: log N = -1.7136/?+ 4.284 and log N = 2 .8 127/?-12214 fo r/? < 1.25 fo r/? >1.25

(Eq. 3.29)

(Eq. 3.30)

where N is the number o f wheel load coverages to crack 50% o f the slabs, and R is the ratio o f flexural stress to mean modulus o f rupture o f the concrete. This relationship has been found to be highly representative o f pavement performance observed under actual traffic conditions.

3-61

3.4.2.3

Typical Approaches The design process for a rigid pavement is an iterative process. Mechanistic design procedures for concrete pavements have generally been based on limiting the flexural stress induced in the slab, so that fatigue cracking can be avoided. The following is a brief description o f a typical mechanistic-empirical procedure for designing concrete pavements. (a) Determine the design period, predicted traffic and traffic distribution. Decide on a slab size and slab thickness. For each traffic load determine the critical tensile stress in the pavement. Generally, the critical stress is taken as that due to the load placed at the edge midway between the joints. For each load determine the stress ratio, which is the applied stress divided by the modulus o f rupture o f the concrete. Use a fatigue relationship to obtain the allowable load repetitions for each stress ratio. Compute the fatigue damage for each load level by dividing the applied load by the allowable load. Sum the fatigue damage due to all loads. I f the sum o f fatigue damage is more than 1, increase the slab thickness and repeat the procedure. I f it is much less than 1, reduce the thickness o f the slab and repeat the procedure.

(b) (c)

(d)

(e)

(f)

(g) (h)

The above procedure is intended to prevent fa tigue failure in the concrete slab. The Portland Cement Association thickness design criteria addresses subbase and subgrade erosion as well

3-62

as fatigue. The PCA method bases it's thickness design on the most critical o f these two failure criteria. In addition to the selection o f an appropriate cri teria for damage or fatigue, there are many other controversial points in the mechanistic design o f PCC pavements. These points o f controversy include inclusion and calculation o f warping and curling stresses; location o f the critical stress at the comer, edge, or transverse joint; and the inclusion o f erosion/loss o f support. The user should be aware o f these weaknesses when they use these or similar models to perform designs.

3.5 EXISTING OVERLAY AND MECHANISTICEMPIRICAL DESIGN PROCEDURES 3.5.1 Introduction Several pavement design methods will be briefly over viewed. The pavement design procedures for all-new or reconstructed pavements includes the Asphalt Institute MS-1 and the Shell method for AC pavements, and the PCA procedure for PCC. TheW SD O T mechanistic-emp irical overlay design procedure is also shown. All o f the design procedures which follow use layer moduli in some manner, with the exception o f the Asphalt Institute deflection based overlay design procedure. 3.5.2 New Design Procedures 3.5.2.1 Asphalt Institute M S -1 [3.9] The Asphalt Institute presents a mechanistic based design procedure for streets and highways. The following steps are used in this approach: (a) Determine the initial inputs:

(i)

Compute the expected number o f ESALs for the design period.

3-63

(ii)

Determine the design resilient modulus o f the subgrade. This is where the conservatism is built into the design. One selects a resilient modulus value which falls below a specified percentile o f test results for that section o f road. Select combinations o f layer materials. These may include: Full-Depth Asphalt Concrete AC over Emulsified Asphalt Base AC over Untreated Aggregate Base

(iii)

(b)

Find layer thickness combinations for the materials selected. Account for stage construction, if used. This is done on the basis o f remaining life in the pavement. Perform an economic analysis o f the vari ous pavement sections. Determine the final design.

(c)

(d)

(e)

An example design chart from MS-1 is shown in Figure 3.19 which can be used to determine the asphalt concrete thickness for a 150 mm (6 in.) untreated granular base course. 3.5.2.2 Shell Pavement Design Manual 13.15] This is probably the earliest procedure that relies primarily on elastic layer analysis, with the origi nal Shell Design Charts published in 1963 [3.15], These charts were based on linear elastic analy ses, as was the method outlined in 1977 [3.15], The procedure is a fairly general version o f the mechanistic-empirical process employing Shell's BISAR elastic layer program, and allows effects such as temperature to be considered.

3-64

Untreated Aggregate Base, 6.0 in. Thickness

Subgrade Resilient Modulus, MR , psi

Figure 3.19 - Example o f an Asphalt Institute Design Chart [after Ref. 3.91

A three layer structure is used in the analyses, with the traditional distress indicators o f asphalt tensile strain and subgrade compressive strain being used as the primary design criteria. H ow ever, secondary criteria include asphalt rutting considerations as well as stress levels in cemented base layers. Poisson's ratio is 0.35 for asphalt and unbound materials; and 0.25 for cemented bases. The typical approach involves develop ment o f design curves such as those shown in Figure 3.20. The current version o f the Shell Design proce dure is available in Europe as a personal com puter program [3.20], The Shell Design Manual is available for use on PC computers at this time, as is BISAR. The name and phone number o f the Shell representative in Houston appears in Appendix E. 3.5.2.3 PCA Design Procedure [3.21] This procedure is based on theoretical stress analyses originally developed by Westergaard, and subsequently modified by finite-element pro cedures. PCC slab material is characterized in terms o f modulus o f rupture (MR, in MPa, or psi) and the support material represented by the modulus o f subgrade reaction (k in MPa/m or psi/in). The distress indicator for fatigue cracking is the stress ratio (SR) o f horizontal tensile stress to MR. Subgrade rutting is not an issue, but the second distress criterion relates to erosion o f foundation and shoulder materials, which is con trolled by limiting deflections. Computer pro grams are available to perform analyses on trial sections. Alternatively, the tabulated values shown in Table 3.7 are used in conjunction with an assumed design M R and Figure 3.21 to evaluate fatigue and erosion performance. The procedure is illustrated in Table 3.8 (the erosion tables and nomographs have not been included in this text).

3-66

DENSE BITUMEN MACADAM (80/100) WEIGHTED MAAT = 12 C

SUBGRADE MODULUS = 5 * 107 N/m2 TOTAL ASPHALT THICKNESS, mm

TOTAL THICKNESS OF UNBOUND BASE LAYERS, mm

F i g u r e 3.20 - Typical Shell Design Curve

3-67

Table 3.7. PCA Slab Stress [3.211


Equivalent Stress - No Concrete Shoulder (Single Axle/Tandem Axle)

Slab Thickness, in.


4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 12 12.5 13 13.5 14

k of subgrade-subbase, pci 50
825/679 699/586 602/516
526/461

100
726/585 616/500 531/436
4 6 4 /3 8 ?

150
671/542 571/460 493/399 431/353 382/316 341/286 307/262 279/241 255/223 234/208 216/195 200/183 186/173 174/164 163/155 153/148 144/141 136/135 129/129 122/123 116/118

200
634/516 540/435 467/376 409/331 382/296 324/267 292/244 265/224 242/208 222/193 205/181 190/170 177/160 165/151 154/143 145/136 137/130 129/124 122/119 116/114 110/109

300
584/486 498/406 432/349 379/305 336/271 300/244 271/222 246/203 225/188 206/174 190/163 176/153 164/144 153/136 144/129 135/122 127/116 120/111 113/106 107/102 102/98

500
523/457 448/378 390/321 343/778 304/246 273/220 246/199 224/181 205/167 188/154 174/144 161/134 150/126 140/119 131/113 123/107 116/"02 109/97 103/93 98/89 93/85

700
484/443 417/363 363/307 320/264 285/232 256/207 231/186 210/169 192/155 177/143 163/133 151/124 141/117 132/110 123/104 116/98 109/93 103/89 97/85 92/81 88/78

465/416 417/380 375/349 340/323 311/300 285/281 264/264 245/248 228/235 213/222 200/211 188/201 177/192 168/183 159/176 152/168 144/162

411/348 367/317 331/290 300/268 274/249 252/232 232/218 215/205 200/193 187/183 17S/174 165/165 155/158 147/151 139/144 132/138 125/133

Equivalent Stress - Concrete Shoulder (Single Axle/Tandem Axle)

S!ab Thickness, in.


4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11 11.5 12 12.5 13 13.5 14

k of subgrade-subbase, pci 50
640/534 547/461 475/404 418/360 372/325 334/295 302/270 275/250 252/232 232/216 215/202 200/190 186/179 174/170 164/161 154/153 145/146 137/139 130/133 124/127 118/122

100
559/468 479/400 417/349 368/303 327/277 294/251 266/230 243/211 222/196 205/182 190/171 176/160 164/151 154/143 144/135 136/128 128/122 121/117 115/112 109/107 104/103

150
517/439 444/372 387/323 342/285 304/255 274/230 248/210 226/193 207/179 191/166 177/155 164/146 153/137 144/; 30 135/123 127/117 120/111 113/106 107/101 102/97 97/93

200
489/422 421/356 367/308 324/271 289/241 260/218 236/198 215/182 197/168 182/156 169/146 157/137 146/129 137/121 129/115 121/109 114/104 108/99 102/95 97/91 93/87

300
452/403 390/338 341/290 302/254 270/225 243/203 220/184 201/168 185/155 170/144 158/134 147/126 137/118 128/111 120/105 113/100 107/95 101/91 96/86 91/83 87/79

500
409/388 355/322 311/274 276/238 247/210 223/188 203/170 185/155 170/142 157/131 146/122 136/114 127/107 119/101 112/95 105/90 99/86 94/82 89/78 85/74 81/71

700
383/384 333/316 294/267 261/231 234/203 212/180 192/162 176/148 162/135 150/125 139/116 129/108 121/101 113/95 106/90 100/85 95/81 90/77 85/73 81/70 77/67

3-68

10,000 ,000
4

6-

2I.OOQOOO
6-

e-

100,0006-

eREPETITIONS ALLOWABLE LOAD

4-

10,000 6t4-

2-

1000e6-

F a tig u e a n a ly s t s a llo w a b le toad rep etition s based o n stresk ratio fa c t o r (w it h a n d w ith o u t c o n cre te sh o u ld e r).

Figure 3.21 - PC A Fatigue Nomograph [3.211

3-69

Table 3.8. PCA Design Example [3.211

P io je c t

lA j

j^ D L /r - / L n CL.
* n

Jr7i,er-r-//z/ct fTSrct/._______________ _
Ooweiefl |omtt yes no _______ Concrete IhOutOer. yes ______ no Design pfnoO

I n a i t h i c k n e s s _____________ ^

Su b b a s e - s u b g r a o e k _______ / ? /? _____ pci

Moduim ftf ruptijf MR Load safely fa c to r. ISF

jijS ~ L
S -2-

pst

^O

years

V-./n i/r7Jr-Ce/c.cS

c*s<e

Fatigue analysis A ite load kips M u ltip lie d by IS F Enpeciet) re petition s

E 'o s o n ana'ysis

Allowable repetitions

fatig ue. peicent

/
1 2

A novvrftie repet.t.ons

D ilig e peiceni

? 3 4 S C 7

1 S in g le A xles

F q mw a l p n l I f f

9 Stress ratio lactO'

2 O> & -3 /

10. E'OSiOn lactO ' 7

2 S

3D ? 2& 2V< 7? 2n / /U fT.

*6 n 31 Z/. 7 PA.fi P i, OW s> P/ /*? ?


/ / /< ^ /?

? //7 y</ =?/9 /

Q/o/-) -7 ir f in n 1r>v P or> *<7? *Tnr>


/V ? /, /

f &? . 77 . 2 2 dd ,/?L ____ 2 3 0 ,0 3 0 .. J2J.. ____3 0OD / O o n or>n __ i f e / ? / 0 D i O /n//m / / d/ d orx~>


....
C / // o

Z7 0 d o

213

V > * O
7

9 //
/

to

GJZQjDnn
/S flA -7 7 / /s~/-t/ //

OS > D O

Ct-

>nr>
11. Equivalent tress /9 2 12. Stress ratio ta c to '. /?

/ >

T andem A xles S 2.

2 ? 4~
..

,7 O-T 7 /:

*T? J < -S o s?
3 > f} <*-

7/ 0-7 # 7 0 / ? O
3 - 7 7 C>r?r? / J ? /S ~ S 7 n n >

/e > . P A D

-J 3
0

.....

.52\2..

c / in o
n

/ r S i , a e v 7
...

7.3 24
i" z >

>/ /!

... -J2

.. K .j S 2 4 . Z

A / f.3 3 9

..... 3<
S ?/5 2 V -

S) 7 / 0 / 5
/ > f 5~Z

33 2 8 J? ?

n n /r >

*? 2 . O C 'Q O O O s'r;. / r r : y f / ^ - J f
//

/ O
0

< ? /? < { 9 & r> / fV -J O O D /

7 r> /A

On

r?<on
Total

Js?

Total

3 6 *

3-70

3.5.2.4

Other New Design Methods Many other mechanistic-empirical or purely emp irical pavement design methods are currently in use in various locations, but it is beyond the scope o f this course to discuss them. The NHI Course on Pavement Analysis and Design Checks provides an in-depth discussion o f some o f these methods. It should be noted, however, that for a number o f procedures the basis o f design is the same in that a layered elastic analysis is per formed to determine critical stresses and strains. The differences between these procedures typi cally lies in the choice o f the distress criteria which are based on typical material types and related performance requirements.

3-71

3.5.3

Overlay Design Procedures 3.5.3.1 Asphalt Institute M S -17 [3.12] The Asphalt Institute M S -17 is used to describe two separate flexible pavement AC overlay design procedures: one is the effective thickness procedure and the other is a deflection-based process. Both will be briefly described. Effective Thickness Procedure This procedure is based on the fundamental assumption that a pavement structure becomes "thinner" as it ages and is sub jected to traffic. Thus, each pavement layer is converted to an equivalent thickness o f AC in such a way as to account for such deteriorating effects. The following briefly overviews the necessary steps to use this procedure. (a) Subgrade Analysis Determine the subgrade resilient modulus (M r ). A variety o f ways to do this are sug gested in M S -17. This input, along with an estimate o f future traffic (ESALs) is required to determine the thickness for a "new" pavement structure. (b) Traffic Analysis Determine the design ESALs. (c) Effective Thickness o f Existing Pavement Structure (Te) Two methods for determining the effective thickness o f the existing pavement structure are described in MS-17. In Method 1, the existing pavement is converted to an equivalent thickness using Present Service ability Index (PSI) and equivalency factors presented in the manual. This method is limited to pavements consisting o f asphalt

3-72

concrete and emulsified asphalt bases. If other material types are present, Method 2 must be used. In Method 2, some estimate o f the condition and thickness o f each layer must be made. Then each layer is con verted to an equivalent thickness o f AC through the use o f "equivalency factors" (conversion factors) as shown in Table 3.9. The use o f these will be illustrated in an ex ample at the end o f this section. (d) Determine Thickness o f New Pavement

(Tn )
Using the design subgrade resilient modulus and ESALs, determine the "full-depth" AC thickness for a new pavement structure. Figure 3.22 can be used to do this. (e) Determine AC Overlay Thickness (T0) The required overlay thickness is T0 = Tn - Te Deflection Procedure This procedure is based simply on the use o f pavement surface deflections obtained with the Benkelman Beam. The basic steps include the following. (a) Establish the pavement section to uv sur veyed. Perform deflection survey. The procedure in M S -17 ip based on re bound deflections measured with the Benkelman Beam; however, the deflections can be obtained from other devices and "converted" to an equivalent Penkelman Beam deflection. (Eq. 3.31)

(b)

3-73

Table 3.9. Asphalt Institute Conversion Factors [3.121

CONVERSION FACTORS FOR CONVERTING THICKNESS OF EXISTING PAVEMENT COMPONENTS TO EFFECTIVE THICKNESS (Te)
(These conversion factors apply ONLY to pavement evaluation for overlay design. In no
Classification of Material Description of Material Native subgrade in all cases Improved Subgrade** - predominantly granular materials - may contain some silt and clay but have P.l. of 10 or less Lime modified subgrade constructed from high plasticity soils P.l. greater than 10. Conversion Factors*

a) b) c)

0.0

II

Granular Subbase or Base - Reasonably well-graded, hard aggregates with some plastic fines and CBR not less than 20. Use upper part of range if P.l. is 6 or less; lower part of range if P.l. is more than 6. Cement or lime-fly ash stabilized subbases and bases** constructed from low plasticity soils - P.l. of 10 or less. a) Emulsified or cutback asphalt surfaces and bases that show ex tensive cracking, considerable raveling or aggregate degradation, appreciable deformation in the wheel paths, and lack of stability. Portland cement concrete pavements, (including those under as phalt surfaces) that have been broken into small pieces 0.6 m (2 ft.) or less in maximum dimension, prior to overlay construction. Use upper part of range when subbase is present; lower part of range when slab is on subgrade. Cement or lime-fly ash stabilized bases** that have developed pat tern cracking, as shown by reflected surface cracks. Use upper part of range when cracks are narrow and tight; lower part of range with wide cracks. Dumping or evidence of instability.

0.1 -0.2

III

IV

0.3-0.5

b)

c)

3-74

Table 3.9 (cont'd) - Asphalt Institute Conversion Factors [3.121

(These conversion factors apply ONLY to pavement evaluation for overlay design. In no case are they applicable to original thickness design.)_______________________________
Classification Description Conversion of Material___________________________of Material____________________________________ Factors* V a) b) Asphalt concrete surface and base that exhibit appreciable cracking and crack patterns. Emulsified or cutback asphalt surface and bases that exhibit some line cracking, some raveling or aggregate degradation, and slight deformation in the wheel paths but remain stable. Appreciably cracked and faulted portland cement concrete pavement (including such under asphalt surfaces) that cannot be effectively undersealed. Slab fragments, ranging in size from approximately one to four square meters (yards), and have been well-seated on the subgrade by heavy pneumatic-tired rolling. Asphalt concrete surfaces and bases that exhibit some fine cracking, have small intermittent cracking patterns and slight deformation in the wheel paths but remain stable. Emulsified or cutback asphalt surface and bases that are stable, generally uncracked, show no bleeding, and exhibit little deformation in the wheel paths. Portland cement concrete pavements (including such under asphalt surfaces) that are stable and undersealed, have some cracking but contain no pieces smaller than about one square meter (yard). 0.5-0.7

c)

VI

a)

0.7-0.9

b)

c)

VII

a) b) c)

Asphalt concrete, including asphalt concrete base, generally un0.9-1.0 cracked, and with little deformation in the wheel paths. Portland cement concrete pavement that is stable, undersealed and generally uncracked. Portland cement concrete base, under asphalt surface, that is stable, non-pumping and exhibits little reflected surface cracking.______________

3-75

FULL-DEPTH ASPHALT CONCRETE

EouXttonl .0 0 0 -t Smgi^AiK Lead (LAU

F igure 3.22 - Design Chart for Full-Depth Asphalt Concrete (from Ref. [3.12])

3-76

(c)

Calculate Representative Rebound Deflec tion (RRD). The RRD is calculated from the deflection data for the design section by use o f the following: RRD = (x + 2s) (f) (c) where RRD = representative rebound deflection (in.), x = mean o f the individual deflection measurements (in.), = standard deviation o f the deflection measurements (in.), = temperature adjustment factor, and = critical period adjustment factor.

(Eq. 3.32)

The use o f appropriate adjustment factors (" f ' and "c") are very important in calculat ing a realistic RRD. If all deflection meas urements were obtained under uniform pavement temperature conditions, then only one value o f " f 1is needed to adjust the de flections to a standard temperature o f 21.1 C (70F). I f this is not the case, then indi vidual measurements must be adjusted. In any case, Figure 3.23 (a) can be used to obtain the necessary " f 1value (or values) and Figure 3.23 (b) provides an estimate o f pavement temperature. The critical period adjustment factor ("c") is a bit more difficult to estimate. This factor is intended to adjust the deflections to the most "critical" period o f the year for a specific pavement section. If for example, the most critical period happens to be the spring thaw period (if this even occurs for a given pavement), then that is the set o f deflections which must be estimated.

3-77

T H IC KN ESS OF UNTREATED AGGR EG ATE BASE

s .00 o
}

01 ill

10

0 MILLIMETERS INCHES ?>

TEMPERATURE, C

PAVEMENT

MEAN

TEMPERATURE ADJU STM EN T FACTOR (F)

Figure 3.23 (a) - A verage Pavem ent T em perature versus B enkelm an Beam Deflection A d ju stm en t Factors for Full-Depth and T hree-L ayered A sp h alt C oncrete Pavements [from Ref. 3.121

3-78

MEAN

PAVEMENT

TEMPERATURE, #F

PA V EM EN T SURFACE TEMPERATURE PLUS 5-DAY MEAN AIR TEMPERATURE. *>F 0 20 40 60 80 100 120 140 160 180 200 220 240 260

TEMPERATURE

P AVEM ENT SURFACE TEMPERATURE PLUS 5-DAY MEAN AIR TEMPERATURE, C

F igure 3.23 (b) - Predicted Pavem ent T em p eratu re [from Ref. 3 J 2

3-79

TEMPERATURE

A DEPTH, F T

A DEPTH, T

Thus, the spring time is the preferred deflection testing time and "c" would equal 1.0. If the deflections were obtained in the summer, then "c" would presumably be greater than 1.0 (say 1.5 or so). Unfortu nately, each unique pavement section has its own unique "c" value. (d) Calculate overlay thickness. Use Figure 3.24 along with the design ESALs and the RRD to determine the nec essary overlay thickness. 3.5.3.2 The Shell Method D.lOl An overlay design procedure was developed by Shell Research, in The Netherlands, based on re sults of FWD testing. The deflection measure ments are used along with a knowledge o f past traffic and the environment to estimate the re maining life o f the existing pavement structure. The remaining life in combination with future traffic requirements are used to determine the re quired thickness o f overlay. The failure criterion in this procedure is based on fatigue. A flow chart o f the procedure is shown in Figure 3.25, and an example o f the overlay design procedure is given in Figure 3.26. The interpretation o f the FWD results in this case is not done by backcalculation. Instead, the sub grade modulus and effective thickness o f the asphalt layer are determined by the maximum deflection, a deflection ratio between the deflec tion at 600 mm from the load to the maximum, assumed Poissons ratios, thickness o f the granu lar base, assumed ratio o f base to subgrade modulus, and the asphalt mix stiffness.

3-80

RRD, MM

0 50

1.00

1.50

2 00

2.50

3 00

3.50

4.00

4.50
EAL 10 , 000,000

OVERLAY THICKNESS, MILLIMETERS O ASPHALT CONCRETE F

5,000,000

2 , 000,000

1, 000,000
500.000

200.000 100,000 50.000 20.000 10,000 5,000


0.000 0.020 0.040 0.060 0.080 0.100 RRD, INCHES 0.120 0.140 0.160 0.180

Figure 3.24 - Asphalt Concrete O v e r l a y Thickness Required to Reduce Pavement D e f l e c t i o n s from a Measured to a Design Deflection V a l u e (Rebound Test) [From Ref. 3.12]

3-81

Figure 3 . 2 5 F 1 o w Chart o f Shell Overlay Design Method [after Ref. 3.10]

3-82

RRO, M M

OVERLAY THICKNESS, MILLIMETERS O ASPHALT CONCRETE F

RRD, INCHES

Figure 3.24 Asphalt Concrete O ve rl ay Thickness Required to Reduce Pavement Deflections from a Measured to a Design Deflection Va lue (Rebound Test) [From Ref. 3.12]

3-81

Verification

Selection of optimum overlay

Figure 3 . 2 5F1ow Chart of SheU Overlay Design Method [after Ref. 3.10]

3-82

a M

3.5.3.3

Washington State Department o f Transportation

Another mechanistic-empirical overlay design procedure (EVERPAVE) was developed by the Washington State Department o f Transportation that is based on the backcalculation o f material properties and fatigue and rutting failures. In this approach, layer moduli can be calculated for each deflection test point. The asphalt concrete modulus is corrected for temperature according to data for typical Washington mixtures (Figure 3.27). Next, an iterative process is used to determine an appropriate overlay thickness for each deflection test point as shown in Figure

3.28.
Both the unstabilized base course (subbase) and subgrade moduli can be non-linear with stress state, i.e., the base, subbase, and subgrade layer moduli can take the following form: E = K, (q) or
K4

K2

(Eq. 3.33)

E = K3 (sd) 4 with the exponents being either positive or nega tive. The failure criteria used in EVERPAVE are based on two basic criteria: rutting and AC fatigue cracking. The rutting criterion was adopted from the Asphalt Institute [3.9. 3.1-7]:
Nf =

(Eq. 3.34)

1.05 x 1(T

4.4843

(Eq. 3.35)

where

Nf

allowable number o f 18,000 lb. (80 kN) single axles so that rutting at the pavement surface should not exceed 0.5 in. (12.7 mm), and vertical compressive strain at the top o f the subgrade layer.

Ey

3-84

Temperature (C)

Rosiliert

Modulus (psl)

Temperature (eF)

Figure 3.27 - Genera! StilTness-Temperature Relationship for Class B (D ense Graded) Asphalt Concrete in W ashington State [from Ref. 3.131

3-85

Read Input Data M aterial properties Seasonal variation Traffic Assume overlay thickness Determine seasonal material properties Analyze pavement Structure fs^, e ) ,,~ 'Calculate performance life (NL N.__________ Determine Damage Ratio Compute Sum of Damage RaUo (SDR) Calculate seasonal tra ffic volume

Increase overlay thickness

^S D R < K Z .

Produce overlay design

Figure 3.28 - W SDO T Overlay Design Flow Chart

3-86

The fatigue cracking failure criterion is based on Monismith's laboratory based model [3.14] and the subsequent work by Finn, et al. f3.8] and Mahoney, et al. [3.131. Fatigue cracking: Nfieid = (Niab) (SF) where Nfieid = Niab = number o f load applications o f con stant stress to cause fatigue cracking, relationship from laboratory data [3.13,3.141, (E 14.82 - 3.291 log ! L - 0.854 log Uo can range from about 4 to 10, depends on AC thickness, ESAL level, climate. (Eq. 3.36)

10

(Eq. 3.37)

SF

The Nfieid applications is estimated to result in about 10 percent or less fatigue cracking in the wheel path area. The original Finn, et al. [3.81 model based on the Monismith laboratory work [3.14] and the results o f the AASHO Road Test is: log N f = where Nf = 5.947 - 3.291 log { - ^ 1 - 0.854 log J O "6 number o f axle applications to result in 10 percent or less fatigue cracking in the wheel path area, horizontal tensile strain at the bottom of the AC layer, and modulus o f the AC layer (psi).
(Eq. 3.38)

et

Eac

The difference between the Finn equation above and the Monismith laboratory based relationship is about 13.4. Thus, the laboratory fatigue rela tionship was "shifted" by a factor o f 13.4 to more realistically represent a field fatigue prediction for the accelerated loading conditions at the AASHO Road Test. WSDOT studies [3.13] have shown, as stated above, that realistic shift factors for inservice WSDOT pavements are less than 13.4

3-87

(more like 4 to 10). Generally, the shift factor is increased for high traffic conditions on say 100 to 150 mm (4 to 6 in.) AC. The shift factor is lower for flexible pavements with AC thicknesses o f about 175 to 200 mm (7 to 8 in.) or thicker. It is appropriate to note that Finn, et al. [3.8] only analyzed the 100, 125 and 150 mm (4, 5, and 6 in.) thick AC flexible pavement sections from Loop 4 (7 sections) and Loop 6 (10 sections) from the AASHO Road Test data. 3.5.3.4 Other Overlay Design Procedures There are a number o f other overlay procedures in use today that are based on mechanisticempirical techniques. For many o f these, the overlay design itself follows the "Typical ap proach" outlined in Section 3.4.1.4, with distress criteria chosen to represent conditions relevant to the particular application. This overlay design is performed after evaluation o f the existing struc ture, often by the use o f backcalculation. The WSDOT EVERPAVE procedure described in Section 3.5.3.3 is a good example o f this approach.

3.6 EXAMPLE 3.6.1 Introduction The medium AC thickness section (B) shown in Figure 3.8 will be used to illustrate the use o f some o f the various AC overlay design procedures. Even though this is only a hypothetical pavement, we will make the necessary as sumptions to make use o f this "pavement section." We assume that Section B can have two levels o f AC stiffness, E a c = 3450 MPa (500,000 psi), which assumes no cracking, and E ac ~ 1035 MPa (150,000 psi), which implies extensive fatigue cracking o f the AC surfacing. The necessary material properties, layer thicknesses, and deflections (calculated using ELSYM5) are summarized in Table 3.10. Further, we will design AC overlays for three assumed ESAL levels: 1,000,000; 2,000,000; and 5,000,000.
3-88

3.6.2 Asphalt Institute Effective Thickness Procedure


(a)

Subgrade M r M r = 52 MPa (7500 psi), a given

(b)

Traffic analysis Use ES AL levels of: 1,000,000 2 , 000,000 5,000,000 Therefore, obtain three overlay thicknesses.

(c)

Effective thickness o f existing pavement structure

(Te )
Use Table 3.9 for equivalency factors. L AC @ 3450 MPa 500.000 psj) (125 mm)*(1.0)= 125 mm (5 in.) (200 mm)*(0.2)= 40 mm (1.6 in.)
Total: 165 mm (6.6 in.)

AC @ 1035 MPa (150.000 psD (125 mm)*(0.5)= 63 mm (2.5 in.) (200 mm)*(0.2)= 40 mm (1.6 in.)
103 mm (4.1 in.)

AC Base

(d)

Determine thickness o f new pavement (Tn). Use Figure 3.22 and appropriate M r and ESAL levels. ______MR__________ ESAL Level____________ Tn_______ 52 MPa (7500 psi) 1,000,000 2,000,000 5,000,000 231 mm(9.1 in.) 262 mm(10.3 in.) 312 mm(12.3 in.)

3-89

Table 3.10 - Pavem ent R esponse Summary for the M edium Thickness Section (Section B)*

125 mm (5) AC 200 mm (8) Base

Fine grained subgrade


AC Moduli, MPa (psi) 1,035 (150,000) 3,450 (500,000) C alculated D eflections, fim (in.) D 0896.4
d 8-

Load

Di2 569.2

D24 351.3

D36 237.2

d 48

P = 40 kN (9,000 lb)

685.0

173.0

(0.03529) (0.02697) (0.02241) (0.01383) (0.00934) (0.00681) 690.4 589.0 517.9 348.2 242.1 177.3

a = 150mm (5.9 in.)

(0.02718) (0.02319) (0.02039) (0.01371) (0.00953) (0.00698)

* Pavement responses obtained from ELSYM5 program

3-90

(e)

Determine AC overlay thickness (T0). ESAL Level 1.000.000 2.000.000 5.000.000 1.000.000 2.000.000 5.000.000

Eac, MPa (psi)___________ T0 = Tn - Te


3450 (500,000) 3450 (500,000) 3450 (500,000) 1035 (150,000) 1035 (150,000) 1035 (150,000) 231 - 165 = 66 mm (2.6 in.) 262 - 165 = 97 mm (3.8 in.) 312 - 165 = 147 mm (5.8 in.) 231 - 103 - 128 mm (5 in.) 262 - 103 = 159 mm (6.3 in.) 312 - 103 = 209 mm (8.2 in.)

3.6.3 Asphalt Institute Deflection Procedure (a)

Eac = 1035 MPa (150,000 psi)


Assume that Do from Table 3.10 represents the mean deflection with a standard deviation about Vi as large as the mean. Therefore, x s = = 896.4 pm (0.3529 in.) 448.2 |jm (0.01764 in.)

Further, assume the deflections were obtained for an average pavement temperature o f 60F and the criti cal period adjustment factor (c) = 1.25. The tem perature adjustment factor (f) is 1.1 from Fig ure 3.23. RRD = = (0.03529 + 2(0.01764))*(1.1)*(1.25) 2464.7 nm (0.09703 in.) = = = 122 mm (4.8 in.) for 1,000,000 ESALs 147 mm (5.8 in.) for 2,000,000 ESALs 190 mm (7.5 in.) for 5,000,000 ESALs

Overlay thickness (from Figure 3.24)

(b)

Eac = 3450 MPa (500,000 psi)


All calculations and estimations will be the same as (a) except Do = 690.4|im (0.02718 in.). Thus, x s f c = = = = 690.4 |am (0.02718 in.) 345.2 urn (0.01359 in.) 1.1 1.25

3-91

Therefore, RRD = (0.02718 + 2(0.01359))*(1.1) (1.25) = 1898.5 nm (0.07475 in.) Overlay thickness (from Figure 3.24 ) = = = 97 mm (3.8 in.) for 1,000,000 ESALs 122 mm (4.8 in.) for 2,000,000 ESALs 157 mm (6.2 in.) for 5,000,000 ESALs

(c)

E ac = 1035 MPa (150,000 psi) and E ac ~ 3450 MPa (500,000 psi) Revised If the pavement deflections had been taken during the critical period and no temperature adjustment was needed (i.e., measurements obtained at 21.1C (70F)) and all measurements were the same (i.e., s = 0, which is highly unlikely), then the resulting overlays would be the following: E ac = E ac = E ac = 1035 MPa (150,000 psi) RRD = 896.4 |am (0.03529 in.) 3450 MPa (500,000 psi) RRD - 690.4 nm (0 02718 in.) 1035 MPa (150,000 psi) = = = 10 mm (0.4 in.) for 1,000,000 ESALs 43 mm (1.7 in.) for 2,000,000 ESALs 74 mm (2.9 in.) for 5,000,000 ESALs

Overlay thickness (from Figure 3.24)


E ac =

3450 MPa (500,000 psi) = = = 0 mm (0.0 in.) for 1,000,000 ESALs 0 mm (0.0 in.) for 2,000,000 ESALs 46 mm (1.8 in.) for 5,000,000 ESALs

Overlay thickness (from Figure 3.24)

3.6.4 WSDOT Mechanistic-Empirical (a) The EVERPAVE program was used with the fol lowing inputs (assumed location for seasonal effects is Spokane, Washington)

Case 1 Existing pavement moduli E ac = 1035 MPa (150,000 psi) EBs = 172 MPa (25,000 psi) E sg = 52 MPa (7,500 psi)
3-92

New AC modulus = 3450 MPa (500,000 psi) Fatigue shift factor = 10 ESAL levels 1,000,000 2 , 000,000 5,000,000

Case 2 Same as Case 1 but existing EAC = 3450 MPa (500,000 psi) (i.e., no initial fatigue cracking) Case 3 Existing pavement moduli EAc = 1035 MPa (150,000 psi) EBs = 8,000 (0)0-375 E sg = 52 MPa (7,500 psi) New AC modulus = 2760 MPa (400,000 psi) Fatigue shift factor =10 ESAL levels 1,000,000 2 , 000,000 5,000,000

Case 4 Same as Case 3 but existing E ac = 3450 MPa (500,000 psi) (i.e., no initial fatigue cracking) (b) Results Case 1 (Original surfacing E ac = 1035 MPa (150,000 psi)) "cracked AC" ESAL Level 1.000.000 2.000.000 5.000.000
3-93

AC Overlay Thickness, inches (mm) 30 mm (1.2 in.) 51 mm (2.0 in.) 89 mm (3.5 in.)

Case 2 (Original surfacing E ac = 3450 MPa (500,000 psi)) "uncracked AC" ESAL Level
1.000.000

AC Overlay Thickness, inches (mm)


0 mm (0 in.)

2.000.000 5.000.000

13 mm 43 mm

(0.5 in.) (1.7 in.)

Case 3 (Original surfacing E ac = 1035 MPa (150,000 psi)) "cracked AC" ESAL Level 1.000.000 2.000.000 5.000.000 AC Overlay Thickness, inches (mm) 56 mm (2.2 in.) 81 mm (3.2 in.) 127 mm (5.0 in.)

Case 4 (Original surfacing E ac = 3450 MPa (500,000 psi)) "uncracked AC" ESAL Level 1.000.000 2.000.000 5.000.000 3.6.5 Summary Refer to Table 3.11 for a summary of the various overlay thicknesses for the cases used in this example. Given that the "pavement section" used was purely hypothetical and required numerous assumptions, one should not expect the various overlay design procedures to result in similar solutions; however, there is a modest amount of agree ment among the design procedures used. AC Overlay Thickness, inches ('mm') 5 mm 30 mm 71 mm (0.2 in.) (1.2 in.) (2.8 in.)

3.7 USE OF ELASTIC ANALYSIS SOFTWARE 3.7.1 Introduction In order to make use of the backcalculation results, we will become acquainted with a layered elastic analysis computer program, ELSYM5. Originally developed at the
3-94

Table 3.11 - Summary o f Overlay Thicknesses for M edium Thickness AC


D esign P ro cedures
Al - Eff. Tk. Al - Eff. Tk. Al - Deflection Al - Deflection @ e ac = @ E AC = @ eac = 1035 MPa 3450 MPa 3450 MPa (150 ksi) (500 ksi) (500 ksi) uncracked cracked uncracked

E S A L Level

@ E ac =
1035MPa (150 ksi) cracked

1,000,000

127 mm (5.0")

64 mm (2.5)

122 mm (4.8 )

97 mm (3.8 )

2,000,000

157 mm (6.2 )

94 mm (3.7 )

147 mm (5.8)

122 mm (4.8 )

5,000,000

208 mm (8.2 )

145 mm (5.7 )

190 mm (7.5 )

157 mm (6.2)

D esign P rocedures
W SDOT EV E R P A V E @ e ac = 1035MPa cracked E BS = 25,000 psi 30 mm (1.2 ) W SDOT EV E R P A V E @ e ac = 3450 MPa uncracked E BS = 25,000 psi 0 mm (0) W SDOT EV E R P A V E
@ E ac =

E S A L Level

W SDOT EV E R P A V E @ E Ac = 3450 MPa E bs 8,000 ( 0 ) 575

1035 MPa E BS = 8,000 (0)0.376

1,000,000

56 mm (2.2 )

5 mm (0.2)

2,000,000

51 mm (2.0 )

13 mm (0.5)

81 mm (3.2 )

30 mm (1.2)

5,000,000

89 mm (3.5 )

43 mm (1.7 )

127 mm (5.0)

71 mm (62.8)

3-95

University o f California, Berkeley for use on a main frame, this version of ELSYM5 was adapted for use on microcomputers by the Federal Highway Administration. It is menu driven and, for the most part, self-explanatory. Until recently an integration error in ELSYM5 (as well as the CHEV program series) generated errors which were significant under certain conditions. This has been cor rected at Cornell University and care should be taken that the correct version is being used. A rectangular coordinate system (X,Y,Z) is used for in put and output data. The horizontal plane is described by X and Y with Z defining the vertical axis. In this program, Z is positive in the downward direction. The loading conditions are defined by any two of three pa rameters: load, contact pressure, or radius of the loaded area. The other value is computed by the program. Each pavement layer is described by its modulus of elas ticity, Poisson's ratio, and thickness. The layers are num bered from the top downward. The subgrade layer is given a thickness of 0 to indicate a semi-infinite depth. 3.7.2 Software Demonstration In this section, we will describe the use of ELSYM5 in a step-by-step example including locating the program in the computer, inputting the data, running the problem, and retrieving the data. The user must first locate the disk drive and, if necessary, the sub directory in which ELSYM5 resides. For instance, if we are currently in the "C" drive and ELSYM5 is located in the "A" drive in a directory named "elsym", then type: C:>a: The computer will respond with an A prompt, and we must tell it to go to the "elsym" directory: A:>cd\elsym This puts us in the proper place for accessing the pro gram. The various screens used in ELSYM5 are shown in Figures 3.29 through 3.38.

3-96

Next, we begin the process of entering the data by typing: A:ELSYM>/5y/w5 The title block and main menu for the program should appear as shown in Figure 3.29. One can choose to receive some abbreviated instructions on the program by typing 1 for the selection. Creating a data file or modify ing an existing file can be done by typing a 2 or 3, respec tively. Selecting 4 will allow you to run the program with the current data file, and 5 will allow you to exit from the program. In our example we will create a new data file, so we select 2. Screen 1.2 (.Figure 3.30 ) appears, and we can now enter the data for our problem. The data for the example comes from Section A of the "Typical Pavements" described ear lier. Option 1 is selected and the title is entered. After the RUN TITLE has been selected, Screen 1.2 returns. Next the pavement layer data are entered by choosing number 2 on Screen 1.2. Then Screen 1.2.2 appears (Figure 3.31), and the user inputs the number of layers, layer thicknesses, Poisson's ratios, and moduli. Layer numbers are entered automatically by the program. To go from one data field to the next, simply press the ENTER key on the computer. The user can incorporate a rigid layer below the subgrade by giving the subgrade a finite thickness. If this is not desired, then give the subgrade a thickness of 0. The data from Section A in the example are shown in Figure 3.31. Screen 1.2 will come back after the user completes entering the data and presses the F2 key. Then Option 3 is chosen in order to provide the load data. This is done on Screen 1.2.3 (Figure 3.32). In this case, we have entered the load and pressure according to the example. The program will calculate the radius of the loaded area. The number of load locations (up to 10) is selected, and the X,Y coordinates of the center of the loads are input. When this is complete, the user presses F2 and Screen 1.2 appears again. The last items to be specified are the locations of interest for the analysis. This is done by selecting Option 4 on Screen 1.2. Then Screen 1.2.4 comes up as shown in

3-97

Figure 3.33. The user first enters the number of horizon tal positions for evaluation. In our example, we are inter ested in various points under the centerline of the tire, so 1 is entered. We want to evaluate the deflection at the pavement surface, the horizontal tensile strain at the bot tom of the AC, and the vertical compressive strain at the top of the subgrade; so three Z locations are specified as shown in Figure 3.33. In order to obtain the vertical strain at the top of the subgrade, a point just below the base/subgrade interface must be specified. In this case, it was 203 mm (8.01 in.). Again, the key F2 returns us to Screen 1.2.

Next, the user can choose to store the data in a file by selecting Option 5 on Screen 1.2. If this is done, the pro gram will prompt you for a file name. After this has been accomplished, return to the Main Menu by typing 6 on Screen 1.2. After the data have been entered or modified, select Op tion 4 on the Main Menu to have the program run the problem. When this is done, a message will appear on the screen as shown in Figure 3.34. The program will prompt you for a file name if you want to save the results of your run. Otherwise, it will tell you that it is performing the calculations. A results menu for the first Z location will appear as shown in Figure 3.35. Notice that this menu is for layer 1 at the pavement surface. Since we are interested in the deflection at this point, we type a 3, and the displacements appear as shown in Figure 3.36. At this point, we look under the heading UZ to find the vertical displacement. Once this value has been noted, a 4 is typed to move on to the next analysis point, which is layer 1, at a depth of 50 mm (2 in.). Here, we want to know the horizontal tensile strain at the bottom of the AC. So, a 2 is typed on the results menu and the results appear as in Figure 3.37. The horizontal strains are listed under the heading EXX or EYY (.467E-03 or 467 x 10"6 in./in.). For the next loca tion a 4 is typed and the next results menu for layer 3 at a depth of 203 mm (8.01 in.) is displayed. We are inter ested in the vertical strain at this point, so we type a 2 and Figure 3.38 appears. This value is found under the head ing EZZ (-.224E-02 or -224 x 10*6 in./in.).

3-98

MAIN MENU
-ELSYM5Interactive Input Processor Version 1.1, Released 04/93 Developed by SRA Technologies, Inc. Updated by Cornell Local Roads Program Under Contract to Federal Highway Administration MAIN MENU 1. 2. 3. 4. 5. Instructions Create a New Data File Modify an Existing Data File Perform analysis Exit - Return to DOS Selection :

Figure 3.29 - ELSY M 5 M ain M enu

DATA FILE MENU


Create a New Data File Menu 1. Enter/Modify Run Title 2. Enter/Modify Elastic Layer Data 3. Enter/Modify Load Data 4. Enter/Modify Evaluation Location Data 5. Write Data to an Output File 6. Return to Main Menu Selection:

Figure 3.30 - Screen 1.2

3-99

Screen 1.2.2

ELASTIC LAYER DATA


Humber of layers: 3 Layer (top to Number bottom) 1 2 3 Thickness (inches) 2.00 6.00 .00 Poisson's Ratio .35 .40 .45 Modulus of Elasticity 500000.00 25000.00 7500.00

Not:

Enter zero thickness when bottom layer is semi-infinit

FI: Modify This Screen; F 2 : Return To Screen 1.2

Figure 3.31 - Elastic Layer Data Screen

Screen 1.2.3'

LOAD DATA
Enter two of the following, the third is calculated. Load: 9000 lbs pressure: 80 psi Load Radius:.00 in 1 Coordinates X =
1 0

Number of load locations: Location number =

Y =
0

Fl: Modify This Screen; F2: Return To Screen 1.2

Figure 3.32 - Load Data Screen

3-100

Screen 1.2.3'

EVALUATION LOCATION DATA


Results are evaluated for all combinations of X-Y coordinates and Depths of Z. Humber of X-Y positions : 1 Humber of Z positions : 3

Figure 3.33 - Evaluation Location Data Screen

ANALYSIS MODE OF ELSYM5 Do you wish the Results Saved on a File (Y/N)==> n *** PERFORMING CALCULATIONS ***

Figure 3.34 - Analysis M ode o f ELSYM 5

3-101

Description o f "Standard" Sections

Earlier we discussed the evaluation of multi-layer systems using layered elastic theory. Now each group will run one of the cases for each of the standard sections (A, B, and C) which were described in Figure 3.8. The layer thick nesses in the sections were: __________ Thickness, in. (mm)__________ Laver AC Base Stab. Subg.* Section A 2 (50) 6(150) 6(150) Section B 5 (125) 8 (200) 6(150) Section C 9 (230) 6(150) 6(150)

*For Group No. 4 .4 Classroom Exercise Find your assigned group number, and run the case below for Sections A, B, and C on ELSYM5. Evaluate the sur face deflection, horizontal strain at the bottom of the AC, and the vertical strain on top of the subgrade at the center of the load. Compare your results with those listed from Table 3.2. If you notice a discrepancy in the results, please notify one of the instructors. Also, calculate the number of repetitions to failure for AC fatigue (Equation 3.25) and subgrade rutting (Equation 3.26) that you would expect in each case. These calculations are the basis of the mechanistic-empirical design approach. Repetitions to failure, combined with traffic (repetitions per year) provides the design life in years. RESULTS MENU FOR ELSYM5 LAYER = 1 Z = .00

1. 2. 3. 4.

- Stresses

Normal

& &

Shear Shear

& &

Principal Principal

- Strains Normal - Displacements

- Return or Continue with Next Layer Selection ==>

Figure 3.35 - Results Menu for ELSYM5

Displacements XP .00 YP .00 UX .00E+00 UY .00E+00 UZ .483E-01

RESULTS MENU FOR ELSYM5 LAYER 1. 2. 3. 4. - Stresses - Strains = 1 Normal Normal & & Z Shear Shear = & & .00 Principal Principal

- Displacements - Return or Continue Selection ==>

with Next Layer

Figure 3.36 - First O utput Location Requested

Normal Strains XP .00 XP .00 YP .00 YP .00 EXX EYY Strains PE 2 PE 3 EZZ

Shear Strains EXY EXZ EYZ .000E+0 PSE3 .104E-0

.467E-03 .467E-03 -.572E-03 .000E+00 .000E+00 Shear Strains PSE1 PSE2 PE 1

Principal

.467E-03 .467E-03 -.572E-03 .104E-02 .000E+00

RESULTS MENU FOR ELSYM5

1. 2. 3.
4.

LAYER = 1 Z = 2.00 - Stresses Normal Shear i Principal - Strains Normal t Shear I Principal - Displacements - Return or Continue with Next Layer Selection ==>

Figure 3.37 - Second Location Strains

3-103

Normal Strains

Shear Strains EYY EZZ EXY EXZ EYZ

XP .00 XP .00

YP .00 YP .00

EXX

.970E-03 .970E-03 -.224E-02 .OOOE+OO .OOOE+OO .000E+0 Strains PE 2 PE 3 Shear Strains PSE1 PSE2 PSE3 PE 1

Principal

.970E-03 .970E-03 -.224E-02 .321E-02 .000E+00 .321E-0

RESULTS MENU FOR ELSYM5

1. 2. 3. 4.

LAYER = 3 Z = 8.01 - Stresses Normal i Shear L Principal - Strains Normal Shear t Principal - Displacements - Return or Continue with Next Layer Selection ==>

Figure 3.38 - Third Location Strains

3-104

Group No.: 1 Number of Layers: 3 Material Properties: Laver Number 1 2 3 Load Data: Number of Loads: 1 Load: 9,000 lb. (40 kN) Contact Pressure: 80 psi (552 kPa) Results: Your Results Vertical Defl. @ Surface, in. Section A Section B Section C ___________ ___________ ___________ 0.048 in. (1.219 mm) 0.027 in. (0.686 mm) 0.018 in. (0.457 mm) Table 3.2 Results Poissons Ratio .35 .40 .45 Modulus of Elasticity, psi (MPa) 500,000 25,000 7,500 (3450) (172) (52)

Horizontal Strain in AC, in/in x 10-6 Reps, to failure Section A Section B Section C ___________ ___________ ___________ _______ _______ _______ 467 279 145

Vertical Strain on Subgrade, in/in x 10-6 Section A Section B


Section C

___________ ___________

_______ _______

2,220

-747
-370

____________ _________________ 3-105

Group No.: 2 Number o f Layers: 3 Material Properties: Laver Number Poissons Ratio Modulus o f Elasticity, psi (MPa')

1 2 3 Load Data: Number of Loads: 1 Load: 900 lb. (4 kN) Contact Pressure: 30 psi (207 kPa) Results: Your Results Vertical Defl. @ Surface, in. Section A Section B Section C Horizontal Strain in AC, in/in x 10-6 Section A Section B Section C Vertical Strain on Subgrade, in/in x 10"6 Section A Section B Section C

.35 .40 .45

500,000 (3450) 25,000 (172) 7,500 (52)

Table 3 .2 Results

0.006 in. (0.152 mm) 0.003 in. (0.076 mm) 0.002 in. (0.051 mm) Reps, to failure 121 44 18

-280 -81 -40

3-106

Group No.: 3 Number of Layers : 3 Material Properties: Laver Number 1 2 3 Load Data: Number of Loads: 1 Load: 9,000 lb. (40 kN) Contact Pressure: 140 psi (965 kPa) Results: Your Results Vertical Defl. @ Surface, in. Section A Section B Section C Horizontal Strain in AC, in/in x 10-6 Section A Section B Section C Vertical Strain on Subgrade, in/in x 10-6 Section A Section B Section C -2,520 -786 -384 Reps, to failure 735 352 163 0.052 in (1.321mm) 0.028 in. (0.711 mm) 0.019 in. (0.483 mm) Table 3 .2 Results Poissons Ratio .35 .40 .45 Modulus of Elasticity, psi (MPa) 500,000 (3450) (172) 25,000 (52) 7,500

3-107

Group No.: 4 Number o f Layers: 4 Material Properties:

Laver Number 1 2 3 4 Load Data: Number of Loads: 1

Poissons Ratio .35 .40 .40 .45

Modulus o f Elasticity, psi (MPa)

500,000 25,000 50,000 7,500

(172) (345) (52)

Load: 9,000 lb. (40 kN) Contact Pressure: 80 psi (552 kPa) Results: Your Results Vertical Defl. @ Surface, in. Section A Section B Section C Horizontal Strain in AC, in/in x 10-6 Section A Section B Section C ___________ ___________ ___________ Reps, to failure _______ _______ _______ 368 246 128 ___________ ___________ 0.036 in. (0.914 mm) 0.023 in. (0.584 mm) 0.016 in. (0.406 mm) Table 3.2 Results

Vertical Strain on Subgrade, in/in x 10'6 Section A Section B


Section C

___________ ________________ ___________ ________________

-957 -437
-253

3-108

Group No.: 5 Number o f Layers: 3 Material Properties: Layer Number Poissons Ratio Modulus o f Elasticity, psi (MPa')

1 2 3 Load Data: Number of Loads: 1 Load: 9,000 lb. (40 kN)

.35 .35 .45

500,000 (3450) 500,000 (3450) 7,500 (52)

Contact Pressure: 80 psi (552 kPa) Results: Your Results Vertical Defl. @ Surface, in. Section A Section B Section C Horizontal Strain in AC, in/in x 10-6 Section A Section B Section C Vertical Strain on Subgrade, in/in x 10*6 Section A Section B Section C -514 -229 -177 Reps, to failure 196 91 71 0.021 in. (0.533 mm) 0.014 in. (0.356 mm) 0.012 in. (0.305 mm) Table 3.2 Results

3-109

Group No.: 6 Number o f Layers: 3 Material Properties: Laver Number Poissons Ratio Modulus o f Elasticity, psi (MPa')

1 2 3 Load Data: Number of Loads: 1 Load: 9,000 lb. (40 kN)

.35 .40 .45

200,000 (1380) 25,000 (172) 7,500 (52)

Contact Pressure: 80 psi (552 kPa) Results: Your Results Vertical Defl. @ Surface, in. Section A Section B Section C Horizontal Strain in AC, in/in x 10*6 Section A Section B Section C Vertical Strain on Subgrade, in/in x 10-6 Section A Section B Section C -2,580 -1,030 -608 Reps, to failure 482 433 257 0.053 in. (1.346 mm) 0.033 in. (0.838 mm) 0.024 in. (0.610 mm) Table 3.2 Results

3-110

SECTION 3.0
REFERENCES

Hveem, F.N. and Carmany, R.M., "The Factors Un derlying a Rational Design of Pavements," Proceedings. Highway Research Board, 1948. Yoder, E.J., Principles of Pavement Design. 1st Ed., John Wiley & Sons, Inc., New York, 1959. Boussinesq, V.J., "Application des Potentiels a l'etude de l'equilibre, et du mouvement des solides elastiques avec notes sur divers points de physique mathematique et d'analyse," Paris, 1885. (Gauthier-Villars) Burmister, D.M., "The Theory of Stresses and Dis placements in Layered Systems and Applica tions to the Design of Airport Runways," Proceedings. Highway Research Board, Vol. 23, 1943. Yoder, E.J. and Witczak, M.W., Principles of Pave ment Design. 2nd Ed., John Wiley & Sons, New York, 1975. Peattie, K.R., "Stress and Strain Factors for ThreeLayer Elastic Systems," Highway Research Board Bulletin 342. Highway Research Board, 1962. Jones, A., "Tables of Stresses in Three-Layer Elastic Systems," Highway Research Board Bulletin 342. Highway Research Board, 1962. Finn, F.N., et al., "The Use of Distress Prediction Subsystems for the Design of Pavement Structures," Proceedings. 4th International Conference on the Structural Design of Asphalt Pavements, University of Michigan, Ann Arbor, 1977. The Asphalt Institute, Thickness Design Asphalt Pavements for Highways and Streets. Man ual Series No. 1 (MS-1), College Park, Maryland, September 1981.
3-111

3.10

Classen, A.I.M. and Ditsmarsch, R., "Pavement Evaluation and Overlay Design The Shell Method," Proceedings. 4th International Conference on the Structural Design of As phalt Pavements, University of Michigan, Ann Arbor, 1977. Highway Research Board, "The AASHO Road Test, Report 5, Pavement Research," Special Re port 6 IE, Highway Research Board, Na tional Academy of Sciences, Washington, D.C., 1962. The Asphalt Institute, Asphalt Overlays for Highway and Street Rehabilitation. Manual Series No. 17 (MS-17), The Asphalt Institute, College Park, Maryland, June 1983. Mahoney, J.P., Lee, S.W., Jackson, N.C., and New comb, D.E., "Mechanistic Based Overlay Design Procedure for Washington State Flexible Pavements," Research Report WA RD 170.1, Washington State Department of Transportation, Olympia, Washington, 1989. Monismith, C.L., and Epps, J.A., "Asphalt Mixture Behavior in Repeated Flexure," Institute of Transportation and Traffic Engineering, Uni versity of California, Berkeley, 1969. Claessen, A.I.M., Edwards, J.M., Sommer, P., and Uge, P., "Asphalt Pavement Design The Shell Method," Proceedings. Fourth Interna tional Conference on the Structural Design of Asphalt Pavements, University of Michi gan, Ann Arbor, 1977. AASHTO, AASHTO Guide for Design of Pavement Structures. American Association of State Highway and Transportation Officials, Washington, D C., 1986. Shook, J.F., Finn, F.N., Witczak, M.W., and Mon ismith, C.L., "Thickness Design of Asphalt Pavements The Asphalt Institute Method," Proceedings. Fifth International Conference on the Structural Design of As phalt Pavements, The Delft University of Technology, The Netherlands, 1982. Ullidtz, P., "Pavement Analysis", Developments in Civil Engineering, Elsevier, 1987.

3.11

3.12

3.13

3.14

3.15

3 .16

3.17

3.18

3-112

de Beer, M., "Developments in the Failure Criteria of the South African Mechanistic Design Procedure for Asphalt Pavements", 7th In ternational Conference on Asphalt Pave ments, Nottingham, U.K., 1992. Valkering, C.P. and Stapel, F.D.R., "The Shell Pavement Design Method on a Personal Computer", 7th International Conference on Asphalt Pavements, Nottingham, U.K., 1992. Portland Cement Association, "Thickness Design for Concrete Highway and street Pavements", PCA, Skokie, IL Ioannides, A. M., Thompson, M. R., and Barenberg, E. J., "Westergaard Solutions Reconsid ered," Transportation Research Record 1043, Transportation Research Board, 1985, pp. 13-22. Foxworthy, P. T. and Darter, M. I., "Preliminary Concepts for FWD Testing and Evaluation of Rigid Airfield Pavements," Transportation Research Record 1070, Transportation Research Board, 1986, pp. 77-88. Ioannides, A. M., Thompson, M. R., and Barenberg, E. J., "Finite Element Analysis of Slabs on Grade Using a Variety of Support Models," Proceedings, 3rd International Conference on Concrete Pavement Design and Rehabili tation, Purdue University, West Lafayette, Indiana, 1985, pp. 309-324.

Majidzadeh, K., lives, G.J., and Sklyut, H., "RISC A Mechanistic Method of Rigid Pavement Design," Proceedings, 3rd International Conference on Concrete Pavement Design and Rehabilitation, Purdue University, West Lafayette, Indiana, 1985, pp. 325-339.

3-113

Finn, F., "Factors Involved in the Design of Asphaltic Pavement Surfaces", NCHRP #39, Highway Research Board, 1967. "Calibrated Mechanistic Structural Analysis Proce dures of Pavements", Phase II, Volume 1, Final Report, NCHRP 1-26, University of IllinoisUrbana-Champaign; The Asphalt Institute; Construction Technical Laborato ries, December 1992. Majizadeh, K., "A Mechanistic Approach to Rigid Pavement Design", Chapter Two of "Concrete Pavements", edited by A.F. Stock, 1988.

3-114

SECTION 4.0 N O N DESTR U CTIV E TESTING DEVICES

4.1 INTRODUCTION 4.1.1 Types of Data Collected The following data are generally collected for pavement evaluation and monitoring purposes. (a) (b) (c) (d) Roughness (ride) Surface distress Structural evaluation (surface deflection) Skid resistance

Period of 1940s and 1950s During this period highway maintenance personnel relied heavily on visual inspections to establish type, extent, and severity of distress, and on experience or judgment to establish maintenance programs. Unfortunately, experi ence is difficult to transfer from one person to another, and individual decisions made from similar data are often inconsistent. Period of late 1950s and early 1960s During this period the increased use of roughness meters and deflection and skid test equipment permitted objective data to be collected and used both alone and with visual distress surveys to aid in making maintenance and rehabili tation decisions. Period of 1970s and early 1980s During this period highway personnel could no longer rely on the luxury of managing roadways solely on the basis of field personnel experience. Because of limited resources,
4-1

it was essential to develop rapid, objective means to establish: (a) (b) Projects in need of maintenance or rehabilitation. Types of maintenance or rehabilitation techniques currently required. Types and schedule of maintenance or rehabilitation to be undertaken in the future to minimize life-cycle costs (construction, maintenance, and user costs) or to maximize the net benefit.

(c)

Present At present, three specific applications for pavement con dition data can be identified. (a) Establish priorities Condition data such as ride, distress, skid and deflection are used to identify the projects most in need of maintenance and rehabilitation. Often only ride and/or distress data are used; at other times ride, distress, and deflection data are combined into a single rating. Skid resistance data are often used separately. Once identified, the projects in the worst condition (lowest rating) will be more closely evaluated to determine repair strategies. (b) Establish maintenance and rehabilitation strategies Data from visual distress surveys are used to de velop an action plan on a year-to-year basis; i.e., which strategy (repairs, surface treatments, overlays, recycling, etc.) is most appropriate for a given pavement condition. (c) Predict pavement performance Data, such as ride, skid resistance, distress, or a combined rating, are projected into the future to assist in preparing long-range budgets or to estimate
4-2

the condition of the pavements in a network given a fixed budget.


4.1.2 Benefits

(a) (b) (c)

Allocation of maintenance and rehabilitation funds. Determination of structural adequacy. Indication of highway network pavement condition and performance (city, county, state). Measure of "year-to-year" differences in pavement condition and performance. Overview of current practices.

(d)

(e)

4.2 SURFACE DEFLECTION MEASUREMENTS (NDT FOR STRUCTURAL EVALUATION) 4.2.1 Deflection Measurement Uses Surface deflections are a primary indicator of pavement structural response to applied loads. As such they are typically used for: pavement evaluation overlay or rehabilitation design load restrictions (both seasonal and overall) overload permit procedures pavement management applications evaluation of anomalies (the most common being void detection in PCC pavements)

The following general points can be made about surface deflections: (a) A tolerable level of deflection is a function of traffic (type and volume) and the pavement structural sec tion.

4-3

(b)

Overlaying a pavement will reduce its deflection. The thickness needed to reduce the deflection to a tolerable level can be established. The deflections exhibited by a pavement varies throughout the year depending on the type of pave ment (rigid or flexible), effects of temperature and moisture changes (including frost and thaw effects). For a flexible pavement structure, the magnitude of surface deflections increases with an increase in the temperature of the bituminous surfacing material (due to decreasing stiffness of bituminous binder with increasing temperature).

(c)

4.2.2 Categories of Nondestructive Testing Equipment (a) (b) (c) Static or slow moving load deflection devices Steady state deflection (vibratory) devices Impact load deflection devices (FWD) The usage of static, steady state, and impact NDT equipment by the 50 state highway agencies (SHAs) in the U.S. is shown in Table 4.1.
Table 4.1. SHA Deflection Equipm ent [after Ref. 4.41

Device Benkelman Beam Dynaflect Road Rater FWD

Number of SHAs Using Device for Various Time Periods mid-1990s 1990 mid-1980s (estimated) 18 18 5 5 3 11 4 30 4 11 5 31

4-4

4.2.3

Typical NDT Patterns

This will vary with intended application of the data and authority for whom data is being collected. For the most part, the current approach is to test primarily in wheel paths since the pavement response at these locations reflect the effect of damage that has been accumulated. In some cases, specific undamaged locations are tested for calibrating the damage equations using historical traffic data. Testing at the project level often involves deflection measurements in all traffic lanes, although it is usual to use higher test densities in the more heavily trafficked lanes (usually the outer lanes). Tests are usually uni formly spaced at 15m (50') to 60m (200') intervals, depending on project length and the expected uniformity of the section. Test locations are often staggered between lanes to provide improved statistical coverage. A mini mum of 7 to 8 test locations per uniform section is desir able for statistical purposes. Multiple load level applica tion at each location allows evaluation of non-linear material response. At the network level test spacing may be on the order of 150 m (500') to 450 m (1,500') in one lane only; but again some thought should be given to sec tion uniformity and statistical coverage. For research applications, test spacing may be as low as 1.5 m (5 ft.) and more load applications may be used. SHRP uses 50 ft. spacing on its LTPP section, and applies 23 loads of which 16 are recorded. For jointed PCC pavements, test spacing may be similar, but each location (slab) may also involve joint and comer testing. On highway pavements, if only one side of a transverse joint is tested the load should be located on the leave slab since this is the side where loss of support is likely to occur. Typical production test rates using an FWD or HWD is about 30 locations an hour, if 3 or 4 test loads are applied at each location. Research oriented testing approaches should be based on the research objec tives. Generally speaking, acquisition of deflection data using an FWD is relatively inexpensive and it is probably better to perform more than the minimum number of tests once the NDT equipment has been mobilized than to have to return for additional data.

4-5

4.3 STATIC OR SLOW MOVING DEFLECTION EQUIPMENT Static or slow moving deflection measuring equipment includes plate bearing tests as well as the Benkelman Beam. This equipment provides deflection measurements at one point under a non-moving or slow-moving load. Plate bearing tests are too time consuming and labor intensive for use in modern NDT testing but are covered here for sake of completeness. 4.3.1 Benkelman Beam (a) Most widely used device (developed at WASHO Road Test 1952). However, its use has declined in recent years in technologically advanced areas. Operates on a lever arm principle (refer to Figure 4.1). Must be used with a loaded truck or aircraft. Truck weight normally used is 80 kN (18,000 lb.) on a single axle with dual tires inflated to 0.48 to 0.55 MPa (70 to 80 psi). Measurements made by placing tip of beam between dual tires and measuring deflection as the vehicle (truck) is moved away. Measurements made with a dial gage. Standard Test Methods (i) Asphalt Institute procedure [4.3] requires placement of tip of beam between dual tires even with the centerline of the rear axle prior to movement of the vehicle. AASHTO T 256-77 (Pavement Deflection Measurements) requires that the tip of the beam be placed between the dual tires 1.4 m (4.5 ft) forward of the rear axle prior to movement of the vehicle.

(b)

(c)

(d)

(e)

(ii)

4-6

s u r ^ . , . s .

,Ch" B r ic c o

"'p"'nnorB enk^ B ean


4-7

(iii) ASTM D4695-87 (Standard Guide for General Pavement Deflection Measurements) recom mends that the standard load for Benkelman Beam measurements be 80 kN ( 18,000 lb.) on a single axle with dual 279 x 572 mm (11.00 x 22.5 in.) 12-ply tires inflated to 0.48 MPa (70 psi). Pavement deflection to be measured with a dial gage or LVDT to within 0.025 mm (0.001 in.). (e) WASHO measurement involved load moving to wards deflection beam. Manufactured by: Soiltest Inc. Materials Testing Division 2205 Lee Street Evanston, Illinois 60202 4.3.2 Plate Bearing Test (a) Standard test methods (i) AASHTO T222-81 (Nonrepetitive Static Plate Load for Soils and Flexible Pavement Compo nents, for Use in Evaluation and Design of Airport and Highway Pavements) ASTM D 1196-77 (same title as AASHTO T222)

(f)

(ii) (b) Uses (i) (ii)

To determine modulus of subgrade reaction (k value). Typical uses of static (or near static) pavement surface deflections.

(iii) Asphalt Institute: overlay design and/or determination of remaining life. (iv) California DOT: overlay design.

4-8

4.3.3 Automated Beams [4.3] Various approaches have been developed to automate Benkelman beam measurements, typically by mounting the deflection beams on the truck that provides the axle load. These move slowly (2 to 4 km/hr or approx. 1 to 2 mph) and measure deflections at 3.5 to 6 m (approx. 10 to 20 ft.) spacing in one or both wheel paths. The most com mon of these is the La Croix Deflectograph, manufactured in Switzerland. The British Transport and Road Research Laboratory (TRRL) uses the Pavement Deflection Data Logging (PDDL) machine which is a modified version of the La Croix. CalTrans used the California Traveling Deflectometer, but this was never commercially produced and is no longer in service. 4.3.4 Curvature Meters [4.3] This is a simple portable device consisting of a long bar supported at each end with a dial gauge in the middle. It is used to estimate the curvature of the pavement surface caused by an applied wheel load from a measure of the middle ordinate for a fixed chord length of 0.3 m (1 ft.). 4.3.5 Typical Applications (a) Structural adequacy and overlays to reduce deflec tion levels [4.3], Network level relative structural response. Deflection basin evaluation (some cases).

(b) (c)

4.3.6 Advantages/Disadvantages of Static or Slow Moving Load Deflection Equipment (a) Advantages (i) Widely used and hence numerous analysis pro cedures available to use with such data Simplicity (deflection beam, plate load)

(ii)

4-9

(iii) Instrument cost low (about $1,000 for Benkelman Beam) (iv) High coverage (automated beams) (V) Move with traffic (automated beams)

(vi) Realistic load levels possible Disadvantages (0 (ii) Slow, requires traffic control Labor intensive

(iii) Typically does not provide deflection basin (iv) Fixed reference necessary (V) Load duration may be unrealistic

(vi) Measurement may depend on technique (rebound vs. WASHO) (vii) High cost (automated beams) (viii) Repeatability is poor in comparison to more modern methods.

4.4 DYNAMIC VIBRATORY LOAD (STEADY STATE DEFLECTIONS) 4.4.1 General (a) Several types of steady state deflection equipment are available. Primarily these include: (i) (ii) Dynaflect (electro-mechanical) Road Rater (electro-hydraulic)

(iii) WES Heavy Vibrator


4-10

(iv)

FHWA Cox Van

(b)

Equipment induces a steady state (non changing) vibration to the pavement with a dynamic force gen erator. Pavement deflections measured with velocity trans ducers.

(c)

4.4.2 Dynaflect (a) Standard test methods (i) (ii) (b) AASHTO T256-77 ASTM D4695-87

Manufactured by: Geolog Inc. 103 Industrial Boulevard Granbury, Texas

(c) (d)

Mounted on a two wheel trailer. Dynaflect is stationary when measurements are taken. Force generator (counter rotating weights) started and deflection sensors (velocity transducers) lowered to the pavement surface. Refer to Figure 4.2 (plot of typical force output) and Figure 4.3 (location of Dynaflect loading wheels and five ve locity transducers). The peak-to-peak dynamic force is 4.4 kN (1,000 lb.) at a fixed frequency of 8 Hz. This load is applied through two 102 mm (4 inch) wide, 406 mm (16 inch) diameter rubbercoated steel wheels which are placed 508 mm (20 inches) apart.

4-11

FORCE

EXERTED

O N

PAVEMENT

TIME

Figure 4.2. Typical Force O utput o f Steady State D ynam ic Deflection Devices (4.2]

4-12

4 in.

4 in.

10 in.

12 in. 10 in.

12 in.

12 in.

12 in.

4 in. 4 in.
Loading W heel C ontact Area G eophone (D eflection S ensor)

Figure 4.3 - Standard Location of Dynaflect Loading W heels and G eophones [4.21

4-13

(e)

Disadvantages (i) (ii) Requires traffic control. Dynamic load significantly less than normal truck traffic.

(iii) Relatively large static preload (816 kg (1800 lb.)). (iv) Pavement resonance may affect measurements. (v) Relatively small load may not produce ade quate deflections on heavy pavements.

(f)

Advantages (i) (ii) High reliability (low maintenance) Can be used to obtain a deflection basin.

4.4.3 Road Rater (a) Standard test methods (i) (ii) (b) AASHTO T256-77 (for Model 400 only) ASTM D4695-87

Manufactured by: Foundation Mechanics, Inc. 421 East El Segundo Boulevard El Segundo, California 90245

(c)

Two production models available as of November 1991: (i) (ii) Model 400 B-l Model 2000 A-1

4-14

(d)

Force generator consists of a steel mass, hydraulic actuated vibrator. Driving frequencies range between 5 and 60 Hz. Load ranges for various models: (i) Model 400 B: 2.2 to 13.3 kN (500 to 3,000 lb.) Model 2000: lb.) 2.2 to 22.2 kN (500 to 5,000

(ii)

The loading footprints for the two models are shown in Figure 4.4 (due to differences, one must be CAREFUL in comparing data between the models). (e) Deflections are measured with four velocity trans ducers. Disadvantages (i) (ii) Requires traffic control Low load level relative to truck traffic (Models 400 B and 2000).

(f)

(iii) Relatively large static preload required [4.2] (iv) Pavement resonance may affect measurements. (g) Advantage (i) (ii) Can measure deflection basin. Widely used - performance history correlation data widely available

(iii) Reliable

4-15

R O A D R A T E R M O D E L 400B
3.5 in. 12 in. 12 in. 12 in. 12 in.

G eophone (Deflection Sensor)

ROAD R A TER M ODEL

2000

12

in.

12

in.

12

in.

12

in.

Loading W heel Contact Area f Geophone (Deflection Sensor)

Figure 4.4 - Standard Location o f Loading Plate(s) and G eophones for the Road R ater M odel 400B and M odel 2000 [4.21

4-16

4.4.4 WES Heavy Vibrator [4.2] This was developed by the Corps of Engineers for airfield pavement evaluation. A 71 kN (16 kip) preload is ap plied, with a peak-to-peak vibratory load of 130 kN (30 kip) possible at a frequency of 15 Hz. It is a large unit mounted in a semi-trailer and is not commercially available. (a) Advantages (i) (ii) (iii) (b) Load variable up to 30,000 lb. peak-to-peak. Load frequency variable from 5 to 100 Hz. Can be used on heavy pavements.

Disadvantages (i) (i) Mounted in 36' trailer. Not commercially available.

4.4.5 FHWA Cox Van (Thumper) [4.2] This is an experimental device developed for FHWA that can apply static, dynamic or intermittent pulse loading. The Thumper is a research oriented device designed to emulate the characteristics of other devices. Load magni tudes of up to 45 kN (10 kips) are possible at frequencies ranging from .1 to 110 Hz. Deflections are measured using 6 LVDTs spaced at 0, 300, 460, 600, 910, and 1200 mm (0, 12, 18, 24, 36, and 48 in.). (a) Advantages (i) (ii) Can emulate most deflection devices Can apply a variable load at multiple frequencies. Multi-frequency loading.

(iii)

4-17

(b)

Disadvantages (i) Not commercially available.

4.4.6 Typical Uses of Steady State Pavement Surface Deflections (a) Correlation with static deflections (Benkelman Beam). Estimation of layer elastic modulus values. Overlay design and/or determination of remaining life.

(b) (c)

4.5 IMPACT (IMPULSE) LOAD RESPONSE DEVICES 4.5.1 General (a) All impact load NDT devices deliver a transient im pulse load to the pavement surface. The subsequent pavement response (deflection) is measured. Standard test methods (i) ASTM 4694-87: Standard Test Method for Deflections with a Falling Weight Type Impulse Load Device Related test method ASTM D4695-87: Standard Guide for Gen eral Pavement Deflection Measurement (iii) Significant features of ASTM D4694 Falling Weight ("force-generating device") Force pulse will approximate a haversine or half-sine wave
4-18

(b)

(ii)

Peak force at least 11,000 lb. (50 kN) Force-pulse duration should be within range of 20 to 60 ms. Rise time in range of 10 to 30 ms. Loading plates Standard sizes are 12 in. (300 mm) and 18 in. (450 mm) Deflection sensors Can be seismometers, velocity trans ducers, or accelerometers. Used to measure the maximum vertical move ment of the pavement. Signal conditioning and recorder sys tem

Load measurements Accurate to at least 2 percent or 160 N ( 36 lb.), whichever is greater.

Deflection measurements Accurate to at least 2 percent or 2\xm ( 0.08 mils), whichever is greater. Recall that 0.00008 inch = 0.08 mils and 0.002 mm = 2\im.

Precision and bias Precision guide When a device is operated by a single operator in repetitive tests at the same location, the test results are questionable if the difference in
4-19

the measured center deflection (Do) between two consecutive tests at the same drop height (or force level) is greater than 5 per cent. For example, if Do = 0.254 mm (10 mils) then the next load must result in a Do range less than 0.241 mm to 0.267 mm (9.5 to 10.5 mils). (c) (d) (e) Measurements obtained very rapidly. Impact load easily varied. Pavement responses are measured with geophones or velocity transducers (Dynatest, Phonix, Foundation Mechanics) and seismometers or LVDT/accelerometer combination (KUAB). The primary impact deflection equipment currently marketed in the U.S. include: (i) Dynatest Dynatest Consulting Ojai, California U.S.A. KUAB KUAB Konsult and Utveckling AB Box 10 79500 Rittvik, Sweden KUAB America 1401 Regency Drive East Savoy, Illinois 61874 U.S.A.

(f)

(ii)

(iii) Foundation Mechanics, Inc. (Jils) 421 East El Segundo Boulevard El Segundo, California 90245 (iv) Resource International (Phonix) 281 Enterprise Drive Westerville, OH 43081

4-20

4.5.2 Dynatest Falling Weight Deflectometer (FWD)

(a)

Two models are primarily availableDynatest Model 8000 (FWD) and Model 8081 (HWD). Most widely used FWD in U.S. (as of 1993) FWD is trailer mounted. By use of different drop weights and heights can vary the impulse load to the pavement structure from about 6.7 to 120 kN (1,500 to 27,000 lb.). The weights are dropped onto a rubber buffer sys tem resulting in a load pulse of 0.025 to 0.030 sec onds (refer to Figure 4.5). The standard load plate has an 300 mm (11.8 in.) diameter. A heavy weight version (HWD) with a load range of about 20 to 240 kN (6,000 to 54,000 lb.) is available. Typical location of the loading plate and seven ve locity transducers is shown in Figure 4.6. Ideally, transducers should be located to ensure that the positions are reasonable relative to the pavement structure being tested. The WSDOT sensor spacings with the 300 mm (11.8 in.) load plate are:
mm inches

(b) (c) (d)

(e)

0 203 305 610 914 1,219

0 8 12 24 36 48

4-21

TIM E (Tim e from A to B is Variable, Depending on Drop Height)

A - Time at which weights are released B - Time at which weight package make first contact load plate C - Peak load reached

Figure 4.5 - Typical Force O utput of Falling W eight Deflectom eter

4-22

H ------------------

12 in.

Loading Wheel Contact Area # Sensor

Figure 4.6 - Typical Location o f Loading Plate and Deflections Sensors for Falling W eight Deflectometers

4-23

The Strategic Highway Research Program (SHRP) sensor spacings with the 300 mm (11.8 in.) load plate are:
mm inches

0 203 305 457 610 914 1,524

0 8 12 18 24 36 60

Texas sensor spacings with the 300 mm (11.8 in.) plate are:

mm

inches

0 305 610 914 1219 1524 1829

0 12 24 36 48 60 72

(f)

SHRP FWD's T4.10. 4.111: Dynatest Model 8000E (i) (ii) Loading plate: 300 mm (11.8 in.) Loads (flexible pavements): Drops result in loads of approximately 27 kN (6,000 lb.), 40 kN (9,000 lb.), 53 kN (12,000 lb.), and 71 kN (16,000 lb.).

(iii) Maximum deflections recorded at each sensor for all four drops. A complete time load and time deflection "history" is recorded for the last drop at each of the four load levels.

4-24

(iv) SHRP Regional LTPP contractors use two computer programs to check FWD data [from Ref 4.101: FWDSCAN: checks FWD data files for completeness and readability. FWDCHECK: checks for section uni formity based on subgrade and pavement strength.

4.5.3 KUAB Falling Weight Deflectometer [4.5. 4.6. 4.9] (a) Model 50 Load range: 7 to 65 kN (1,500 to 15,000 lbs.) (b) Model 150 Load range: 14 to 150 kN (3,000 to 34,000 lbs.) (c) Total of five models are available (as of 1991). The heaviest load model has a range of 14 to 290 kN (3,000 to 66,000 lbs ). KUAB models are completely enclosed for protec tion during towing. The impulse force is the result of a unique two-mass system. The deflection sensors are called seismometers and use LVDTs along with a mass-spring reference system (standard KUAB FWD equipped with seven deflection sensors). Each sensor has micrometer making static field calibra tions possible. The seismometers have three ranges: low (0-50 mils), medium (0-100 mils), and high (0200 mils). Worldwide distribution of the KUAB FWD began in 1976 and subsequently over 60 units have been sold (as of 1991) [after Ref. 4.9~|. The basic weight of a KUAB FWD and associated trailer is about 1,800 kg (4,000 lb.). A load of about
4-25

(d)

(e)

(f)

320 kg (700 lb.) is applied to the pavement by the loading plate prior to testing [4.9], (g) An original feature of the KUAB FWD is the seg mented load plate (four quarter - circle segments). This provides a more uniform pressure distribution to the pavement surface [4.9], Accuracy and precision of KUAB 50 (based on KUAB product literature mostly from Ref. 4.9) (i) Accuracy Deflection sensors: 2 ^m (0.08 mils), 2 percent Load cell: 20 kg ( 44 lb.), 2 percent

(h)

(ii)

Precision Deflection sensors: 1 ^m ( 0.04 mils), 1 percent Load cell: 10 kg ( 22 lb.), 1 percent

(iii) Range Deflection sensors range 0 -5 .0 8 mm (0 - 200 mils)

4.5.4 Foundation Mechanics Falling Weight Deflectorneter [4J2] (a) (b) Current model is the JILS-20-FWD. As of November 1991, three of these models have been in service for two years. Uses seven velocity transducers to measure the deflection basin (location of sensors is variable). Load is measured with a transducer.
4-26

(c)

(d)

Load range: Approximately 7 to 107 kN (1,500 to 24,000 lbs.) JILS-FWD is trailer mounted (tandem axle trailer) and includes a 16 horsepower gasoline engine to provide all necessary hydraulic and electrical power for operation. The gross weight of the unit is about 1,300 kg (2,800 lbs.). The FWD unit can be en closed with an available cover.

(e)

4.5.5 PhonixFWD [45, 4T3] (a) (b) Current models are ML6, ML11 and ML25. As of 1992, it is estimated that there are three Phonix FWD's in the USA. Six geophones are used for deflection measure ments, with variable locations. Load is measured with a load cell. Load range is approximately 10 kN to 110 kN (2.2 24.7 kips) for the ML 11. The Phonix units are trailer-mounted and independ ent trailer power source is optional.

(c)

(d)

(e)

4.5.6 SASW Approaches Spectral analysis of surface waves (SASW) approaches evaluate pavements from Rayleigh wave measurements involving low strain levels. Until very recently, both data acquisition and analysis was cumbersome and time con suming. Recent equipment and software development under the SHRP-IDEA program appears to have made significant advances in the application of SASW tech niques [4.14], A brief description of the SASW proce dures has been included as Appendix E at the end of the participant workbook. SASW is of potential interest to the highway engineer for a number of reasons:

4-27

It can provide information to determine the ap proximate thickness of individual pavement layers without coring. It can provide a starting point for estimating modu li, and modular ratios of the pavement layers during the backcalculation process. It can detect rigid layers and provide an estimate of depth. It can more accurately determine & quantify modular values of thin ACP layers on the surface, as compared to the FWD.

SASW is a natural complement to the FWD. It cannot replace the FWD however, as it cannot predict moduli of paving layers under traffic loads, as most materials behave in a non-linear fashion. 4.5.7 Typical Uses of Impulse Pavement Surface Deflection (a) (b) Estimation of layer elastic moduli Overlay design and/or determination of remaining life Network level monitoring Correlation with static deflections

(c) (d)

4.5.8 Advantages and Disadvantages of Impulse Load Equipment (a) Advantages 0) (ii) Gaining worldwide use Best simulates actual wheel loads

(iii) Can measure deflection basin (iv) Relatively fast data acquisition 4-28

(v)

Only a small preload is placed upon the pave ment surface

(vi) No fixed reference required (c) Disadvantages (i) (ii) High initial cost Traffic control required

(iii) Relatively complex system

4.6 COMPARISONS AND CORRELATIONS BETWEEN FWD AND OTHER DEVICES 4.6.1 Introduction Numerous comparisons between deflection devices have been made and published to date, using various approaches and evaluation criteria [References 4,2 and 4^5 are examples]. It does seem, however, that the calibration center approach adopted by SHRP will become essentially an absolute reference standard by which the impulse load equipment will be evaluated. Four SHRP regional cali bration centers have been established, operated by State DOT personnel at each location. Some of the published comparison data has been included in this section. Corre lations between FWD deflections and those measured by other devices have also been included. It should be em
phasized that such correlations may be misleading and should be used with care. Generally speaking, a

correlation equation between different devices is techni cally valid only for the specific location for which it was developed at the time o f development, due to the effects of temperature, moisture, load-level, time-of-loading, material non-linearity, etc. From a practical standpoint, however, there may be situations where there are no alternatives but to rely on correlation equations.

4-29

4.6.2

Comparisons Between Devices

4.6.2.1

Comparisons Between KUAB 150 and Dynatest 8000 (KUAB Literature) a) Refer to Table 4.2 for a comparison of model specifications. Refer to Figure 4.7 for a comparison of the falling weight systems for the KUAB and Dynatest models.

b)

Table 4.2 - Equipm ent Specifications [after Ref. 4.61

KUAB
Load Range Load rise time Load duration Load generator Load plate 7-150 kN 28 ms 56 ms Two-mass system Segmented or nonseg mented with rubberized pads (300 & 450 mm diameter) Seismometer with static field calibration device 0-1.8 m 7 (all available positions) 5 mm (200 mils) 1 (im (0.04 mils) 2 nm 2% Unlimited, user selected 35 secs. HP 85B or IBM compatible, MS DOS

DYNATEST
7-125 kN Variable 25-30 ms One-mass system Rigid with rubberized pad or split plate, tilts 6" (300 & 450 mm diameter) Geophones with or without dynamic cali bration device 0-2.25 m Up to nine 2 mm (80 mils) or 2.5 mm (100 mils) Same Same < 64 drops 25 secs. Same

PH NIX
10-250 kN 12-15 ms 25-30 ms One-mass system Segmented with rubber pads (300 mm)

JILS
8-107 kN selectable selectable One-mass system Rigid with rubberized pad (300 & 400 mm diameter) Geophones

Deflection sensors

Geophones

Deflection sensor positions Number o f sensors Deflection sensor range Deflection resolution Relative accuracy Test sequence Test time sequence (4 loads) Computer

0-2.5 m 6 to 9 2 mm (80 mils) 1 micron 2 microns 2% User selected 20 sec. (per drop) IBM compatible

0-2.4 m 7 2 mm (80 mils) 1 micron 2 microns 2% User selected 30 secs. IBM compatible

4-30

Figure 4 .7

Sketches of Dynatest and KUAB Falling Mass Systems [from Ref. 4 .6 )

4-31

4.6 .2.2

Selected comparisons for seven NDT devices [after Ref. 4.5] The U.S. Army Corps of Engineers, Waterways Experiment Station (WES) in Vicksburg, Mississippi, evaluated the following NDT devices: KUAB Model 50 FWD Dynatest 8000 FWD Dynatest HWD (Heavy Weight Deflectometer) Phonix FWD Dynaflect Road Rater 2008 WES 16 kip vibrator

The basic equipment characteristics are shown in Table 4.3 Selected results from this evaluation include: (i) Measured deflections on a short-term repeatability experiment based on 25 tests (measurements) on AC and PCC surfaced pavements: Figure 4.8. Comparison of output of a "standard" load cell and each device in terms of absolute sum of percent difference: Figure 4.9.

(ii)

(iii) Plots of typical load pulses for the FWD's evaluated: Figure 4.10.

4-32

Tabic 4 .3

NDT Device Characteristics (WES Evaluation) [from R ef.45|

D ev ice N am e K uab FW D

D ynam ic F orce R i n g t . IbP 3000 to >5 000

L oad T ransm itted

N um ber an d T ype of D eflection

by
S cctior.jlucd circular plate 11 8 i n d n C ircular plate 1 1 8 or 17.7 a. dia T w o 16-in dia by 2 in width urethanecoated steel wheels C ircular plate II 8 or 17.7 in. dia C ircular plate 18 in. dia C ircular plate 18 in. dia C ircular plate 1 1 8 in. d ii

Senvon
7 te is m o m e te n

D e f le c tio n S e n so r S p a c in j F ix e d at 0 .8 .1 2 .1 8 . ' 2 4 ,3 6 .4 8 in. V a r ia b le , 12 to 96

D y n a te s t H W D

10 000 to 55 000

7 le o p h o n e i

in.
5 je o p h o n e V a r ia b le , 0 to 48

D y n a fle c t

1000 peak to p eak

in.

D y n a le s t F W D

1500 to 27 000

7 je o p h o n e

V a r ia b le , 12 to 96

in. 4 je o p h o n e *
5 je o p h o n e 6 je o p h o n e t V a r ia b le , 24 to 48

R o a d R a te r 2006

W E S 16-Kip
PhonLx F W D

500 to 7000 p eak to p eak 500 to 30 000 p e a k to peak 2300 to 23 000

io.
V a n a b l e . 12 to 60

in.
V a r ia b le . 8.3 lo 58 in

1 Ib f - 4 448 N. * 1 in . 2.54 cm .

4-33

SHORT TERM REPEATABILITY

c o e f f ic i e n t

of v a r i a t i o n

DEVICE

ESI 0 0

E 2

Di

CZZ 02

E S

03

C Z 3 04

C~7

DJ

I----- 1 DO

SHORT TERM REPEATABILITY

c o e f f ic i e n t

or

v a ria tio n

DO

E2

DI

EZ3 02

E3

DEVICE 03

E2

04

EH2

05

I os

Figure 4 .8

Coefficient of Variation of Deflections from Short-term Repeatability Experiment (WES) [fromRef. 4 .5 ]

4-34

LOAD R E CO RD IN G A C C U R A C Y
A B S O L U T E SU M OF PERCEhfT D I F F E R E N C E

PERCENT

DEVICE

k i

sk

c v a 7*

rm

iok

ijk

t\ / i

20/21

r7~i

ok

Figure 4 .9

Average Absolute Sura of Percent Difference from Load Measurement Accuracy Experiment (WES) [from Ref. 4 . 5 ]

4-35

K U A B

F W D

D Y K A T E S T

H W D

7lMt. m*f c

D Y N A T E S T

F W D

P H O N IX

F W D

TIKE.

TIME, m sec

Figure 4.10 - Typical FWD Load Pulse Plots (WES) [from Ref. 4.51

4-36

Measured loading times for the load cell (standard) experiment performed by WES: Device Loading Time

(ms) KUABFWD Dynatest FWD Dynatest HWD Phonix FWD 79.8 30.4 28.1 40.7

Loading times are important in explaining differ ences in backcalculated moduli, particularly for visco-elastic materials such as asphalt concrete. Shorter loading times result in higher backcalcu lated moduli for AC (as evidenced by the Asphalt Institute equation (Equation 2.4)). This would help explain observed differences in deflections obtained with the various FWD types available. 4.6 .2.3 More comparisons between deflection measuring equipment [after Ref. 4.7] (a) Operating characteristics of nondestructive equipment (refer to Table 4.4) First costs and operating costs of nonde structive equipment (refer to Table 4.5)

(b)

4.6.3 Correlations between deflection measuring equipment 4.6.3.1 Introduction In general, correlations between deflection de vices should be used with caution. Too often, a correlation is developed for a specific set of conditions that may not exist at the time the cor relation equation is used. It appears that the best approach is to obtain pavement parameters (such as layer moduli) from the specific NDT device being used. However, that said, a few of many such correlations that have been developed fol low.

4-37

Table 4.4 - O p e ra tin g Ch ar a c t e r i s t i c s of N o n d e s t r u c t i v e Eq u i p m e n t [4 . 7 ] . (Note: This I n f o r m at io n wa s P u b l i s h e d in 1986.)

Ajmey Benkelman Beam kaho fi 1not* 1 Louisiana VIchlgan MIssourl He Jersey Ne York O lahoma v Of fon South Carolina Teas Summary

Data Point per Day Average Range 40 >00 so so 400 10 1(0 ISO >00 10 so us >0 - 0 Ito - >40

Lant M of iles Pavement per D ay Range Average 40 10 s 4 1 1 >0 % 1 10 41 S O 1 49 10 .0 IS 41 >4 10 IS 49 4 *0 IS ss s 40 70 10 10 ss so 11 . 75 S S IS > - to 0 S - 1

Data Points per Lana M ila 1 >0 s 10 100 so ISI s S3 10 St SS ) s I 1 % 1 3 3 > 1 s 1 30 1 >6 11 SS 10 9 IS 11 11 % 3 2(0 1 10 s

Equipment Utilization (Days per Year) Average Range S t 10 40 10 10 30 no ISO IS 10 37 100 100 ISO no 9S 110 no 30 100 170 no iso no 95 100 110 13 no 70 40 4S 1(0 7S 71 ISO >0 100 130 too to 1 - >0 T - 11 J>0 10 US 10 IS S O 150 I7S >0

>0 1)0 )>S ISO 40 40 >0 -

so >IS 17S >00 100 10 >00

I.7S >0 1t I.7S I.7S -

1.7S 40 10 10 1.7S 10

1 us
to - 140 100 - >00 to - 100 to - >00 IS to 0it 100 40 140 360 no >00

Dyna fleet - Network Maniement System Arkansas ISO - >so 17S kJiho to to - no Hebruka 10 to - 100 Oregon s South Dakota 41 40 - SS Utah 4S0 >S0 - SS0 Summary 140 40 - SS0 Dynafleet * Project Aritona Arkansas California Kansas Nebraska Nevada South Dakota Teas Utah Virginia Summary Road Rater Illinois Kentucky Louisiana Maryland Pennsylvania Summary Management System 4S >s - ss 17S ISO - >so >S2 47 - 420 no to - ISO 10 10 - 100 4S0 41 40 - S S >10 170 - >S0 4S0 >50 - SS0 >71 S7 - 2 >00 > - SO S S 17S 400 >10 >00 no 0 ISO - >00 ISO - 300 ISO >00

35 - 10 >0 - ts to - 100 40 - S S 70 - 110 > - 110 0 10 35 110 to >0 10 >0 30 100

40 - S S 3- S 70 - 110 1 - >0 1 * no >7

90 - 100 to - no 4S - 14 IS - >00 S - 10S S to - >00 33 - >00

I - IS S - 70

Fa1Hnj-We!|hl Deflectometer Alaska ISO Arliona 3S Florida 410 Tennessee ISO Washington ISO Sum mary 110

100 >50 100 - >50

S 40 O 10 - > 0 It - 40

4-38

Table 4.5. First Costs and Operating Costs of Nondestructive Equipment [4.71 (Note: This information was Published in 1986)
O p e r a t i n g C o i l t (t) P a r C a l a Poi nt Agmey Brnk tlrr i n B e t m l d*ho Dl moU Loulllina V i t i our l H e York Oklihoma Oregon South Carolina T titi Summary 10.00 I.SS 4 . 17 1.50
1.00 >.31 i.oo >.1S a . io >.00

Per Lane Mile of Pa v e me nt Avenge Range

A v e r ag e

Range

Operatlnf Personnel Required

Pu r c h a s e Pr i ce
(S)

10. 00 >1.00 ISO. 00 i.eo 1.10 1.00

1.00

IS.00

> 1 T
1

>. 1 0 4.00 >.00 >. 0 0

1.00 1.00 -

4.>o
.00

11. 10 10. 00 100.00 I S . 00

>.00 1.00 II.IS 11.30 4 . 0 0 - >20. 00

1.00 1.00

I S . 00 - y s . o o
>.00 - >70.00

9 1 >00 1,000

Dynafleet * Network M i n t j e m e n t Syilem Ark i a u i kuho Oregon South Dakota Utah Summary 1.14 n.oo 110.00 1. 1. >.11
1.30 >.50 .SI

>.IS -

.>0

1 0 . 0 0 - >0.00
.00 .so

110. 00 .IS
l.ts

0.00 - 130.00

9 I
s

1.1 -

1.1

1.00 1.1 -

0 .so 1. 3

9
>

1 . 1 - >0 . 00

4. 14

1. 1 - I SO. 00

> 0 , 0 0 0 - >3,000

Dynafleel - Project Management Syitem Ar i z o n a Arkanm California K inid Hevad a South Dakota Ten* Utah Vi r gi nia Summary Ro a d R t t e r loa Mirylind Pen nsy lv an ia Summary 1. >.71 . 00 .II
11.11

>0.00 1.14 1.1 1 S .00 . 71 IS 1.00


1.1

1.50 .IS 10.00 .SI .00

- .0 - .so - 11.00 - 1.17 - 1.50


i.40 l.)S 1.00

1.00 .)

e.io l.ts .10 -

10. 00 . SI II.IS TS.00 . I.1S 0.00 1.1S 17.30


>1.11

M l 1.10 11. 13 - >47. 00 17. 41 .00 >0.00 1. 13 >4. 11 .30


11.00

9 9 >
1

1 . 13
13.00

10.00 -

9 9 9 9 I 9 >0 , 0 0 0 - >3,000

. 7 1 - 1 1 . 00

1. 1 - X 7 . 0 0

4 43. 00 11. 11 (3.00 11.00


9 9 9

1.10 1.10 -

>. 00 1.00

S3.11 11.00 4.07

> 3 , 0 0 0 - >3,000

F a l l l n g - We t g h t D e f l e t t o m e t e r A l t ska Arizona Tmnene W j h l n g l o n Summary

t.oo
>s.oo

t.oo
>.S0 >.>4 >.00 .10 TS .00 11.00 4.00
>7.SO

>.00 >.7S .4)

10. 00 - 14. 00 >1.00 - 111. 00


T.oo i i i .o o

9 9 9 9 9 90,000 - 111,000

1 . 0 0 - > S .00

4-39

4.6 .3.2

Benkelman Beam to Falling Weight Deflectometer (based on unpublished data collected by WSDOT Materials Laboratory in 1982-1983) BB 1.33269 + 0.93748 (FWD)
0.86
(Eq. 4.1)

Std Error Sample Size where BB FWD

= 3.20 mils = 713 matched deflection points = Benkelman Beam _3 deflection (in. x 1 0 ), = FWp deflection (in. x 1 0 ' ) corrected to a 9,000 lb. load applied on a 1 1 . 8 inch diameter plate

4.6 3.3

Benkelman Beam to Dynaflect (a) Arizona [after Reference 4.2] BB where BB = 22.5 (DMD) = Benkelman Beam _3 deflection (in. x 1 0 ), = Dynaflect Maximum Deflection (in. x 10'3).

(Eq. 4.2)

DMD

4-40

(b)

Asphalt Institute [after Reference 4.2] BB = 22.30 (D) - 2.73 = Benkelman Beam 3 deflection (in. x 1 0 ), = Dynaflect center deflec tion (in. x 1 0 " ), same as DMD (Arizona).
(Eq. 4.3)

where

BB D

(c)

Louisiana [after Reference 4.2] BB R2 = 20.63(D) = 0.72 = Benkelman Beam .3 deflection (in. x 1 0 ), = Dynaflect^deflection (in. x 1 0 ), same as DMD (Arizona).
(Eq. 4.4)

where

BB D

4.6 .3.4

Benkelman Beam to Road Rater [from Reference 4.8] (a) Stabilized pavements: for Benkelman Beam load at 9,000 pounds on dual tires with 7080 psi inflated tires and Road Rater at 8,000 pound peak-to-peak load at 15 Hz on a 1 2 inch diameter plate BB R2 where BB = 2.57+ 1.27 RR = 0.66 = Benkelman Beam .3 deflection (in x 1 0 ),
(Eq. 4.5)

4-41

RR

= Road Rater (Model 2008) deflection at 8 , 0 0 0 pounds and 15 Hz (in x 10).

(b)

Asphalt Institute [4.3]

Recommends that correlation between Benkelman Beam and Dynaflect be used to correlate Benkelman Beam to Road Rater Model 400 (with caution). (c) Western Direct Federal Division, Federal Highway Administration, Vancouver, Washington

Correlation for Benkelman Beam to Road Rater Model 400 BB where BB = 8.0+ 9.1026 (D0) = Benkelman Beam deflection (in. x 1 0 "^) = Maximum deflection from Road Rater Model 400 (deflection location between load pads) at a load of 1,300 pounds at 25 Hz.

(Eq. 4.6)

Do

4.6 .3.5

Falling Weight Deflectometer to Road Rater [from Reference 4.8] For Road Rater at 8,000 pound peak-to-peak load at 15 Hz on a 12 inch diameter plate and FWD at 8,000 pounds (+ 5%) on a 12 inch di ameter plate

4-42

-3.40+ 1.21 RR 0.94 Std Error = 3.23 mils

(Eq. 4.7)

n where Do

95 Maximum FWD de flection middle of loading plate at 8 , 0 0 0 lb. load on 1 2 in. diameterj)late (in x 1 0 " ). Road Rater (Model 2008) deflection at 8 , 0 0 0 pounds and 15 Hz (in x 10" ).

RR

This equation is shown in Figure 4.11. 4.6 .3.6 Use of Correlation Equations Generally, the use of correlation equations should be avoided when possible, and treated with a high degree of caution when it is necessary to use them. Because the correlation model is a func tion of pavement type, time of testing, material properties, and a whole host of other variables under which it was developed, it is impossible to estimate it's accuracy on any given project. Sev eral suggestions are offered here: 1. The decision to use correlation equations should be based on the sensitivity of the bottom line of the analysis to errors in the model. For example, if the objective is to design an overlay, then the sensitivity of the overlay thickness to errors in the correlation equation should be used as a basis to decide if it may be used safely.

4-43

Foiling

Weight

Deflectom*?er

O * f le c ! lo n ,m ll *

Rood

R o K r Dcfltclion,mil*

Figure 4 .1 1 . Falling Weight Deflectometer versus Road Rater Deflections 1 4 .8 ]

4-44

2. Correlation equations are generally docu mented in the literature. Track down the source and determine the r^ , or correlation coefficient for the equation, and the condi tions under which it was derived (pavement type, season, type of equipment, etc.). Compare the conditions under which it was developed to those in which it will be used. 3. Cross check pavement or overlay designs with results using other procedures, (e.g. AASHTO) until confident that use of the correlation equation provides reasonable results. 4. Formulate plans to eliminate the need for correlation equations. Generally, if you have to use them, either your data collection or data analysis methods are outdated.

4.7 CALIBRATION OF LOAD CELL AND DEFLECTION SENSORS The periodic calibration of measuring devices (i.e., load cell & deflection sensors) on NDT equipment is essential in the acquisition of meaningful pavement deflection data. In spite of its importance, little has been written on the calibration procedures. Until recently, no calibration procedures existed except for established manufacturer procedures. SHRP has developed reference calibration procedures which will be published pending further verification and adjustment. SHRP's calibration process consists of three stages: 1) ref erence calibration of the load cell, 2 ) reference calibration of the sensors, and 3) relative calibration of the sensors. Ref erence calibration requires the establishment of an inde pendent reference measurement system which can accu rately measure the load and deflections. Both test and ref erence systems are then set up to measure loads and
4-45

deflections under the same conditions. A specially designed load cell, which will be calibrated annually at the National Bureau of Standards will be used for the calibration of the FWD load cell. A commercial LVDT will be used to cali brate the sensors. Relative calibration procedures can further improve the accuracy of the sensors. In this process, all of the sensors are stacked in a special frame and are subjected to, and thus measure, the same pavement deflection simultaneously. Several test are usually performed at the same test point. Adjustment factors are determined by dividing the overall mean of all the deflection readings obtained by the mean for each individual sensor. Relative calibration should be per formed on a clean, distress-free pavement. ASTM D 4694-87 and 4695-87 provide guidelines for deflection testing and address the calibration issue. How ever, no specific calibration procedures are recommended. According to these procedures, calibration of the load cell and sensors should be carried out once per month during continuous operation, or before testing begins whenever the equipment is used on an intermittent basis. Experience with the FWD that has been in service at Cornell University con tinuously since 1981 has shown that the calibration should be performed every 6 to 1 2 months. States are able to calibrate their FWDs at the SHRP Regional centers in Harrisburg, Pa., Reno, Nev., College Station, TX, and St. Paul, Minn. Alternatively, states can establish sites of their own especially if they own more than one FWD. Regardless of the calibration site used (SHRP, or in-house site) all users must perform a full calibration of their FWDs at least once per year to be eligible to collect data for the LTPP program.

4-46

SECTIO N 4.0

REFERENCES Hicks, R.G. and Mahoney, J.P., "Collection and Use of Pavement Condition Data," NCHRP Synthesis No. 76, Transportation Research Board, Washington, D C., July 1981. Smith, R E. and Lytton, R.L., "Synthesis Study of Nondestructive Testing Devices for Use in Overlay Thickness Design of Flexible Pavements," Report No. FHWA/RD-83/097, Federal Highway Admini stration, U.S. Department of Transportation, Washington, D.C., April 1984. The Asphalt Institute, "Asphalt Overlays for High way and Street Rehabilitation," Manual Series No. 17, The Asphalt Institute, College Park, Maryland, June 1983. Federal Highway Administration, "Automated Pavement Distress Data Collection Equipment Seminar," Volume I, Appendix D, Seminar at Iowa State University, Ames, Iowa, June 12-15, 1990. Bentsen, R.A., Nazarian, S., and Harrison, J.A., "Reliability Testing of Seven Nondestructive Pave ment Testing Devices," Nondestructive Testing of Pavements and Backcalculation of Moduli, ASTM STP 1026, American Society for Testing and Mate rials, Philadelphia, Pennsylvania, 1989, pp. 41-58. Crovetti, J.A., Shahin, M.Y., Touma, B.E., "Comparison of Two Falling Weight Deflectometer Devices, Dynatest 8000 and KUAB 2M-FWD," Nondestructive Testing of Pavements and Backcal culation of Moduli, ASTM STP 1026, American Society for Testing and Materials, Philadelphia, Pennsylvania, 1989, pp. 59-69. Epps, J A. and Monismith, C.L., "Equipment for Obtaining Pavement Condition and Traffic Loading Data," NCHRP Synthesis No. 126, Transportation Research Board, Washington, D C., September 1986.
4-47

SECTION 5.0 DEFLECTION ANALYSIS TECHNIQUES

5.1 INTRODUCTION 5.1.1 General This section discusses techniques that are used for pave ment deflection analysis, and also introduces the concepts and basis of layer modulus backcalculation from measured deflection basins. A manual approach is illustrated using ELSYM5 in a trial and error mode, followed by a discus sion of typical automated approaches. Backcalculation techniques that are applicable to rigid pavements are also discussed. 5.1.2 Deflection Basin Parameters (including maximum deflections) When loads are placed on the surface of a pavement, such as a truck, aircraft, or passenger car wheel, the pavement will deflect downward to form a bowl shaped depression known as a deflection basin. The size, depth, and shape of the deflection basin is a function of several variables, including the thickness and stiffness of the pavement, the underlying materials, and the magnitude of the load. A PCC slab with a high elastic modulus will spread the wheel load over a large area resulting in a shallow deflection basin. Flexible pavements are much less stiff and tend not to spread the load as much resulting in a deeper basin. The difference in the deflec tion basins will be most noticeable within a 610 mm (24") radius with respect to the center of the load. In addition to stiffness, loading has a definite effect on the deflection basin. For example, as the load is increased, the pavement deflection will increase. However, many times this increase in deflection is not linear as many aggregates and subgrade materials are stress dependent.

5-1

The deflections are measured at various radial offsets, r, with respect to the center of the load plate. These deflec tion measurements define the basin. The load, P, plate radius, a, and plate pressure, p, must also be measured or known. These parameters, when analyzed with the deflection basin, enable us to estimate the stiffness profile of the pavement with respect to depth below the surface. Studies have shown that the outer deflection sensors respond primarily to the subgrade characteristics, while the inner sensors respond to the subgrade and upper pavement layers. The slope of the deflection basin at close proximity to the load is largely a function of the stiffness of the upper pavement layers. Over the years numerous techniques have been developed to analyze deflection data from various kinds of pavement deflection equipment. A summary of Deflection Basin Parameters was provided by Horak at the Sixth Interna tional Conference on Structural Design of Asphalt Pave ment [5.1] and is shown in Table 5.1. All of these parameters tend to focus on four major areas: (a) Deflection below the center of the load which repre sents the total deflection of the pavement. This was obviously the first deflection parameter which was developed for use with the Benkelman Beam. It has been used for many years as the primary input for several overlay design procedures. The slope or deflection differences close to the load such as Radius of Curvature (R), Shape Factor (Fj), and Surface Curvature Index (SCI). These parame ters tend to reflect the relative stiffness of the bound or upper regions of the pavement section.

(b)

5-2

Table 5,L Summary of Deflection Basin


Parameters (modified from Ref. S.U

P aram eter Maximum Deflection


^0 r 2

Form ula

M easuring Device Benkelman Beam, Lacroix deflectometer, FWD

Radius of Curvature

R=

/ 2 D 0 ( D0 / Dr -

Curvature Meter

r = 127mm | ((D 0 + D j + D 2 + D 3 ) / 5)100

Spreadability
~ L
D j ...

D
spaced at 305 mm

J
d0

Dynaflect

A = e [ l + 2( D , / d 0 ) + 2(D2/ D 0 ) + (D3/

)]

Area
( sensor spacing = 305 mm ( 12M )) Fl = ( D 0 - D 2 ) / D l

FWD

Shape Factors

FWD
f2

=(Dl - D 3)/D l

Surface Curvature Index

SCI = D q - D r , where r = 305m/w, or r = 500 mm

Benkelman Beam Road Rater FWD

Base Curvature Index Base Damage Index Deflection Ratio Bending Index Slope of Deflection

BCI = D6\0 ~ D 9\5 BDI = D 305 - D 610

Road Rater Road Rater


/ 2

Q r = D f / D q , where D f s

FWD Benkelman Beam Benkelman Beam

Bl = D / a, where a = Deflection basin SD = tan (


dq

- Df ) / r

where r = 610 mm

D0= center deflection (r=0), D p D2, D3 = first, second, third sensor from the load respectively D 305 = deflection at 305 mm, etc

5-3

(c)

The slope or deflection differences in the middle of the basin about 300 mm (11.8 in.) to 900 mm (35.4 in.) from the center of the load. These parameters tend to reflect the relative stiffness of the base or lower regions of the pavement section. The deflections toward the end of the basin. Deflections in this region relate to the stiffness of the subgrade below the pavement surfacing.

(d)

Correlations developed between basin parameters and pavement structural condition have been used for pave ment evaluation application. The parameters presented in this section were developed to provide a relative stiffness index, or a means of obtain ing the resilient modulus values of the surfacing layers in lieu of the more rigorous backcalculation process. In gen eral the success of these indices to accurately relate to the resilient modulus of the surfacing layers has been limited. There is a clear consensus; however, that the deflections measured at the outer deflection sensors relate quite well to the resilient modulus of the subgrade, (E sg ), and this forms the basic premise for most backcalculation tech niques. 5.1.3 Regression Equations for Predicting Moduli Several researchers have developed regression equations to predict layer moduli directly from deflections. Similar relationships can be derived fairly easily from theoretical considerations [5.9], Regression equations are typically developed to reduce the effort involved in backcalculation for production purposes. Sources of error in regression include ( 1 ) the analysis programs on which they are based, (2 ) the quantity of data used in development of the equa tion, and (3) the degree to which the model (usually linear elastic) simulates actual material behavior. Some of the regression approaches include:

5-4

(a)

WSDOT Equations

Newcomb developed regression equations to predict E sg as part of an overall effort to develop a mecha nistic empirical overlay design procedure for WSDOT [5.2]. For two layer cases, the subgrade modulus can be estimated from:
E Sg E sg E sg

= -466 + 0.00762 (P/D3), = -198 + 0.00577 (P/D4), = -371 + 0.00671 (2 P/(D3 + D4)),

(Eq. 5.1) (Eq. 5.2) (Eq. 5.3)

[Note: Variables are defined after Eq. 5.9] and for three layer cases
E sg E sg E sg

= -530 + 0.00877 (P/D3), = -H I +0.00577 (P/D4), = -346 + 0.00676 (2 P/(D3 + D4)) applied load, lbs

(Eq. 5.4) (Eq. 5.5) (Eq. 5.6)

Where P =

D3 = third sensor from the load D4 = fourth sensor from the load etc... The R2 ~ 99% for all equations and the sample sizes were 180 (two layer case) and 1,620 (three layer case). Figures 5.1 and 5.2 are used to illustrate typical results from Equations 5.4 and 5.5. These equations were developed from generated data using ELSYM5 on the following input data:

5-5

Two Laver Cases

Load, P, kN (lb.)
22

Surface Thickness Surface Modulus, Subgrade Modulus, hAc, mm (in.) E ac, MPa (psi) E sg, MPa (psi) 50 (2 ) 150 (6 ) 300 (12) 450 (18) 13,800 (2 , 0 0 0 , 0 0 0 ) 3,450 (500,000) 690 ( 1 0 0 ,0 0 0 ) 345 (50,000) 207 (30,000) 69 ( 1 0 ,0 0 0 ) 35 (5,000) 17 (2,500)

(5,000) 44 ( 1 0 ,0 0 0 ) 67 (15,000)

Three Laver Cases

Surface Base Load, P, Thickness, Thickness, kN (lb.) hAC mm (in.) he, mm (in.) ,
22

Surface Modulus, E a c ,MPa (psi) 13,800 (2 ,0 0 0 ,0 0 0 ) 3,450 (500,000) 690 ( 1 0 0 ,0 0 0 )

Base Subgrade Modulus, Modulus, E sg , MPa (psi) E b ,MPa (psi) 690 ( 1 0 0 ,0 0 0 ) 345 (50,000) 207 (30,000) 69 ( 1 0 ,0 0 0 ) 345 (50,000) 207 (30,000) 69 ( 1 0 ,0 0 0 ) 35 (5,000) 17 (2,500)

(5,000) 44 ( 1 0 ,0 0 0 ) 67 (15,000)

50 (2 ) 150 (6 ) 300 (12)

100

(4) 250 (1 0) 450 (18)

(assumed that load applied on a 300 mm (11.8 in.) diameter load plate)

WSDOT DEFLECTION vs SU3GRADE MODULUS EQUATIONS

DEFLECTION AT 3 FT. (MILS)

Figure 5.1 - Deflection vs. Subgrade Modulus for Equation 5.4 (Three Layer Case - Deflection Measured at 3')[after Newcomb 5.2]

MODULUS (MPa)

MODULUS (KSI)

WSDOT DEFLECTION vs SUBGRADE MODULUS EQUATIONS

Figure 5.2 - Deflection vs. Subgrade Modulus for Equation 5.5 (Three-Layer Case - Deflection Measured at 4 ') [after Newcomb 5JJ

From this generated data (no stiff layer), regression equations were also developed for estimating the surface modulus (AC) for a two layer case (for ex ample a "full-depth" pavement): log Eac = -0.53740 - 0.95144 logio E sg
2 ___

(Eq. 5.7)

-1.21181 VhAC+ 1.78046 logio (PAi/D02) where R2 = 0.83 For a three layer case, equations were developed for both E a c and E b as follows: If both E a c and E b unknown: log E a c = -4.13464 + 0.25726 (5.9/hAc) + 0.92874 V5.9/hB - 0.69727 ^ h AC^B 0.96687 logio Esg + 1.88298 logio (PAi/Do2) where R2 = 0.78. log Eb = 0.50634 + 0.03474 (5.9/hAc) + 0.12541 V 5 -9/hB - 0.09416 VhAC/hB + 0.51386 log E sg + 0.25424 logio (PAi/Do2) where R2 = 0.70.
(E q. 5.9) (E q. 5.8)

The following variables were used in equations 5.1 through 5.9: P = applied load (lbs.) on a 300 mm ( 1 1 . 8 in.) plate, surface course thickness (in.), base course thickness (in.),

hAC ~ hfi =

5-9

Ea c =

surface course modulus (psi), base course modulus (psi), subgrade modulus (psi), deflection under center of applied load (in.), deflection at 8 in. (0.67 ft.) from center of applied load (in.), deflection at 1 ft. from center of applied load (in.), deflection at 2 ft. from center of applied load (in.), deflection at 3 ft. from center of applied load (in.), deflection at 4 ft. from center of applied load (in.), and approximate area under deflection basin out to 3 ft.

Eb

E sg = Do =

Do.6 7 =

Di

D2

D3

D4

Ai

2 [2 (Do + Do.6 7 ) + (Do.67 + Di) + 3(Di + D2) + 3(D2 + D3)] = (b) 4Do + 6Do,67 + 8 D 1 +
12

D2 + 6 D 3

AASHTO Equations Witczak presented a regression equation in Part III of the 1986 AASHTO Guide for Design of Pave ment Structures [5.3] to predict the subgrade modulus. That equation is similar to the theoretical surface moduli presented in the next section, and has the following form:

5-10

Ec E sg where P Sf = =

(P)(sf ) /
M ) ( r )

(Eq. 5.10)

plate load (lb.), subgrade modulus prediction factor, 0.2686 for p = 0.50 0.2792 for p = 0.45 0.2892 for p = 0.40 0.2874 for p = 0.35 0.2969 for p = 0.30

Dr

pavement surface deflection (in.) measured at r distance, from the load, and distance from the load to Dr (in.).

Using an Sf value of 0.2892 where the Poisson's ra tio is 0.40, the equation reduces to the following equations for deflections measured at 610 mm ( 2 ft.), 914 mm (3 ft.), and 1,219 mm (4 ft.).
e Sg

0.01205(p /d 2) 0.00803(p /d 3) 0.00603(p /d 4)

(E q. 5.11) (E q. 5.12) (E q. 5.13)

E sg =
e Sg

In the AASHTO Guide, detailed procedures are provided to insure that the deflections used to de termine E sg are outside the pressure bulb from the test load. To most accurately represent the subgrade stiffness; however, the deflections closest to the load without being directly effected by the pressure bulb should be used. Stiff underlying layers will have the greatest effect on deflections furthermost from the test load. For example in cases where total pavement thicknesses are around 300 mm (11.8 in.), deflec tions taken around 600 mm (23.6 in.) should be used
5-11

to determine E sg - This ensures that the modulus obtained is not contaminated by the effects of the upper pavement layers (See page 111-86 of the AASHTO Guide for the Design of Pavement Structures). (c) South African Equations The following relation between deflections taken at 2000 mm (78.7 in.) was published by Horak [5 JJ:
logio Esg = 9,727 - 0.989 logio d2000
(E q. 5.14)

where

E sg d2 0 0 0

= subgrade elastic modulus (Pa), and = deflection at a distance of 2 0 0 0 mm from the (point) of loading
( n m ).

5.1.4

Surface Moduli These are based on Boussinesq or Boussinesq-Love equa tions, and are defined by Ullidtz [5.9] as "The 'weighted mean modulus' of the half space calculated from the surface deflection". Surface moduli calculated from deflections measured at some distance from the applied load can be considered representative of the subgrade response. In a recent NCHRP study [5.8] which will be used to revise Part III of the AASHTO Pavement Guide [5.3], it is rec ommended that the Boussinesq point-load equation be used to solve for subgrade modulus:
Mr

= P(1 - |i 2 )/(7t)(Dr)(r) = backcalculated subgrade resil ient modulus (psi), = applied load (lbs.),

(E q. 5.15)

where

Mr

5-12

pavement surface deflection a distance r from the center of the load plate (inches), and r = distance from center of load plate to Dr (inches).

Using a Poisson's ratio of 0.40, Equation 5.15 reduces to


M r

= 0.01114 (P/D2)
= = 0.00743 0.00557

(Eq. 5.16) (Eq. 5.17) (Eq. 5.18)

Mr Mr

(P/D3) (P/D4)

for sensor spacings of 610 mm (2 ft.), 914 mm (3 ft ), and 1,219 mm (4 ft.). If a Poisson's ratio of 0.45 is used instead for the same sensor spacings, the equations become:
M
r

= 0.01058(P/D2) = 0.00705 (P/D3)


= 0.00529

(Eq. 5.19) (Eq. 5.20) (Eq. 5.21)

Mr
M
r

(P/D4)

Darter et al. [5.8] recommends that the deflection used for subgrade modulus determination should be taken at a distance at least 0.7 times r/ae where r is the radial dis tance to the deflection sensor and ae is the radial dimen sion of the applied stress bulb at the subgrade "surface." The ae dimension can be determined from the following:

ae where ae = radius of stress bulb at the subgrade-pavement interface, (in.) NDT load plate radius (in.), D; = thickness of pavement layers i (in.)
5-13

(Eq. 5.22)

I.

)
n = number o f pavement layers

Ep

= effective pavement modulus (psi), and = backcalculated subgrade resil ient modulus.

M r

The effective pavement modulus, Ep may be derived from the following equation:

(E q. 5.23)

do

= 1.5pa
f Mr.

+
I ------

1 + --3

D (e7 a VM

where

d0 = deflection measured at the center of the load plate (and adjusted to a standard temperature of 6 8 F), inches p= a= NDT load plate pressure, psi NDT plate radius, in.

D = total thickness of pavement layers above the subgrade, inches Mr = subgrade resilient modulus, psi Ep = effective modulus of all pavement layers above the subgrade, psi

5-14

5.1.5 Backcalculation

Backcalculation involves estimating pavement layer moduli from measured surface deflections, and, usually, known layer thicknesses. The moduli which are obtained can be used in fundamental engineering analyses of the pavement using mechanistic approaches. Knowledge of material type for each layer can also allow one to use the modulus as an indication of material condition. Both manual and automated backcalculation procedures are discussed in detail in subsequent sections. Most backcal culation procedures are based on the assumption that deflections measured at some relatively large distance from the load primarily reflect subgrade response. This allows use of the outer sensor deflections for estimating subgrade modulus as a starting point, and the solution for all layers is typically derived iteratively from there. 5.1.6 Combining Indices for Project Analysis In many cases, use of more than one analytical approach can provide complimentary and supporting information, as illustrated by the Washington State procedure described below. Over the years WSDOT has found that the use of selected indices and algorithms provide a fairly complete and nec essary picture of the relative conditions found throughout a project. This picture is very useful in performing back calculation and may at times be used by themselves on projects with large variations in surfacing layers. WSDOT is currently using deflections measured at the center of the test load combined with Area values and Esg computed fl-om deflections measured at 610 mm (24 in.) presented in a linear plot to provide a visual picture of the conditions found along the length of any project (as illustrated in Figure 5.3). The Area value tends to provide a fairly good indication of the relative stiffness of the pavement section, particu larly the bound layer, because it is largely insensitive to subgrade stiffness. Combining the Area value in a plot with E sg and Do provides a good picture of the relative

5-1.5

stiffness of both the surfacing and subgrade that interact to produce the measured deflections. A more direct method, which is fairly commonly used in practice, is to consider the variation of deflections, layer moduli (including subgrade) and performance indicator such as remaining life or overlay along the project in a lin ear plot similar to Figure 5.3. This has the advantage of identifying relative performance of each pavement layer rather than the combined response reflected by the AREA parameter. 5.1.6.1 Area Parameter The Area value represents the normalized area of a slice taken through any deflection basin be tween the center of the test load and 914 mm (3 ft.). By normalized, it is meant that the area of the slice is divided by the deflection measured at the center of the test load, DO. Thus the Area value is the length of one side of a rectangle where the other side of the rectangle is D0. The actual area of the rectangle is equal to the area of the slice of the deflection basin between 0 and 914 mm (3 ft.). The original Area equation is: A =
6

(Do + 2Di + 2 D2 + D3 )/Do

(E q. 5.24)

where Do = surface deflection at center of test load, Di = surface deflection at 305 mm (1 ft.), D2 = surface deflection at 610 mm and D 3 = surface deflection at 914 mm (3 ft.).
(2

ft.),

5-16

W S DOT
S R 5 X O

N o n De a t r u e
M P 0 . 0 - ~ 7 . 6 0 E B

i ve
C a .s e

Pa vcme n t .
# 1 at

Te a t i ng .

NOTE: * Sumsa r y v a i u e s a r e n o r a a l i z e c to a 9 , COO pound l o a d anc aa j u s t e o f o r p a v e me n t t h i c k n e s s and t e i p e r a u r c . u o a u l u * det ermi nat ion : s based . on tr.* a e i l e c t i o n at the 4th s e n s o r '2 feet f r o * l o a a

Date Mile 2.900 3.100 3.300 3.500

e ste o

= C2 / 0 9 / 8 4
: = Area =s : = sx*>

:o : : emmmm
: 3 C H H

' - ===D e i l e c t : o n : s s * >

30

40

50

20

25

30

35 2 i

<=Suograoe Moduiu**> 0 20 30 4 0 50

7E M B +

3.700 3.750 4.300 4.500 4.700 4.900 5 . 100 5.300 5.500 5.700 5.900 6.100 6.300 6.510 6.700 6.900 7.100 7.300 7.500

zsmmmm smm c QB xsam


5 9 1r H H c JHH

im

6fl B i l a m 32B H M m 2:h o 2 S M H

.. liwmm .......... :+ma ii warn ii mm \:m .......... it mm naaamammmm . 2 'wmmmmw arn ... * C iwarn ) w arn 2 :: ammm *C wm HHH 2 'H H flB H H H mam 2 - H H H H H H im 1: H H i 1: B B liH B *C H M 8 i i f l HH ............. . . . . . T ------30 35

zammmm zaaaaa izmm


- flHBBH

i:mm 2 2 mmmmm 2 cwmmmmmm i ~mmm 2 :mtmmm 3 smaammmammi - waaam 2 1aaaamm 34 M H H B H H I 2 zmmmmm 2 smmmmmm -wm 3iwmmmmmmm 1 2wmm . ... ltmmmm lAwmm m
0 .0 20

*^

20

40

ID

:=

20

25

---------------30 4 0 50

Figure 5.3. Illustration of Basic NDT Parameters as Used by W'SDOT (SR 510 MP 2.9 to MP 7.5)
5-17

The approximate metric equivalent of this equa tion is: A = 150(Do + 2 D 3 0 0 + 2D600 + D9oo)/Do where Do = deflection at center of test load,
D 3 0 0 = deflection at 300 mm, D600 = deflection at 600 mm, and D 9 0 0 = deflection at 900 mm.

Figure 5.4 shows the development of the normal ized area for the Area value using the Trapezoidal Rule to estimate area under a curve.
The basic Trapezoidal Rule is:
K = h (y2y 0 + y, + y 2 + l 2y ,) /

(Eq. 5.25)

where

K= any planar area, yo - initial chord,

yj, y2 = immediate chords, y3 = last chord, and

h = common distance between chords. Thus, to estimate the area of a deflection basin using Do, Di, D2 , and D3 , and h = 12 in., so that h/ 2 = 6 in., then: K = 6 (D0 + 2Di + 2D2 + D3) Further, normalize by dividing by D q:

(Eq. 5.26)

(Eq. 5.27) 5-18

o*

12*

24"

36"

(Section B "Standard Pavement")

b
o o

C \J
o

* \j O
o
a

o o o

12

'

N O

C5

12

Estimated Area Using Trapezoidal Rule


A - 23.3

Equal Area Bounded by D o and Area Parameter

Figure 5.4. C om puting an Area Param eter

5-19

Thus, since we normalized by Do, the Area Pa rameter's unit of measure is inches (or mm) since it is in fact in2/in i.e., basin area per unit of center deflection. The maximum value for Area is 915 mm (36) and occurs when all four deflection measure ments are equal (not likely to actually occur) as follows: If, Do = Di = D 2 = D 3 Then, Area = 6(1 + 2 + 2 + 1) = 36.0 in. For all four deflection measurements to be equal (or nearly equal) would indicate an extremely stiff pavement system (like Portland cement concrete slabs or thick, full-depth asphalt concrete.) The minimum Area value should be no less than 280 mm ( 1 1 . 0 in.). This value can be calculated for a one-layer sys tem which is analogous to testing (or deflecting) the top of the subgrade (i.e., no pavement struc ture). Using appropriate equations, the ratios of

always results in 0.25, 0.125, and 0.083, respec tively. Putting these ratios in the Area equation results in Area = 6(1+ 2(0.25) + 2(0.125) + 0.083) = 1 1 . 0 in. Further, this value of Area suggests that the elas tic moduli of any pavement system would all be equal (e.g., Ei = E 2 = E 3 ). This is highly unlikely for actual, in service pavement structures. Low area values suggest that the pavement structure is not much different from the underlying subgrade material (this is not always a bad thing if the sub grade is extremely stiff which doesn't occur very often). Typical Area values were computed for pavement Sections A, B, and C (refer to Figure 5.5) and are shown in Table 5.2 (along with the calculated
5-20

Pavement Cases Standard Pavement Section A (thin) Section B (med.) Section C (thick) Stabilize Subgrade Section A (thin) Section B (med.) Section C (thick) Asphalt Treated Base Section A (thin) Section B (med.) Section C (thick) Moisture Sensitivity Section A (thin) Section B (med.) Section C (thick)

Table 5.2. Estimates of Area from Pavement . Sections___ Cases A, B, and C_____ _________ r:---- rrr ,-----? ---- ---------Pavement Surface Deflections, mm (inches) Area Do D305 D610 D915 mm (in.)
1.219 (0.048) .686 (0.027) .457 (0.018) .914 (0.036) .584 (0.023) .406 (0.016) .533 (0.021) .356 (0.014) .305 (0.012) 1.346 (0.053) .838 (0.033) .610 (0.024) .660 (0.026) .508 (0.020) .381 (0.015) .508 (0.020) .432 (0.017) .330 (0.013) .457 (0.018) .305 (0.012) .279 (0.011) .660 (0.026) .559 (0.022) .457 (0.018) .229 .356 (0.014) (0.009) .356 .254 (0.014) (0.010) .305 .229 (0.012) (0.009) .330 .229 (0.013) (0.009) .303 .229 (0.012) (0.009) .229 .279 (0.011) (0.009) .330 .254 (0.013) (0.010) .229 .254 (0.010) (0.009) .229 .203 (0.009) (0.008) .356 .229 (0.014) (0.009) .356 .229 (0.014) (0.009) .254 .330 (0.013) (0.010) 434 592 686 (17.1) (23.3) (27.0)

470 597 696

(18.5) (23.5) (27.4)

676 729 762

(26.6) (28.7) (30.0)

406 526 610

(16.1) (20.7) (24.0)

5-21

surface deflections (Do, Di, D2 , D 3 )). Table 5.3 provides a general guide in the use of Area values obtained from FWD pavement surface deflec tions. As mentioned previously the "typical" sec tions shown in Figure 5.5 are used throughout these notes for illustrative purposes. Typical Area values for different pavement structures and conditions are: Pavement PCCP Thick ACP Thin ACP BST flexible pavement (relatively thin structure) Weak BST Area, mm (in.) 610-840 (24-33) 530-760 (21-30) 410-530(16-21) 380-430 (15-17) 300-380 (12-15)

The following example in the use of the plot of Do, Area Parameter, and Esg, comes from State Route 510 just north of Olympia Washington (refer to Figures 5.3 and 5.6). MP 2.9 to MP 3.75 (Figure 5.3) The pavement through this area consists of various layers of bituminous surfacing (BST) totaling about 75 mm (3 in.) over about 300 mm (12 in.) of good quality gravel base. This is confirmed by Area Parameters ranging from 300 to 400 mm (12 to 16 in.). Within this section the sub grade consists of a sandy gravel with a modulus value of 150 MPa (22,000 psi) at MP 2.9 transitioning to 35 MPa (5,000 psi) at MP 3.55 (wet sandy silt) and increasing to 100 MPa (15,000 psi) near MP 3.7. It is clear that the large variation in D0 through this section is due primarily to variations in the subgrade stiffness.

5-22

Table 5.3 - Trends of D q and A rea Values

FWD Based Parameter


Area Low Low High High Maximum surface Deflection (D0)
Low High Low High

Generalized Conclusions*
Weak structure, strong subgrade Weak structure, weak subgrade Strong structure, strong subgrade Strong structure, weak subgrade

* Naturally, exceptions can occur

SECTION A (THIN)

SECTION B 50 mm (2 ) mm

SECTION C -----------

mm

AC BASE SU BG RA D E (FINE GRAINED)

Figure 5.5 5-23

W S D O T

N o n - D e s t r u c t M F > 1 0 . 0

i v e

P a v e m e n t

T e s t i n g

S R 5 L O

E B

C a .se

# 2 .

NOTE: Summary values are normalized to a 9,000 pound l o a d and adjusted f o r pavement tnicxness and temperature. Modulus determination i s b a s e d

on t h e
Date
Mi l e

aerlection

a t th e 4th s e n s o r

(2 f e e t f r o a l o a d )

Tested

= 12/18/89 <=======Area==s==ss> 1C 15 20 25 30 35

< = = = = De : l e c t i o n = = s*>
0 10 :C 20 40 50

<=Subgrade Modulus=> 10 20 30 40 50

10.003 10.009 10.015 10.021 10.027 10.033 10.039 10.045 10.051 10.057 10.063 10.069 10.071 10.077 10.081 10.087 10.093 10.099 10.105 10.111 10.117 10.123 10.129

2 61 3 51 301 2 31 1C|
8]

ci
CB

121 3 59 351 4 21 4 Cl
44|

361 141 ^ A1 221 241 221

10

20

40

50

10

25

30

35

10

20

30

40

50

Figure 5.6

Illustration of Basic NDT Parameters as Used by WSDOT (SR 5 1 0 MP 1 0 . 0 0 3 to 10.129)

5-24

M P 4.3 to M P 6.1 (Figure 5.3)

The pavement through this area consists of an older Portland Cement Concrete Pave ment (PCCP) which had been overlaid with about 2 in. (2 in.) AC ten to fifteen years ago. The Area Parameters of about 700 mm (28 in.) confirms the existence of the underlying PCCP except at MP 5.3 where the roadway was widened with ACP to build a center left turn lane, and the deflec tion tests were taken on the ACP widening. The subgrade consists predominantly of silty sandy gravel 110 to 260 MPa (16,000 to 38,000 psi) with isolated deposits of both silty sands and sandy gravels. Through this area the low Do values are due largely to the PCCP.

MP 10.105 to MP 10.123 (Figure 5.6)


The Do deflections drop to the 360 to 610 Hm (14 to 24 mil) range and the Area Pa rameters increase to between 380 to 510 mm (15 to 20 in.) indicating that the ACP is only fatigue cracked in spotty areas. The subgrade modulus values also increase to between 138 to 165 MPa (20,000 to 24,000 psi).

5-25

5.1.7 Joint Evaluation in Rigid Pavements

5.1.7.1

Introduction Deflection testing using a FWD can be used to determine the condition of the transverse joints in concrete pavements. When a load is applied near a joint, the loaded as well as the unloaded slabs deflect. This occurs because a portion of the load that is applied to the loaded slab is carried by the unloaded slab through load transfer. The magni tude of the tensile stress induced in the loaded slab depends on the amount of load transfer at the joint. If the joint is performing perfectly, both the loaded and unloaded slabs show equal deflec tion near the joint. A perfectly efficient system for transferring load from one side of the joint to the other can reduce the free edge stress by nearly 50%. Reducing the edge stress reduces fatigue damage while reducing deflections minimizes pumping potential. Therefore, good load transfer at the joints is essential for satisfactory perform ance of rigid pavements. The following factors affect the load transfer at joints. a) b) c) d) aggregate interlock subbase/subgrade support load transfer devices temperature

A brief description of how each of these factors affect the load transfer follows. a) Aggregate Interlock Interlocking of aggregate particles of the fractured surface below the saw cut at a joint provides load transfer between the
5-26

slabs at the joint. The main factors affect ing load transfer at the fractured surface are, the width of crack opening and the texture of the crack face. As the joint opens, there is less contact area between the two faces of the joint and the load transfer reduces. When the joint opens completely a minimum load transfer is available through the base course or subgrade. The texture of the cracked face de pends on size of coarse aggregate, maturity of concrete at time of fracture, and strength of concrete [5.20], Angular, rough surfaced aggregates (such as crushed stone) gener ally provide better interlock and load trans fer than do rounded and smooth surfaced aggregate (natural gravel) [5.20], The main factor which determines the texture of a crack face is the mode of fracture. Concrete can fracture around the aggregate or through the aggregate. When fracture oc curs around the aggregate, many pullouts of aggregate particles exist, resulting in a rough interface. Early fracture (when the aggregate-paste bond) is weak results in many pullouts. When the concrete strength increases pullouts diminishes and more ag gregate fractures occur [5.20], b) Subbase Support Some load transfer is provided by the sub grade or the subbase below the pavement. The amount of load transfer will depend on the type of subbase. Stabilized or lean con crete subbases will provide more load trans fer than an unstabilized subbase. Studies on undoweled airport pavements by Foxwor thy [5.32] have shown that generally a mini mum of 25% load transfer is provided by the subgrade.
5-27

c)

Load Transfer Devices Pavements with adequate dowel bars at joints provide increased load transfer across the joint. Pavements with dowels generally show good load transfer. However, with load repetitions looseness of the dowels can occur and this can lead to reduced load transfer.

d)

Temperature The temperature has a significant effect on load transfer. When a joint opens as the temperature decreases there is less contact between the two faces of the slabs at the joint. This can significantly affect the load transfer between slabs. The effect of tem perature on load transfer will be described in detail in a later section.

5.1.7.2

Determination of Load Transfer Nondestructive deflection testing can be used to evaluate the load transfer at joints in rigid pave ments. The test is conducted by applying a load near the joint and measuring deflections near the joint on the loaded and the unloaded slabs. The test can be conducted by using the FWD, and one of the common load and sensor configurations for measuring load transfer of the approach slab is shown in Figure 5.7(a). For this approach the sensor configuration for testing the leave slab load transfer is shown in Figure 5.7(b). The load transfer at the joint based on deflection is computed from the following equation: d = (du/dj) * where, d = joint efficiency (deflection) du = deflection of the unloaded slab dj = deflection of the loaded slab
100

(Eq. 5.28)

5-28

a)
A- 3

Approach Side
305 mm (12) 305 mm ( 12)

Traffic

Load Transfer = ^(100)

Ai 3* q

&

Joint or C r a c k s 0 * o>.

b) Leave Side

Load Transfer 305 mm 305 mm ^2

^ ( 100)
A 1

A3

Figure 5.7 - Arrangement of Deflection Sensors for Determining Load Transfer Efficiency at Approach and Leave Sides of a Joint or Crack

5-29

Another method of calculating load transfer efficiency is given by: d = 7 ------ y T (d u + d |) where du and dj were previously defined [5.44]. The theoretical joint efficiency (deflection) may range from 0 percent (none) to 1 0 0 percent (full). These two conditions are illustrated in Figure 5.8. The joint condition can be generally classi fied based on the following deflection transfer ef ficiencies: Good 75 - 100 %, Fair 50 - 75% and Poor < 50%. If joint performance is poor, fault ing is likely. The load transfer efficiency that was described previously was based on deflections. A joint load transfer efficiency based on stress can be defined by the following expression: S = (Su/Si) * where, S = load transfer efficiency (stress) Su = stress in the unloaded slab Sj = stress in the loaded slab This stress based load transfer efficiency indicates the stress carried by the unloaded slab in relation to the stress carried by the loaded slab. Studies have indicated that a one to one relationship be tween deflection efficiency and stress efficiency does not exist [5.3], A relationship that has been developed between these two parameters is shown in Figure 5.9 [5.3], As it is difficult to measure stresses in concrete slabs, the deflection based efficiency is commonly used to measure load transfer in concrete slabs.
100

(Eq. 5.29)

(Eq. 5.30)

5-30

Load

0 mm

Load

.51 mm ( 0 .020 )
A A

A *
A

.51 mm ( 0 .020 ) '


A A A

Ci.
A

C*.
A A

t.

t.

.51 Good Load Transfer= : = 1.00 .51

Figure 5.8 - Illustration of Poor and Good Load Transfer

5-31

20

40

60

80

100

Deflection Eff.cierKy (DTE 6

* 100) -

Fig 5. 9 Relationship Between Joint Efficiency For Flexural Stress and Deflection Methods of
M easurem ent (By I . Korbus and E. J. Barenberg: From D O T/FA A /R D -7914. IV). "Longitudinal Joint Systems In Slip-Formed Rigid Pavements Volum e IV*'.

5-32

The same procedures that were used for testing the load transfer across joints can be used to determine load transfer across cracks. Generally the load transfer efficiency decreases as the num ber of load applications the pavement is subjected to increases [5.20], Foxworthy [5.32] carried out a study to determine the effect of different ap plied load levels on joint efficiency. He tested airport joints with an FWD at three load levels which ranged from 6,500 - 23,000 and found that load transfer was consistent for the three load levels. 5.1.7.3 Effect of Temperature on Joint Testing When the temperature decreases, the concrete contracts and the joint opens. This causes less contact to occur between the fractured faces at the joint, and as a result the load transfer between slabs decreases. As the temperature increases, the slabs expand and more contact occurs at the fractured faces of the two slabs. This causes the load transfer between slabs to increase. There fore, the amount of load transfer depends on the temperature at the time of testing. Edwards et al [5.27] have shown that on a summer day when the pavement temperature rises substantially, load transfer may increase on the same joint from 50% in the morning to 90% in the afternoon. Greer [5.28] report that limited tests on Memphis Inter national Airport indicated that the transverse joint efficiency went up from 16% (December 1987) to 84% (May 1988) when the weather warmed. Foxworthy [5.32] conducted a series of tests on undowelled airport pavements to study the varia tion of load transfer efficiencies with temperature. Figure 5.10 shows the variation of joint efficien cies with temperature for four slabs having the same thickness. The load transfer efficiency gen erally approached 1 0 0 % with increasing tempera ture while with decreasing temperature the effi ciency approach a minimum of 20-25%. Foxwor thy [5.32] found that generally all joints showed the S-shape relationship shown in Figure 5.10 with a horizontal shift between the curves. In some instances, good load transfer existed throughout the temperature range, presumably due to small joint openings while some slabs
5-33

showed a poor load transfer and showed a flat response throughout the temperature range.

Figure 5.10

Relationship between air temperature and joint load transfer. [5.32]

5-34

Several researchers have developed methods to correct the load transfer to a reference tempera ture and to predict the load transfer at different times or temperatures. Shahin [5.231 presented a chart shown in Figure 5.11, which can be used to adjust the load transfer to a reference time. Obvi ously, this chart would only be valid for use under the conditions used to develop it. Foxwor thy and Darter [5.26] developed a relationship to predict the load transfer efficiency at any tem perature once the load transfer at one tempera ture is known. 5.1.7.4 Joint Test Considerations The following guidelines should be followed when conducting tests to determine load transfer across joints or cracks. (a) Testing should be performed when the joints are open. The best time to perform the test is during the night or in the early morning hours. Testing should be avoided during midday to minimize the possibility of joint lockup. On cool overcast days, joint testing may be performed throughout the day. A load approaching 40 kN (9,000 lbf) or more should be used for testing on highway pavements. The joint test should be carried out along the outer wheelpath. Generally in highway pavements the load transfer efficiency of approach and leave slabs are different. Usually the lower load transfer occurs in the leave slab. When con ducting joint testing it is recommended that both the approach and leave slabs be tested to determine their load transfers.

(b)

(c) (d)

5-35

TIM E

Or

DAY

Figure 5.11 - Exam ple Chart for Correcting Load Transfer for a Referenced Time o f Loading of 2:00 p.m.

LO D TRANSrCR A

(L.T.1 CORRECTION

TA R ( D CTO

5-36

5.1.7.5

Example As an illustration o f Load Transfer Efficiency (LTE) measurements, results for WSDOT routes 1-5 and 1-90 will be illustrated. The FWD deflec tion data was originally obtained during 1986 and 1987. The 1-5 location illustrates a PCC pave ment with generally few joint problems and good load transfer. The 1-90 location has badly faulted transverse joints and generally poor load transfer. (a) 1-5 MP 176.35 Northbound (Seattle) Deflections were measured across trans verse joints, with the load applied on both the approach and leave slabs, across the longitudinal joint, and across two longitudi nal cracks. The air temperature during testing was approximately 15.6C (60F) and the testing was conducted between 11 p.m. and 4 a.m. The measured deflections were used to calculate load transfer effi ciencies at each location using Equation 5.28. The load transfer efficiencies are summarized in Table 5.4. The average load transfer efficiency is the average deter mined for four load levels (approximately 26.7, 40.0, 53.4, and 66.7 kN (6,000, 9,000, 12,000, and 15,000 lbs.)). The results shown in Table 5.4 show that the average load transfer efficiency for the approach side o f the transverse joints was 91.2 percent and for the leave side was 92.1 percent. These load transfer efficiencies were high for the temperature at which the joints were tested, as well as for the pave ment age (22 years) and number o f load applications (approximately 13,000,000 ESALs). The load transfer efficiencies at the site showed very little variation, as evi denced by the low standard deviations and coefficients o f variation.

5-37

Table 5.4. Summary of Load Transfer Efficiencies 1-5

Row 1 (outer lane) 2 (outer lane) 3 (outer lane) 4 (inner lane) 5 (inner lane)

Location TJ-A TJ-L TJ-A TJ-L TJ-A TJ-L TJ-A TJ-L TJ-A TJ-L LJ-A LJ-L LC-A LC-L

Mean LTE (%) 90.7 91.4 92.8 91.2 92.0 94.4 88.9 91.8 91.5 91.8 63.3 69.3 74.7 59.2

Standard Deviation 11.4 8.6 5.8 4.1 3.0 2.9 12.3 7.7 4.8 2.9 28.6 19.2 20.4 29.6

COV (%) 12.6 9.4 6.2 4.5 3.3 3.1 13.7 8.4 5.2 3.2 45.2 27.7 27.3 50.1

Maximum LTE (%) 100.0 98.8 100.0 97.0 96.6 99.2 100.0 100.0 97.7 95.4 97.7 91.5 95.2 92.3

Minimum LTE (%) 52.8 63.5 81.2 84.2 87.2 89.4 49.0 69.9 80.8 85.8 21.5 39.0 51.2 13.5

Notes: 1)

2)

TJ = transverse joint LJ = longitudinal joint LC = longitudinal crack A = approach L = leave Approach and leave on longitudinal joints refer to direction o f FWD movement between Lanes 1 and 2.

5-38

The load transfer efficiencies across the longitudinal joint and longitudinal cracks were much lower and more variable than those for the transverse joints. The average load transfer efficiency for the longitudinal joint, when the load was applied on Lane 2, was 63 .3 percent for the load applied on Lane 1, was 69.3 percent. There was sig nificant variation between the maximum and minimum load transfer efficiencies meas ured (21.5 to 97.7 percent). The mean load transfer efficiencies meas ured on either side o f the longitudinal crack were 74.7 percent and 59.2 percent. One explanation for the difference in load trans fers may be that the crack faces were not vertical. If this were the case, when loaded on one side, the loaded slab would have been supported by the unloaded slab, which would have resulted in a higher measured load transfer efficiency. Even though several longitudinal joint and crack locations showed low load transfer efficiency, these joints and cracks were not faulted. They experienced little, if any, stress reversal because wheel loads moved parallel to the cracks rather than across them. (b) 1-90 MP 278.60 Westbound (Spokane) Deflections were measured across each transverse joint and crack, with the FWD load applied to both the approach and leave slabs. The air temperature during testing ranged from 12.8C (55F) for Row 3 to 22.2C (72F) for Row 5. Testing was conducted between 4 a.m. and 10 a.m. With the FWD data, load transfer efficien cies for each joint and crack were 5-39

calculated using Equation 5.28. Table 5.5 summarizes the load transfer efficiencies measured at this site. The results in Table 5.5 show that the average transverse joint approach site load transfer efficiency was 59.7 percent, and that o f the joint leave side was 74.0 percent. The average load transfer efficiency for the transverse crack approach side was 72.8 percent, and that for the crack leave side was 76.6 percent. There was also a large variation in the load transfer efficiencies in each row, as evidenced by the high coeffi cients o f variation. These load transfer efficiencies were, on the average, much lower than those measured at the 1-5 site. The lowest load transfer efficiencies measured in each row at the I90 test site were about 25 percent. Some researchers have suggested that a load transfer efficiency between 15 and 25 per cent will be measured across a joint even if the joint faces are not in contact because the subgrade provides some shear resis tance. As described above, the 1-90 joint conditions, (i.e. badly faulted) are consistent with the poorer LTE than the 1-5 joints, where little or no faulting was observed, joint faulting is generally attributed to traffic load associated damage. Reasons for greater damage at the 1-90 site are probably due to differences in traffic loading (quantity and magnitude); differences in support conditions and climatic differences.

5-40

Table 5.5. Summary of Load Transfer Efficiencies 1-90

Row 1 (inner lane)

Location TJ-A TJ-L TC-A TC-L

Mean LTE (%) 73.3 84.3 87.5 94.5 69.5 84.5 89.9 93.7 53.7 75.9 70.5 72.2 55.2 67.7 56.0 62.2 46.9 57.7 60.2 60.6

Standard Deviation 15.2 5.6 1.3 0.8 24.0 10.5 0.7 1.5 32.1 20.0 17.6 20.6 21.1 16.3 22.1 16.4 17.3 14.5 16.7 17.1

COV (%) 20.7 6.6 1.5 0.9 34.6 12.5 0.8 1.6 59.8 26.3 25.0 28.5 38.2 24.1 39.6 26.4 36.8 25.2 27.7 28.2

Maximum LTE (%) 90.6 91.2


-

Minimum LTE (%) 46.1 72.1


-

93.7 95.7
-

29.8 65.7
-

2 (inner lane)

TJ-A TJ-L TC-A TC-L

93.8 94.2 89.2 92.9 82.5 84.5 84.0 89.1 75.0 78.5 79.3 84.5

11.4 38.0 38.3 37.5 21.3 32.5 21.4 41.0 24.6 33.8 28.8 36.9

3 (outer lane)

TJ-A TJ-L TC-A TC-L

4 (outer lane)

TJ-A TJ-L TC-A TC-L

5 (outer lane)

TJ-A TJ-L TC-A TC-L

Notes:

TJ = transverse joint TC = transverse crack A = approach L = leave

5-41

5.1.8

Void Detection in Rigid Pavements 5.1.8.1 Introduction Voids are generally created below slab corners as a result o f pumping and erosion o f sub base/subgrade material. Nondestructive deflection testing can be used to detect the presence of voids. These tests consist o f measuring deflec tions at the slab corner. Corner testing should not be conducted during winter as water below the corner can freeze and the corner can show good support. The deflection o f the slab corners is very much influenced by slab curling. Before methods to detect the presence o f the voids are described, the effect o f slab curling on deflection measurements will be described. 5.1.8.2 Effect o f Slab Curling on Deflection When the temperature at the slab surface is greater than at the bottom o f the slab, the central portion o f the slab tends to lift off so that firmer contact with the subgrade is obtained at edges and corners. This condition usually occurs during the warm mid-day period due to thermal expan sion of the concrete. If the temperature at the surface o f the slab is less than that at the bottom, the pavement will lift off the joints and edges. In this case firmer contact is obtained at the center of the slab and this condition usually occurs in the early morning hours. Curling and warping may also occur as a result o f changes in moisture content o f the slab. Figure 5.12 illustrates this effect. Deflection measurements at slab corners when the slab is not in contact with the surface do not relate to pavement deflections. Figure 5.13 shows a load-deflection relationship at different 5-42

VOID

NIGHTTIME CURLING

Figure 5.12 - Slab Curling Due to Temperature Differentials in Slab. [5.451

5-43

LOAD

( F )

DEF LECT I ON

( m m . IO3 )

Figure 5.13. Typical Load Deflection Relations at Slab Corner

5-44

temperature differentials beneath the top and the bottom o f the slab obtained by Larsen [5.29]. In this figure a positive temperature differential cor responds to the case where the temperature at the top o f the slab is greater than at the bottom, while a negative temperature differential occurs when the top o f the slab is cooler than the bot tom. This figure shows that the load-deflection relationship is essentially linear when there is firm contact between the slab and the subbase and non-linear when there is a loss o f contact be tween the corner and the subbase. When there is a negative temperature differential in the slab, greater deflections are obtained at the corner when compared to deflections obtained for a positive temperature differential for the same load. Therefore, corner testing should be avoided when the slab is curled concave side upward. When testing corners, the ambient temperature as well as the range o f temperatures during the sea son in which testing is performed should be con sidered so that testing can be avoided when the slabs are curled concave upwards. Generally, corner testing should be avoided during the early morning hours. Larsen [5.29] investigated the ef fect o f test temperature on corner deflections and report that uniformity in test data was obtained when tests were conducted in the night. 5.1.8.3 Methods for Void Detection The following methods can be used to detect the presence of voids. a) b) c) d) Comer Deflection Profile Variable Load Corner Deflection Analysis Void Size Estimation Procedure Mechanistic Based Approach

5-45

A description o f each o f these methods follows. a) Corner Deflection Profile This is an approximate method for detect ing voids and is described in the AASHTO Guide [5.3], In this method corner deflec tions are measured under a constant load (preferably 9 kips). The deflections at the approach and leave corners are then plotted as shown in Figure 5.14. This figure shows that the leave corners show higher deflec tion than the approach corners. Usually approach corners have less voids than leave corners and show less deflection. The cor ners which exhibit the lowest deflections are expected to have full support value. A maximum allowable deflection, which is somewhat higher than the full support value can then be selected and used as a criteria to determine corners which have voids. For example, in the figure a reasonable maxi mum deflection would be ,508mm (0.020 in.). A single value o f maximum allowable deflection used in this method may not be appropriate if load transfer varies from joint to joint. Because o f this factor as well as the influence o f test temperature on the results, this method must be viewed as an approximate method. b) Variable Load Corner Deflection Analysis This method is described in the AASHTO Guide [5.3], In this method the comer de flections are measured at three load levels (e.g., 27, 40, 53 kN (6, 9, 12 kips)) to es tablish a load-deflection relationship at each corner.

5-46

JR C P J O IN T DEFLECTIO N PROFILE

FW CORNER DEFL. (MICRONS) D

JOINT NO. ALONG PROJECT

Figure 5.14 - Profile of Corner Deflection for JRCP (60 ft. Jt. Space)

FW CORNER DEFL. (MILS) D

Typically locations with no voids cross the axis very near the origin (less than or equal to 50 microns (0.002 in.)) as shown in Fig ure 5. 15 (for the approach slab). If the line crosses the deflection axis at a distance greater than 50 microns (0.002 in.), a void can be suspected (see leave slab before grout in Figure 5.15). The response before and after subsealing is shown in Figure 5.15. c) Void Size Estimation Procedure A method to estimate the size o f voids be low a slab corner was developed in the NCHRP Project 1-21 [5.34], This method requires deflection testing at the slab center (including the deflection basin), corner de flection and transverse joint load transfer to estimate the size o f voids. d) Mechanistic Based Approach Several methods which use mechanistic based approaches to determine the presence of voids at slab corners have been devel oped. Shahin [5.23] presented a mechanistic procedure to detect voids by comparing measured corner deflections with theoreti cal corner deflections. The elastic modulus of the slab and the modulus o f subgrade reaction below the slab are needed for this analysis. These values can be determined by conducting a deflection test at the center o f the slab and backcalculating the above val ues (this will be described later). In this method, a finite element program was used to establish a relationship between load transfer, corner deflection and k-value o f the support as shown in Figure 5.16. These relationships were established based on given load position, load magnitude, slab dimensions, slab arrangement and slab elas tic properties. The computed corner deflec tion appropriate to the slab support condi tions is determined from this figure and

5-48

20

FW Load - lb D s

x 10

Corner Deflection - ins x 1 0 '5

Figure

5.15 Joint load deflection where large void under leave corner was suspected (Ohio 1-77)

5-49

o tu -I u. UJ o
U

c w
o

r n: o o

UJ
CL

s o
o

LOAD TRANSFER
{% )

re 5.16. Relationship ofL oad Transfer vs. 6Cfor Various Dynamic Subgrade Modulus (k). 5-50

compared to the measured deflection. The difference between the measured and com puted corner deflections can be used to de tect possible voids. Ullidtz [5.9] describes a method to detect voids by comparing the k-value at the cor ner o f the slab to the k-value at the center of the slab. In this method, the elastic modulus and the k-value at the center o f the slab has to be determined first. This can be accomplished by conducting a deflection test at the center o f the slab and backcalculating the concrete modulus and the k-value of the supporting medium. Thereafter, as suming that the elastic modulus o f the slab is constant throughout the slab, the k-value at the corner is calculated based on the measured corner deflection and the degree of load transfer. Generally if the k-value at the corner is between 0.6 to 0.8 o f the kvalue at the center o f the slab, poor corner support is indicated. 5.1.9 Class Exercise A - Deflection Basin Parameters Three typical pavement structures are shown in Figure 5.5 which represent thin, medium, and thick pavement sections for illustrative purposes. Table 5.6 shows the deflections resulting from a typical FWD deflection test using 40 kN (9,000 lb.) load at 0.55 MPa (80 psi). The deflections were developed using elastic modulus values of 3,445 MPa (500,000 psi) for the AC, 172 MPa (25,000 psi) for the granular base and 52 MPa (7,500 psi) for the subgrade for Cases A, B, and C. The subgrade elastic modulus was increased to 103 MPa (15,000 psi) for cases A l, B l, and C l. The following class exercise (see Table 5.7) is provided to help gain a "hands-on feel" for the relative values o f the various deflection basin parameters we have been discuss ing. To complete the class exercise, determine the Area value, Shape factors Fi and F 2, Surface Curvature Index for r = 305 mm (12 in.) and r = 610 mm. (24 in) for the six Cases shown above from the various formulas covered 5-51

in this section. If time allows, also complete the calcula tion o f subgrade moduli shown in Table 5.8. Solutions are shown in Tables 5.9 and 5.11.

Table 5.6. Pavement Surface Deflections for Class Exercise Sections A (Thin AC), B( Medium AC), and C (Thick AC)

Calculated Deflections, ^m(mils) (inches)


Case A EgQ = 52 MPa (E sg = 7500 psi) A1 E$g = 103 (E sg = 15000 psi) B E sg ^ 52 MPa (E sg = 7500 psi) B1 E sg ~ 1^3 MPa (E sg = 15000 psi) 1226

Do
(48.28) (0.0483) 818 (32.20) (0.0322) 690 (27.18) (0.0272) 465 (18.30) (0.0183) 455 (17.92) (0.0179) 302 (11.90) (0.0119)

D 305 mm (12 in.)

^ 6 1 0 mm (24 in.)

^ 915m m (36 in.)

666

(26.23) (0.0262)

353

(13.91) (0.01391)

226

(8.91) (0.0089)

358

(14.10) (0.0141)

170

(6.69) (0.0067)

110

(4.32) (0.0043)

518

(20.39) (0.0204)

348

(13.71) (0.0137)

242

(9.53) (0.00953)

310

(12.20) (0.0122)

182

(7.18) (0.0072)

118

(4.64) (0.0046)

c
E sg = 52 MPa (E sg = 7500 psi)

385

(15.17) (0.0152)

305

(12.01)

238

(9.36) (0.0094)

(0.0120)
170 (6.70) (0.0067) 123

Cl
ESq =103 MPa (E sg = 15000 psi)

234

(9.20) (0.0092)

(4.86) (0.0049)

5-52

Table 5.7. Class Exercise

Deflection Parameters
SCI12 in. or SCI305 SCI24 in. or SCIio

Case

Area (in.)

Fi

f2

A A1

B B1
C Cl Necessary formulas:

6 Dq (Do + 2D 12 in. + 224 in. + D 36 in.) Dp - P24"


D

Fi

12"

F2
S C I n in . =

P l2 " ~ D36"
D

24"

Do - D

12 in.

SCI24 in. =

D o - D24 in.

1"

= 25.4 mm

5-53

Table 5.8. Class Exercise Subgrade Moduli Subgrade Modulus MPa (psi) Case @6 1 0 mm (24 in.) Eq. 5.10* A Al B B1 C Cl @ 914 mm @ 914 mm (36 in.) Eq. 5.10* (36 in.) Eq. 5.4*

*Use Poisson's Ratio = 0.45 Necessary formulas: Equation 5.4: Equation 5.10: E sg = -530 + 0.00877 (P/D 36 in.) E Sg = (P)(Sf)/(Dr)(r) where Sf = 0.2792 for jo = 0.45 . 1 MPa = 145 psi

5-54

Table 5.9 - Solutions Class Exercise Deflection Parameters


SC Iir or SCI 3 0 5 , SCI 2 4 " or s c i 610,

Case

A re a , m m (in .)

Fi

f2

microns (mils)
560 (22.05) 460 (18.10) 172 (6.79) 155 ( 6 .1 0 ) 70 (2.75) 69 (2.70)

microns (mils)
873 (34.37) 648 (25.51) 342 (13.47) 282 ( 1 1 .1 2 ) 150 (5.91) 132 (5.20)

434 (17.08) 397 (14.55) 588 (23.16) 514 (20.23) 694 (27.33) 622 (24.48)

1.31

1.25

A1

1.81

1.46

0 .6 6

0.79

B1

0.91

1.05

0.39

0.48

C1

0.57

0.65

5-55

Table 5.10. Solutions Class Exercise Subgrade Moduli Subgrade Modulus, MPa (psi) Case @ 610 mm (24 in.) 51.9 (7,527) 107.9 (15,650) 52.7 (7,637) 100.5 (14,582) 60.1 (8,718) 107.7 (15,627) @ 914 mm @9 1 4 mm (36 in.) (36 in.) 54.0 (7,834) 111.4 (16,157) 50.5 (7,324) 103.7 (15,043) 51.4 (7,457) 99.0 (14,362) 57.4 (8,329) 122.3 (17,741) 53.4 (7,752) 113.6 (16,481) 54.5 (7,903) 108.3 (15,711)

Al

B1

Cl

5.1.10Class Exercise B - Load Transfer Efficiency The deflections in Table 5.11 were obtained on a jointed concrete pavement in the early morning hours. Calculate the load transfer efficiency at each point using the proce dure illustrated in Figure 5 .7, Arrangement o f Deflection Sensors.

5-56

Table 5.11 - Class Exercise "B" - Load Transfer Efficiency (LTE)

Test Slab 1

Load Position A L

D1

D2

D3

LTE (%)

6.21 5.92

5.23 5.76

6.11 5.23

A L

5.13 4.67

4.72 4.60

4.99 4.48

A L

7.36 8.62

7.21 8.36

7.15 8.23

A L

5.40 6.13

2.54 3.77

3.60 3.43

A L

4.45 6.31

2.67 4.36

3.31 3.91

A = Approach Side L = Leave Side

5-57

Table 5.12 - Solutions - Class Exercise "B"

Test Slab 1

Load Position A L A L A L A L

D1

D2

D3

LTE (%)

6.21 5.92 5.13 4.67 7.36 8.62 5.40 6.13 4.45 6.31

5.23 5.76 4.72 4.60 7.21 8.36 2.54 3.77 2.67 4.36

6.11 5.23 4.99 4.48 7.15 8.23 3.60 3.43 3.31 3.91

84 88 92 96 98 95 47 56 61 62

A L A = Approach Side L = Leave Side 5

In summary, this illustrates that it is not always necessary to perform full elastic layered backcalculation for evalua tion purposes. Simple and effective procedures and equations have been used for many years to provide rela tively quick pavement condition information from deflec tion measurements.

5.2 MANUAL BACKCALCULATION Manual backcalculation basically involves a trial and error approach using one o f the elastic layered programs to match a set o f measured deflections. An initial set o f moduli are assumed and surface deflections are calculated using the program and compared with measured deflections. Moduli are adjusted based on the comparison, and the procedure is

5-58

repeated until an acceptable match between measured and calculated deflection basins is achieved. Layer thicknesses are necessary to use the approach. Although the approach is cumbersome, it provides an excellent learning experience. Two examples are presented. The first is a very simple problem using a theoretical deflection basin developed with ELSYM5. The second involves actual field measured FWD deflections from one o f the SHRP sites. The first example uses deflections generated for the medium thickness, typical section defined in Section 3.2.6 and shown in Figure 3.8, i.e. 125 mm (5") AC on 200 mm (8") base on a fine-grained subgrade. Surface deflections for a standard 40 kN (9,000 lb.) wheel at 0.55 MPa (80 psi) contact pres sure are (for a constant 305 mm (12") between deflection sensors) in microns (mils): D0 691 (27.2) Dj 518 (20.4) D2 348 (13.7) D3 242 (9.53) D4 177 (6.98) D5 137 (5.40) D6 111 (4.38)

We know, o f course, that the layer moduli used to generate these deflections are 3450 MPa (500 ksi) for the AC, 172 MPa (25 ksi) for the base and 52 MPa (7.5 ksi) for the sub grade. To illustrate the approach estimate the following values as seed moduli: E ac = 2069 MPa (300 ksi) 172 MPa (25 ksi) 103 MPa (15 ksi)

E base = E sg =

Poisson's ratios are the same as chosen in Section 3.2.6, i.e. .35, .4 and .45 for AC, base and subgrade. Using these val ues in ELSYM5 gives the following deflections in microns (mils):

5-59

D0

Di

D2

D3

D4

D5

D6

528 (20.8)

323 (12.7)

180 (7.09)

CALCULATED 116 84 4.55) (3.30) TARGET 242 177 (9.53) (6.98) ERROR % -52 -53

66 (2.58)

54 (2.13)

691 (27.2)

518 (20.4)

348 (13.7)

137 (5.40)

111 (4.38)

-24

-38

-48

-52

-51

Starting with Dg, it is clear that the subgrade modulus of 104 MPa (15 ksi) is too high since D6 is too low. One method for adjusting moduli is to multiply by the ratio of the calculated deflection to the target deflection. In this case 54/111 = 0.49 so that the adjusted subgrade modulus becomes .49 * 104 MPa = 51 MPa (7.35 ksi). [Note: This approach works well here because the assumption o f elastic behavior is correct for the theoretical basin derived with ELSYM5. Also, most o f the deflection at D6 can be attributed to the subgrade.] The next trial, using E ac = 2069 MPa (300 ksi) 172 MPa (25 ksi) 51 MPa (7.35 ksi)

E Ba se =

E sg

provides the following:

5-60

D0

Dj

D2

D3

D4

D5

D6

782 (30.8)

551 (21.7)

CALCULATED 358 245 179 (14.1) (9.63) (7.04) TARGET 242 177 (9.53) (6.98) ERROR % +1 +1

138 (5.45)

113 (4.44)

691 (27.2)

518 (20.4)

348 (13.7)

137 (5.40)

111 (4.38)

+13

+6

+3

+1

-2

The match is fairly good, with the largest error at D0, sug gesting that the AC modulus is too low. Note that the deflection ratio adjustment factor does not work well at DO since most o f the deflection is generated in the subgrade (For example, for this trial, 625 microns (24.5 mils) o f the total deflection o f 782 microns (30.8 mils) comes from the subgrade, i.e. 80%). Various methods are used for this ad justment. We will simply guess for the third trial and use an AC modulus twice as large as previously used, i.e. 4138 MPa (600 ksi) to get the following deflections: Dq D! D2 D3 D4 D5 D6

671 (26.4)

516 (20.3)

CALCULATED 353 248 182 (13.9) (9.76) (7.17) TARGET 242 177 (9.53) (6.98) ERROR % +2 +3

141 (5.54)

114 (4.49)

691 (27.2)

518 (20.4)

348 (13.7)

137 (5.40)

111 (4.38)

-3

+1

+3

+3

The error is still large at D0, but now the AC modulus is too high. Since trial runs 2 and 3 are identical except for AC modulus, we can try interpolating for a new AC modulus to get the following

5-61

4138 -

(413 8 - 2069) * 7 67 I| (7 8 2 -6 7 1 )

= 4138 - 373 = 3765M Pa(546ksi). Using this with ELSYM5 gives: D0 D2 D3 D4 D5 D6

686 (27.0)

521 (20.5)

353 (13.9)

CALCULATED 247 182 (9.74) (7.15) TARGET 242 177 (9.53) (6.98) ERROR % +2 +2

140 (5.53)

114 (4.48)

691 (27.2)

518 (20.4)

348 (13.7)

137 (5.40)

111 (4.38)

-1

+1

+1

+2

+3

This looks quite good, but now D6 shows the largest error. Additional iteration would involve adjusting the subgrade 114 modulus to 51 x = 52M Pa(7500ksi) to get: Do Dj D2 D3 D4 D5 D6

678 (26.7)

513 (20.2)

348 (13.7)

CALCULATED 242 178 (9.54) (7.00) TARGET 242 177 (9.53) (6.98) ERROR %

137 (5.41)

111 (4.38)

691 (27.2)

518 (20.4)

348 (13.7)

137 (5.40)

111 (4.38)

-2

-1

The largest error is approximately 2% at D0, which is usu ally considered acceptable. We have converged fairly quickly in this example to E Ac = 2069 MPa (300 ksi), Ebs = 172 MPa (25 ksi), and E Sg = 52 MPa (7.5 ksi), and would have arrived at the correct AC modulus o f 3450 MPa (500 ksi) in one or two more iterations.

5-62

In the second example actual pavement thickness data and deflections from Figure 5.17 and Table 5.13 are used with the ELSYM5 computer program to estimate a "reasonable" set o f layer moduli. This section is from the SHRP/LTPP GPS sites (simply described as Section F). The selected load is 9,512 lb. Use a "standard" FWD load plate with 300 mm (11.8 in.) diameter. It is noted that this is a relatively straightforward appearing pavement section; however... 5.2.1 Initial Estimates First, based on what we know from the prior information covered in this course, estimate/guess the approximate layer moduli. (a) Asphalt Concrete From Figure 3.25 (SECTION 3.0) and for an aver age temperature (mid-depth) o f about 15.6C (60 F), estimate E a C ~ 6,900 MPa (1,000,000 psi). (b) Base Course Since the base is crushed limestone, start with a E b 40,000 psi (276 MPa). This is just a marginally educated guess. Use o f the Shell criterion from SECTION 2.0 results in a base modulus of about 965 MPa (140,000 psi) which seems high (however, later on, a modulus o f 965 MPa (140,000 psi) will not seem so high). (c) Subgrade Using Newcomb's equation for estimating subgrade modulus with a deflection measured 914 mm (36 in.) from the center o f the load plate (refer to Section 5.1.3):

5-63

Asphalt Concrete 194 mm (7.65) Crushed Limestone Base 368 mm (14.47)

Silty Sand Subgrade 4,467 mm (175.88) or oo

Possible Shale Rock Layer @ 5 m (16.5(198))

re 5.17 - SHRP Pavem ent Section F (GPS-6A: AC Overlay of AC Pavem ent - Section Located in K entucky)

Table 5.13 - Test Temperatures and Surface Deflections for SHRP Section F.

Air Temperature: Pavement Temperatures: Surface: Depth = 25 mm (1): Depth = 100 mm (4): Depth = 173 mm (6.8):

12.8 C (55.0 F)

10 C (50.0 F) 16.7 C (62.0 F) 15.7 C (60.3 F) 16.3 C (61.4 F)


D e fle ctio n @

Load No.

FW D Load kN lb

0 mm
(0 )

203mm i8 )

305mm
( 1 2 ']

457mm

610m m {2 4 )

914m m
{3 6 )

1524mm
(6 0 )

um mils u m m ils

um

n ils

u m mils um m ils

u m m ils 1.09 1.69 2.37 3.29

u m m ils 17.3 0.68 25.7 1.0 1 33.8 1.33 46.5 1.83

1 2
3 4

29.06 6534 83.3 3.28 68.3 2.69 59.2 2.33 47.8 1.88 39.6 1.56 27.7 42.3 9512 128.8 5.07 109.7 4.32 93.2 3.67 75.9 2.99 61.0 2.40 42.9

56.32 12662 184.9 7.28 151.6 5.97 131.3 5.17 108.2 4.26 88.6 3.49 60.2 74.78 16812 246.6 9.71 207.5 8.17 180.6 7.11 149.4 5.88 122.2 4.81 83.6

5-65

= -530 + 0.00877 (P/D3) = -530 + 0.00877 (9,512 lb./0.00169 in.) ^ 48,830 psi (without a stiff layer i.e. subgrade is semi-infinite) . This seems a bit high, so try something lower, say 172 MPa (25,000 psi). (d) Stiff Laver For now, assume there is no stiff layer. 5.2.2 Class Exercise The following set o f ELSYM5 results illustrate a manual approach to analyzing the deflections in Table 5.13 (Pavement section shown in Figure 5.10). Based on these results, each group should analyze three additional struc tures and record results for Runs 12, 13 & 14 below. (N ote: There are no correct answersthis is just to get a "quick" idea o f how your changes affect the deflections.)

i I

Run h

o. 1

Layer

Material Properties E, MPa (ksi) 0 mm (0 in.) 0.35 0.40 0.45 196.6 (7.74) 203 mm
(8 in )

Calculated Deflections, jam (mils) 305 mm 457 mm (12 in.) (18 in.) 150.8 (5.94) 127.0 (5.00) 610 mm (24 in.) 106.9 (4.21) 915 mm 1525 mm (36 in.) (60 in.) 76.7 (3.02) 44.5 (1.75)

AC Base Subg.

6,900 (1,000) 276 (40) 172 (25)

168.1 (6.62)

Measured deflections (mils): 5.07 Comment

4.32

3.67

2.99

2.40

1.69

1.01

Calculated deflections too high; D 36 in. and D60 in. indicate "stiffer" subgrade required.

5-66

(b) Layer

Run No. 2 Calculated Deflections, |am (mils) 0 mm (0 in.) 203 mm (8 in.) 139.7 (5.50) 305 mm 457 mm (12 in.) (18 in.) 122.7 (4.83) 100.6 (3.96) 610 mm (24 in.) 82.3 (3.24) 915 mm 1525 mm (60 in.) (36 in.) 56.1 (2.21) 31.0 (1.22)

Material Properties E, MPa (ksi)

AC Base Subg.

6,900 (1,000) 276 (40) 241 (35)

0.35 0.40 0.45

167.9 (6.61)

Measured deflections (mils): 5.07 Comment:

4.32

3.67

2.99

2.40

1.69

1.01

Calculated deflections still too high, increase base stiffness; D 36 in. and D60 in. improved but also increase subgrade stiffness a bit. (c) Run No. 3 Calculated Deflections, |im (mils) 0 mm (0 in.) 0.35 0.40 0.45 4.32 3.67 2.99 2.40 1.69 1.01 150.4 (5.92) 203 mm (8 in.) 123.2 (4.85) 305 mm 457 mm (12 in.) (18 in.) 107.7 (4.24) 87.4 (3.44) 610 mm (24 in.) 71.1 (2.80) 915 mm 1525 mm (36 in.) (60 in.) 48.5 (1.91) 27.2 (1.07)

Layer

Material Properties E, MPa (ksi)

AC Base Subg.

6,900 (1,000) 345 (50) 276 (40)

Measured deflections (mils): 5.07 Comment:

Calculated deflections still too high, increase base stiffness a bit more; D 36 in. improved, D60 in. much better thus leave subgrade modulus as is for now.

5-67

(d) Layer Material Properties E, MPa (ksi) AC Base Subg. 6,900 (1,000) 414 (60) 276 (40) 0.35 0.40 0.45

Run No. 4

Calculated Deflections, 0 mm (Oin.) 144.0 (5.68) 203 mm


(8 in.)

(mils) 915 mm 1525 mm (36 in.) (60 in.) 47.8 (1.88) 27.2 (1.07)

305 mm 457 mm (12 in.) (18 in.) 103.1 (4.06) 84.1 (3.31)

610 mm (24 in.) 68.8 (2.71)

117.9 (4.64)

Measured deflections (mils): 5.07 Comment:

4.32

3.67

2.99

2.40

1.69

1.01

Calculated deflections still too high, try a stiff layer located at 198 in. (5.03 m) depth. Reduce base and subgrade moduli since the stiff layer will reduce the deflections somewhat. (e) Run No. 5 Calculated Deflections, jj.ni (mils) 0 mm (0 in.) 0.35 0.40 0.45 0.45 168.4 (6.63) 203 mm (8 in.) 140.7 (5.54) 305 mm 457 mm (12 in.) (18 in.) 124.0 (4.88) 102.1 (4.02) 610 mm (24 in.) 83.3 (3.28) 915 mm (36 in.) 55.6 (2.19) 1525 mm (60 in.) 25.7 (1.01)

Layer

Material Properties E, MPa


(ksi)

AC Base Subg.

6,900 (1,000) 345 (50) 172 (25)

6,900 Stiff Layer (1,000)

Measured deflections (mils): 5.07 Comment:

4.32

3.67

2.99

2.40

1.69

1.01

Try a stiffer base and subgrade.

5-68

(f) Layer Material Properties E, MPa


(ksi)

Run No. 6 Calculated Deflections, jam (mils) 0 mm (0 in.) 203 mm


(8 .in.)

305 mm 457 mm (12 in.) (18 in.) 90.9 (3.58) 71.9 (2.83)

610 mm (24 in.) 56.6 (2.23)

915 mm 1525 mm (36 in.) (60 in.) 35.8 (1.41) 15.5 (0.62)

AC Base Subg. Stiff Layer

6,900 (1,000) 414 (60) 272 (40) 6,900 (1,000)

0.35 0.40 0.45 0.45

132.1 (5.20)

105.7 (4.16)

Measured deflections (mils): 5.07

4.32

3.67

2.99

2.40

1.69

1.01

Comment: D 36 in. and D60 in. too low. Increase base stiffness and decrease subgrade stiffness.
(g )

Run No. 7 Calculated Deflections, |im (mils) 0 mm


(0 in.)

Layer

Material Properties E, MPa (ksi) 203 mm (8 in.) 124.5 (4.90)

305 mm 457 mm (12 in.) (18 in.) 110.0 (4.33) 91.4 (3.60)

610 mm (24 in.) 75.9 (2.99)

915 mm 1525 mm (36 in.) (60 in.) 52.8 (2.08) 26.2 (1.03)

AC Base Subg. Stiff Layer

6,900 (1,000) 552 (80) 172 (25) 6,900 (1,000)

0.35 0.40 0.45 0.45

150.1 (5.91)

Measured deflections (mils): 5.07 Comment:

4.32

3.67

2.99

2.40

1.69

1.01

Increase subgrade stiffness a bit to see if Do in. through D60 in can be reduced.

5-69

(h) Layer Material Properties E, MPa .. il-i6,900 (1,000) 552 (80) 207 (30) 6,900 (1,000)

Run No. 8 Calculated Deflections, |im (mils) 0 mm (0 in ) 203 mm ( Si n) 113.0 (4.45) 305 mm 457 mm (12 in.) (18 in.) 99.1 (3.90) 81.0 (3.19) 610 mm (24 in.) 66.3 (2.61) 915 mm 1525 mm (36 in.) (60 in.) 45.0 (1.77) 21.6 (0.85)

AC Base Subg. Stiff Layer

0.35 0.40 0.45 0.45

138.7 (5.46)

Measured deflections (mils): 5.07 Comment:

4.32

3.67

2.99

2.40

1.69

1.01

Now try to reduce Do in. and increase D60 in. (slightly) by increasing the AC and reducing the subgrade stiffnesses. (i) Run No. 9 Calculated Deflections, fim (mils) 0 mm
(0 ,n )

Layer

Material Properties E, MPa (ksi) M . 0.35 0.40 0.45 0.45 203 mm (8 in.) 114.3 (4.50)

305 mm 457 mm (12 in.) (18 in.) 103.1 (4.06) 87.6 (3.45)

610 mm (24 in.) 73.9 (2.91)

915 mm 1525 mm (36 in.) (60 in.) 52.6 (2.07) 26.7 (1.05)

AC Base Subg. Stiff Layer

10,350 (1,500) 552 (80) 172 (25) 6,900 (1,000)

133.4 (5.25)

Measured deflections (mils): 5.07 Comment:

4.32

3.67

2.99

2.40

1.69

1.01

Do in. improved but D 12 in. through D 36 in. increased a bit too much. Try increasing the base stiffness.

5-70

(j) Layer Material Properties E, MPa (ksi) AC Base Subg. Stiff Layer 10,350 (1,500) 690 (100) 172 (25) 6,900 (1,000)

Run No. 10 Calculated Deflections, ^m (mils) 0 mm


(0 in.)

203 mm
(8 in.)

305 mm 457 mm (12 in.) (18 in.) 97.5 (3.84) 83.1 (3.27)

610 mm (24 in.) 70.9 (2.79)

915 mm 1525 mm (36 in.) (60 in.) 51.1 (2.01) 26.7 (1.05)

0.35 0.40 0.45 0.45

126.2 (4.97)

108.2 (4.26)

Measured deflections (mils): 5.07 Comment:

4.32

3.67

2.99

2.40

1.69

1.01

Do in., D 12 in., D l8 in., D24 in. and D 36 in. all improved but D 36 in. not close enough. Try a slightly higher subgrade modulus. (k) Run No. II Calculated Deflections, Jim (mils) 0 mm
(0 in.)

Layer

Material Properties E, MPa (ksi) 203 mm


(8 in.)

305 mm 457 mm (18 in.) (12 in.) 92.2 (3.63) 78.2 (3.08)

610 mm (24 in.)


66.0

915 mm 1525 mm (36 in.) (60 in.) 47.2


(1.86)

AC Base Subg. Stiff Layer

10,350 (1,500) 690 (100) 190 (27.5) 6,900 (1,000)

0.35 0.40 0.45 0.45

120.9 (4.76)

102.9 (4.05)

(2.60)

24.1 (0.95)

Measured deflections (mils): 5.07 Comment:

4.32

3.67

2.99

2.40

1.69

1.01

It appears that Run No. 10 provided a slightly better match of the calculated and measured deflections.

5-71

(1)

Run No. 12

Layer

Material Properties E, MPa 0 mm


(0 in )

Calculated Deflections, urn (mils) 203 mm


(8 in.)

305 mm 457 mm
(12 m.) (18 in.)

610 mm (24 in.)

915 mm 1525 mm (36 in.) (60 in.)

AC Base Subg. Stiff Layer

0.35 0.40 0.45 0.45

(m) Run No. 13

Layer

Material Properties E, MPa (ksi) 0 mm


(0 in ;

Calculated Deflections, |im (mils) 203 mm


{8 in )

305 mm 457 mm (12 in.) (18 in.)

610 mm (24 in.)

915 mm 1525 mm (36 in.) (60 in.)

AC Base Subg. Stiff Layer

0.35 0.40 0.45 0.45

5-72

(n)

Run No. 14

Layer

Material Properties E, MPa (ksi) 0 mm (0 in.) 0.35 0.40 0.45 0.45 203 mm (8 in.)

Calculated Deflections, |am (mils) 305 mm 457 mm (12 in.) (18 in.) 610 mm (24 in.) 915 mm 1525 mm (60 in.) (36 in.)

AC Base Subg. Stiff Layer

5.2.3 Problem Summary After 11 ELSYM5 runs and about an hour or so o f effort, Run No. 10 appears to match the measured deflection the best. Thus, what in the beginning appeared to be a rather straightforward pavement section is not. The resulting moduli are a bit unusual and are: EAC Eb ESG = = = 10,342 MPa (1,500,000 psi) 690 MPa (100,000 psi) 172 MPa (25,000 psi) 6,895 MPa (1,000,000 psi)

RB @ 198 in. =

It should be apparent that a better method for performing backcalculation is desirable. Automated backcalculation programs perform this exercise more efficiently.

5-73

5.3 AUTOMATED BACKCALCULATION 5.3.1 Introduction This portion of SECTION 5 will be used to illustrate some fundamental characteristics about backcalculation computer programs. This will include: a) b) c) d) e) Typical Approaches Measures o f Convergence Convergence Techniques Layers Summary o f Existing Backcalculation Programs

The information presented below is generally applicable to flexible pavements and partially applicable to rigid pave ments. For instance, the techniques can be used on con tinuous PCC pavements and for slab center tests on jointed PCC. However, PCC and overlaid PCC structures may pose problems so special consideration for backcal culation o f rigid pavements is addressed in Section 5.4. 5.3.2 Typical Flowchart A basic flowchart which represents the fundamental ele ments in most known backcalculation programs is shown as Figure 5.18. This flowchart was patterned after one shown by Lytton i5.10~ Briefly, these elements include: |. (a) Measured deflections Includes the measured pavement surface deflections and associated distances from the load. (b) Laver thicknesses and loads Includes all layer thicknesses and load levels for a specific test location.

5-74

O c c a s i o n a l P a th U s u a l P a th

Figure 5.18 - Common Elem ents of Backcalculation Program s (modified after Lytton 15.11)

5-75

(C)

Seed moduli and Poisson's Ratio The seed moduli are the initial moduli used in the computer program to calculate surface deflections. These moduli are usually estimated from user expe rience or various equations (as illustrated in SEC TION 2.0). Typical values for Poisson's Ratio are given in Section 2.8, Page 2-47.

(d)

Deflection calculation Layered elastic computer programs such as WESLEA, CHEVRON, BISAR, or ELSYM5 are generally used to calculate a deflection basin.

(e)

Error check This element simply compares the measured and cal culated basins. There are various error measures which can be used to make such comparisons (more on this in a subsequent paragraph in this section).

(f)

Search for new moduli Various methods have been employed within the various backcalculation programs to converge on a set o f layer moduli which produces an acceptable error between the measured and calculated deflec tion basins.

(g)

Controls on the range o f moduli In some of the backcalculation programs, a range (minimum and maximum) o f moduli are selected or calculated to prevent program convergence to un reasonable moduli levels (either too high or low).

5.3.3 Measures o f Deflection Basin Convergence In backcalculating layer moduli, the measure o f how well the calculated deflection basin matches (or converges to) the measured deflection basin was previously described as the "error check". This is also referred to as the "goodness o f fit" or "convergence error." As computer speed has increased allowing greater numbers o f iterations

5-76

the accepted level o f convergence error has decreased quite substantially. It should also be mentioned that some programs may report errors based only on those deflec tion sensors used in the basin matching routine. Conver gence errors are the primary measure o f how well the backcalculation routine has matched measured values. In some cases one should not expect low errors. For instance, a badly cracked pavement violates the funda mental assumption o f continuity, so that one should ex pect a difference between theoretical and measured deflections. Three o f the more common ways to calculate such measures include: (a) Average o f absolute relative differences (ABS)

(Eq. 5.31)
where ABS = average of absolute relative differences between the calculated and measured deflection basin, expressed as a percentage, calculated pavement surface deflection at sensor i, measured pavement surface de flection at sensor i, and number o f deflection sensors used in the backcalculation process.

dci

dmi

nd

5-77

(b)

Root Mean Square (RMS)


1 "
m

RMS (%) =

1 Z
nd

r d ci -d m i V

(100)

(Eq. 5.32)

where RMS = dci dmi nd = Root mean square error, as defined in 5.3.3(a), as defined in 5.3.3(a), and as defined in 5.3.3(a).

(c)

Sum o f absolute values o f the relative differences (ARS) d -d


i= l

(100)

(Eq. 5.33)

where ARS = sum o f the absolute values of relative differences between the calculated and measured deflec tion basin, expressed as a per centage, [Note: ARS = nd * ABS] dci dmi as defined in 5.3.3(a), and as defined in 5.3.3(a).

It has been suggested in the literature that an RMS (%) error must be less than 1% [5.41], However, this is not always achievable in practice due to the inability to accurately characterize pavement layer thickness variations throughout the project. Often

5-78

the practicing engineer must seek to minimize the error to the greatest extent possible, realizing he must accept errors significantly greater. A 1-2% error per sensor is generally considered to be acceptable for RMS or ABS. ARS should be approximately 1.5 * no. of sensors used. (d) Example 1 From Table 5.13, calculate the ABS, RMS, and ARS for Load 2 and Run No. 10. The following deflections apply: Deflections |am (mils) nd 1 2 3 4 5 6 (0mm, 0") (203mm, 8") (305mm, 12") (457mm, 18") (610mm, 24") (915mm, 36") Measured 128.8 109.7 93.2 75.9 61.0 42.9 25.7 (5.07)
( 4 . 32 ) ( 3 .67)

Calculated 126.2 108.2 97.5 83.1 70.9 51.1 26.7 (4.97) (4.26) (3.84) (3.27) (2.79) (2.01) (1.05)

(2.99) (2.40) (1.69)


d.oi)

7 (1,525mm, 60") (i) ABS ABS % = 7

4 .9 7 -5 .0 7 5.07 3.84 - 3.67 3.67

4 .2 6 -4 .3 2 4.32

3.27 - 2.99 2.99 2.01 - 1.69 1.69

2.79 - 2.40 2.40 1.05-1.01


1.01

(100)

8.1% (a high error!)

5-79

(ii)

RMS
I f f 4.97 - 5 . 0 7 V

RMS (%) =

( 4.26

- 4.32'

5.07

4.32

3.84 - 3 .6 7 f + p . 27 - 2 9 9 V + ( 2.79 - 2.40V


V

3.67

2.99
1/2

2.40

+ p . 01 - 1.69V +

f 1.05 - 1 . 0 l V A
V 101

1.69 = (iii) ARS ARS (%) = = (e) Example 2

( 100)

10.4% (also a high error!)

5.14 - 5.07 5.07

4- +

1.05 - 1.01 1.01

( 100)

56.5%

(still a high error!)

Compare Run No. 4 to Run No. 10 for Load 2. Use an ABS convergence measure. We already know ABS = 8.0% and ARS = 55.9% for Run No. 10. Calculate ABS and ARS for Run No. 4.

Deflections |im (mils) nd 1 2 3 4 5 6 (0mm, 0") (203mm, 8") (305mm, 12") (457mm, 18") (610mm, 24") (915mm, 36") Measured 128.8 109.7 93.2 75.9 61.0 42.9 25.7
(5.07)
( 4 . 32 )

Calculated 144.0 117.9 103.1 84.1 68.8 47.8 27.2


(5.68)

(4.64) (4.06)
(3.31) (2.71)

(3 67)

(2.99)
(2 40 )
(

1 69)

(1.88)
(107)

7 (1,525mm, 60")

.(L 0 .il..

5-80

ABS (%)

=y

5.68 - 5.07 5.07 4 .0 6 -3 .6 7 3.67

4 .6 4 -4 .3 2 4.32

3.31 -2 .9 9 2.99

2.71 -2 .4 0 2.40 1.88 - 1.69 1.69


10 . 1%

1.07 - 1.01 \ 1.01 /

( 100)

ARS (%)

f 5.68 - 5.07 -f . . + 1.07 - 1.01 1.01 5.07 V 70.9%

( 100)

(Thus, based on the ABS and ARS measures, Run No. 4 has higher convergence errors than Run No. 10; however, both convergence errors are substan tially higher than normally acceptable when using a backcalculation computer program (recall the "fits" were done manually).) (f) Example 3 As a final example o f convergence measures, com pare RMSs for Run No. 10 and Run No. 11. This will conclusively tell us whether Run No. 10 was actually "better" than Run No. 11.

5-81

Recall for Run No. 10: RMS = 10.6% For Run No. 11 (Load 2): Deflections mm (mils)
nd

Measured 128.8 (5.07) (4 32) 109.7 (367) 93.2 75.9 (2.99) 61.0 (2,1) 42.9 (1 69) 25.7 ..... (.10,1)
1 " 4.76 - 5 .07 V vv +

Calculated 120.9 102.9 92.2 78.2 66.0 47.2 24.1 (4.76)


(4.05)

1 (0mm, 0") 2 (203 mm, 8") 3 (305mm, 12") 4 (457mm, 18") 5 (610mm, 24") 6 (915mm, 36") 7 (1,525mm, 60") RMS (%)

(3.63) (3.08) (2.60) (1.86)


(0.95)

( 4.05 - 4.32
v

5.07

4.32
+

^3.63 - 3.67V 3.67 / 1.86 - 1.69 1.69

f3.08 - 2.99x2 2.99 0.95 -1.01 1.01

2.60 - 2.40 2.40 V

2> ' 1/2

(100)
/.

6.5%

Oops, Run No. 11 was "better" than Run No. 10 after all. This possibly suggests another reason for using convergence measures. Also calculate ARS: ARS: ARS (%) = 4.05-4.32 4.32 3.08 - 2.99

240
+

0.95 - 1.01

1.01
40.8%

5-82

5.3.4 Class Exercise Convergence Error The exercise is based on SHRP Section F, as shown Figure 5.17. Exercise No. 1 Load = 42.3 kN (9512 lb.) r = 150 mm (5.91 in.) Estimated moduli
e ac = 7,094 MPa (1028.8 ksi) E bs = 480 MPa (69.7 ksi) E sg = 305 MPa (44.3 ksi)

Deflections, um (mils') 0 in. 128.8 (5.07) 131.8 (5.19) 8 in. 109.7 (4.32) 106.9 (4.21) 12 in. 93.2 (3.67) 93.0 (3.66) 18 in. 75.9 (2.99) 75.4 (2.97) 24 in. 61.0 (2.40) 61.7 (2.43) 36 in. 42.9 (1.69) 42.9 (1.69) 60 in. 25.7 (1.01) 24.6 (0.97)

Measured: Calculated:

Calculate ARS = ? see Section 5.3.3(c) (Answer: ARS = 11.06%) RM S= ? see Section 5.3.3(b) (Answer: RMS = 2.07%)

Exercise No. 2 Same section as above, except a different backcalculation attempt Estimated moduli E a c = 5,929 MPa (859.9 ksi) EBs = 888 MPa (128.8 ksi) E sg = 193 MPa (27.9 ksi) Rigid Base = 6,895 MPa (1000 ks i ) @ 198 in. (5.0 m) depth

5-83

Deflections, jim (mils) 0 in. Measured: Calculated: 128.8 (5.07) 130.8 (5.15) 8 in. 109.7 (4.32) 104.9 (4.13) 12 in. 93.2 (3.67) 91.9 (3.62) 18 in. 75.9 (2.99) 76.4 (3.01) 24 in. 61.0 (2.40) 64.0 (2.52) 36 in. 42.9 (1.69) 45.7 (1.80) 60 in. 25.7 (1.01) 23.9 (0.94)

Calculate ARS = ? RMS = ? (Answer: ARS = 26.5%) (Answer: RMS = 4.5%)

5.3.5 Measure o f Modulus Convergence Some backcalculation computer programs (such as EVERCALC and MODCOMP) have another conver gence criterion which can terminate the backcalculation process. This criterion checks the moduli changes from one iteration to the next. If the change o f each layer modulus is below some preselected limit, the deflection matching process is overridden and the backcalculation process is stopped. This modulus check takes the follow ing form: Ej(k+1) -EjOO Ej(k)

Modulus Tolerance > where Modulus Tolerance

(100)

(Eq. 5.34)

= difference in layer moduli from one itera tion (k) to the next

(k+1),
EiOO

= a specific layer modulus for the i-th layer at the k-th iteration, and = a specific layer modulus for the i-th layer at the (k+l)-th itera tion.

Ei(k+1)

This criterion can be particularly helpful if due to model ing errors (such as incorrect layer thicknesses) or

5-84

deflection measurement errors the normal deflection basin convergence criterion cannot be achieved. In this manner, the program will terminate before the maximum number o f allowable iterations is achieved. 5.3.6 Convergence Techniques The straightforward goal o f the backcalculation process is to estimate a set o f layer elastic moduli that best match the measured and calculated deflections. A number o f com puter programs have been developed during the past dec ade for this purpose, some o f which are discussed below. Most o f the currently used programs use elastic pavement analysis for the iterative deflection basin matching process which assume that the layer thicknesses and Poisson's ra tios are known. As was briefly described in Section 5.3.2, the process is started with initial (seed) moduli which are used to calculate a first deflection basin. The measured and calculated deflections are then compared and if they are different more than a preselected convergence error allows (such as ABS or RMS), the process is repeated until an acceptable convergence is achieved. This subsec tion is an attempt to describe how this iterative process typically is structured in backcalculation computer programs. (a) BISDEFAVESDEF Programs The convergence technique used in these two U.S. Army Waterways Experiment Station developed backcalculation programs will be briefly described [5.12. 5.131. Basically, the iterative process in volves development of a set o f equations which define the slope and intercept for each deflection and unknown layer modulus as follows: log (deflectionj) = Aji + Sji (log Ei) where A = intercept, S = slope, j i = 1 , 2 , . . . , ND (where ND = number o f deflections), and = 1,2, ..., NL (where NL = number o f layers with unknown moduli)

(Eq. 5.35)

5-85

This is further illustrated by use o f Figure 5.19 which shows how the deflection-modulus relation ship is developed from layered elastic analysis (originally done with the BISAR program, but more recently using the layered elastic system recently de veloped at WES, which is called WESLEA).

LOG MODULUS

Figure 5.19. Basic Process for Matching deflection Basins [from Ref. 5.431

5-86

(b)

MODULUS Program The MODULUS program is quite unique in it's ap proach to backcalculation in that it first performs a series o f forward layered-elastic program runs to construct a database o f deflections based on user supplied moduli ranges. It then utilizes a pattern search procedure to match each observed deflection basin in the field data file to a calculated deflection basin. If the measured deflection basin falls between two calculated basins, the program utilizes an inter polation procedure to arrive at a final calculated deflection basin. This negates the need to iterate through a series of forward layered-elastic computer runs. MODULUS seeks to minimize the error be tween the calculated and measured deflection basins using the following equation:

(Eq. 5.36)
where = squared error w wc t s
w ei

= measured deflection at sensor i = computed deflection at sensor i = number o f sensors = user supplied weighting factor for sensor i

The user supplied weighting factor can be used to lessen or completely omit consideration o f any sen sor in the backcalculation process. This can be useful in many instances, for example when the sub grade is behaving in a non-linear fashion, the weighting of outer sensors can be set to cause MODULUS to fit the inner sensors only.

5-87

Although a rigorous discussion o f the pattern search procedure is beyond the scope o f this course, it is necessary to discuss the "convexity test" and the meaning o f the results o f this test because it does appear on the MODULUS output. As stated earlier, MODULUS generates a database o f calculated de flection basins for use in matching the observed, or measured basins. The error term, shown previously, represents the error between the measured and cal culated deflection basins. MODULUS searches for the combination o f layer stiffnesses which minimizes the error term. The pattern search procedure will always converge in MODULUS, i.e. it will always find an answer. The question becomes: is it the best answer? Not always. Figure 5.20 shows what may happen in the course o f backcalculation. (MODULUS prefers to work in terms o f modular ratios with respect to the subgrade. On one axis, the ratio o f layer 1 with re spect to the subgrade is plotted. On the other axis, the modular ratio o f layer 2 with respect to the sub grade is shown). As can be seen from the plot, the error term reaches a minimum at point A and B. When MODULUS converges to an answer, it searches the surrounding area to see if it is indeed at the lowest point. If the area near the solution is not convex, such as around point A or B, the program will indicate that the convexity test has failed. This means that the results obtained are suspect. Convexity errors will occur for two reasons: 1) the operator has confined the ranges for the seed moduli too tightly or 2) the deflection basin is irregular and does not conform to layered-elastic theory (measuring across pavement cracks will result in irregular basins).

5-88

5.3.7

Summary of Backcalculation Programs (a) Backcalculation Programs Shown in Table 5.14 is a summary o f the most commonly used backcalculation computer programs used or developed mostly in the U.S. (as o f Novem ber 1990). This summary was prepared by SHRP/PCS/Law for use by the SHRP LTPP Expert Task Group for Deflection Testing and Backcalcu lation. The summary identifies 13 separate computer programs along with some o f their basic features. There are a number o f other programs described in the literature, and the field keeps growing. (b) Typical Program Results The "typical" pavement in sections used in SEC TION 3.0 are shown, for convenience, as Fig ure 5.21. The ELSYM5 program was used to gen erate a "manufactured" deflection basin for each o f the three sections. The four backcalculation pro grams (BOUSDEF, EVERCALC, MODULUS and MODCOMP) used in this course were then run with the known layer thicknesses and the "manufactured" deflection basins to illustrate typical program results. The assumed layer thicknesses, moduli, and Poisson's ratios are shown in Table 5.15. Further, the "manufactured" deflection basins obtained with ELSYM5 are shown. The results from the ELSYM5, BOUSDEF, EVERCALC, MODULUS and MODCOMP programs are shown in Table 5.16. User's guides, program demonstration and application are included in SECTION 6 and the appendices.

5-90

Table 5.14 - Partial List of Layer Moduli Backcalculation Programs [Rada et al]
Program Name Developed Forward Calculation Method Forward Calculation Subroutine Backcalculation Method NonLinear Analysis Rigid Layer Analysis flayer Interface Analysis Maximum Number of Layers Seed Moduli Range of Acceptable Modulus Ability to Fix Modulus Convergence Routine E rror Conver gence Function Yes

By
USACE-WES

BISDEF

Muhi-Layer Elastic Theory Method of Equiv. Thickness MuKi-Layer Elastic Theory Method of Equiv. Thickness Multi-Layer Elastic Theory MuHi-Layer Elastic Theory Muhi-Layer Elastic Theory Muhi-Layer Elastic Theory Muhi-Layer Elastic Theory Muhi-Layer Elastic Theory

BISAR (Proprietary)

Iterative

No

Yes

Variable

BOUSDEI

ZHOU, etal. OREGON STATE UNIV. USACE-WES

MET

Iterative

Yes

Yes

Fixed (Rough)

Cannot Exceed No. of Deflec., Works Best For 3 Unknown* 5, Works Best for 3 Unknowns

Required

Required

Yet

Sum of Squares of Absolute Error Sum of Percent Errors

Required

Required

Yes

Yes

CHEVDEF

CHEVRON

Iterative

No

Yes

Fixed (Rough)

ELMOD/ ELCON

P. ULL1DTZ DYNATEST

MET

Iterative

Yes (Subgrade

Yes

Fixed (Rough) Fixed (Rough)

Cannot Exceed No. of Deflec., Works Best For 3 Unknowns Up to 4, Exclusive of Rigid Layer

Required

Required

Yes

Sum of Squares of Absolute Error Relative Error on 3 Sensors

Yes

None

No

Yes

No

ELSDEF TEXAS AAM UNIV., USACE-WES PCS/LAW ELSYM5 Iterative No Yes

EMOD

CHEVRON

Iterative

Yes (Subgrade
Only)

No

Fixed (Rough)

Cannot Exceed No. of Deflec., Works Best For 3 Unknowns 3

Required

Required

Yes

Sum of Squares of Absolute Error Sum of Relative Squared Error Sum of Absolute Error Unknown

Ye

Required

Required

Yes

No

EVERCALC

J. MAHONEY, etal. W. UDDIN

CHEVRON

Iterative

Yes

Yes

Fixed (Rough) Fixed (Rough)

3 Exclusive of Rigid Layer Unknown

Required

Required

Yes

No

FPEDDI

BASINPT

Iterative

Yes

Yes (Variable)

Program Gener ated Required

Unknown

Unknown

No

ISSEM4

P. ULUDTZ, R. STUBSTAD

ELSYM5

Iterative

MODCOMP 3

L. IRWIN, SZEBENYI

CHEVRON

Iterative

Yes (Finite Cylinder Concept) Yes

No

Fixed (Rough)

Required

Yes

Relative Deflec. Error

No

Yes

Fixed (Rough)

2 to 15 layers, Max 5 Unknown Layers

Required

Required

Yes

Relative Deflec. Error at Sensors

No

Table 5.14 (cont'd.) - Partial List o f Layer Moduli Backcalculation Programs [Rada et al]

Program Name MODULI'S

Developed

By
TEXAS TRANS. INSTITUTE S.F. BROWN, et al.

Fortran! Calculation Method Mufti-Layer Elastic Theory Multi-Layer Elastic THeofy Mu hi-Layer Elastic Theory Multi-Layer Elastic Theory

Forward Calculation Subroutine WESLEA

Back calculation Method DataBase

NonLinear Analyib Yes?

Rigid Layer Analysis Yes (Variable)

Layer Interface Analysis Fixed?

Maximum Number of Layers

Seed Moduli

Range of Acceptable Modulus Required

Ability to Fix Modulus Yes

Convergence Routine

Up to 4 Unknown plus Stiff Layer

Required

Sum of Relative Squared Error Sum of Relative Squared Error Sum of Squares of Absolute Error Sum of Relative Squared Error

E rror Conver gence Function Yes

PADAL

UNKNOWN

Iterative

Yes (Sub grade


Only)

Unknown

Fixed?

Unknown

Required

Unknown

Unknown

Unknown

WESDEF

USACE-WES

WESLEA

Iterative

No

Yes

Variable

Up to 5 Layers

Required

Required

Yes

Yes

5-92

MICHBAK

MICHIGAN STATE

CHEVRON

Iterative

No

Yes

Fixed

Up to 4 Unknown plus Stiff Layer

Required

Optional

Yes

Yes

2' (50 mm) ACP

6* (150 mm) Base

Fine-grained subgrade

Section A (Thin Thickness Section)

5* (125 mm) ACP

6* (200 mm) Base

Fine-graned subgrade

Section B (Medium Thickness Section)

9* (230 mm) ACP

6 (150 mm) Base

Fine-grained subgrade Section C (Thick Section)

F ig u r e

5.21

Typical" Pavement Sections

5-93

Table 5.15 "Typical" Sections Roadway Sections and Backcalculation Results

Typical Pavement Sections


Thickness, mm (in.) Layer Thin (A) 51 (2.0) 152 (6.0) Medium (B) 127 (5.0) 203 (8.0) Thick (C) 229 (9.0) 152 (6.0) Poisson's Ratio Modulus, MPa (psi) 3467 (500,000) 172 (25,000) 52 (7,500)

AC

0.35

Base

0.40

Subgrade

--

0.45

Deflection Measurements Tire Load = 40kN (9000 lbs.) Tire Radius = 150 mm (5.9")
Offset Sensor No. 1 2 3 4 5 6 7 8 mm 0 203 305 457 610 914 1219 1524 (in.) (0) (8) (12) (18) (24) (36) (48) (60) 1226.3 869.4 666.2 471.4 353.3 226.3 164.6 129.8 Thin (48.28) (34.23) (26.23) (18.56) (13.91) (8.91) (6.48) (5.11) Deflection, nm (mils) Medium 690.4 589.0 517.9 423.9 348.2 242.1 177.3 137.2 (27.18) (23.19) (20.39) (16.69) (13.71) (9.53) (6.98) (5.40) 455.2 412.0 385.3 343.9 305.0 237.7 185.9 147.3 Thick (17.92) (16.22) (15.17) (13.54) (12.01) (9.36) (7.32) (5.80)

5-94

Table 5.16 "Typical" Sections Roadway Sections and Backcalculation Results (Continued)

Backcalculation Results
Layer Modulus, MPa(psi) Program BOUSDEF Thin Medium Thick EVERCALC Thin Medium Thick MODULUS Thin Medium Thick MODCOMP Thin Medium Thick 3520 3570 4130 (510,400) (517,650) (598,850) 174 170 54.3 (25,230) (24,650) (7,900) 51.8 51.8 53.4 (7,500) (7,500) (7,700) 0.7 1.1 2.8 0.1 0.2 0.6 3517 3261 3618 (510,100) (473,000) (524,700) 172 184 153 (25,000) (26,700) (22,200) 52 52 52 (7,500) (7,500) (7,500) 0.3 1.3 2.2 3295 3403 3390 (477,900) (493,500) (491,600) 179 174 185 (26,000) (25,300) (26,800) 51 52 52 (7,400) (7,500) (7,500) 5.7 1.0 1.6 0.9 0.2 0.2 5516 3228 3490 (800,000) (468,100) (506,200) 109 205 396 (15,800) (29,800) (57,500) 50 50 48 (7,300) (7,200) (6,900) 6.3 7.9 2.6 AC Base Subgrade ARS (%) RMS (%)

5.3.8. Verification of Backcalculation Results There have been and, undoubtedly, will be a number of attempts to verify that backcalculated layer moduli are "reasonable." To date, these attempts can be illustrated as those which examine measured, in-situ strains which are compared to calculated strains based on layered elastic analysis and backcalculated moduli and the other in which laboratory and backcalculated moduli are compared. (a) Comparison Based on Strain Lenngren [5.14] in a recently completed study for RST Sweden which used backcalculated layer moduli (from a modified version of the EVERCALC program) to estimate tensile strains at the bottom of

5-95

two thicknesses of AC (80 mm (3.1 in.) and 150 mm (5.9 in.)) pavement. For these actual pavement structures (located in Finland), actual, in-situ tensile strains were measured simultaneously along the de flection basins from an FWD. The instrumented pavement sections were developed by the Road and Traffic Laboratory (VTT) in Fin land. The strain gages were attached to the 150 mm (6 in.) cores which were, in turn, replaced into the AC surfacing. The deflections were induced in the two test pavements by use of a KUAB 50 FWD equipped with seven sensors (at 0 mm (0 in.), 200 mm (7.9 in.), 300 mm (11.8 in.), 450 mm. (17.7 in.), 600 mm (23.6 in.), 900 mm (35.4 in.), and 1,200 mm (47.2 in.)). The load plate had a diameter of 300 mm (11.8 in.). The applied load levels were ap proximately 12.5 kN (2,810 lb.), 25 kN (5,620 lb.), and 50 kN (11,240 lb.). The two pavement sections had either 80 mm (3.1 in.) or 150 mm (5.9 in.) AC surface course overlying a gravel and sand base (ranged from 550 to 620 mm (22 to 24 in.)), which, in turn, overlaid a "dry granular" subgrade (again, according to Lenngren [5.14]'). The computed results are shown in Tables 5.17 and 5.18. One can see that the backcalculated moduli were generally within 5 percent of the measured values. (b) Comparison of Laboratory and Backcalculated Moduli An attempt was made to compare resilient moduli from laboratory tasks (ASTM D4123 for AC cores and a simplified version of AASHTO T274 for un stabilized triaxial tests (base and subgrade)) to backcalculated moduli for five pavement sites in Wash ington state [5.15], The partial results of these com parisons are shown in Table 5.19. In reviewing these comparisons, the following apply: The backcalculated moduli were not adjusted for temperature or load rate effects. No stiff lay ers were used in the subgrade.

5-96

The laboratory measured AC moduli were ad justed to a temperature estimated at the mid dle of the AC layer at the time of FWD testing. The laboratory moduli were all obtained at a load duration of 100 ms (about three to four times longer than the FWD load pulse). The laboratory tests on the unstabilized materials were done on remolded specimens. Note that neither the backcalculated or laboratory moduli can be taken as the "truth." Why can such differences occur? There are numer ous reasons, the following are but a few: Remolding effects in laboratory compacted specimens (can include inability to reproduce field soil structure, density and/or moisture content). Load (stress) differences between FWD testing and laboratory tests. Nonrepresentative field sample obtained for laboratory testing (e.g., sampling a subbase or improved fill material when attempting to sample the "true" subgrade soils.) Non homogeneity of in-situ pavement materials. Backcalculation results represent the "mean" valve of a large volume of soil. Within this vol ume, the stresses differ significantly as may the soil properties. Other sources of error which will be discussed later in SECTION 5.0.

5-97

Table 5.17 - Backcalculated and Measured Tensile Strains - 80mm (3.1") AC Section (after Lenngren [5.141)

Tensile Strain Bottom of AC (x 10-6) Time of Day (a.m. or p.m.) Backcalculated* Measured % Difference
a.m. a.m. a.m. a.m. p.m. p.m. p.m* p.m. p.m. p.m. 119 119 74 60 284 284 167 167 87 81 123 122 64.9 64.7 292 283 159 158 84.8 84.2 -3 -2 +14 -8 -3 ~0 +5 +6 +2 -4

*BackcaIculation process used sensors @ D0, D300, D600, D900 and D1200

5-98

Table 5.18 - Backcalculated and Measured Tensile Strains 150mm (5.9") AC section (after Lenngren [5.141)

Tensile Strain Bottom of AC (x 10*6) Time of Day (a.m. or p.m.) Backcalculated* % Difference Measured
a.m. a.m. a.m.
a.m. a.m. a.m. p.m. p.m. p.m. p.m. p.m. 66 71 68 38 127 119 178 182 104 51 56 69.5 69 68.7 34.7 130 130 185 183 95.9 48.0 48.5 -6 +3 -1 +9 -2 -8 -4 -1 +8 +6 +14

*Backcalculation process used sensors @ D0, D300, D600, D900 and D1200

5-99

Table 5.19 - Com parison o f Backcalculated and Laboratory M oduli for Five Test Sites in W ashington State [from Ref. 5.15 along with updated Laboratory Results]

Location Section 1 Section 4 Section 5 Section 11 Section 15


Notes:

Moduli, MPa (ksi) AC


5250* 2790** 4730* 1570** 5770* 3360** 4260* 3520** 7370* (761) (405) (685) (228) (836) (487) (617) (510) (1069)

Base
159* 186** 331* 221** 276* 186** 186* 221** 269* 193** (23) (27) (48) (32) (40) (27) (27) (32) (39) (28)

Subg.
186* (27) 131** (19) 186* (27)

Convergence Error (RMS)


1.2%* n/a** 1.7%* n/a** 2.3%* n/a** 1.4%* n/a** 1.0%* n/a**

179** (26) 248* (36)

145** (21) 186* (27)

193** (28) 131* (19)

10900** (1580)

69** (10)

*Backcalculation Method; Backcalculated moduli obtained from the EVERCALC program. Laboratory Method.

5-100

5.4 BACKCALCULATION OF RIGID PAVEMENTS

5.4.1 Introduction In rigid pavement evaluation, the factors that are of inter est are the elastic modulus of Portland cement concrete (PCC) and the modulus of subgrade reaction (k-value) of the supporting medium. The elastic modulus of PCC and the k-value of the support can be backcalculated using the deflections obtained from a deflection test conducted on the interior of a rigid pavement. The elastic modulus of PCC can be used to evaluate the structural condition of the PCC slab, while the k-value of the supporting medium can be used to evaluate the supporting medium. The elas tic modulus of the PCC and the k-value of the supporting medium are required inputs for most overlay design meth ods. When conducting deflection testing on rigid pavements with the FWD, a load of 9,000 lb. or more should be applied to obtain the deflection basin. The effect of tem perature on the deflection basin measured at the slab inte rior has been found to be small [5.20], The maximum effect of temperature on deflections measured at the inte rior of the slab occurs during the warm mid-day period, when the slab is curled concave downward. If the effect of temperature on deflection is found to be significant, test ing during this period should be avoided. For backcalculation, rigid pavements are generally treated as two-layer systems because the base or subbase will have little influence on the shape of the deflection basin when compared to the influence of the PCC and subgrade. An estimate of the modulus of subgrade reaction (k-value) below the rigid pavement can be determined by computing the volume of the deflection basin as shown in Figure 5.22 [5.35], The k-value obtained in this manner is only an estimate.

5-101

Sensor Offset, Inches


20 40 60 80 100 120

140
"i

160

180

200 220

A7

This point determined as the extension of a straight line through the outermost two deflection readings (maximum of 200 in. used in basin)

Volumetric k (PCI) =

Force (lbs.) Arean (Sq. in.) x (Distance to Centroid ) x 2 x n


n=1

Deflection, Mils

Figure 5.22

5-102

The backcalculation of composite pavements, which are concrete pavements overlaid by asphalt concrete creates some special problems. The backcalculation of these pavements are explained separately in Section 5.4.3. 5.4.2 Backcalculation Methods for Rigid Pavements Several methods are available to backcalculate the elastic modulus of PCC and the k-value of the supporting me dium for rigid pavements. These methods can be classified as: a) b) c) d) e) Closed Form Solutions Elastic Layer Theory Method of Equivalent Thickness Database Approach Finite Element Methods

In all these methods the deflections are measured at the center of the slab. A description of each of these methods follows. a) Closed Form Solution Hall and Mohseni [5.36] describe a closed-form solution for determining the elastic modulus of the PCC slab, the modulus of subgrade reaction and the elastic modulus of the subgrade from the data from a deflection test. This method is applicable to a two layer system. In order to use this method deflections have to be obtained at distances of 0, 305, 610 and 915 mm (0, 12, 24 and 36 in.) from the load. This procedure uses the maximum deflection and the 'AREA' parameter which is computed from the de flection basin. The method uses relationships which were developed by Ioannides [5.35. 5.36], (Note: These are regression equations developed in U.S. units. To convert modulus values in psi to MPa divide by 145.)

5-103

The following steps must be used in this procedure: STEP 1: Determine the AREA parameter from the deflection basin using the following equation: AREA (in) = 6 ( 1 + 2 (Di/Do) + 2 (D2/D0) + (D3/D0)) where, Dq, D j, D 2 , D 3 are deflections at 0, 12, 24 and 36 inches from the load. STEP 2: Determine the dense liquid radius of rela tive stiffness (10 and the elastic solid radius of rela tive stiffness (le) using the following equations. lk = {In [(36-AREA)/1812.279]/(-2.559)}1/0-228 le = {ln(36-AREA)/4521.676)/(-3.645)}1/0187 These equations are valid for 1^ and le values be tween 15 and 80. STEP 3: Using the value of AREA calculated from the measured deflection basin and 1 calculated from ^ the previous step the k-value can be obtained from the Westergaard equation for center deflection which is: k = (P/8d0 l2k) * {1 +(l/27t)[ln(a/21k) + y-1.25] (a/lk)2} where, P = applied load (lbs.) d0 = maximum deflection at the center of the load (in) a = load radius (in) y = Euler's constant, 0.57721566490 STEP 4: Calculate the elastic modulus of the sub grade by using Losberg's deflection equation f5.39].
(Eq. 5.39) (Eq. 5.37) (Eq. 5.38)

5-104

(Eq. 5.40)

Es = [2P(l-n2s)/d0le] [0.19245 + 0.0272(a/le)2 + 0.0199(a/le)2 ln(a/le)] STEP 5: The elastic modulus of the concrete can be computed from either of the following equations:
[ l 2 1 < ( l - V Pc c ) k ]

(Eq. 5.41)
-' pcc L

where, lk

= dense liquid radius of relative stiffness , in. (from Eq. 5.37)

Epcc = PCC elastic modulus, psi Dpcc = PCC thickness, in (ipC = PCC Poisson's ratio ( assume a value) C k = k value ,psi/in (calculated from Eq. 5.39)

6e 0 1
le D 3p c c ( l - H s ) 2

(Eq. 5.42)

where, le = elastic solid radius of relative stiffness, in. (from Eq. 5.38)

Hs = subgrade Poisson's ratio (assume a value) Es = subgrade resilient modulus, psi

5-105

b)

Elastic Laver Theory

The approach used to backcalculate layer moduli of rigid pavements is essentially the same as that used for backcalculating flexible pavements. Backcalculation programs include a program for computing de flections as a subroutine. When backcalculating rigid pavements this subroutine should preferably be capable of handling variable interface conditions. Some programs cannot handle variable interface conditions and assume that full friction is present at all interfaces. At the interface between the PCC slab and the subbase full friction is usually not present. Friction can vary widely, ranging from 0 or near 0 for a PCC slab on subgrade to high friction for a PCC slab on cement treated base or for a bonded concrete overlay on PCC. Therefore, it is preferable to use programs that give the flexibility of choosing the interface condition. Once the layer moduli have been obtained from backcaiculation, the k-value below the slab can then be obtained by any of the following procedures. (i) Using Correlations A method of obtaining the modulus of the subgrade reaction from the elastic moduli of the subgrade and subbase, and subbase thickness is described in the Navy Manual for Nondestructive Evaluation [5.35]. The k-value of the subgrade is obtained from the elastic modulus of the subgrade using the following equations;
k = 10x

X = (Log 10E - 1.415)/l .284 where, k = modulus of subgrade reaction, psi/in E = subgrade modulus, psi

(Eq. 5.43)

5-106

If an untreated granular base or subbase is present above the subgrade, the effective k is determined using Figure 5.23. (ii) Westergaard Equations Use the Westergaard equation for center deflection with the backcalculated PCC modulus and obtain k.

c)

Method of Equivalent Thickness The approach is similar to that used for flexible pavements. ELCON [5.33] is a program that is similar to ELMOD, which can be used to analyze deflection measurements on two and three layer rigid pavements. Once the program determines the modulus of concrete using the method of equivalent thickness, it calculates the k-value of the support from the backcalculated subgrade modulus at slab centers. Edge and corner calculations use the Westergaard equations.

d)

Database Approach The drawback of programs using the database ap proach is that they can only be applied to the pave ment system configurations for which they were developed. DBCONPAS (Database for Concrete Pavement Systems) developed by Tia et al [5.25] uses a database of analytical results generated by FEACONS (Finite Element Analysis of Concrete Slabs) program. MODULUS is another database program that can be used for backcalculating rigid pavements.

5-107

EFFECTIVE k A SURFACE O BASE O SUBBASE. IBS/CU T F R

IN.

THICKNESS OF BASE OR SUBBASE, IN.

F ig u re 5 . 23

5-108

e)

Finite Element Programs

A method to backcalculate rigid pavement moduli using results from a finite element program is de scribed by Foxworthy and Darter [5.26], In this method deflections generated by the finite element ILLI-SLAB program is used to backcalculate the elastic modulus of PCC and the k-value of the sup porting medium. In order to use this method, de flections have to be obtained by the FWD with the seven sensors being placed at 12 in. intervals from each other. In this method, the deflection basin is characterized by the parameter 'AREA' which is obtained by the following equation. Area (in) = 6 * (1 + 2Di/Do + 2D 9/D 0 + 2D 3/D 0 + 2D4/D 0 +2D 5/D 0 + d 6/ d 0) In this equation Dq is the deflection obtained below the load while D j, D 2 , D 3 , D 4 , D 5 and D 5 are de flections at 12, 24, 36, 48, 60 and 72 inches from the load. The program ILLI-SLAB is used to generate deflection basins for a range of PCC moduli and support k values. These results in terms of Dq and AREA are plotted as shown in Figure 5.24. This figure is used to obtain the elastic modulus of PCC and the k-value of the supporting medium by using Dq and AREA from the measured deflection basin. This graphical procedure is time consuming as the plot shown in the figure has to be generated for each slab thickness. In addition errors can arise during interpolation. An iterative computer program to carry out this procedure was developed later. 5.4.3 Backcalculation of Composite Pavements In this section, composite pavements refer only to asphalt overlays on rigid pavements. Most distresses in composite pavements occur due to deterioration of the concrete slab below the asphalt. PCC distresses that are most responsi ble for distresses in the asphalt overlay are slab cracking, punchouts, joint deterioration, deterioration resulting from poor PCC durability (D cracking and reactive aggregate distress), and deterioration of PCC and asphalt patches [5.241. Deflection testing can be used to evaluate the

5-109

condition of the PCC slab that is not visible and to obtain the k-value below the pavement.

Deflection

Basin "Area"

(in i)

Monim um D eflection, DO (mils)

Figure 5.24. A Typical ILLI-SLAB Grid for the Backcalculation of E and k.

5-110

Generally it is difficult to achieve a solution when using orograms based on multi-layer elastic theory to backcalculate moduli of composite pavements, where the two upper layers are stiff when compared to underlying material. In general, iterative elastic layer backcalculation programs do not perform well in analyzing composite pavements [5.24], Their tendency is to under predict the modulus of the asphalt surface often going to the lower limits of the asphalt modulus range allowed by the user, and overpre dicting the modulus of the PCC [5.241. Methods for backcalculating the moduli of composite pavements have been presented by Anderson [5.241 and Hall and Mohseni [136] Anderson [5.24] developed a program called COMDEF to backcalculate moduli of a three layer pavement which consists of an asphalt layer, concrete layer and a uniform subgrade. COMDEF can only backcalculate moduli based on deflections measured by the FWD, with seven sensors spaced 12 inches apart. The program cannot accommo date fewer sensors or different spacings. The program uses pre-calculated solutions stored in database files to backcalculate moduli. These deflection basins have been calculated using elastic layer theory. Interpolation tech niques are used with the database of precalculated solu tions to obtain deflections for cases not covered in the database. COMDEF includes 33 database files, with each standard database file containing deflections correspond ing to a fixed asphalt thickness. The deflections in one database file has been generated for the following matrix. PCC thickness (in) 4, 6, 9, 14, 20, 30 Asphalt Modulus (ksi) 33, 82, 205, 512, 1,280, 3,200 PCC Modulus (ksi) 82, 205, 512, 1,280, 3,200, 8,000, 20,000 Subgrade Modulus (ksi) 2, 6, 18, 54, 162

5-111

COMDEF uses multiple application of two techniques to backcalculate. These two methods are: (a) stepwise direct optimization and (b) a iterative relaxation technique which uses gradient matrices. An option allows the user to en force to reasonable limits on the asphalt modulus based on the temperature. A typical output of the program is shown in Figure 5.25 [5.24], Hall and Mohseni 5.361 presented a method to backcal culate moduli of composite pavements. This procedure utilizes closed-form solutions for backcalculation of bare PCC pavements, with adjustments made to measured de flections to account for the influence of the asphalt con crete layer.

5.5 CRITICAL SENSITIVITY ISSUES IN BACKCALCULATION Lytton [5.10] discusses the need for experience both in analysis and with materials and deflections to ensure that the backcalculation process yields the most acceptable set of moduli for a given deflection basin. Many "well behaved" data sets will pose no problem, but in many cases the data is likely to be irregular in some way, making backcalculation difficult. Irregularities may result from a number of reasons, including pavement distress, variations in layer thicknesses, presence of bedrock or other stiff layer, or moisture. Effects of these irregularities can be compensated for by recognizing prob able causes and adjusting during backcalculation. The ob jective in applying these adjustments to the pavement model is to yield more representative or reasonable values for layer moduli, not simply to fit the deflection basin more closely. It should be pointed out that, as backcalculation techniques mature many of these critical issues are being addressed by software modification. The intent of this section is to dis cuss some of the reasons that may cause backcalculation problems, and procedures that may provide better solutions.

5-112

206

NONUESTnuCTIVE TESTING OF PAVEMENTS

PAVEMENT FACILITY OR FEATURE 10: EXAMPLE DATA 1 KDT LOADINGS PER TEST LOCATION STATION TRACK DATE TEHP LOAD 01 02 03

DS

06

07

1.0

87042)

70.0

25 025.

20.9

18.1 16. 2 14. 1 1 2 . 0 1 0 . 0

8.2

PAVEMENT FACILITY OR FEATURE ID: EXAMPLE DATA

SOLUTION FOR PROBLEM

1 O F

I FO FILE exmpTe.out R

STATION NUMBER 1.00 TRACK NUMBER I DATE OF TEST 870421 SURFACE TEMPERATURE - 70.0 DECREES F THICKNESS OF AC - 6.00 INCHES THICKNESS OF PCC - 7.00 INCHES DYNAMIC LOAO 25025. POUNDS MODULUS OF AC MODULUS OF PCC MODULUS OF SUBGRADE DISTANCE FROM LOAD (INCHES) 0. 12. 24. 36. 48. 60. 72. 55520B. PSI 50907B8. PSI 7975. PSI PREDICTED ACTUAL DEFLECTION DEFLECTION (NILS) (MILS) 20.90 20.90 18.10 I B . 10 16.23 16.20 14.09 14.10 11.97 12.00 9.99 10.00 8.22 8.20

SENSOR NUMBER 1 2 3 4 S 6 7

SUM OF ABSOLUTE VALUE OF ERRORS IN DEFLECTION TOTAL PERCENTAGE ERROR IN DEFLECTION BASIN -

0 .8 8 X

F ig u r e

5.25

E xam ple of COMDEF Input File and C O M D E F O utput File.

5-113

The primary rule is to always inspect the backcalculated moduli and apply a little engineering knowledge to the problem, if there is one. It should also always be kept in mind that if a layer stiffness is such that it has relatively little effect on the surface deflections, then backcalculation can provide little or no information about that layer. This will be discussed further in Section 5.5.5. 5.5.1 Input Data These include seed moduli, moduli limits and layer thick nesses as well as program controls such as number of it erations allowed and convergence criteria. Due to the non-uniqueness of the solution it is possible to obtain dif ferent layer modulus estimates for a given deflection basin by using different seed moduli, or limit for instance. Some programs, such as EVERCALC, have routines that will estimate seed moduli based on deflection basin char acteristics, which should provide a fairly reliable starting point. The basic approach is to choose seed moduli and moduli limits consistent with the materials and conditions in the pavement section at the time of test. Limits on moduli and convergence criteria should be set fairly wide initially, for a low number of iterations, to provide an ini tial indication of how reasonable the seed moduli are for a given basin. Note that if limits are not wide enough, and the iteration procedure fixes one of the lower layers at a value lower or higher than it appears to be in the deflection data, the compensating layer effect discussed below will result. Layer thickness effects are discussed in a subsequent sec tion. 5.5.2 Compensating Layer and Non-Linearity Effects This is an effect that essentially results from incorrect modeling of the pavement material response and the se quential nature of the backcalculation iterative procedure, as well as the geometry of a deflection basin test. A typi cal result may show, as an example, subgrade modulus that is significantly higher than expected for the material

5-114

type, while the base layer modulus is far too low and the surfacing modulus is too high. This probably occurs most commonly for a significantly stress softening subgrade, where the subgrade stress level for the outer sensors in a FWD test is very much lower than the subgrade stress level directly beneath the load plate. The apparent sub grade modulus for the outer sensor location is therefore higher than the apparent subgrade modulus directly beneath the load plate. If the subgrade is modeled as a linear elastic material, then, since most backcalculation routines first calculate subgrade modulus from the outer sensors, the higher modulus value is calculated and assumed to be constant throughout. At the next iteration, when the base modulus is being calculated, the too high subgrade modulus is compensated for by calculating a modulus that is too low for the base, in order to match the deflections measured in this region. In other words, alter nating layers exhibit a high or low compensating effect. Ideally, correctly modeling non-linear material response will remove this type of error, and this is becoming more and more common (e.g., MODCOMP3, EVERCALC, BOUSDEF can all use non-linear material models). If an elastic subgrade is used, then the inclusion of a stiff layer, (see 5.5.3) or the use of a layered subgrade, can help alleviate the problem. This is at least partially the reason why some backcalculation routines include a stiff layer by default at some depth (usually 20 ft ). 5.5.3 Subgrade "Stiff1Layers 5.5.3.1 General For the purposes of a general definition, a "stiff1 layer is one below which there is little or no ap parent contribution to the measured surface deflections. "Stiff1 layers can be real or "apparent" and are possibly the most common problem encountered during the evaluation of deflection basins. For the purposes of backcalculation, a stiff layer can be considered as any layer below the sub-

5-115

grade appearing to have a stiffness which is 10 times the modulus of the subgrade. A stiff layer may be comprised of bedrock underlying the sub grade. Stress sensitive subgrade materials, whose modulus increases as deviator stress decreases may form an "apparent" stiff layer, a phenomenon also referred to as "non-linearity". Granular ma terials tend to increase in modulus as the confin ing stresses increase as well. These materials will exhibit an increase in stiffness with depth. Stiff layer effects, whether real or apparent, mani fest themselves in the outer deflection measure ments. Typically, they result in an attenuation of the deflections at the outer radii leading to unrealistically high and subsequently inaccurate moduli values for the subgrade. This error will invariably result in inaccurate base and surface moduli values. For the case where an actual rigid layer exists, a variety computer backcalculation programs such as MODULUS, BISDEF, and WESDEF have a rigid layer subroutine built in. If non-linearity is not severe, the stiff layer rou tine can handle this as well. Bedrock information can be obtained from geologic maps, or by per forming cone penetrometer tests or drilling on the shoulder of the road. One solution to the "apparent" stiff layer prob lem, if a layered-elastic backcalculation program is used, is to divide the subgrade into two or more layers, allowing the backcalculation pro gram to assign modular ratios which achieve the best fit. Typically, for the case of the "apparent" stiff layer, these effects can be overcome by dividing the subgrade into two layers, the top of which is 300 mm (12) or more, depending on the reasonableness of the resulting layer moduli. This results in longer computer run times, addi tional operator input, and the need to combine upper layers if the total number of layers exceeds four, if MODULUS is being used.

5-116

An alternative solution would be to utilize a backcalculation program such as MODCOMP3, which addresses non-linearity and can analyze more than 4 layers. It should be noted however, that this introduces additional complexity to the procedure resulting in slower computer run times and additional input variables. It should be noted that high water tables can give the appearance of a stiff layer beneath the sub grade. Water is an incompressible fluid. When an impulse load is applied to a saturated soil, and the water has nowhere to go, it will build up pressure sufficient to resist the pressure exerted by the load. This pressure is known as pore water pressure, is equal and opposite in direction to the load, and can build up instantaneously. Thus, the saturated subgrade material is not deformed under the influence of a dynamic sur face load and the backcalculated subgrade modulus is unrealistically high leading to prob lems identical to the stiff layer case. Overcoming this problem can be handled the same way as the stiff layer case, however, an additional problem is present in this case. The stiff layer effect due to pore water pressure occurs only under dynamic loading. A slow-moving, heavy vehicle may impart vertical stresses in the subgrade of sufficient duration for the pore water pressure to dissipate, leaving only the soil skeleton to carry the load. The load could exceed the load carry ing capacity of the soil skeleton resulting in a deep foundation failure and depression of the pavement surface. Water tables tend to fluctuate throughout the year and from year to year in many parts of the country, a fact which further complicates the issue. It should be noted that the inclusion of a stiff layer, or subdivision of the subgrade into multiple layers is not a "cure-all" for backcalculation

5-117

problems. These approaches should be used only after all other options have been exhausted.
5.5.3.2

An Example It is often necessary to have a "deep," stiff layer within the subgrade in order to achieve reason able backcalculation results. This may be due to a rock layer or some other kind of "stiff1 condi tion. This can be illustrated using one of the SHRP/LTPP GPS sites (and associated deflection data). Figure 5.26 shows the two cases backcalculated one with a stiff layer at a depth of 6.1 m (240 in.) (shown as Case 1) and the other which assumes a uniform subgrade'with a semiinfinite depth (Case 2). The mid-depth AC tem perature at the time of testing was about 21C (70F) and the specific load level and associated deflections used in this illustration were: Load = 75.9 kN (17,054 lb.) Deflections: 0 mm (0 in.) 382.2 (im (15.05 mils)

203 mm (8 in.) 301.5 nm (11.87 mils) 305 mm (12 457 mm (18 610 mm (24 914 mm (36
n .)

257.0 |im (10.12 mils) |jm (7.92 mils) fim (6.36 mils) (im (4.14 mils)

n .) 201.2 n .) 161.5 n .) 105.2

1,524 mm (60 in.) 52.3 |im (2.06 mils)

5-118

Asphalt Concrete 126 mm (4.95 in.) Crushed Limestone Base 340 mm (13.40 in.)

Asphalt Concrete 126 mm (4.95 in.) Crushed Limestone Base 340 mm (13.40 in.)

Soil/Aggregate Subbase 305 mm (12 in.) Sand Subgrade 5.6 m (222 in.)

Soil/Aggregate Subbase 305 mm (12 in.) Sand Subgrade oo

Case 1

Case 2

Figure 5.26 - SHRP Pavem ent Section A (GPS-1: Asphalt Concrete Pavem ent with Granular Base - Section located in Florida)

5-119

The backcalculated results are:

Case 1: Layer AC Base Combined Subbase/ Subgrade StifFLayer Modulus. MPa (psi) 10,474 396 177 (1,519,000) (57,400) (25,700)

6,895

(1,000,000)

whereas the stiff layer modulus was preselected (or fixed) at 6,895 MPa (1,000,000 psi). Case 2: Laver AC Base Combined Subbase/ Subgrade Modulus. MPa (psi) 13,900 216 239 (2,016,000) (31,400) (34,700)

Thus, one can see that the combined sub base/subgrade stiffness is actually a bit higher than the base in Case 2. Normally, this is an un reasonable result and suggests that the use of a stiff layer (such as Case 1) provides a more "reasonable" set of layer moduli. 5.5.3.3 Load and Geostatic Stresses The need for stiff layers within the subgrade do main can certainly be due to rock layers or ex tremely stiff soils such as some glacial tills. However, there may be other conditions, not so immediately apparent, which warrant the use of a

5-120

stiff layer within the subgrade. First, we should look at some typical stresses in the subgrade due to an applied load and geostatic conditions. SHRP/LTPP Section F (recall Figure 5.17) was modeled with ELSYM5. Vertical and horizontal stresses were estimated under a 40 kN (9,000 lb.) load with a 0.69 MPa (100 psi) contact pressure. Two moduli conditions for Section F were used. The associated moduli and stresses follow:

Case A (a) Laver AC (7.65 in.) Base (14.47 in.) Subgrade (oo) Case Material Properties Modulus. MPa (psi) 6,895 (1,000,000) 345 (50,000) 276 (40,000) Poisson's Ratio 0.35 0.40 0.45

5-121

(b)

Results

Load Stresses* kPa (psi) Depth, mm (in.) 1,525 (60) 3,050 (120) 4,575 (180) 5,030 (198) 6,100 (240) 7,625 (300) 12,200 (480) 24,400 (960) 25,375 (999) -6.2 -2.1 -1.4 -1.4 -0.7 -0.7 oz (-0.9) (-0.3) (-0.2) (-0.2) (-0.1) (-0.1) ~0 ~0 ~0 ox or ov -0.7 (-0.1) ~0 ~0 ~0 ~o ~0 ~0 ~0 ~0
7.6

0
(1.1)

5.5 2.1 1.4 1.4 0.7 0.7

(0.8) (0.3) (0.2) (0.2) (0.1) (0.1) ~0 ~0 ~0

2.1 1.4 1.4 0.7 0.7

(0.3) (0.2) (0.2) (0.1) (0.1) ~0 ~0 ~0

* Due to 40 kN (9,000 lb.) only. Stresses: (-) compression, (+) tension. Case B (a) Layer AC (7.65 in.) Base (14.47 in.) Subgrade (175.88 in.) Stiff Layer @ 198 in. Depth Case Material Properties Modulus. MPa (psi) 6,895 621 207 6,895 (1,000,000) (90,000) (30,000) (1,000,00) Poisson's Ratio 0.35 0.40 0.45 0.45

5-122

(b)

Results

Depth, mm (in.) 1,525 (60) 3,050 (120) 4,575 (180) 5,030 (198) 6,100 (240) 7,625 (300) 12,200 (480) 24,400 (960) 25,375 (999) -5.5 -2.8 -2.1 -2.1 -1.4 -0.7 (-0.8) (-0.4)
(-03) (-0.3)

Load Stresses* kPa (psi) 0 GX Or Gy -0.7 (-0.1) -0.7 (-0.1) -1.4 (-0.2) -1.4 (-0.2) ~0 ~0 ~0 ~0 ~0 6.9 4.1 4.8 4.8 1.4 0.7 (1.0) (0.6) (0.7)
(0.7)

<7d 0.7 2.1 0.7 0.7 1.4 0.7 ~0 ~0 ~0 (0.1) (03) (0.1) (0.1) (0.2) (0.1)

(-0.2) (-0.1) ~0 ~0 ~0

(0.2)
. (o .i)

~0 ~0 ~0

Geostatic stresses are caused by the weight of the soil. Vertical geostatic stress, a v, can be straight forwardly calculated as follows [after Lambe and Whitman (5.2)]: av = (z) (y) where a v z y = = = vertical stress, depth, and total unit weight of the soil

(Eq. 5.44)

If we let g =100 lb./ft3 and a v units of psi, then ov where z = = 0.694 (z) depth (ft)

Horizontal geostatic stress, .h, is related to the vertical geostatic stress by a factor called the co efficient of lateral stress, which is designated Kq:

5-123

K0 ~ 0.5 for normally consolidated sedimentary soils but can approach 3 for heavily preloaded soils (over consolidated). When K0 < 1, ctv = oi and ah = 03 . When K0 > 1, Oh = oi and ov = 03 . If we assume the subgrade soils have a density (y) = 100 lb./ft3, then using the prior, preselected depths: Geostatic Stresses, MPa (psi) Ko = 0.5 Depth, mm (in.) 1,525 (60) 3,050 (120) 4,575 (180) 5,030 (198) 6,100 (240) 7,625 (300) 12,200 (480) 24,400 (960) 25,375 (999)
V oh

Kn = 3 0-d 0 0.12 (17.5) .05 Od (7.0)

(K = 0.5) .01 (1.8) .02 (3.4) .04 (5.2) .04 (5.8) .05 (7.0) .06 (8.7) 0.1 (13.9) 0.19(27.8) 0.2 (28.9)

Oil (K = 3)

0 .05 (7.1)

.02 (3.5) .05 (6.9) .07(10.4) .08(11.5) 0.1 (13.9) 0.12(17.4) 0.19(27.8) 0.38(55.5) 0.40(57.8)

.07 (10.5) 0.14 (20.7)

.01 (1.7) .02 (3.5) .04 (5.2) .04 (5.7) .05 (6.9) .06 (8.7) 0.1 (13.9) 0.19(27.7) 0.2 (28.9)

.09 (13.7)

0.24 (34.5) 0.09 (13.8) 0.36 (52.0) 0.14 (20.8) 0.4 (57.5) 0.16 (23.0)

0.22 (31.2) 0.14 (20.8) 0.24 (34.5) 0.16 (23.1) 0.29 (41.7) 0.19 (27.9) 0.36 (52.2) 0.24 (34.8) 0.58 (83.4) 0.38 (55.6) 1.15 (166.5) 0.77(111.1) 1.2 (173.4) 0.80(115.6)

0.48 (69.5) 0.19 (27.8) 0.6 (87.0) 0.24 (34.8)

0.96(139.0) 0.38 (55.6) 1.91 (277.5) 0.77(111.0) 1.99 (289.0) 0.8 (115.6)

From the above, it can be easily seen that the geostatic stresses dominate the stresses within the deeper portions of the subgrade (with or without a stiff layer in the system). If the modulus of a subgrade layer was stress sensitive, the magni tude of the stresses shown could greatly affect the backcalculation results and hence the need for a multi-layer subgrade; however, the geostatic stresses are static and it is not clear how impor tant these stresses are.

5-124

5.5.3.4

Depth to Stiff Layer

Recent literature (as of 1991) provides at least two approaches for estimating the depth to stiff layer [Rohde and Scullion [5.16], Hossain and Zaniewski [5.17]]. The approach used by Rohde and Scullion [5.16] will be summarized below. There are two reasons for this selection: (a) ini tial verification of the validity of the approach is documented, and (b) the approach is used in MODULUS 4.0 a backcalculation program widely used in the U.S. (and used in SECTION 6 ). (a) Basic Assumptions and Description A fundamental assumption is that the meas ured pavement surface deflection is a result of deformation of the various materials in the applied stress zone: therefore, the measured surface deflection at any distance from the load plate is the direct result of the deflection below a specific depth in the pavement structure (which is determined by the stress zone). This is to say that only that portion of the pavement structure which is stressed contributes to the meas ured surface deflections. Further, no sur face deflection will occur beyond the offset (measured from the load plate) which cor responds to the intercept of the applied stress zone and the stiff layer (the stiff layer modulus being 100 times larger than the subgrade modulus). Thus, the method for estimating the depth to stiff layer assumes that the depth at which zero deflection oc curs (presumably due to a stiff layer) is re lated to the offset at which a zero surface deflection occurs. This is illustrated in Fig ure 5.27 where the surface deflection Dc is zero.

5-125

Figure 5 . 2 7

Illustration of Zero Deflection Due to a Stiff Layer

5-126

An estimate of the depth at which zero deflection occurs can be obtained from a plot of measured surface deflections and the inverse of the corresponding offsets This is illustrated in Figure 5.28 and is based on theoretical consideration of Boussinesq equations, as originally devel oped by Ullidtz.. The middle portion of the plot is linear with either end curved due to nonlinearities associated with the upper lay ers and the subgrade. The zero surface deflection is estimated by extending the lin ear portion of the D vs. ~ plot to D = 0, with the ~ intercept being designated as r0. Due to various pavement section-specific factors, the depth to stiff layer estimated from r0 may not be reliable. In an attempt to improve this, additional factors were considered and regression equations were developed based on BISAR computer pro gram generated data for the following fac tors and associated values:

Load = P = 40 kN (9,000 lbs..) (only load level considered) Moduli ratios: = 10, 30, 100

cS g

0.3, 1.0, 3.0, 10.0

^ESg = 100
Sg

5-127

Figure

5 . 2 8 Plot o f Inverse of Deflection Offset vs. Measured Deflection

5-128

Thickness levels: Ti = 25, 75, 125, 250 mm (1, 3, 5, and 10 in.) T 2 = 150, 250, 375 mm (6 , 10, and 15 in.) B = 1.5, 3.0, 4.5, 6.0, 7.5, 9.0, 15.0 m (5,10, 15, 20, 25, 30, and 50 ft.) where Ej = elastic modulus of layer i, Ti = thickness of layer i, B = depth of the rigid (stiff) layer measured from the pavement surface (ft). The resulting regression equations follow in (b). (b) Regression Equations Four separate equations were developed for various AC layer thicknesses. The depend ent variable is ^ and the independent vari ables are r0 (and powers of r0) and various deflection basin shape factors such as SCI, BCI, and BDI (discussed earlier in SECTION 5.0). (i) AC less than 50 mm (2 in.) thick (Eq. 5.45) 0.0362 - 0.3242 (r0) + 10.2717 (r02) 23.6609 (r03) - 0.0037 (BCI)
R2 = 0.98

5-129

(ii)

AC 50 to 100 mm (2 to 4 in.) thick (Eq. 5.46)

0.0065 +0.1652 (r0) +

5.4290 (r02) - 11.0026 (r03) - 0.0004 (BDI) R 2 = 0.98 (iii) AC 100 to 150 mm (4 to 6 in.) thick (Eq. 5.47) ^ = 0.0413 + 0.9929 (r0) -

0.0012 (SC I)+ 0.0063 (BDI) - 0.0778 (BCI) R 2 = 0.94 (iv) AC greater than 150 mm (6 in.) thick (Eq. 5.48) ^ = 0.0409 +0.5669 (r0) +

3.0137 (r02) + 0.0033 (BDI) - 0.0665 log (BCI) R 2 = 0.97

where

= ~ intercept (extrapolation of the steepest section of the D vs. ~ plot) in units of ^ ,

SCI = Do - D305 mm (Do ' ^12 in )> Surface Curvature Index, BDI = (D 305 - D 610 mm) (D 12 Base Damage Index, - D 24 in ),

(Eq. 5.49) (Eq. 5.50)

5-130

BCI = D6io - D914 mm (D 24 in. - D36 in ) Base Curvature Index, Di = surface deflections (mils) normalized to a 40 kN (9,000 lb.) load at an offset i in feet. (c) Example Use the deflection data from SHRP Section F (Figure 5.17) to estimate B (depth to stiff layer). The drillers log suggests a stiff layer might be encountered at a depth of 5.0 m (198 in.). (i) First, calculate normalized deflections (40 kN (9,000 lb.) basis).__________

(Eq. 5.51)

Deflections, mm (mils)
Load Level, kN (lb.) 42.3 (9,512) 40 29 (9,000) (6,534)

D0

Dg

D 12 93.2 (3.67) 87.4 (3.44) 59.2 (2.33)

D 18 75.9 (2.99) 71.1 (2.80) 47.8 (1.88)

24 61.0 (2.40)

D36 42.9 (1.69) (1.59) (1.09)

D60 25.7 (1.01) 24.1 (0.95) 17.3 (0.68)

128.8 (5.07) 109.7 (4.32) 120.9 (4.76) 102.6 (4.04) 83.3 (3.28) 68.3 (2.69)

57.4 (2.26) 40.4 39.6 (1.56) 27.7

(ii)

Second, estimate r0. Plot Dr vs. ~ (refer to Figure 5.27):

Dr mm (mils) 4.76 4.04 3.44 2.80 2.26 1.59 0.95

r, mm (in.) 0 203 305 457 610 915 1,525 0" 8" 12" 18" 24" 36" 60"

1 1 r m 0
-------

4.93 3.28 2.19 1.64 1.09 0.66

(1.50) (1.00) (0.67) (0.50) (0.33) (0.20)

where all Dr normalized to 40 kN (9,000 lb.)


5-131

PLOT OF DEFLECTION vs 1/R

DEFLECTION (MICRONS)

1/R (INVERSE OF DEFL. OFFSET 1/FT)

Figure 5.29 - Plot of Inverse of Deflection Offset vs. M easured Deflection for SHRP Section F

5-132

DEFLECTION (MILS)

(in) Third, use regression equation in (b)(iv) (for AC = 7.65 in. (194 mm)) to calculate B: ^ = 0.0409 +0.5669 (r0) + 3.0137

(r02) + 0.0033 (BDI) - 0.0665 log (BCI) where r0 = ~ intercept (refer to Figure
5.27)

= 0 (used steepest part of deflection basin which is for sensors at 24 and 36 inches) BDI = D 12 in. D 24 in. = 3.44 2.26= 1.18 mils BCI = D 24 in. * D 36 in. = 2.26 1.59 = 0.67 mils > = 0.0409 + 0.5669 (0) + 3.0137 (02) + 0.0033 (1.18) - 0.0665 log (0.67) = B 0.0564 = 0~0564~ 17'^ ^eet (213 inches or 5.4 m)

This value agrees fairly well with "expected" stiff layer conditions at 5.0 m (16.5 ft) as indicated by the drillers log even though the ~ value equaled zero.

5-133

5.5.3.5

Example of Depth to Stiff Layer Estimates Road Z-675 (Sweden) (a) Overview This road located in south central Sweden is used to illustrate calculated and measured depths to stiff layers (the stiff layer appar ently being rock for the specific road). (b) Measurement of Measured Depth The depth to stiff layer was measured using borings (steel drill) and a mechanical ham mer. The hammer was used to drive the drill to "refusal." Thus, the measured depths could be to bedrock, a large stone, or hard till (glacially deposited material). Further, the measured depths were obtained independently of the FWD deflection data (time difference of several years). (c) Deflection Measurements A KUAB 50 FWD was used to obtain the deflection basins. All basins were obtained within 16 ft (5 m) of a specific borehole. The deflection sensor locations were set at 0, 200, 300, 450, 600, 900, and 1200 mm (0, 7.9, 11.8, 17.7, 23.6,35.4, and 47.2 in.) from the center of the load plate. (d) Calculations The equations described in Section 5.5.3.3 were used to calculate the depth to stiff layer. Since the process requires a 40 kN (9,000 lb.) load and 305 mm (1 ft.) deflec tion sensor spacings, the measured deflec tions were adjusted linearly according to the ratio of the actual load to a 40 kN (9,000 lb.) load.

5-134

(e)

Results

The results of this comparison are shown in Figure 5.30. Given all the uncertainties concerning the measured depths, the agreement is quite good. 5.5.4 Pavement Layer Thickness Effects Due to limitations in the backcalculation software, and the limited time available to perform backcalculation activities in a production environment, pavement layer thicknesses are assumed to be constant over the pavement section un der test. This is never the case. Pavement layer thickness variations result from poor construction quality control during initial construction, periodic overlays over existing rough pavements, and spot level-ups over short distances. In Texas, spot level ups pose the greatest problem be cause they are performed by in-house maintenance per sonnel and no detailed records are kept to indicate the thickness or extent of the level-up. On Texas SHRP sections, it has been found that asphalt concrete thicknesses may vary up to 2 in. within 500 ft. Pavement layer thickness variations will produce varia tions in the deflections from point to point which are in distinguishable from layer moduli variations. The net result is that this variation manifests itself in the backcalculated moduli for the various layers. A detailed research study found that if the pavement layer thickness variations are considered during the backcalculation process, the variation in layer modulus for the various layers was reduced significantly. In addition, more realistic moduli values were found for the various layers. Several methods exist to quantify pavement layer thick ness variations within a project. All are either expensive or time consuming. Examples discussed here include cone penetrometers, coring and drilling, and Ground Penetrat ing Radar.

5-135

ROAD Z-675,SWEDEN - STIFF LAYER COMPARISONS


3 T
T

10 9

2.5 O J
8
On

7
DEPTH (m) DEPTH (FT)

1.5

5 4 3

0.5

2
1

0 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
POINT NUMBER
FIELD

22 23 24 25

EQUATIONS

Figure 5.30 - Plot of Measured and Calculated Depths to Stiff Layer

The cheapest, slowest, and most painful is the Dynamic Cone Penetrometer. This device consists of a metal rod with a sharp metal cone at the bottom which penetrates the pavement layers in small increments as a weight is repeatedly lifted by hand and dropped. The speed of penetration of the cone is inversely proportional to the layer stiffness. By recording the depth of penetration, and noting the depth at which the penetration rate changes, the operator can approximate the thickness of each layer. Infiltration of fine material from the subgrade into the base can result in difficulty distinguishing the base/subgrade interface, as the change in rate of penetration is more gradual. Small vehicle mounted cone penetrometers exist, but are expensive to operate. They are faster than the hand op erated cone penetrometer but are still quite a bit slower than the FWD. These devices also provide an approxi mation of layer thicknesses. The most common method of verifying layer thicknesses is by coring. This is the most accurate method available for measuring asphalt thickness, but is time consuming and expensive. It is a destructive process and requires backfilling the hole with quick setting concrete or coldmix asphalt patching material. If the base material does not contrast visually with the subbase or subgrade, it can be difficult to get an accurate measurement of base thick ness. Usually, a 100mm (4") core is required to obtain adequate visibility for measurement purposes. Granular base materials tend to fall apart when extracted from the hole. Large stones in the base can also roll around under the core barrel during the drilling process and disturb the surrounding material, making measurements even more difficult. Cone penetrometer testing and coring require extensive traffic control and closure of the lane in which the work is being performed. Neither can keep pace with the FWD. Finally, the quickest method of obtaining pavement layer thicknesses is with Ground Penetrating Radar (GPR).

5-137

GPR testing can be performed at highway speeds, so no lane closures or traffic control is required. Many miles of pavement can be surveyed during the day and GPR is much faster than the FWD. GPR can measure asphalt concrete thickness quite accurately and reliably. Reli ability is improved if limited coring is used for calibration purposes. Limitations of GPR are as follows: a) If two adjacent pavement layers are constructed of similar materials, the interface between the two will not show up in the data and the thickness of the up per layer cannot be determined (example: Portland cement concrete slab and cement stabilized base) GPR is not accurate on reinforced Portland cement pavement due to signal attenuation by reinforcing steel. For non-reinforced PCC the accuracy improves if the underlying base or subbase is com posed of materials having a dielectric constant dif ferent than that of the slab. The presence of moisture adversely affects GPR per formance, so the pavement layers must be relatively dry Infiltration of fine material from the subgrade to the base may result in an indefinite base/subgrade inter face which may render GPR useless for base thick ness determination Software which analyses GPR data and reports layer thicknesses is still under development and available from limited sources GPR equipment costs several hundred thousand dollars and still requires interpretation by an experi enced technician.

b)

c)

d)

e)

f)

Before employing any of the above methods in an NDT investigation, one should first determine if accurate layer thicknesses are critical to the analysis. If the objective of the analysis is to determine remaining life of a pavement based on a cracking or rutting model which utilizes

5-138

pavement strains, then layer thickness determination is probably justified. However, if the objective is to deter mine how long an overlay might last, using tensile strains within that new layer, the thickness of the underlying lay ers is not critical. Any reasonable combination of layer thickness and modulus for the underlying layers which yields an equivalent deflection basin will work. This is due to the compensating effect whereby if the thickness of a layer is under estimated, the backcalculated modulus will be greater by the amount required to compensate for the error. However, it should be kept in mind that stress or strain levels are more affected by thickness than modulus, as shown by Odemark's transformation, in which the effect of the layer is a function of its thickness cubed, so some care should be taken in the approach. Finally, although the various methods for determining layer thicknesses are either laborious, time consuming, or expensive, one can employ various strategies to minimize their use, yet still acquire accurate information. One such strategy would be to use the deflection data or GPR results to separate the project into uniform sections . One or two cone penetrometer tests or cores could be taken within each uniform section. The thicknesses at these points could then be assumed to be representative of the entire section. It should be noted here that the MICHBACK program (See Table 5.14) includes an option to backcalculate the thickness as well as stiffness of one layer provided that the total number of layers is five or less. Minimum Pavement Laver Thicknesses for Backcalculation It should be noted that asphalt surface layer thicknesses of less than three inches cannot be reliably characterized with the Falling Weight Deflectometer data. It is standard practice in Texas, for instance, to assume a base value of 2069 MPa (300,000 psi) then apply the Asphalt Institute temperature correction formula. It is also standard practice to combine any pavement layer below the surface, and less than 3 inches in thickness, with another pavement layer as these cannot be reliably charac terized with the FWD. Backcalculation can describe a pavement layer's stiffness only to the degree to which that layer affects the deflec tions. Thin layers contribute only a small portion to the

5-139

overall deflection and as a result, the accuracy of their backcalculated values is reduced. 5.5.5 Relative Layer Stiffness Effects The technique of subgrade subdivision (including the use of a default stiff layer) was discussed under the preceding section. To some extent, the layer thickness discussion covers relative layer stiffness effects. However, the intent of this section is to emphasize that the layer stiffness (i.e., combination of thickness and modulus) needs to be rela tively significant (compared with other pavement compo nents) for it to influence the surface deflections. If this is not the case, then backcalculation approaches will not be successful in identifying the effect of the layer. As an ex ample, consider a 200 mm (8 in.) thick natural gravel base course. If this layer is placed on an average subgrade and surfaced with a chip seal, it is relatively stiff and backcal culation will easily evaluate the difference in modulus between the base and subgrade. On the other hand, if this base material occurs beneath a 406 mm (16 in.) PCC slab, it is not relatively stiff and it is unlikely that the backcalcu lation process will be able to reliably separate the contri bution of this layer from the subgrade effect. Figure 5.31 illustrates this point. The deflections on the surface treated pavement are sensitive to the value of the base modulus. In the case of the PCC slab, the base modulus does not appreciably affect the magnitude of the deflec tions or the shape of deflection basin. It would be quite difficult to derive a meaningful value for the base modulus in the case of the PCC pavement. This situation is further complicated by low load stress levels beneath the PCC, response would result in base moduli significantly differ ent than the material would exhibit under a flexible surface.

5-140

Deflection Basin Sensitivity Comparison

800 0

1000

1200

1400

1600

-200

-400
Deflection, Microns

-600
N ote t h a t varying th e base moduli fo r each o f th e tw o pavement s tru c tu re s from 2 0 6 to 412 and Z4 MPa, (3 0 . 6 0 . & 120 KSI) re s u lts in a profound change fo r th e flexible pavement b u t very little change fo r th e rigid pavement

25.4 mm (1 in.) Surface Treatm ent a t 2 0 6 7 MPa (3 0 0 K.SI) + 2 0 3 mm (0 in.) Granular Base

-800

-1000

4 0 6 mm (16 in.) PCC Slab <> 2 7 5 6 0 MPa S (4 .0 0 0 K.SI) + 2 03 mm (& in.) Granular Base

-1200
Load = 4 0 K (9 0 0 0 lbs.) . Plate Pressure = 551 KPa (5 0 PSI) . Subgrade Modulus = 52 MPa (7 5 0 0 PSI) .N

-1400
Deflection Radius, mm

Figure 5.31

Figure 5.31 shows an extreme example, but illustrates the point. Similar problems occur for many unbound base/subbase combinations. These materials often differ only in terms of gradation and indicator specifications and their moduli are relatively similar, so that their contribu tions to the deflection response are difficult to separate. Similarly, if the surfacing is made up of more than one AC layer, these should be considered as a single layer. There is generally not enough difference between the response of an asphalt concrete surfacing layer and an asphalt treated base to identify these layers.

5.5.6 Seasonal Effects Moisture Effects Moisture can and does have a profound effect on pave ment deflections and subsequent backcalculated moduli. Generally, one would think that higher moisture contents result in weaker pavements. However, an extensive study undertaken in Texas showed that, although subgrade and base moduli varied with season, the subgrade varying more than the base, no strong correlation could be estab lished between rainfall patterns and moduli variations. Ten asphaltic concrete sites in Texas were selected for the study. Five were near Abilene and five were near Brownsville. Generally, for the Brownsville sections, sub grade moduli were at their weakest during March, April, and May, a period of greater rainfall. However, the Abilene sections exhibited their weakest subgrades in Au gust, a period of little rainfall. This was probably due to cracking of the clay subgrades. In summary, pavement moduli are affected by moisture content. Reduced moduli may be a result of excessive moisture, or due to cracking as a result of insufficient moisture. It appears that rela tionships between moisture and pavement deflections will need additional study. Since base modulus is often a function of subgrade modulus, no relationship between base course modulus and moisture could be determined either.

5-142

For the aforementioned reasons, no correction factors have been developed to adjust either base or subgrade moduli with moisture for Texas conditions. Other regions show significant changes in deflection with moisture variations. Temperature Effects Normal deflection measurements should not be conducted when the subgrade is frozen. Temperature effects in Texas have been found to be pre dictable on higher type asphaltic pavements, i.e. deflec tions increase with temperature. However, for low vol ume roads (150mm (6") granular base + surface treat ment) it has been found that granular base moduli increase with increasing temperature. It is thought that the base expands with temperature causing the angular particles to lock up and become stiffer. However, this trend is de pendent on materials used and is not consistent. Sub grades have found to be insensitive to temperature. At this time, the asphalt concrete is the only pavement mate rial corrected for temperature, and the Asphalt Institute method is used. This is shown on Figure 3.23. Each agency should develop or select a method of temperature correction applicable to the materials and range of condi tions present in their State. For rigid pavements, the deflection at joints and cracks increase as the temperature decreases as joints open due to contraction of PCC slabs. As the temperature increases, the slabs expand and the deflections reduces as more contact occurs between the fractured faces of the slabs. Therefore, the load transfer at joints and cracks should be tested in the night or early morning hours, to obtain the critical load transfer. Corner testing in PCC slabs to detect voids should be avoided when the corners are curled concave upwards due to the temperature differ ential within the slabs. The ambient temperature as well as the range of temperatures during the season in which testing is mentioned should be considered when selecting a time for corner testing. Generally testing should be

5-143

avoided during the early morning hours. Deflections at the center of the slab are most affected during the mid day when the temperature at the top of the slab is higher than at the bottom. However, generally the effect of tempera ture on center slab testing is small. Freeze-Thaw Effects As discussed in SECTION 2, this usually refers to the thaw weakening observed during spring. The combina tion of low temperature and poor drainage often results in very poorly performing pavements under applied loads. For backcalculation applications the main point to keep in mind is that a varying depth stiff layer exists as the thaw ing front progresses and that critical damage locations can vary extremely rapidly. 5.5.7 Fixing Layer Moduli There are certain instances where it is extremely difficult to backcalculate layer moduli with any degree of certainty from typical deflection basin data. Possibly the most common of these involves a relatively thin surfacing layer, with approximately 75 mm (3 in.) generally considered the lower limit for backcalculation purposes. This limitation is related to the typical test geometry involving a 300 mm (12 in.) diameter plate, which provides inadequate resolu tion in surface shape for thin layers. In such cases a better estimate for surface layer modulus would be based on knowledge of the material type and test conditions. Thin surfacing usually consists of AC, and a temperature meas urement at the time of test can be used, with knowledge of the test load duration (usually 25-35 ms for an FWD) to estimate AC modulus using the techniques discussed in SECTION 2. A similar situation exists for PCC overlaid with AC. The AC overlay stiffness has relatively little effect on the deflection basin if the PCC slab is 150 mm or thicker (6 in. +), so that fixing the AC modulus based on temperature provides the best solution.

5-144

5.5.8 Rules o f Thumb (a) U se as few unknow n layers as possible to adequately define the problem . D o not try to separate layers that are likely to have a very similar m odulus. Also, consider relative stiffness to decide if a layer is sig nificant. F o r P C C pavem ents a tw o-layer system (PC C on subgrade) is likely to provide the best solution. A void attem pting to backcalculate m oduli for thin layers. U se seed m oduli and ranges that are consistent w ith test condition. C onsider such factors as tem pera ture, m oisture condition, pavem ent age and crack ing. U se a range o f 0.5 to 2.0 tim es the m ost p ro b able m oduli values. I f the program has the option o f estim ating seed m oduli from deflection basins, use it. G ather as m uch inform ation as possible for each test location. A s-built inform ation is rarely available and construction plans rarely provide an accurate depic tion o f as-built conditions. A ttem pt initial analyses w ith few iterations and w ide limits to identify possible solutions. In other w ords, three runs w ith three iterations each is probably better than starting the first analysis w ith 10 itera tions. E valuate the ou tp u t critically before proceeding. H igh erro r levels may suggest that the problem definition is incorrect. E ven w hen errors are low th ere may be problem s. U se m aterials inform ation to evaluate the feasibility o f the solution. S ubgrade m oduli are critical to the rest o f the analy sis. C heck for rigid layer and/or non-linearity.

(b)

(c)

(d)

(e)

(f)

(g)

(h)

Som e specific pointers are as follows: (a) Input D ata (seed m oduli, moduli limits, convergence criteria, etc.) Seed M o d u li: E n ter normally expected values for first run. These are:

5-145

2070 M P a (300 ksi) to 4140 M P a (600 ksi) (check tem perature!) A sphalt Stabilized B ase - 3450 M P a (500 ksi) to 6,900 M P a (1,000 ksi) (check tem perature!) G ranular B ase - 138 M P a (20 ksi) to 690 M P a (100 ksi) S ubgrade - 21 M P a (3 ksi) to 104 M Pa (15 ksi) P ortland Cem ent C oncrete - 27,600 M Pa (4,000 ksi) M oduli Limits: U pper range - tw ice the seed value L ow er range - half the seed value

A sp h a lt-

T he w ider the range the longer the M odulus pro gram takes to run. I f you have a 486/50 it only tak es a m inute o r tw o to com plete 200 o r 300 fo r w ard runs for a four layer pavem ent system. In this case you can be very liberal w ith the limits. H o w ever, if you are running on a 286/8 o r 286/12 it will tak e all day to com plete this m any runs. T herefore it will be necessary to proceed cautiously w ith very narrow limits, say a range o f plus o r minus 50% o f the seed value. Modulus ranges, practically

speaking, are a function of the computer power at your disposal.


C onvergence C riteria: C onvergence criteria you choose, like m oduli limits, are a function o f how m uch tim e you have to com m it to the backcalculation process. Occasionally, you may find it easy to achieve a predeterm ined convergence criteria. M ost often, how ever, you m ust rerun the problem m any tim es and accept the best fit even though it does not m eet the target criteria. Rule: Shoot for 2% per sensor but accept less if it looks as though the data is ill-behaved. (b) C om pensating laver effects M any tim es during the backcalculation process, the m odulus o f a given layer is obviously to o high. In o rd er to com pensate, the program assigns a low er than expected moduli for another layer. Rule: D is regard the com pensating layer effect if stresses or

5-146

strains are not to be com puted for those layers in question. I f the objective is to determ ine stresses or strains at the bo tto m o f a proposed overlay, the com pensating layer effect does not hinder the analy sis. H ow ever, if the rem aining life calculation involves subgrade strain criteria, the com pensating layer effect can adversely affect the results. (c) S tiff L aver E ffects B edrock, non-linear subgrades, and high w ater ta bles can give the appearance o f a stiff layer underly ing the subgrade. The presence o f bedrock affects the deflection basins in tw o ways: it decreases the m agnitudes o f the outer deflections as well as the slope o f the o u ter basin. This usually results in an inverted m odular ratio during the backcalculation process (e.g. asphalt m odulus low er than base m odulus). T he m agnitudes o f the deflections o f the o u ter sensors can be used to separate non-linear or w ater table effects from bedrock effects. Rule: If

the outer deflection (R =l,800 mm (72 in.)) is small (less than 25 microns (one mil)), you can be reasonably sure that bedrock is present. This is tru e for m ost cases although there may be a load re lated effect and relative deflection m agnitudes should be kept in mind. I f inverted m odular ratios are obtained for a flexible pavem ent during the backcalculation process and the o u ter sensor regis ters deflections in excess o f 25 m icrons for a 40 kN load, (1 mil at 9,000 lb. load), it is m ost likely a m anifestation o f nonlinearity in the subgrade soil and should be handled as recom m ended below.
(d) N onlinearitv Effects N onlinearity is mainly a characteristic o f subgrades and often show s up under m oderate to heavy FW D loads on thin pavem ents. N onlinearity implies that the m odulus o f the m aterial is dependent on, or changes w ith, the m agnitude o f the deviator stress (FW D load). T he m odulus o f the m aterial can either increase, o r decrease, w ith increasing stress. The

5-147

m ost com m on case is a subgrade w hich becom es softer w ith increasing stress. B ase m aterial can often exhibit stress hardening behavior, identifiable by a low deflection vs. load ratio as the load is increased. Subgrade nonlinearity results in high deflection basin curvature and deflection m agnitudes in the im m ediate vicinity o f the load plate. W hen a layered elastic program is used to characterize pavem ent layer m oduli, the only w ay it can replicate the steepness and m agnitude o f the deflection basin near the load is to invert the moduli o f the paving layers (soft over hard). Rule: W hen encountering

nonlinearity, it is often helpful model the upper subgrade as a separate pavement layer, 300 mm (12) or thicker, and allow its stiffness to decrease as needed to obtain reasonable deflec tion basin matches.
It should be noted here that if a subbase, or sub grade is highly non-linear, its value will vary greatly betw een a PC C slab that is intact and one that is rubblized. This is because the deviator stresses un der an intact slab are m uch low er than under a slab w hich has been broken up. (e) L aver Thickness Effect

In most cases, the following should be assigned moduli values in the backcalculation process. If their moduli are backcalculated, the results should be evaluated with great care: 1. Any surface layer under 76 mm (3) in thickness 2. Any pavement layer directly under a PCC slab whose thickness is less than or equal to the slab
These layers do not appreciably affect the shape o f the m easured deflection basin or the m agnitude o f the deflections. C onsequently, during the backcal culation process, these moduli may vary wildly w ith little effect on the degree o f fit.

5-148

(f)

Sensor Spacing

Rule: The softer the pavement the closer the sensor spacing should be.
A spacing ofO , 200, 300, 450, 600, 900 and 1500 mm (0, 8, 12, 18, 24, 36 and 60 inches) is often used for light pavem ents such as surface treated roads w ith thin base courses (150 to 300 mm (6 to 12 )

Corollary: The stiffer the pavement the further the deflection sensor spacing should be.
A spacing ofO , 300, 600, 900, 1500 and 1800 mm (0, 12, 24, 36, 48, 60, and 72 inches) w orks well on a Portland cem ent concrete or heavily stabilized asphalt pavem ent. E xperience has show n that life is m uch easier for the E ngineer and the FW D o p erato r if one sensor spac ing configuration is adopted and used for all pave m ent types. This elim inates the possibility o f e rro neous sensor spacings being stored in the FW D data file, and frees the op erato r from having to m ove sen sors. A com prom ise spacing can be adopted w hich accom m odates both pavem ent types. F o r example, a spacing ofO , 200, 300, 600, 900, 1200 and 1800 mm (0, 8, 12, 24, 36, 48 and 72 inches) might represent a reasonable com prom ise. This provides deflection inform ation close to the load plate for light pavem ents while providing a good spread for rigid pavem ents. The E ngineer should try a variety o f settings and select that which best suits his par ticular needs. It should be noted, how ever, that som e backcalculation program s require a certain sensor spacing and this should be taken into consid eration as well. (g) Seasonal and T em perature Effects

Rule: The only predictable aspect of seasonal and temperature effects is that they are unpre dictable.
It has been generally accepted in the industry that subgrade m oduli becom e stiffer during periods o f low rainfall and less stiff w ith high rainfall. W hile true in som e cases, dram atic low ering o f the sub grade m odulus has been observed during sum m er

5-149

low rainfall periods, possibly due to cracking o f clay m aterials. H igher rainfall can cause w ater tables to rise, result ing in an apparent decrease in subgrade stiffness. G ranular bases have been observed to stiffen w ith increasing tem perature during the day, possibly due to higher confining pressures as a result o f m aterial expansion and subsequent aggregate lockup. A sphalt concrete pavem ent (an engineered m aterial) behaves predictably under tem perature variations, becom ing softer at high tem peratures and harder at low er tem peratures. (h) Fixing L ayer M oduli (1) I f an AC surface layer is less than 3 inches thick the m odulus should be fixed according to som e tem perature m odulus relationship. On a PC C pavem ent, w hen no distresses are present, and the deflections are varying widely, the PC C m odulus should be fixed at 3,000 to 6,000 ksi.

(2)

(i)

C om bining Lavers P avem ent layers should be com bined when: (1) Layers are o f similar m aterial, i.e. asphalt surfacing and asphalt stabilized base Portland cem ent concrete and cem ent stabilized base Layers are o f similar stiffness (as a result o f previous backcalculation run)

(2)

5-150

5.6 R E L IA B IL IT Y A N D E R R O R S IN D E F L E C T IO N A N A L Y SIS

5.6.1 In troduction This section will describe the various sources o f m eas urem ent errors and their effects on back-calculation. It has been established in previous sections that layer m oduli can be back-calculated provided that a num ber o f surface deflections are m easured, that the force causing the deflections is know n, and that the layer thicknesses are know n. B ut w hat if there are errors in m easuring these input param eters? W hat are the consequences o f such errors on the back-calculated moduli? 5.6.2 T ypes o f M easurem ent E rrors M easurem ent errors can be subdivided into tw o general categories: system atic errors and random errors. The system atic erro r is also som etim es called a "bias" or "offset." S y ste m a tic e rro rs are internal to the m easurem ent device, and they can only be detected by com parison to a separate, independently calibrated m easurem ent system. F o r instance, suppose that a m anufacturer m ade a m eter stick by em bossing m arks at one centim eter intervals on a piece o f green w ood. As the w ood ages, it dries out and shrinks. F o r this hypothetical exam ple let us suppose that the shrinkage is uniform , linear, and that there is no w arping o f the w ood. [A dm ittedly, this is rather ideal!] I f the shrinkage am ounted to five percent o f the original length o f the m eter stick, then every m easurem ent w ould be in erro r (i.e., short) by five percent. I f the stick w ere used to m ake m arks at one m eter intervals, the m arks w ould, in fact, be only 95 centim eters apart. To detect this erro r it w ould be necessary to com pare the m easure m ents using a device o f know n accuracy. A nother m eter stick printed on green w ood w ould not suffice. A general rule is that the reference device should be calibrated to an

5-151

accuracy that is ten tim es better than the device under com parison. This is not alw ays achievable, but certainly the reference device m ust be m ore accurate than the device that is being calibrated. To correct for the system atic error, provided that it is uniform and linear, an adjustm ent factor can be deter mined. C ontinuing the preceding exam ple, presum e that the com parison betw een the m eter stick and the reference device (perhaps an electronic distance device) establishes that the system atic error is exactly 5.00 percent on the low side. T herefore, to correct the m eter stick m easure m ents its indicated readings could be multiplied by 0.95, leading to the equation:
R c = 0.950 * R m

w here R is the corrected reading, and R, is the indicated reading from the m eter stick.

Random errors are often associated w ith the repeatability and least-count o f a m easurem ent device. They can be detected by m aking repeated m easurem ents using the sam e m easurem ent system. In so doing it will be noticed that there are slight, but noticeable, differences from one reading to the next. I f there seem s to be no general trend to the readings, neither upw ard nor dow nw ard, then the readings are said to be stochastic, or random .
T he relationship betw een the m easured value and the ran dom erro r is given by the follow ing equation: M easured V alue = E xact V alue + R andom E rro r T he random errors can be either positive o r negative, and they are spread out random ly about the exact value. P re sum ably each tim e a m easured value is obtained it is slightly different than the previous m easurem ent. This is due to the fact that the random erro r changes from one m easurem ent to the next. T he exact value is unchanging. W e w ould like to know the exact value, but w e cannot

5-152

measure it directly, so we can only estimate it from the measured values. To illustrate the concept o f random errors, presume that we had the following ten measurements o f the diameter o f a ground and polished shaft, all taken at exactly the same point (to rule out point to point variability) using a machinist's micrometer which is capable o f measuring up to 200 mm with a least-count o f 0.01 millimeter. The data may be presumed to be random because they are neither steadily increasing nor decreasing. [There are sta tistical tests that can be used to confirm that the data are random, but they will not be discussed herein.] Since each measurement has a small random error in it, no single observation can be taken to be the exact diameter o f the shaft. Reading Number
1 2

Shaft Diameter, millimeters 100.51 100.48 100.53 100.55 100.49 100.45 100.51 100.50 100.48 100.47

3 4 5
6

7
8

9
10

H owever, an excellent estimate o f the shaft diameter could be calculated by determining the average o f the ten readings. In this example the average is 100.497 millime ters, or perhaps 100.50 millimeters (considering signifi

5-153

cant figures). The average is calculated according to the following equation: x = I Xi n where n is the number o f observations, Xj is a given obser vation, and the symbol E signifies a summation o f all o f the observations. Thus, in plain English, you would add up the ten observations o f shaft diameter in the table above, and divide by 10, the number o f observations. The spread o f the random errors contained in a set o f individual measurements can be determined by calculating the standard deviation, a. The standard deviation is more or less the root mean square o f the random errors. When the standard deviation is large, one or more o f the random errors are large, and vice versa. The standard deviation is calculated according to the fol lowing equation:

where the terms are as previously defined. The standard deviation o f the above ten measurements is 0.0294 mil limeters. I f we use the standard deviation to represent the random error, then it could be said that in this example the random measurement error is 0.03 millimeters (rounded to tw o decimal places). The standard deviation defines a range within which the exact shaft diameter could be expected to fall about 68 percent o f the time. It would fall within tw o times the standard deviation 95 percent o f the time. Presuming that you knew the standard deviation o f the shaft diameter was 0.03 mm from previous studies, then if you made only a single measurement c f the shaft diameter, and if you read 100.51 millimeters, there is a 95 percent probability that the true diameter is 100.51 0.06 mm, or somewhere between 100.45 and 100.57 millimeters.

5-154

There is a big advantage o f making multiple m easure ments. Doing so will result in a more accurate estimate o f the exact value (the mean), and reduce the size o f the uncertainty band. The standard deviation o f the mean, a m , can be computed from a knowledge o f the standard deviation o f a single measurement according to: a vn where all terms remain as defined previously. The mean, or average, in the example, has been reported earlier as 100.497 millimeters. The standard deviation, a , is 0.029. The standard deviation o f the mean can be computed by dividing a by the square root o f 10 (the number o f obser vations), to get a m = 0.0093 millimeters. Thus it can be said that there is a 95 percent probability that the true shaft diameter is 100.497 0.019 millimeters, or between 100.48 and 100.52, which is the average plus or minus tw o times the standard deviation o f the mean. [Statistical purists would point out that since there are only ten measurements involved, we should use Student's t distribution instead o f the normal distribution to deter mine the probability. The net effect would be to reduce the probability from 95 percent to 92 percent. We will not w orry about the small stuff.] From the original set o f ten readings it would at first appear that the shaft diameter could be anywhere from 100.45 to 100.55 millimeters. We determined from the data that the standard error o f a single measurement was about 0.03 millimeters. By averaging the readings we arrived at a better estimate o f the exact value for the shaft diameter, namely 100.497 millimeters. And because ten readings were involved, each with a small amount o f ran dom error in the measurement, we conclude that there is a 95 percent probability that the exact value is 100.497 0.019 millimeters. In summary, to minimize the effect o f system atic errors it is necessary to perform a calibration using an independent

5-155

measurement system. This removes the bias, or offset, attributable to the systematic error. To minimize the effect o f random errors it is necessary to make multiple measurements. The average is a better value than any single reading, and the uncertainty due to the random error is reduced by the square root o f the number o f observations. Thus when four observations are averaged, the error band is reduced by half. 5.6.3 Sources o f Errors in Back-calculation Each o f the many input parameters that are required for back-calculation have been described in previous sections. Potentially each input param eter is a separate source for error. But some errors are more significant than others. For instance it is likely that there will be a high degree o f uncertainty regarding the seed moduli that are used to begin the back-calculation. However, if the backcalculation com puter program is good, such "errors" will have very little influence on the back-calculated moduli. It has also been noted previously that the selection o f Poisson's ratio for most materials has relatively little influ ence on the back-calculated moduli. One important input param eter that is difficult to know with certainty is the layer thickness. For one thing, con struction variability from point-to-point makes a knowl edge o f layer thickness very difficult to obtain, although the technology o f ground-penetrating radar may poten tially offer some hope for future improvement. Today, layer thickness is typically determined by pavement coring methods. However, FWD testing can be done at many more points on a pavement than it is practical (or afford able) to do coring. Thus it is almost always the case that the actual layer thicknesses at a particular test point will differ slightly from the data that is entered for backcalculation. A nother source o f error is the FWD measurements. Both the load and the deflections have several possible kinds o f measurement errors. Typical specifications for FW Ds require an accuracy within 2 percent for the indicated

5-156

load and deflections. This is a systematic error which does not vary from time to time, but its magnitude will vary from one transducer to another, depending on the accuracy o f calibration o f the transducer. It can be reduced by performing reference calibration according to the procedure developed by the Strategic Highway Research Program (SHRP) [5.401. This will be discussed in section 5.6.5. A second source o f FW D measurement error comes from the analog-to-digital conversion in the data acquisition system, typically with a range o f not more than 2 bits (a "bit" is digital measure o f voltage, nominally about 5 mil livolts). This is a random error that continuously varies within its range at the speed o f the a/d converter (e.g., from one reading to the next), but the range is equal for all transducers. It can be reduced by making multiple readings and averaging the results. A third source o f error is mechanical in nature, due to an initial lack o f seating o f the points o f contact o f the deflection sensors. M ost pavement surfaces are rough, and there are grains o f dirt and sand on the surface. It usually takes only one or two drops o f the FW D mass to cause the deflection sensors to move into firm contact with the pavement. This random error can be minimized, if not eliminated, by performing an extra tw o or three drops after the plate is lowered at each test location with out recording the data. 5.6.4 Effects o f Errors on Back-calculated Moduli The consequence o f the systematic error in the indicated load is relatively small. If the load is registered tw o per cent higher than it really is, the back-calculated moduli will come out tw o percent high. The percentage effect on the modulus is the same for each layer. The random error is o f even less concern. One bit typically corresponds to 45 N (10 pounds) or so, thus the accuracy o f a load measured as 27 kN (6000 pounds) would ordinarily be limited by the systematic error, not the random error.

5-157

The consequence o f errors in the indicated deflections is o f greater concern. For a deflection sensor, one bit usu ally corresponds to 1 micron (0.04 mils). Therefore the random error o f an indicated deflection (represented by the standard deviation) is 2 microns. The center deflec tion on a deflection basin, which might be 500 microns (20 mils) in amplitude for a medium-strength asphalt pavement, could potentially have an error as large as 10 microns (due to a 2 percent systematic error). The accu racy o f an outer deflection, which might have an ampli tude o f 50 microns or less, would be limited by the ran dom error. I f the systematic error is 2 percent, then any deflection that is measured to be less than 100 microns would have its accuracy controlled by the 2 micron ran dom error. The effect o f the random deflection error on backcalculated moduli was investigated by Irwin, Yang and Stubstad [5.41], Using elastic layer theory, they calcu lated the deflection basin for a medium strength pavement, simulating the data that might be obtained with an FWD. They used a normally-distributed random number genera to r to modify the deflection data, simulating the random error, assuming a standard deviation o f 2 microns. Thirty deflection basins were produced in this fashion, and the data sets w ere processed using a back-calculation program to determine the layer moduli. Since the moduli that w ere used to generate the original deflection basin w ere known, they could be compared to the thirty sets o f back-calculated moduli. The results are summarized in T ab le 5.19,

5-158

Table 5.19. Effect of 2 Micron Random Measurement Error on Back-Calculated Pavement Layer Moduli [5.411 Layer Layer Thickness, in 3 6 12
oo

Original Modulus, psi 300,000 45,000 21,000 7,500

Range of BackCalculated Moduli, psi 196,000 to 426,000 32,300 to 59,900 18,700 to 25,000 7,390 to 7,670

1 2 3 4

The data in T able 5.19 show that the influence o f the ran dom deflection error is greatest for layer 1 and least for layer 4. The effect on the back-calculated subgrade modulus was essentially nil, while the error in the modulus o f the surface layer am ounted to 35-45 percent. These results w ere based solely on the 2 micron random error. I f a 2 percent systematic error had been incorporated also, the effect would have been even more dramatic. Irwin, et al, also studied the effect o f uncertainties in pavement layer thicknesses. Assuming that the layer thickness was measured to the nearest quarter-inch (6 mm), and assuming a scenario where the in situ variability o f the surface (asphalt) layer was 0.25 inch (6.4 mm), the variability o f the base layer was 1.0 inch (25 mm), and the variability o f the subbase was 1.5 inches (38 mm), the thickness o f each layer was allowed to vary ran domly. The outcom e was very comparable to that shown in T ab le 5.19 for random deflection errors. Thus the authors concluded that the random variability o f layer thickness has about the same effect on back-calculated moduli as the 2 bit random error o f deflection m easure ments. While the consequence on moduli was very noticeable, when the various sets o f moduli w ere used to calculate overlay thicknesses the effect was greatly dimin-

5-159

ished, amounting to only about 0.10 inches (2.5 mm) o f asphalt concrete. 5.6.5 Calibration o f FW Ds Procedures for the calibration o f FW Ds were developed by the Strategic Highway Research Program (SHRP) as a part o f their Long-Term Pavement Performance study [5.401. SHRP assigned four FW Ds to do pavement deflection tests in their respective regions o f the country, and it was important to calibrate them so that it would not m atter which unit tested any given pavement section. Specifically, the objective was to remove any systematic error, or "bias." The earliest SHRP efforts to develop a calibration proce dure for FW Ds were carried out at the research laboratory o f the Indiana Departm ent o f Transportation in the fall o f 1988 [5.42], Purdue University advised and assisted the effort. A prototype instrumentation system, involving a com puter-based data acquisition system, a reference load measurement system, and a reference deflection measure ment system, was developed. The four SHRP Dynatest FW Ds w ere run through reference calibration o f the load cell and each o f the seven deflection sensors. From this pilot study several things were learned [5.42]: It is feasible to do reference calibration o f FW Ds, but improvements in the equipment, particularly the reference load and deflection measurement systems, were necessary. It is necessary to conduct the calibration indoors so that the equipment and the operators are protected from the weather. The reference deflection measurement system must be isolated from transient vibrations generated by the FW D. A special test pavement must be constructed for conducting the calibration tests to assure that the
I

5-160

deflections are large enough to get beyond the range of the 2 micron random error and into the range o f the systematic error. A standardized procedure for calibration o f the FW Ds must be developed so that calibration could be carried out regionally. After calibration, the four FW Ds yielded data from specified test points that was highly comparable. Before calibration the results were within the specification tolerances, but the deflection data was statistically significantly different.

The latter finding is a result o f the fact that the m easure ments from any one FWD are very highly repeatable. Therefore a very small difference in the means o f the measurements from tw o different FW Ds can be statisti cally significant. D ue to this fact it was determined by SHRP that it would be desirable to develop an FWD calibration protocol and to establish four regional FWD calibration centers. Vari ous state departments o f transportation w ere contacted to see if they would be willing to host and to operate the calibration centers. As a result, four regional calibration centers are now operating under the management o f the Pennsylvania D epartm ent o f Transportation, the M inne sota D epartm ent o f Transportation, the Nevada D epart ment o f Transportation, and the Texas D epartm ent o f Transportation. The SHRP FWD reference calibration protocol [5.40] calls for five drops at each o f four load levels. One trans ducer, either the FW D load cell or a deflection sensor, is calibrated at a time. The twenty data pairs (FWD system reading as abscissa, reference system reading as ordinate) are subjected to a linear regression that is forced through zero. The resulting slope is the reference calibration fac tor. It is a multiplier which, when applied to the FWD reading, yields a reading that is corrected to agree with the reference system.

5-161

Each sensor (typically the load cell plus seven deflection sensors) gets an individual reference calibration factor. For convenience, these numbers must be entered into the FW D software, and thus it is necessary that there be a place to put the calibration factors. The FW D manufac turer supplies the software to make this possible. After the reference calibration factors have been entered into the FW D software, then the FWD deflection sensors are run through a relative calibration. This test is per formed twice, and the results are averaged. Relative cali bration involves stacking the geophones in a stand and subjecting them to a series o f 35 drops. The ratio o f the average deflection for all seven sensors to the average deflection for an individual sensor is the relative calibra tion factor. An analysis o f variance program is used to evaluate the data. The relative calibration factor for each sensor is multiplied times the corresponding reference calibration factor to arrive at the final calibration factor. These final values are then entered into the FW D soft ware. Since the systematic error is specified to be less than 2 percent, it is expected that the calibration factors will be between 0.98 and 1.02. With relatively new equipment this has generally been found to be the case. It usually takes about one day to complete the testing, perform the calculations and enter the results [5.42], This includes the time it takes to remove the deflection sensors from their holders on the FW D and return them to posi tion. This time requirement may vary, o f course, depend ing on the preparedness o f the FW D and the experience o f the calibration center operator. Copies o f the calibration protocol and o f the accompany ing FW DREFCL and FW DCAL software are available from the FHW A LTPP Division which is located at the Turner-Fairbank Highway Research Center, 6300 G eorgetow n Pike, M cLean, VA.

5-162

SHRP has established a policy that the relative calibration shall be repeated monthly [5.40], This makes it possible to determine if an individual geophone has failed, in which case it can be replaced with a spare. Thereafter the refer ence calibration must be repeated. A ccording to the SHRP protocol, reference calibration must be performed annually [5.40], Several studies have been conducted w here a SHRP FW D was given a com plete reference calibration more frequently, and the results indicate that once per year is sufficiently often. The pro cedure has not been in place long enough to verify this fully, however. 5.6.6 Procedures to Minimize Errors The previous material in SECTION 5.6 has established that there are a variety o f sources o f errors that can have an effect on the moduli that are back-calculated from deflection data. To summarize, the following are a series o f suggestions that could be implemented to minimize these effects. Calibrate the FW D annually using the SHRP reference calibration protocol. Be sure that all sources o f sys tematic error have been eliminated as thoroughly as possible. Calibrate the FWD monthly using the SHRP relative calibration protocol. This will enable the detection o f any sensor that is steadily changing, which might indi cate the need for reference calibration or replacement. Perform tw o or three seating drops at each test point, for which the data do not need to be recorded. This will eliminate seating errors. Perform multiple drops at each drop height (preferably four or more drops), and calculate an average deflec tion for each sensor. This will reduce the uncertainty o f the load and deflection measurements due to the random errors. For this to be a valid approach the

5-163

edge should be included in a general frame w ork that can be modified easily as better understanding o f the problems develop or new research findings emerge. Such a careful verification process is necessary before the current "prototype" systems can become production systems. They should also be easily accessed by any engineer who attem pts to perform backcalculation, so that both the procedures and results would be standardized. The most widely used method o f representing domain knowledge in an expert system is the use o f production rules. In this method, knowledge is decomposed to many IF (condition) THEN (action) statements. For example, IF the pavement surface tem perature is greater than 90 F AND the asphalt layer is not aged, TH EN the asphalt concrete modulus should be less than 600 ksi. Many such facts can be included in the expert system to assist the designer in coming to the most cost effective solution.

5-165

SECTION 5.0 REFERENCES

H orak, E., "The Use o f Surface Deflection Basin M easurements in the M echanistic Analysis o f Flexible Pavements," Proceedings, Vol. 1, Sixth International Conference Structural Design o f Asphalt Pavements, University o f Michigan, Ann Arbor, Michigan, U S A 1987. Newcomb, D. E., "Development and Evaluation o f Regression M ethods to Interpret Dynamic Pavement Deflections," Ph.D. Dissertation, Departm ent o f Civil Engineering, University o f Washington, Seattle, Washington, 1986. AASHTO, "AASHTO Guide for Design o f Pave ment Structures," American Association o f State Highway and Transportation Officials, W ashington, D.C., 1986. Chou, Y. J., Uzan, J., and Lytton, R. L., "Backcalculation o f Layer Moduli from N ondestructive Pavement Deflection D ata Using the Expert System Approach," N ondestructive Testing o f Pavements and Backcalculation o f Moduli, ASTM STP 1026, American Society for Testing and Materials, Philadelphia, 1989, pp. 341 - 354.

Smith. B. E. and Witczak, M. W., "Equivalent Granular Base Moduli: Prediction," Journal o f Transportation Engineering, Vol. 107, TE6, American Society o f Civil Engineering, Novem ber 1981. Shook, J. F., Finn, F. N , Witczak, M. W., and Monismith, C. L., "Thickness Design o f Asphalt Pavements The Asphalt Institute M ethod," Proceedings, Fifth International Conference on the Structural Design o f Asphalt Pavements, Vol. I, The Delft University o f Technology, The Netherlands, August 23-26, 1982, pp. 17-44.

5-166

The Asphalt Institute, "Research and Development o f The Asphalt Institute's Thickness Design M anual (M S-1) Ninth Edition," Research Report No. 82-2, The Asphalt Institute, College Park, Maryland, August 1982. D arter, M .I., Elliott, R.P., and Hall, K.T., "Revision o f AASHTO Pavement Overlay Design Procedures," Preliminary Draft Final R eport, Project 20-7/39, National Cooperative Highway Research Program, Transportation Research Board, Washington, D.C., June 1991. Ullidtz, P., "Pavement Analysis", Elsevier, 1987. Lytton, R.L., "Backcalculation o f Pavement Layer Properties," N ondestructive Testing o f Pavements and Backcalculation o f Moduli, ASTM STP 1026, American Society for Testing and Materials, Philadelphia, 1989, pp. 7-38. Lambe, T.W ., and Whitman, R.V., Soil Mechanics. John Wiley and Sons, N ew York, 1969. Van Cauwelaert, F.J., Alexander, D R., White, T.D., and Barker, W .R., "Multilayer Elastic Program for Backcalculating Layer Moduli in Pavement Evaluation," N ondestructive Testing o f Pavements and Backcalculation o f M oduli, ASTM STP 1026, American Society for Testing and M aterials, Philadelphia, 1989, pp. 171-188. Bush, A.J., III, "Nondestructive Testing for Light Aircraft Pavements, Phase II," Research Report FAA-RD-80-9-II, Federal Aviation Administration, U.S. Departm ent of Transportation, Washington, D.C., 1980. Lenngren, C.A., "Relating Bearing Capacity to Pavement Condition," Ph.D. Dissertation, Royal Institute o f Technology, D epartm ent o f Highway Engineering, Stockholm, Sweden, 1990. M ahoney, J.P., Coetzee, N.F., and Stubstad, R.N., "A Performance Comparison o f Selected

5-167

Backcalculation Com puter Programs," N ondestructive Testing o f Pavements and Backcalculation o f Moduli, A STM STP 1026, American Society for Testing and Materials, Philadelphia, 1989, pp. 452-467. 5.16 Rohde, G.T., and Scullion, T., "MODULUS 4.0: Expansion and Validation o f the M ODULUS Backcalculation System," Research R eport No. 1123-3, Texas Transportation Institute, Texas A&M University System, College Station, Texas, N ovem ber 1990. Hossain, A.S.M ., and Zaniewski, J.P., "Detection and Determination o f Depth o f Rigid Bottom in Backcalculation o f Layer Moduli from Falling Weight Deflectometer Data," Transportation Research Record No. 1293, Transportation Research Board, Washington, D.C., 1991. Neville, A.M., Properties o f Concrete. John Wiley and Sons, N ew York, Second Edition, 1973, p. 313, 318, 319. Kosm atka, S., and Panarese, W .C., Design and Control o f Concrete M ixtures. Portland Cement Association, Skokie, Illinois, 13th Edition (1988), p. 48. Raja, Z. I. and Snyder, M. B., "Factors Affecting D eterioration o f Transverse Cracks in Jointed Reinforced Concrete Pavements," Transportation Research Record 1307, Transportation Research Board, 1991, pp. 162-168. Ioannides, A. M ., Thompson, M. R., and Barenberg, E. J., "Finite Element Analysis o f Slabs on Grade Using a Variety o f Support M odels," Proceedings, 3rd International Conference on Concrete Pavement Design and Rehabilitation, Purdue University, W est Lafayette, Indiana, 1985, pp. 309-324. M ajidzadeh, K., lives, G.J., and Sklyut, H., "RISC A Mechanistic M ethod o f Rigid Pavement Design," Proceedings, 3rd International Conference on Concrete Pavement Design

5.17

5.18

5.19

5.20

5.21

5.22

5-168

and Rehabilitation, Purdue University, W est Lafayette, Indiana, 1985, pp. 325-339. 5.23 Shahin, M. Y., "Use o f the Falling W eight Deflectom eter for the N ondestructive Deflection Testing o f Jointed Airfield Pavements," Proceedings, 3rd International Conference on Concrete Pavement Design and Rehabilitation, Purdue University, W est Lafayette, Indiana, 1985, pp. 549-556. Anderson, M., "A Database M ethod for Backcalcu lation o f Composite Pavement Layer Moduli," N ondestructive Testing of Pavements and Backcalculation o f Moduli, ASTM STP 1026, American Society for Testing and Materials, Philadelphia, 1989, pp. 201-216. Tia, M., Eom, K.S., and Ruth, B., "Development o f the DBCONPAS Com puter Program for Estimation of Concrete Pavement Param eters from FW D Data," N ondestructive Testing o f Pavements and Backcalculation o f Moduli, ASTM STP 1026, American Society for Testing and Materials, Philadelphia, 1989, pp. 291-312. Foxworthy, P. T. and D arter, M. I., "Preliminary Concepts for FW D Testing and Evaluation o f Rigid Airfield Pavements," Transportation Research Record 1070, Transportation Research Board, 1986, pp. 77-88. Edwards, W. F., Green, R. L., and Gilfert, J., "Implementation o f a Dynamic Deflection System for Rigid and Flexible pavements in Ohio," Report No. FHW A/OH-89/020. Greer, W. C., "Seasonal Variation in Joint Efficiency o f Dowelled Concrete Pavements," 3rd International Conference on Bearing Capacity o f Roads and Airfields, Trondheim, Norway, 1990, pp. 65-74. Larsen, T. J., "Test Procedures Rigid Pavements Using Falling W eight Deflectometer," WD. Torbjorn J. Larsen, Trondheim III, 3rd International Conference on Bearing

5.24

5.25

5.26

5.27

5.28

5.29

5-169

Capacity o f Roads and Airfields, Trondheim, N orway, 1990, pp. 139-147. Majidzadeh, K., "A M echanistic Approach to Rigid Pavement Design," Concrete Pavements, Ed. A. F. Stock, 1988, Elsevier Applied Science, pp. 11-56. Ioannides, A. M., Thompson, M. R., and Barenberg, E. J., "W estergaard Solutions Reconsidered," Transportation Research Record 1043, Transportation Research Board, 1985, pp. 13-22. Foxworthy, P., "Concepts for the Development o f a N ondestructive Testing and Evaluation System for Rigid Airfield Pavements," Ph.D. Dissertation, Departm ent of Civil Engineering, University o f Illinois at UrbanaChampaign, 1985. Ullidtz, P. and Stubstad, R. N , "Structural Evalua tion o f Highway and Airfield PCC Pavements Using the Falling W eight Deflectometer," Proceedings, 3rd International Conference on Concrete Pavement Design and Rehabilitation, Purdue University, W est Lafayette, Indiana, 1985, pp. 567-574. D arter, M. I., Barenberg, E. J., and Yrjanson, W. A., Joint Repair M ethods for Portland Cement Concrete Pavements, National Cooperative Highway Research program report 281, Transportation Research Board, 1985. Section 5, N ondestructive Evaluation and Strength ening o f Existing Airfield Pavements. Hall, K. T. and Mohseni, A., "Backcalculation o f Asphalt Concrete-Overlaid Portland Cement Concrete Pavement Layer Moduli," Transportation Research Record 1293, Transportation Research Board, 1991, pp. 112-123.

5-170

5.37

Ioannides, A. M., "Dimensional Analysis in N D T Rigid Pavement Evaluation," Transportation Engineering Journal, ASCE, Vol. 116. Ioannides, A. M., Barenberg, E. J., and Lary, J. A., "Interpretation of Falling weight deflectom eter Results Using Principles o f Dimensional Analysis," Proceedings, 4th International Conference on Concrete Pavement Design and Rehabilitation, Purdue University, W est Lafayette, Indiana, 1989, pp. 231-241. Losberg, A., "Structurally Reinforced Concrete Pavements," Doktorsavhandlingar Vid Chalmers Tekniska Hogsko;a, Goteborg, Sweden, 1960. "SHRP FWD Calibration Protocol", Long-Term Pavement Performance Group, Federal Highway Administration, M cLean, VA, April 1993. Irwin, L.H., Yang, W.S. and Stubstad, R.N., "Deflection Reading Accuracy and Layer Thickness Accuracy in Back-Calculation o f Pavement Layer Moduli", Special Technical Publication no. 1026, American Society for testing and Materials, Philadelphia, 1989, pp. 229-244.

5.38

5.39

5.40

5.41

5.42

Irwin, L.H., Cumberledge, G., and Henderson, B., "Implementation o f a Calibration Procedure for Falling W eight Deflectometers", Proceedings, 3rd International Conference on M anaging Pavements, Transportation Research Board, Washington, D.C., in press. Irwin, L.H., "M ODCOM PI User's Guide", Cornell University, 1981. Teller, L.W ., Cashell, H .D ., "Performance o f D ow elled Joints U nder Repetitive Loading", Bulletin 217, HRB, National Research Council, W ashington D.C., 1959. Voight, G.F., Darter, M .I., Carpenter, S., "Field Performance Review o f Unbonded Jointed

5.43

5.44

5.45

5-171

Concrete Overlays", TRB TRR #1227, Transportation Research Board, 1989.

5-172

SECTION 6.0 BACKCALCULATION PROGRAMS 6.1 OVERVIEW To this point in the course, we have discussed the motiva tion for the backcalculation o f elastic moduli in terms o f the pavement design process, the concept o f elastic modulus and how it is determined, types o f nondestructive testing devices, means o f interpreting the results o f deflection tests, and the fundamentals of the backcalculation approach. This portion o f the course is where the students gain "hands-on" experience with the four backcalculation pro grams discussed in SECTION 5, BOUSDEF, EVERCALC, MODULUS and MODCOMP. FWD deflection basins from three actual pavement sections will be used by the class to backcalculate elastic layer moduli in a group approach. Actual FWD data files from several other projects are included for use in the workshop, as described in Section 6.4. These include AC and PCC data, as well as thawing sections, composite sections and an overlaid rubblized PCC structure. A brief summary o f the backcalculation programs is presented to acquaint the student with their basic princi ples and operations. The users manuals for these programs are given in Appendices A, B, C and D. 6.1.1 Computer Programs for Backcalculation A rational model is required to determine the in situ material properties o f a multi-layered pavement system from the surface deflection data. Because o f the complex nature o f the process, many computer programs have been developed to backcalculate the pavement layer moduli. These programs are generally based on either layered elastic or finite element theories. These programs include many different features. The features which are common in nearly all computer programs is shown in Figure 5.19 and can be described as follows:

6-1

a.

Input to Backcalculation Programs

(i)

Seed Moduli These are the starting or assumed initial values o f the layer moduli. In some methods, these are generated from the measured deflections or the regression equations. The first estimation (or range) o f the moduli should be based on the characteristics of materials in each pavement layer, and the local environmental conditions as they have consid erable effect on elastic modulus o f common pavement materials. For example, AASHTO suggests that the spring thaw modulus for sub grade may be as low as 20 to 50% o f the summer modulus. Similarly, temperature de pendency o f asphalt materials is well known. Subgrade modulus may be estimated based on soil classification, environmental conditions (rainfall, depth o f water table, and drainage situation) and material density. Range o f moduli for asphalt concrete, granular base materials, and stabilized materials can also be selected from the available information. Whichever source is used care must be exer cised when selecting a seed moduli for back calculation or when interpreting the results. During the iteration, the layer moduli are gen erally adjusted within the predetermined range to ensure that the modulus for each individual layer remains within the assumed modulus range for that layer. (Possible values o f pave ment materials moduli are noted in the typical pavement materials table shown in Section 2.5.2.)

6-2

(ii)

Poisson's Ratio Assumed values o f Poisson's ratio are used in nearly all analysis methods as any small vari ations in Poisson's ratios generally do not have significant effect on the backcalculated moduli.

(iii) Pavement Thicknesses Layer thicknesses including depth o f the sub grade are generally obtained from the historic data or by coring. When using either o f the methods care should be taken to assure that accurate layer thicknesses are obtained, oth erwise the results will not be representative. Research is currently underway to determine the layer thicknesses from the deflection data. Some progress has been made in this area. One other nondestructive technique, ground penetrating radar, has shown the ability to de termine layer thicknesses within tolerances o f V inch which is considered to be fairly rea 2 sonable accuracy. Radar can also be used to locate reinforcing steel, dowel bars, and voids. Radar surveys can be carried out at speeds ranging from 5 to 40 MPH and measurements have been shown to be repeatable. (iv) Measured Surface Deflections and Load This includes the surface deflections and the distances from the center o f load; load level, and the area over which load is applied (i.e., diameter o f loading plate). The loading condi tions and sensor configurations depend upon the NDT equipment and analytical method used. Surface deflection measurements at each sen sor location, and applied load level are recorded by the NDT devices. b. Deflection Computation A number o f forward analysis computer programs are used to calculate the deflections. The program takes the layer thicknesses, load level and the area over which it is applied, the seed moduli (or the lat
6-3

est set o f layer moduli), and the radii to the deflec tion sensors and calculates the surface deflection at each sensor location. In those programs where ad justments are made for nonlinearity, stresses or strains at selected locations are also calculated. Most o f the backcalculation programs utilize an iterative approach in which a forward deflection computing program is called in to compute deflec tions for each set o f new layer moduli. Some back calculation programs use a forward deflection calcu lation scheme to build a data base from which re gression equations are formulated to estimate the latest set o f moduli or used with interpolation tech niques to compute the deflections, thus avoiding the use o f the deflection computing program for each new set o f moduli in the iterative process. c. Error Check All computer programs use some kind o f error check to reach an acceptable convergence between the measured and calculated deflections. These er ror checks include: (i) The summation o f the squared differences between the measured and calculated deflec tions. (least square) The summation o f the squared o f the absolute differences between the two deflections, (absolute error)

(ii)

(iii) The summation o f the squared relative errors in which the difference between the deflections is divided by the measured deflections before the ratio is squared and summed, (least square relative) If the error check indicated convergence within acceptable levels o f tolerance, the results are printed out. If not, a new iteration is started with a new set o f layer moduli. Normally, three to four (or more) iterations are required to reach an acceptable con vergence.

6-4

d.

Results

Results o f backcalculation usually include the meas ured and calculated deflections, the differences, per cent differences, the error sums, and the final set o f layer moduli. e. Constitutive Relations and Correction for Stress & Strain The relationships between stress, strain, strain-rate, temperature, and moisture o f a material are known as the constitutive equation for that material. A rea sonable knowledge o f this relationship is necessary if corrections are to be made accurately and consis tently to adjust the backcalculated modulus o f each pavement layer to some other conditions (such as standard conditions) in which it can be compared with moduli determined in other places, at other times, or by other NDT equipment. The constitutive equations used in various computer programs vary widely from linear elastic theory with no corrections for nonlinearity to various forms of assumed relations between the stress or strain be neath the load to the modulus o f the layer. Consti tutive equations used may vary from layer to layer according to assumptions such as fine-grained soils becoming less stiff and coarse-grained soils becom ing stiffer with increased levels of stresses. The constitutive equations and corrections along with any calculated stresses or strains are used in conjunction with the "search for new moduli" (next item) to estimate new layer moduli for the next iteration f. Search Algorithm for New Moduli This is one o f the major distinguishing features o f all o f the computer programs. This step involves selec tion of a new set o f moduli based on the magnitude o f the difference in deflections, and the degree of nonlinearity introduced by the constitutive equations and corrections. The search methods attempt, by using efficient search techniques, to find a solution which represents the least error, the best fit o f the measured basin, and the best set o f layer moduli.

6-5

g.

Program Controls

Since several combinations o f pavement layer moduli may yield the same or similar measured deflections, the moduli backcalculated from the measured deflections may not be unique. In order to guide the search toward a set o f calculated moduli that are considered to be acceptable, numerous con trols are programmed to direct the search away from unwanted or unreasonable values o f the moduli. Examples o f such controls include assumptions such as: that moduli decrease with depth, that the sub grade modulus is constant with depth, that a rigid layer exists at a depth below the subgrade, or that a relationship exists between the modulus o f the lower layers and that o f the layer above it. These controls should be used carefully as some o f the layers may not fit with these or other programmed controls. 6.1.2 Selection o f a Backcalculation Computer Program Selection o f a computer program to backcalculate layer moduli depends on a number o f factors including the time it takes to converge, the accuracy o f the results, analysis method used, user friendly features, etc. Some o f the Finite Element programs require so much time that they are only suitable for mainframes. Nearly all elastic layer computer programs can run on microcomputers with a mathematics coprocessor chip. A list o f commonly used backcalculation programs was presented in Table 5.14. Many programs process each deflection basin one at a time requiring 10-15 minutes per deflection basin (on an IBM-PC/AT), and up to 30-45 minutes to analyze a threelayer pavement. A four-layer pavement can require twice as much time to converge to an acceptable set of moduli. Some programs generate a data base ahead o f time (requiring 30-45 minutes for an IBM-PC/AT computer), thereby eliminating the need for deflection calculations for each set o f new moduli. This approach eliminates the need for calling the forward deflection computing pro gram again, and reduces the processing time to 1-2 min utes per deflection basin. Some programs have capability for user-defined depth to bedrock and/or permit assumption o f a finite depth for subgrade in order to reduce memory and computational effort required by the computer. Presence o f rigid layer at
6-6

some depth can also be checked/estimated by some pro grams. Different programs are limited to the different number o f layers they can handle. Some o f the programs can adjust moduli for composite pavements, and still others are more suitable for either flexible or rigid pavement. Some pro grams used for backcalculation can also calculate the residual life and needed overlay for a pavement using the mechanistic-empirical approaches discussed earlier in the workbook. In most o f the programs the seed moduli are user depend ent. One measure o f the precision o f backcalculated layer moduli is that one should get essentially the same results regardless o f the choice o f initial seed moduli. To facili tate this a small tolerance (0.1% or less) between the measured and calculated deflections is generally specified. The Strategic Highway Research Program (SHRP) has developed a general selection criteria for backcalculation software. Agencies are encouraged to consider SHRP's recommendations when selecting software. Some o f the primary factors that should be considered in the selection o f computer programs include the following: Capabilities (i.e., linear elastic, nonlinear elastic)

Number o f layers which can be analyzed Search algorithm/Error check scheme utilized in the program Time required to converge to a satisfactory match Seed moduli (user provided or self generated) Ability to handle thin surface layers, presence o f a rigid bottom at shallow depth, and other unusual field conditions Sensitivity of the backcalculated solution to various factors such as layer thickness, assumed Poisson's ratios, deflection measurements, and seed moduli should be considered when selecting a computer pro gram.

User friendliness

6-7

Software cost

Many o f the existing backcalculation programs are usually written around readily available forward calculation pro grams. The four forward calculation layered linear elastic computer programs from which backcalculation computer programs were developed are: CHEVRON; BISAR; ELSYM5 and WESLEA. BISAR is proprietary and must be obtained by each user from the Shell Oil Company. Although the BISAR pro gram is based on linear elastic theory, an interface friction parameter may be defined. This allows consideration of partial bonding between layers. Horizontal loading caused by braking or turning may also be considered. ELSYM5 allows consideration of multiple loading and interface friction. CHEVRON is a strict application of elastic theory. WESLEA was developed at the U.S. Army Corps of Engineers Waterways Experiment Station. All three of these programs are in the public domain. Finite elements (FEM) programs have also been devel oped which can analyze both linear and nonlinear elastic pavement materials. Some o f the backcalculation computer programs in which the deflection calculations are performed by the above general purpose computer programs are: CHEVDEF, MODULUS, MODCOMP (Chevron); ELSDEF (ELSYM5); BISDEF (BISAR). Names of some of the computer programs which can per form backcalculation (some can also perform analysis for mechanistic-empirical overlay design procedures) for flexible or composite pavements are as follows: MODCOMP3, ISSEM4, LOADRATE, ELMOD, EVERCALC, WESDEF, PADAL, COMDEF, FPEDD1, RPEDD1, WESDEF, MICH-PAVE. A brief description of some o f these backcalculation programs is as follows: M ODULUS This backcalculation program uses a data base generated using the WES5 computer program, although the original versions were based on CHEVRON and BISAR. The first step is to make numerous computer runs to develop a data base containing deflections basins for expected range of modulus combinations assigned to the pavement layers.
6-8

Using an IBM-AT 286 with a 287 math coprocessor chip, approximately 30 minutes are required for MODULUS to compute the data base for a four-layer pavement section. After the generation o f the data base, the deflection com puter program is not required any more. Once the data base is developed, the procedure then uses the interpolation technique coupled with a Hooke-Jeeves' pattern search algorithm, for computing the deflection basin and minimizing the sum o f the squared error between calculated and measured deflections. The error check normally takes two minutes to complete. Because of its fast operation the program can be used in the field, while the data bank can be generated ahead o f time, overnight, or on the way to the field. This program can handle up to four layers and a rigid layer can be specified at any depth. The program is suit able for use when large number o f deflection measure ments are made on the pavements with the same configu ration. The program may also be used for nonlinear analyses. COM PEF This linear elastic program is used to backcalculate layer moduli for composite pavements from deflections meas ured by an FWD. It also uses data base o f precalculated modulus combinations and interpolation scheme to calcu late the layer moduli. E L S P E F a n d BISDEF These programs are identical except that ELSDEF uses ELSYM5 and BISDEF is based on BISAR for the calcu lation o f deflections. An iterative procedure is used to determine the best fit between measured deflections and computed deflections. Seed moduli are required and the number o f layers with unknown moduli cannot exceed the number o f measured deflections. No provision is available for nonlinear pavement material behavior. The program can be run with or without a rigid base. EV ERCA LC This is a mechanistic based pavement analysis program based on CHEVRON and was developed for Washington State DOT. This program uses an iterative procedure o f matching the measured surface deflections with the theo6-9

retical surface deflections calculated from assumed modu li. Absolute error check criteria which falls within a preset allowable tolerance (generally 10% or less using five deflection inputs) is used. A match is reached in about five minutes on a PC for a three-layer system. The program is capable o f handling up to five layers, and can be run with or without a "rigid base". User provided seed moduli are not required. The seed moduli can be estimated using internal regression equations which are based on relationships between layer moduli, load, and various deflection basin parameters. Some provisions are available for nonlinear material behavior. M O P C O M P3 This program utilizes the corrected Chevron elastic layer computer program for calculation o f deflections. Seed moduli are required, and an iterative approach is used for matching the deflections within the specified tolerance. MODCOMP3 assigns deflection to layers and backcalculates layer moduli based on this match. The remaining deflections are then used to check the fit o f the results. Some layers in the pavement system can be assigned to be "known". The known layers can either be linearly elastic or they can be stress dependent, in which case the appropriate constitutive model can be assigned as an input parameter. This program can deal with up to 15 layers. Material behavior may be linearly elastic, or nonlinear. Small toler ance on deflection match (about 0.5%) are required to ensure accurate results. ISSEM 4 This is a mechanistic-based pavement analysis computer program based on the ELSYM5 program. Seed moduli are required in this procedure. It uses an iterative proce dure o f matching the measured surface deflections with the theoretical surface deflections. The ISSEM4 program uses a "sequential cylinder" con cept in order to estimate the stress sensitivity o f the unbound materials from a single FWD load level. It takes about five minutes to complete a typical three-layer run.

6-10

F P E P P 1an d RPEDP1

Linear elastic computer programs FPEDD1 (for flexible and composite pavements) and RPEDD1 (for rigid pave ments) utilize an iterative approach for matching the measured and calculated deflections. User provided seed moduli are not required. The programs generate the seed moduli as the function o f measured deflections and radial distances o f the sensors. A significant feature in these programs is the consideration o f rock layers at a finite thickness of subgrade. The programs have a built-in pro cedure to create an artificial rigid bottom. The programs perform rational analysis to correct moduli for unbound layers and subgrade which exhibit nonlinear behavior. Backcalculation Computer programs for rigid pavements include: ILLI-BACK; WESLIQID; RPEDD1; DBCONPAS; ELCON; and FINITE. ILLI-SLAB was developed at the University o f Illinois for structural analysis o f jointed, one or two-layer concrete pavements with load transfer systems at the joint. This finite element program can be used to backcalculate the two key parameters needed to characterize a classical Westergaard rigid pavement; a Young's Modulus o f the concrete surface, and a composite modulus o f subgrade reaction (k). 6.1.2.1 Remarks Layer moduli predicted from the NDT deflection basin and use o f any o f the computer programs should be compared to the Agency's prior experience with similar materials to ensure that the results are reasonable. Limited destructive testing/lab analysis is recommended to provide spot verification. Keep in mind the differences between field and lab conditions when evaluating the data. Layer moduli adjustment in the backcalcu lation process is enhanced when one or more of the layer moduli can be deter mined accurately by other means and used as input into the analysis.

6-11

Layer moduli determined from an NDT deflection basin can also be verified by comparing with moduli determined by use o f a multidepth deflectometer (MDD). The MDD was developed by the National Institute for Transportation and Research in South Africa and used is extensively in the United states on research projects. The MDDs are installed in specially drilled holes in each pavement layer, and up to six modules (to measure up to six pavement layers) may be placed in a single hole. This device measures the relative deflection o f each layer with respect to an anchor point located in the same hole approximately 2,185 mm (86") below the pavement surface. By measuring MDD response under a deflection device load ing, two independent procedures are available for backcalculating layer modu lus, one with the deflection device sensors and the other with the MDD output.

6.2 SPECIFIC PROGRAMS The concept behind each o f the backcalculation techniques will be briefly discussed here, and the programs will be demonstrated. As the demonstration proceeds, each group should follow along on their computer. This will work best if the person with the least computer experience is the one who actually operates the computer. The data used in the following examples are from SHRP Section C. 6.2.1 BOUSDEF This program was developed at Oregon State University [6.1. 6.2]. and is similar in principle to ELMOD [6.3] in the way it backcalculates layer moduli. BOUSDEF uses the method o f equivalent thickness and Boussinesq the ory. In this approach the layers in a pavement crosssection are converted into one layer with an equivalent thickness as shown in Figure 6.1. The equivalent thick ness is calculated according to the following equation:

6-12

fSh,
i=l

n-1

/. \2 \ /3 -xVE. ' (1 -H ,)2

where he = the equivalent thickness o f the pavement structure f = a correction factor hj = thickness o f ith layer Ei = modulus o f ith layer E n = modulus o f the subgrade Pi = Poisson's ratio for ith layer pn = Poisson's ratio for subgrade This simplifies the calculation o f theoretical deflections in the basin, making the process very quick. The program provides backcalculated layer moduli and in formation regarding the stress dependency o f the materials accounting for the density o f the various layers. There are some limitations to this approach which the user should keep in mind. The greatest limitation is that the layer moduli should decrease with depth, preferably by a factor o f two or more between consecutive layers. Also, the computed equivalent thickness o f a layer should be greater than the radius o f the applied load. More infor mation regarding the program can be found in Appendix A and in Reference [6.2]. Begin running BOUSDEF by switching the computer to the drive which has the program in it, and then enter the proper subdirectory. For the purposes o f illustration, we will assume that the program is located in disk drive A, and that it is in subdirectory BOUSDEF. At this point, type in the name o f the program to start the execution as shown: A :\BO U SD EF>bousdef

6-13

[i

9] a jn io o u s iu3cu3a?j b jo ssauDpny, iu3[*Ajnb3 j o idaoucQ

T 9 a-mSij

Next, the title screen will show up as in Figure 6.2. Press <Enter> and the menu screen will appear as shown in Figure 6.3. Select "[2], Create a Data File" in order to enter your data. The program will request an input file name at this point (ROADl .DAT). The next display is the data input/edit screen as shown in Figure 6.4. The first piece o f information is the number o f layers in the pavement. The file name you specified in the previous screen will appear in the upper right corner. The column labeled "Layer for M" allows you the option o f backcalculating the modulus for a layer by entering 1 or fixing the modulus by entering 0. The thickness and Poisson's ratio are given in the next two columns. In the next three fields, the minimum, maximum, and initial moduli for each layer are entered. The last column is for the density o f the materials. These values are used to account for overburden pressures in stress calculations. Data concerning the FWD and the deflection readings are given in the next part o f the data input screen {Figure 6.4). The load plate radius and number o f sensors on the FWD are the first two items in this part o f the data input. Next, the sensor locations relative to the center o f load are entered. Below this are the data fields for each FWD test at a location. The load and deflection readings for each sensor location are given here. Finally, the tolerance for deflection errors and the maximum number o f itera tions are specified by the user. After the data have been entered, the user can choose to run the problem by pressing <F8>, save the data under the file name specified at the top o f the screen by pressing <F10>, or exiting the screen by pressing <Esc>. Once the program completes the backcalculation proce dure, a summary o f results will appear for each deflection basin in the problem as shown in Figure 6.5. The user may press any key to continue reviewing the results o f each drop. The last item in the output is a summary o f the stress sensitivity coefficients for the base and subgrade, and the moduli computed for each FWD drop (Figure 6.6). At this time, BOUSDEF provides results on-screen only.

6-15

BACKCALCULATION PROGRAM BASED UPON BOUSSINESQ THEORY Developed by Haiping Zhou R.G. Hicks Research Assist. Professor Civil Engineering Dept. Oregon State University Corvallis, OR 97331 Version 2.0 December 1988

Figure 6.2 - Title Screen for BOUSDEF

This program allows user to backcalculate pavement layer moduli from deflection basin data. The program was developed for use with Falling Weight Deflectometer (FWD) data. However, other NDT data may also be used with some modification of the data.

[1] . [2] . [3] .

Edit a Data File Create a Data File Analyze a Data File

Enter your selection

----

Press Esc to Exit ---

Figure 6.3 - BOUSDEF M enu

6-17

Pavement Structure Data Number of Layers: Layer No. 1. 2. 3. 4. 5. Layer for M 1 1 1 0 0 3 Thickness (inch.) 7.92 8.36 Poisson Ratio 0.35 0.35 0.40 Save t o : cROAD1.DAT> Minimum Modulus 636000 100000 1000 0 Maximum Modulus 1750000 3000000 50000 0 0 Initial Modulus 1000000 200000 15000 0 0 Density (pcf) 144.0 130.0 110.0 0.0 0.0

0.00 0.00 0.00

0.00 0.00

Deflection Measurement data Load Plate Radius: Number of Sensors: Sensor Locations: Load (lb) Test 1: Test 2: 6616 9522 5.91 7 0.0 24.0 Deflection Readings at Corresponding Sensor Locations 2.72 3.89 5.30 6.48 2.37 3.40 4.69 5.73 3 Esc = Exit (No save) 2.03 2.95 4.07 4.98 1.77 2.57 3.57 4.39 8.0 12.0 18.0 36.0 1.38 2.03 2.83 3.47 60.0 0.89 1.31 1.79 2.21

3.39 4.87 6.64 8.12 10 F8 = Run

Test 3: 12958 Test 4: 16696 Tolerance (%): F1 = Help

Number of Iterations: F10 = Save

Figure 6.4 - BO USDEF D ata In p u t/E d it Screen

6-18

Summary of Backcalculated Results Test Number: 1 The Final Modulus Values, after 3 iterations, are (psi): 689,064 485.715 36,857 Modulus changes are IN tolerance .................................. -........... Table of Deflections--------------------------------------Radial Position Calcu Defle Measured Defle Difference % Difference 3.39 3.38 0.01 0.43 1 0.09 3.41 2 2.63 2.72 2.37 -0.04 -1.89 3 2.41 2.03 -0.08 -3.77 4 2.11 -0.05 5 1.82 1.77 -2.72 1.35 1.38 0.03 1.84 6 0.89 0.05 6.15 0.84 7 Abs. Sum of Diff: 0.36 Abs. Sum of % Diff: 20.20 Mth. Sum of % Diff: 3.44
Press any key to continue_______________________

Figure 6.5 - BOUSDEF Results Summary for One Load

Summary of Backcalculated Results Summary of Non-Linear Characteristics of Lower Layers For base layer: K1 = 270867 K2 = 0.203 For subgrade: K1 = 35064 K2 = 0.025 RA2 = 0.83 RA2 = 0 . 1 4

Summary of Moduli and Stresses Load (lb) 6,616 9,522 12,958 16,696 E(1) 689,064 636,000 667,868 682,980 E( 2 ) 485,715 536,298 526,109 599,363 536,871 E( 3 ) 36,857 36,787 36,185 37,701 36,882 BSTRS 17.64 25.60 33.59 43.83 DSTRS 5.47 6.68 8.05 9.48

Average 668,978

Press any key to continue

Figure 6.6 - BOUSDEF Summary for all Load Levels

6.2.2

EVERCALC

EVERCALC was written at the University o f Washington for the Washington State Department o f Transportation l~6.4.6.5]. It uses an iterative approach in changing the moduli in a layered elastic solution to match theoretical and measured deflections. A simplified flow diagram of this method is shown in Figure 6.7. The program uses CHEVRON [6.6] as the layered elastic solution to com pute the theoretical deflections. Initially, these deflections are computed from the seed moduli supplied either by the user or EVERCALC. The moduli are then changed using the techniques described in Section 5 until the deflections are within the specified tolerance or the maximum number o f iterations have been reached. The program allows the user to enter the deflection data manually or retrieve it from an FWD data file created in the field. The output includes information regarding the backcalculation results from each FWD drop. The modulus o f the asphalt concrete can be adjusted in two different ways to a temperature of 77F (25C). The coefficients for stress sensitivity o f the base and subgrade are computed automatically by the program. Finally, the moduli are normalized to a 9000 lb. (40 kN) FWD load. The first step in using the program is to locate the disk drive and subdirectory containing EVERCALC. Assume that the program is in drive A in a subdirectory named EVERCALC. Simply type the program name after the prompt to begin the input process: A :\EVERCA LC>evercaic The title screen (Figure 6.8) will appear; strike any key to get the main menu. The next display you should see on the screen will be THE MENU as shown in Figure 6.9. The up and down cursors can be used to move to the desired selection, or key the appropriate menu item num bers. Choose " 1 Edit General Data File" to input general information about the problem. At this point, the program will request a name for the general data file. This screen is not shown in these notes, but we have named it
6 -2 1

SHRTCRS.GEN. Then the GENERAL DATA screen will appear as shown in Figure 6.10. The first input in the general data is the title o f the prob lem. The file name for the general data file is next. The next field is for the number o f layers in the pavement sys tem. Do not add an additional layer for a stiff layer con dition. This will be done internally within the program. EVERCALC allows the user the option o f English (E) or Metric (M) units. The FWD load plate radius is then en tered, followed by the number o f sensors to be used. The locations of the FWD deflection sensors are then entered (radial offsets measured from the center o f the load plate). EVERCALC allows the user to correct the modulus of the asphalt concrete layer to 77F (25C) from the tem perature measured in the pavement. If this is desired, type a Y at the "Temp Correction" prompt and an N if it is not. The temperature correction can either be done directly (D) by the program or by using Southgate's method (S) [6.7], The seed moduli for the layers can be done either internally by the program (I) or by using one's own engi neering judgment (E). The internal equation option (I) should only be used if the number o f layers is less than or equal to three. The next item on the GENERAL DATA screen is whether you wish to use the stiff layer calcula tion option (yes (Y) or no (N)). If you use the stiff layer option, only use the Seed Modulus Option "E" (i.e., enter your own selection o f layer seed moduli). Additionally, if you select the stiff layer option, you need not enter any in formation about the stiff layer into subsequent screens until it is requested (which occurs only after you select THE MENU Option 3 (Perform Backcalculation)). Once you have completed inputting the information for this screen, press the <F1> key to return to THE MENU.

6-22

Figure 6.7 - Simplified Flow Chart for EVERCALC

6-23

Figure 6.8 - Title Screen for EVERCALC

6-24

EVERCALC VERSION 3.3 - February 1992 Washington State Department of Transportation - University of Washington

THE MENU Use <ARROWS> to select <ENTER> to Choose 1 2 3 4 5 6 7 Edit General Data File Enter Deflection Data Interactively Perform Backcalculation Convert FWD raw Data File Plot Bar Chart of Normalized Modulus Execute DOS Commands Exit to DOS

Edit file containing information such as # of layers, # of sensors etc..

Figure 6.9 - Menu for EVERCALC

6-25

GENERAL DATA

INSERT OFF

Title:

SECTION c 3 5.91

File Name: Units: E

SHRTCRS.DAT

No of Layers:

Load Plate Radius:

No of Sensors: 7 36 8 60 9 10

Sensor No: 1 2 3 4 5 6 Radial Offsets: 0 8 12 18 24 Temp Correction: Y I

Teitip Measurement: Stiff Layer Option: RMS Tolerance(%): 1

D N

Seed Modulus Option: Maximum Iteration: 10

Modulus Tolerance(%): 1

Please Enter Percentage Modulus Tolerance for Convergence


Use Arrows, TAB or ENTER to Move Highlight Bar, <F1> When Done Editing Screen

Figure 6.10 - General Data Input Screen for EVERCALC

6-26

Once THE MENU comes back on the screen, select "2 Enter Deflection Data Interactively". Next, the program will ask for the general data file (SHRTCRS.GEN) and the deflection data file (SHRTCRS.DEF). This screen is not shown in the notes. The DEFLECTION DATA screen will then appear (Figure 6.11). The title and file name will show up in the first two fields o f this display. Below this, information concerning the layers are given by the user. The layer numbers are assigned automatically by the program. In the column next to this the user may opt to either fix the modulus of any layer by entering a 0 or choose to have the layer modulus backcalculated by the program by entering a 1. Enter the Poisson's ratio for the material in the third column. The last three columns are for the user to input the fixed modulus o f the material or the seed modulus and associated range if engineering judgment is being used. (Note: moduli range is optional). In this example, we are letting the program assign its own seed moduli, so this column was left blank. (Note: Do not attempt to add the stiff layer information here if using the stiff layer option.) The next set of inputs are the station number, thicknesses o f the pavement layers, number o f deflection basins, and the pavement temperature. (Note: Do not attempt to add the stiff layer option information here either. It will be requested in THE MENU Option 3.) In this example, we are using data from station 0.000 at which 4 data sets were collected. The top layer thickness was 201 mm (7.92") (201 mm), and the base was 212 mm (8.36") thick. The temperature o f the asphalt concrete layer was 12C (53F). After this, data from each FWD test are entered starting with the load level followed by the deflection measurements at each sensor. When the user has finished entering the data on this screen, the <F1> key is pressed to return to THE MENU (Figure 6.9).

6-27

Once THE MENU comes back on the screen, select "2 Enter Deflection Data Interactively". Next, the program will ask for the general data file (SHRTCRS.GEN) and the deflection data file (SHRTCRS.DEF). This screen is not shown in the notes. The DEFLECTION DATA screen will then appear (Figure 6.11). The title and file name will show up in the first two fields o f this display. Below this, information concerning the layers are given by the user. The layer numbers are assigned automatically by the program. In the column next to this the user may opt to either fix the modulus of any layer by entering a 0 or choose to have the layer modulus backcalculated by the program by entering a 1. Enter the Poisson's ratio for the material in the third column. The last three columns are for the user to input the fixed modulus o f the material or the seed modulus and associated range if engineering judgment is being used. (Note: moduli range is optional). In this example, we are letting the program assign its own seed moduli, so this column was left blank. (Note: Do not attempt to add the stiff layer information here if using the stiff layer option.) The next set of inputs are the station number, thicknesses o f the pavement layers, number o f deflection basins, and the pavement temperature. (Note: Do not attempt to add the stiff layer option information here either. It will be requested in THE MENU Option 3.) In this example, we are using data from station 0.000 at which 4 data sets were collected. The top layer thickness was 201 mm (7.92") (201 mm), and the base was 212 mm (8.36") thick. The temperature of the asphalt concrete layer was 12C (53F). After this, data from each FWD test are entered starting with the load level followed by the deflection measurements at each sensor. When the user has finished entering the data on this screen, the <F1> key is pressed to return to THE MENU (Figure 6.9).

6-27

DEFLECTION DATA

INSERT OFF

Route: SECTION C File Name: SHRTCRS.DEF Layer no Known/Unknown Poisson's Ratio Seed Moduli Min Moduli Max Moduli 1 1 0.35 0.0 0.0 0.0 2 1 0.40 0.0 0.0 0.0 3 1 0.45 0.0 0.0 0.0

Station: 0.000 No of Data Sets : 4 Load 6616.0 9522.0 12958.0 16696.0

Thickness : Temperature : 1 3. 39 4 .87 6. 64 8. 12 2 2.72 3. 89 5. 30 6. 48

Data Data Data Data

Set Set Set Set

1 2 3 4

7. 92 53. 0 Sensor No 3 4 2.37 2.03 3.40 2. 95 4. 69 4.07 5.73 4. 98

8.36

5 1.77 2.57 3.57 4.39

6 1.38 2.03 2.83 3.47

7 0.89 1.31 1.79 2.21

Please Enter Route Name or Title of Analysis, 25 Characters Max


Use Arrows, TAB or ENTER to Move Highlight Bar, <F1> When Done Editing Screen

Figure 6.11 - Deflection Data Input Screen for EVERCALC

6-28

After returning to THE MENU, select "3 Perform Backcalculation" to run the analysis. The screen shown in Fig ure 6.12 will appear in which the program requests the file name for the general data (SHRTCRS.GEN), the deflec tion data (SHRTCRS.DEF). The program will automati cally assign file names for the backcalculation or summary outputs, or the user may assign his/her own file names. Once this is done the backcalculation process begins. At this point, if you selected the stiff layer option, you will be prompted for the associated modulus and Poisson's ratio for that layer. Further, you will be presented with esti mated depths to stiff layer for your review (with an option to change the depths if so desired). During the backcalculation, information regarding the progress o f the program will appear on the screen. After the backcalculation is through, the program returns to THE MENU. If you want to print out your results, select "7 Execute DOS Commands", and the prompt for your disk drive will appear. At this point, you can give the following command to get the full output: A:\>print SHRTCRS.OUT Your printout should look like that shown in Figure 6.13. Only the data from one load level is shown here. At the end o f the printout, a summary will appear as shown in the figure. This gives information on the layer moduli and the stress sensitivity. It also lists the asphalt concrete modulus adjusted to 25C (77F). Similarly, a printout of the summary can be obtained by giving the command: A:\>SHRTCRS.SUM Which should give an output like that shown in Figure 6.14. These results are normalized to a 40 kN (9,000 lb.) FWD load.

6-29

Figure 6.12 - File Name Screen for EVERCALC

6-30

BACK CALCULATION BY EVHERCALC VERSION 3 . 3

Rout* Mllepost KarJ^er of L * y % n Thicknsases(In) Fsv*nnt Terop. (F) Load (lbs) :


.00 0

SECTIOK C 3 7.9 53.0 461ft.

8.4

Seed Moduli t)Md (pal) Moduli Calculated (pal) Deflections (mila) NO 1 2 3 4 5 6 7 OrrSET(in) .0 8.0 12.0 18.0 24.0 36.0 40.0

:
I

loooooo.
571830. 4

50000. 430614.

41515. 36571

lifter Iteration

CALCULATED 3.394 2.644 2.370 2.064 1.809 1.396 .866

MEASURED 3.390 2.720 2.370 2.030 1.770 1.380 .890 ABSOL. SUM: ARITH. SUM; JUiS UCROR:

DIFFERENCE -.004 .076 .000 -.034 -.039 -.016 .024 1131

DirF(%) -.1 2.8 .0 -1.7 -2.2 -1.2 2. 10.7

.2

l.t

(pai) Z .00 .00 .00 7.92 12.10 16.28

Straina (10*-6 in/in) VERTICAL TANGENTIAL RADIAL 6.4042 21.4910 t.3892 20.0638 -.3670 32.0249 ULK

STRESS: STRAIN: STRESS: STRAIN: STRESS: STRAIN:

-23.2769 -48.2004 -9.0256 -36.5285 -3.0512 -74.4000

(.4042 21.4910 8.3692 20.0636 -.3670 32.0249

-10.0686 7.7528 -3.7852

Modulus tolerance criteria Is satisfied. Load (lbs) s : : 522. 570830. 560009. 2 DIFFERENCE -.002 .102 -.009 -.037 -.062 -.016 .030 .2578 DIFF (1) .0 2.6 -.3 -1.2 -2.4
-.6

Seed Moduli Csed (psl) Moduli Calculated (pal) Deflections (sills) NO 1 2 3 4 5 4 7 OrrSET(ln) .0 8.0 12.0 18.0 24.0 36.0 40.0

430814. 486207.

36571. 35785.

After Iteration

CALCULATED 4.872 3.786 3.409 2.967 2.632 2.046 1.260

MEASURED 4.670 3.890 3.400 2.950 2.570 2.030 1.310 AB SOL. SUH: ARITH. SUH: RMS ERROR:

2.3

t.7
1.7

.2

(psl) 2 .00 .00 .00 7.92 12.10 14.28

Strains (10~-f in/in) VERTICAL TANGENTIAL RADIAL 4.4209 26.9694 12.4933 26.4252 -.5754 44.1414
VJLK

STRESS: STRAIN: STRESS: STRAIN: STRESS: STRAIN:

-34.9415 -47 .920 4 -13.3235 -48.2883 -4.2134 -103.2731

4.4209 24.9694 12.6933 26.6252 -.5754 44.1414

-24.0996 12.0631 -5.3642

Modulus tolerence criteria is satisfied.

Figure 6.13. Sample Output for EVERCALC 6-31

Summary o f Backc a lc ula tion at Station

.0 0 0

LOAD 6616. 9522. 12958. 16696.

E (l ) 578830. 560009. 582255. 605056.

CAD 200808. 194279. 201997. 209907.

E (2 ) 430814 486207 473830 519403

CSTR (3) asm CSTR 36571. 7 .1 5 .4 7 -7.74 .7 7 -12.04 35785. .0 0 1 0 .5 6 -16.64 35214. .0 0 1 2 .5 9 36852. 00 -21.96 1 .3 7 -.98 0000 -14.59 -11.27 .00 36105. 35926. 35974. 9 .7 8 .4 9 .0 0 1 5

1 .4 3 1 .6 3 1 .8 1 2 .0 5 1 .7 3 1 .6 0 .0 0

KEAN 581537. 201748. 477563 NORM.-* 563389. 476257 195452. to 9000 lbs K l, K2 A R5Q 0 Subgraoe la a C oarse-Craine S o il

M ot:

C (i ) - Modulus o f 1-th layer (pal) LAD - A d j . K x lu li o f asphalt layer for 77 6mq F (psl) BSTR - Bulk stra ss (psl) CSTR C o n fin in g stra ss (psl) DSTR - Devlato r atreaa (pal) K 1,K 2 - Stress s e n s it iv i t y c o effic ien ta R SQ - C o e ffic ie n t of determination -

Figure 6.13. Sample Output for EVERCALC (Continued)

BACKCALCULATION BY EVERCALC VERSION 3.0

SECTION C Modulus are Noramlized to 9000 lbs Mile ARMS .000 195.5 563.4 476.3 -1.0 0. .00 35.9 8.5 35976. .00 1.44 EAD E(1) E(2) BSTR(2) K1(2) K2(2) E(3) BSTR(3) K1(3) K2(3)

Note: E(i) = Modulus of i-th layer (ksi) EAD = Adj. moduli of asphalt layer for 77 deg F (ksi) BSTR = Bulk stress (psi) K1,K2 = Stress Sensitivity Coefficients ARMS = Average RMS Relative Error

Figure 6.14 - Summary O utput for EVERCALC

6-33

6.2.3

MODULUS This backcalculation method was developed at Texas A&M University for the Texas D epartm ent o f Public Transportation and Highways [6.8.6.9] . It differs from the other approaches in that a layered elastic com puter code (W ES5) is used to generate a database o f deflection basins for a range o f layer moduli. A pattern search method and interpolation are employed to minimize the error between the measured and calculated deflection basins. The depth to a stiff layer is automatically calculated by the program, but this can be overridden by the user. M ODULUS will perform detection o f nonlinear subgrade behavior and select the optimum number o f sensors to use in backcalculating moduli. It also contains an option to use default databases generated for common Texas pave ment sections. Up to seven deflections in a basin may be used in the analysis, but as a minimum, four deflections at 0, 12, 24, and 36 inches (0, 305, 610, and 915 mm) from the center o f the load must be used as input. A sensor weighting factor is provided to minimize or eliminate altogether one or more sensors from consideration during the backcalcu lation procedure. This was incorporated to improve esti mates o f subgrade moduli in highly non-linear cases. By eliminating the outer sensors, one can sometimes obtain more realistic answers for backcalculated moduli. f6.10], M ODULUS has several features which will not be used in this example, such as reading FWD deflection data auto matically. The reader is referred to the users manual in Appendix C for more information on these. To start running M ODULUS, first locate the disk drive where the program is located and then enter the subdirec tory. Begin the execution o f the program by typing M ODULUS at the prompt as shown below: A :\M O D U L U S >m odulus

6-34

After this the title screen will appear as shown in Figure 6.15, and the user may strike any key to gain access to the main menu (Figure 6.16). Choose the first selection on the menu, "1) Input D ata Conversion Options" and press <ENTER>. The program will ask if this is correct; type Y for yes, and the IN PU T DATA OPTIONS M EN U will appear {Figure 6.17). On this screen select "2) Enter OUT data manually", and it will again ask if this is the correct choice; type Y. Next, the FALLING W EIGHT D EFLECTOM ETER M ANUAL INPUT SCREEN will appear as shown in Figure 6.18. The top portion o f the screen is for "housekeeping" information. The user first names the file he/she is creating (ROAD1). The number o f deflection basins to be entered is given on the following line (4). For this example we will designate each o f the district, county, and highway as 1 as shown in Figure 6.17. The bottom part o f the screen is used to input the deflection informa tion. The program automatically numbers the deflection basin. The station or milepost in this course will be set at 0.000 for all problems, and the lane will be designated as R. The user next enters the load and the deflection for each sensor in the FWD. Finally, the user checks all the data to see if the values are correct. If the data are cor rect, a Y is entered after "VALIDATE?", and if they are not an N is entered. This latter option gives you a chance to correct any o f the data. This process o f entering de flection data is repeated for each deflection basin auto matically until the specified number o f basins have been completed. The program then returns the user to the main menu {Figure 6.16). At this point, select option "2) Run M odulus Backcalculation program" to perform the analy ses. The IN PU T/OU TPU T FILE INFORM ATION screen {Figure 6.19) will show up next. The file name is entered in the first line followed by the beginning and ending mile points (0.000 in our case) and the number o f deflection basins to be analyzed. The <ENTER> key is then pressed and the M ODULUS BACKCALCULA TIO N M EN U appears {Figure 6.20). On this screen, select "3) Run a full analysis".

6-35

<MODULUS>

VERSION 4.0 TEXAS SDHPT VERSION DEVELOPED BY THE PAVEMENT SYSTEMS PROGRAM TEXAS TRANSPORTATION INSTITUTE COLLEGE STATION, TEXAS

(C) Copyright 1989, Texas Transportation Institute. All Rights Reserved

Press any key to begin . . .

Figure 6.15 - Title Screen for MODULUS

6-36

v4.0

<MODULUS>
Main Program Menu

>1) Input Data Conversion Options< 2) Run Modulus Backcalculation program 3) Plot Deflection and/or Moduli values 4) Print results of latest analysis 5) Exit to DOS

Use the t or i keys or enter the option NUMBER and press <ENTER>
SDHPT version. Developed by the Texas Transportation Institute.

Figure 6.16 - Main Menu for MODULUS

6-37

v4.0

<MODULUS>

INPUT DATA OPTIONS MENU

> 1) Convert FWD to OUT data 2) Enter OUT data manually 3) Return to MAIN MENU

<

Use the t or 1 keys or enter the option NUMBER and press <ENTER>
SDHPT version. Developed by the Texas Transportation Institute.

Figure 6.17 - Input Data Menu for MODULUS

6-38

v4.0

<MODULUS>
FALLING WEIGHT DEFLECTOMETER MANUAL INPUT SCREEN OUTPUT FILE NAME................................................... > R0AD1.0UT NUMBER OF BOWLS TO BE ENTERED.........................................> 4 DISTRICT -> 1 COUNTY > 1 HIGHWAY....................... >1

ENTER INFORMATION FOR BOWL No. - > 2: STATION OR MILEPOST- > W1 ~> 4.87 W5 > 2.57 0.000 LANE > R W3 ~> 3.40 W7 > 1.31 LOAD >9522.00 W4 > 2.95 VALIDATE ? ->Y

W2 > 3.89 W 6-> 2.03

SDHPT version. Developed by the Texas Transportation Institute. Enter Y to validate the entry, or N to Repeat!

Figure 6.18 - FWD M anual Input Screen for MODULUS

6-39

v 4 .0

<MODULUS>

INPUT/OUTPUT FILE INFORMATION NAME OF THE INPUT FILE ............................................... > ROAD1 .OUT BEGINNING MILE POINT................................................................. > 0.000 ENDING MILE POINT------------------------------------------------------------ > 0.000 NUMBER OF BOWLS FOR THIS ANALYSIS ---------------------------- ...> 4

PRESS <ENTER> TO CONTINUE

SDHPT version. Developed by the Texas Transportation Institute.

Figure 6.19 - Input/O utput File Screen for MODULUS

6-40

v4.0

<MODULUS>

MODULUS BACKCALCULATION MENU

1) Use an existing fixed design 2) Input material types > 3) Run a full analysis

<

4) Return to Main Menu

Use the

or

keys or enter the option NUMBER and press <ENTER>

SDHPT version. Developed by the Texas Transportation Institute.

Figure 6.20 - Backcalculation Menu for MODULUS

6-41

An input screen will appear as shown in Figure 6.21. In this part o f the program, the user is given the option to edit information regarding the FWD characteristics, layer thicknesses, and material properties. The FWD informa tion has already been set for this exercise, so it is not nec essary to change it. To input the layer thicknesses, press the <F3> key and enter the surface thickness (H I), the base thickness (H2), the subbase thickness (H3), and the subgrade thickness above a rigid base (H4) if desired. If there is no subbase, enter a 0 for H3, and if you w ant to consider the subgrade to be infinite in depth, enter a 0 for H4. Press the <F4> key to designate minimum and maximum layer moduli and Poisson's ratios for the pave ment layers. Below this, your best guess o f subgrade moduli and Poisson's ratio are entered. Next, messages on the screen will appear as the backcalculation program is running (Figure 6.22). After the pro gram has finished, the PRINT RESULTS M EN U will show up as in Figure 6.23. Choose "1) Print Deflection & M oduli summary table" to have your results printed. M ODULUS does not have a feature that allows the user to view the results on his/her terminal. You may have a problem printing the results depending on the printer setup. I f you need assistance, please contact one o f the instructors. The printed results should look like those given in Figure 6.24.

6-42

v4.0

<MODULUS>
<FK2>
PLATE RADIUS ( in .) > 5.910 SEN SO R No. 1 DISTANCE FROM PLATE -> 0.00 W E IG H T FACTO R ----------> 0.00 <FK3> LAYER THICKN ESSES (in) NUM BER OF SENSO RS > 7

2 8.00 0.00

12.00 18.00

5 24.00

6
36.00

7 60.00

0.00 0.00
-> H1 H2 7.9 8.4

0.00
H3

0.00

0.00

0.0

H4 283.7

M ODULUS RANGES FOR: <FK4>

MINIMUM (KSI) SURFACE L A Y E R -------------------- > 636.00 -------------------- > 100.00

MAXIMUM (KSI) 1750.40 3000.00

PO ISSO NS RATIO 0.35 0.35

BASE LAYER

SUBBASE LAYER-------------------- > (KSI) 30.00 PO IS S O N S RATIO 0.40

SUBG RADE MODULUS (M O ST PRO BABLE VALUE) >

Press <ENTER> to continue, <FK1> = HELP, <FK2-4> = edit, <ESC> = ABORT

Figure 6.21 - FWD/Pavement Structure Input Screen for MODULUS

6-43

THE WES5 PROGRAM IS NOW RUNNING PROBLEM 3 OF 12 PROBLEMS

THE SEARCH PROGRAM IS NOW CALCULATING BOWL 4 OF 4 BOWLS

SEARCH

PROGRAM ANY

TERMINATED KEY TO

NORMALLY!

PRESS

CONTINUE

Figure 6.22 - Program Progress Messages for MODULUS

6-44

v4.0

<MODULUS>

PRINT RESULTS MENU

> 1) Print Deflection & Moduli summary table < 2) Print Estimated Deflection table 3) Print both of the above tables 4) Return to Main Menu

Use the t or 1 keys or enter the option NUMBER and press <ENTER>
SDHPT version. Developed by the Texas Transportation Institute.

Figure 6.23 - Printing Menu for MODULUS

6-45

TTI MODULUS ANALYSIS SYSTEM (SUMMARY REPORT) District: 1 County: 1 Highway/Road: 1

(VERSION 4.0)

Pavement: Base: Subbase: Subgrade:

Thickness (in) 7.90 8.40

MODULI RANGE (psi) Minimum Maximum 636,000 1,750,400 180,000 0 29,700 3,000,001 0

o.oc
283.70

Load Station (lbs)

Measured Deflection (mils) R1 R2 R3 R4 R5 R6

Calculated Moduli value(ksi)

Absolute

Depth to

R7 SURF(E1) BASE(E2) SUBB(E3) SUBG(E4) ERROR/Sens. Bedrock

0.000 0.000

6,615 3.39 9,521 4.87

2.72 3.89 5.30 6.48

2.37 2.03 3.40 2.95 4.69 4.07 5.73 4.98

1.77 2.57 3.57 4.39

1.38 2.03 2.83 3.47

0.8S 636. 1.31 636. 1.79 636. 2.21 636.

495.1 542.4 561.0 645.

0.0 0.0 0.0 0.0

30.3 29.5 28.9 30.2

3.08 2.97 2.31 2.27

300.00 * 300.00 * 300.00 * 300.00 *

0.000 12,957 6.64 0.000 16,695 8.12

Mean: Std. Dev.:

5.76 2.06

4.60 1.64

4.05 3.51 1.47 1.29

3.07 1.14

2.43 0.91

1.55 636. 0.57 0. 0.

560.9 62.6 11.2

0.0 0.0 0.0

29.7 0.7 2.2

2.66 0.43 16.11

300.00 0.00 0.00

VarCoeff(% ):35.83 35.66 36.30 36.73 37.23 37.64 36.99

Figure 6.24 - Sample Printout for MODULUS

6-46

6.2.4

M ODCOM P

The MODCOMP series o f backcalculation programs were developed at Cornell University, and the current version is MODCOMP3, Version 3.6. MODCOMP3 uses the CHEVRON elastic layer program to calculate theoretical deflections for use in the backcalculation process. This code has been modified to correct the integration errors that occurred with the original CHEVRON version, as well as to improve the accuracy of the solution. A fairly detailed description o f the program is included in Appen dix D. The basic approach is similar to the iterative proc ess applied in most backcalculation programs, using a set of seed moduli to initiate the analysis and sequentially adjusting the layer moduli, beginning at the outer sen s o rs) and working inwards, in order to match theoretical and measured deflections. The program can deal with: (i) (ii) up to 10 deflection sensors up to 6 load levels

(iii) Up to 12 layers, although a maximum o f 5 or 6 un known moduli is suggested (iv) non-linear material response (7 different constitutive models are listed in the manual) Seed moduli are required, but moduli limits are not user defined. Data is entered into input files with the extension .DAT, as described in the user's manual. A typical data file is shown for the SHRP Section C data in Figure 6.25. One approach is to use a screen editor to modify an exist ing file. A pre-processing routine for reading data from FWD files is under development. To run the program, enter the drive and directory where MODCOMP and the .DAT files reside and simply type M O D I <ENTER>. MODCOMP will sequentially process all the .DAT files in the directory and create two output files for each. Mes sages such as those shown in Figure 6.26 will appear on the screen during processing. The summary file (.SUM) provides the information shown in Figure 6.27, while complete information on the backcalculation sequence is provided in the .LST file shown in Figure 6.28.

6-47

6.3

CLASS PROJECT DESCRIPTIO N

There are three exercises described in this section. The first o f these is to perform deflection analyses by the backcalcu lation procedures which have been described. The next two optional exercises utilize the moduli calculated in the first for mechanistic-empirical analyses o f a pavement section. 6.3.1 Perform Backcalculation Deflection data, layer thicknesses and descriptions, and surface layer temperatures from three actual pavement sections are given in Tables 6.1 through 6.3. Each group will determine the required input for each program to be used for each section. Forms have been provided for you in the notes to write down certain input parameters. Please use these in order to help the instructors aid you if you have questions about any o f your runs. Next, input the data into the programs using the proce dures demonstrated earlier. Check your input carefully; this is where most mistakes are made! Finally, summarize the output on the results forms which have been provided.

SHRP SECTION C DATA LOAD 1 RUN 1 'M E TR \ 'LONG' 'H \ 1 .0 , 15 3 U \ 0 0 6895 .35 0 0 0.201 0 0 U\ 0 0 1379.25 0 0 0.212 0 0 U \ 0 0 103.40 0 0 0.0 0 0 1 7 0 .203 .305 .457 .610 .914 1.524 86 69 60 52 45 35 23___________________
Figure 6.25 - MODCOMP3 Data Input File for Section C, Load 1, Run 1

.15

0.0 416

6-48

PROCESSING FILE

___

SECCL1R1

MODCOMP3 Pavement Modulus Back-Calculation Program Version 3.60 D 06 December 1992 (C) Copyright Cornell Local Roads Program 1992 (C) Copyright Cornell University 1992 All Rights Reserved SHRP SECTION C DATA LOAD 1 RUN 1 Working on iteration no. 1 Elapsed time: 0:00:03 Working on iteration no. 2 Elapsed time: 0:00:06 Working on iteration no. 3 Elapsed time: 0:00:09

RMS Error: RMS Error: RMS Error:

3.48 percent 3.33 percent 3.22 percent

Computation terminated because deflection match was within .15 percent after 3 iterations. M O D C O M P 3 : Normal Termination - Out of Data

Figure 6.26 - MODCOMP3 Progress Messages

6-49

M 0D C 0M P3 Pavement Modulus Back-Calculation Program Version 3.60 D 06 December 1992 (C) Copyright Cornell Local Roads Program 1992 (C) Copyright Cornell University 1992 All Rights Reserved *** Processed: Apr. 28, 1993 @ 18:40 hrs *** SHRP SECTION C DATA LOAD 1 RUN 1 Summary of Results for Each Iteration Thickness: Using Model: .201 0 .212 0 0 3<E>3 103.0 265.0 267.0 266.0

Itn RMS Err 1 < E > 1 2 < E > 2 Seed 1 2 3 6895.0 4930.0 4680.0 4560.0 1379.0 1760.0 2040.0 2130.0 (

3.48 3.33 3.22

Elapsed time: 0:00:09

9.1 sec. )

Computation terminated because deflection match was within .15 percent after 3 iterations.________________

Figure 6.27 - Typical MODCOMP Summary (.sum file)

6-50

a *
r t

***** a a *

*** a a a a a a * ***

i *

* a a a

a a a a a *****

a a * *****

***** a a a a a a a a a a *****

a****

a* * *

a
a

a a a a a a a a a a a

a *****

a a a a ****

( Version 3.60 0; Released: 06 Decenber 1992 ) Processed: Apr. 28, 1993 9 00:34 hr ***

SHRP

SECTION C D TA L A 1 R N 1 A OD U

Load Nurber

load <kN)

P ressu re <kPa) 416.0

lo ad Radius <) .150

Input Dcflectfona (atcrons) 1<Radius>1 2Radfus>2 3<Radius>3 4<Radius>4 5<Radius>5 6<Radfus>6 7<Radius>7 .000 . 203 .305 .457 .610 .914 1.524
66.

29.4

69.

60.

52.

45.

35.

23.

layer Htrber 1 2 3

Type Model U u u 0 0 0

Seed Modulus (MPa) 6895.0 1379.0 57.0

P oisson's Ratio .35 .25 .40

Thickness (aeters) .201 .212 In fin ite

ro

Density (kg/cu a) .0 .0 .0

Non-linear C h a ra c te ris tic s K 1 K2 (MPs) .00 .00 .00 .000 .000 .000

.00 .00 .00

1.00 Modjlus Precision Tolerance (percent)s .150 D eflection Basin F it Tolerance (percent)s Maximm Murker of Iteratio n s to Achieve Tolerances 15

le y e r Number 1: la y e r Nurber 2: Layer Number 3:

D e f l e c t i o n

A i s l g n i t n t i

* t Radius a t Radius * t Radius .000 .305 .610

Cofrputer Assigned Deflection Nurber 1 Computer Assigned Deflection Nurber 3 Computer Assigned Deflection Nurber S

Figure 6.28

Typica 1 Modcomp list file (.LST)

6-51

OOCONP

I t e r a t i o n

Nuibvr:

|Layer| I I I

Type E (HP*) E (MPa) E (MPa)

Modal

|Coaff. ( 1 I i i ** ** l E Constant l E Constant i E * Constant

C o a i t n t i ( I in ta r la y e r) (lin e a r la y e r) ( lin e a r la y e r)

1 I Unknown * I Unknown 3 I Unknown

4620.0 2050.0 256.0

Calculeted/Supplied Deflection Agreement for Load Nirfeer:

Radius .000 .203 .305 .457 .610 .914 1.524

C alculated D eflectio n 88. 69. 62. 53. 46. 35. 22.

M easured/Interpolated D eflection 86. 69. 60. 52. 45. 35. 23. Measured Measured Measured Measured Measured Measured Measured

Percent Difference 2.22 .68 3.14 2.72 3.07 1.16 4.96

Assigned to la y e r 1 2 3

S e n sitiv ity to Layer Modulus -.245 -.159 -.848

2.88 percent Root Mean Square Error (Based on aeasured deflections only)

Elapsed tin e :

0:00:01

1 .6 sec. )

Figure 6.28

Continued

6-52

* * *

S u a I r y

of

K t i u l t s

of

M0 D C 0 MP

E x o c u t

ion

I Layer| I ! 1

Typ

1 1E 1E 1E
( H P .) (H P ) (H P * )

Model > 4540.0 2160.0 266.0

|Cocff. (* )| 1 1 1 ** ** ** 1 E Constant 1 E * Constant 1 E * Constant

C o a u n t ( lin e a r la y e r) (lin e a r la y e r) ( lin e a r la y e r)

1 1 Unknown 2 1 Unknown 3 1 Unknown

C alculated/Suppl f ed D tflection Agreement for load Nunbcr:

Radius .000 .203 .305 .457 .610 .914 1.524

C alculated O eflection 66. 68. 60. 52. 45. 34. 21.

M easured/Interpolated Oef lection 86. 69. 60. 52. 45. 35. 23. Measured Measured Measured Measured Measured Measured Measured

Percent . Difference .04 2.08 .19 -.23 .11 -1.76 7.74

Assigned to la y e r 1 2 3

S en sitiv ity to layer Modulus -.245 -.159 -.848

to o t Mean Squirt Error 3.10 percent (Based on aeasurtd deflections only)

Elapsed tin e ;

0:00:03

3 .0 se e . )

Figure 6.28

Continued

* * *

S u R i a r y

of

R e s u l t s

of

M0 D C 0 MP

I t . n t l o n

k u i b i r ;

I Layer| I 1 1

Type

1 1 E (MPa) 1 E (MPa) 1 E (MPa)

Model 4500.0 2160.0 266.0

|Coeff. ( 1 1 1 1 * * ** 1 E Constant 1 E Constant 1 E * Constant (lin e a r lay er) (lin e a r lay er) (lin e a r la y er)

1 1 Unknown 2 1 Unknown 3 1 Unknown

C alculated/Supplied Deflection Agreewnt for Load Mrber:

Badius .000 .203 .305 .457 .610 .914 1.524

C alculated D eflectio n 86. 67. 60. 52. 45. 34. 21.

Measur e d /1nt erpo I at ed D eflection 66. 69. 60. 52. 45. 35. 23. Measured Measured Measured Measured Measured Measured Measured

Percent Difference .01 2.22 .04 -.37 .00 1.85 7.81

Assigned to la y e r 1 2 3

S e n sitiv ity to layer Modulus -.245 -.159 -.648

3.15 percent Root Mean Square Error (Based on measured deflections only)

Elapsed tin e :

0:00:05

5 .0 se c . )

Computation term inated because d e fle c tio n natch for the assigned sensors was w ithin t .15 percent a f te r Modulus p recisio n to leran ce of *1.00 percent was also sa tisfie d .

3 ite ra tio n s .

Figure 6.28

Continued

6-54

Table 6.1. Data for Section C Pavement Structure Layer Number 1 2 3 Material _____________ Type________________ Asphalt Concrete Soil Cement Base Silty Sand Subgrade Layer Thickness, mm(in) 201 (7.92) 212 (8.36) ----

Temperature Data Temperature @ 102 mm (4 inches) = 12C (54F)

Deflection Data

Deflection, |im (mils) Test No. 1 2 3 4 Load kN (lbs) 29.4 (6,616) 42.4 (9,522) 57.7 (12,958) 74.3 (16,696) 0mm Oin 86.1 (3.39) 123.7 (4.87) 168.7 (6.64) 206.2 (8.12) 203 mm 8 in 69.1 (2.72) 98 .8 (3.89) 134.6 (5.30) 164.6 (6.48) 305 mm 12 in 60.2 (2.37) 86.4 (3.40) 119.1 (4.69) 145.5 (5.73) 457 mm 18 in 51.6 (2.03) 74.9 (2.95) 103. 4 (4.07) 126.5 (4.98) 610 mm 24 in 45.0 (1.77) 65.3 (2.57) 90.7 (3.57) 111.5 (4.39)

@ 914 mm 36 in 35.1 (1.38) 51.6 (2.03) 71.9 (2.83) 88.1 (3.47) 1,524 mm 60 in 22.6 (0.89) 33.3 (1.31) 45.5 (1.79) 56.1 (2.21)

6-55

Table 6.2. Data for Section F Pavement Structure Layer Number 1 2 3 Material _____________ Type________________ Asphalt Concrete Crushed Limestone Base Silty Sand Subgrade Possible Rock Layer at 5,029 mm(198 in) Layer Thickness, mm(in) 194 (7.65) 368 (14.47) ----

Temperature Data Temperature @ 102 mm(4 inches) = 15C (60F)

Deflection Data Deflection, (im (mils) Test No. 1 2 3 4 Load kN (lbs) 29.1 (6,534) 42.3 (9,512) 56.3 (12,662) 74 .8 (16,812) 0mm Oin 83.3 (3.28) 128.8 (5.07) 184.9 (7.28) 246. 6 (9.71) 203 mm 8 in 68.3 (2.69) 109.7 (4.32) 151. 6 (5.97) 207.5 (8.17) 305 mm 12 in 59.2 (2.33) 93.2 (3.67) 131.3 (5.17) 180. 6 (7.11) 457 mm 18 in 47.8 (1.88) 75.9 (2.99) 108.2 (4.26) 149. 4 (5.88) 610 mm 24 in 39.6 (1.56) 61.0 (2.40) 88.6 (3.49) 122.2 (4.81) @ 914 mm 36 in 27.7 (1.09) 42.9 (1.69) 60.2 (2.37) 83.6 (3.29) 1,524 mm 60 in 17.3 (0.68) 25.7 (1.01) 33.8 (1.33) 46.5 (1.83)

6-56

Table 6.3. Data for Section G Pavement Structure Layer Number 1 2 3 4 Material _____________ Type________________ Asphalt Concrete Portland Cement Concrete Sand Base Silty Sand Subgrade Layer Thickness, mm(in) 135 (5.33) 244 (9.60) 102 (4.00) ----

Temperature Data Temperature @ 64 mm(2.5 inches) = 27C (81F)

Deflection Data

Deflection, nm (mils) Test No. 1 2 3 Load kN (lbs) 41.8 (9,398) 54.5 (12,256) 72.8 (16,350) Omm Oin 71.4 (2.81) 99.1 (3.90) 127.8 (5.03) 203 mm 8 in 61.2 (2.41) 86.4 (3.40) 111.0 (4.37) 305 mm 12 in 59.2 (2.33) 84 .3 (3.32) 107.2 (4.22) 457 mm 18 in 56.9 (2.24) 80.5 (3.17) 102.6 (4.04) 610 mm 24 in 54.4 (2.14) 76.7 (3.02) 97.5 (3.84)

@ 914 mm 36 in 48.3 (1.90) 67.8 (2.67) 85.9 (3.38) 1,524 mm 60 in 35.6 (1.40) 49.8 (1.96) 62.2 (2.45)

6-57

Input Form SECTION F

BOUSDEF

Layer No.

Poisson's Ratio

Min. Mod., MPa (psi)

Max. Mod., MPa (psi)

Seed Mod., MPa (psi)

xxxxxxxxxxxxxx
Tolerance =

xxxxxxxxxxxxxxx

Number of Iterations =

EVERCALC / MODCOMP

Layer No.

Poisson's Ratio

Seed Mod., MPa (psi) Temp. Corr.? Temp. Meas.? Y or N D or S

Seed Mod. Opt.? I or E

Error Tolerance = Number of Iterations =

Modulus Tolerance =

MODULUS

Layer No.

Poisson's Ratio

Min. Mod., MPa (psi)

Max. Mod., MPa (psi)

Seed Mod., MPa (psi)

xxxxxxxxxxxxxxx xxxxxxxxxxxxxxx xxxxxxxxxxxxxxx xxxxxxxxxxxxxx


6-59

xxxxxxxxxxxxxxx

Input Form SECTION G

BOUSDEF

Layer No.

Poisson's Ratio

Min. Mod., MPa (psi)

Max. Mod., MPa (psi)

Seed M o d . , MPa (psi)

Tolerance =

Number of Iterations

EVERCALC / MODCOMP

Layer No.

Poisson's Ratio

Seed Mod. , MPa (psi) ___ ___ ___ __ _ Temp. Corr.? Temp. Meas.? Y or N D or S

Seed Mod. Opt.? I or E Seed Mod. Opt.? I or E


%

Error Tolerance = Number of Iterations =

Modulus Tolerance =

MODULUS

Layer No.

Poisson's Ratio

Min. Mod. MPa (psi)

Max. Mod., MPa (psi)

Seed Mod., MPa (psi) XXXXXXXXXXXXXXX XXXXXXXXXXXXXXX

xxxxxxxxxxxxxxx xxxxxxxxxxxxxx
6-60

xxxxxxxxxxxxxxx

Output Form SECTION C

BOUSDEF Load Level


6616 9522 12958 16696

E^, MPa (psi)

E2 , MPa (psi)____

E 3 , MPa (psi)____

E^, MPa (psi)____

ARS

Base:

K ^ = ________ K ^ = ________ -

I = <2 K2 =

Subgrade:

EVERCALC / MODCOMP Load Level


6616 9522 12958 16696

E ^ MPa -, (psi)

E 2 , MPa (psi) ____

E 3 , MPa (psi) ____

ARS

RMS

Base:

K ^ = ________ K ^ = ________ -

I = < K2 =

Subgrade:

MODULUS Load Level


6616 9522 12958 16696

E^, MPa (psi)

E2 , MPa (psi) _ _

Eg, MPa (psi)

ARS

6-61

Output Form SECTION F

BOUSDEF Load Level


6534 9512 12662 16812

E ^ MPa -, (psi)

E2 , MPa ____ (psi)

E 3 , MPa (psi)____

ARS

Base:

K ^ = ________ K ^ = ________ -

I = <2 K2 =

Subgrade:

EVERCALC Load Level


6534 9512 12662 16812

E ^ MPa -, (psi)

E2 , MPa (psi) ____

Ej, MPa (psi)____

ARS

RMS

Base:

K ^ = ________ K ^ = ________ -

K2 = K2 =

Subgrade:

MODULUS Load Level


6534 9512 12662 16812

Ej, MPa (psi)

E2 , MPa ____ (psi )

E 3 , MPa (psi ) ____

ARS

6-62

Output Form SECTION G

BOUSDEF

Load Level
9398 12256 16350

E ^ MPa -, (psi )

E2 , MPa E 3 , MPa
(psi)______ (psi )

Base:

K ^ = ________ K ^ = ________ -

J = __ <2 K2 =

Subgrade:

EVERCALC Load Level


9398 12256 16350

E^, MPa (psi)

E 2 / MPa E 3 , MPa (psi)_____

(psi)

Base:

Ki = _ K^ -

Subgrade:

MODULUS Load Level


9398 12256 16350

E^, MPa (psi)

E2 , MPa 3 , MPa (psi)______

(psi)

6.3.2

Perform a Basic Mechanistic-Empirical Analysis

Use the backcalculated moduli from Section F to determine the number of load repetitions to failure in rutting and fatigue. In this problem, select the layer moduli calculated at the 9512 lb. (42.33 kN) load level.

Layer inputs: Layer No. 1 2 3 4 Thick., mm (in) 194 (76.5) 368 (14.47) 4,467 (175.88) 6,900 (1,000,000) 0.20 Modulus, MPa (psi) Poisson's Ratio

Load inputs: Number of Loads = 1 Magnitude of Load = 40 kN (9000 lb.) Contact Pressure = 552 kPa (80 psi)

Run ELSYM5, and calculate the strains at critical locations. Horizontal Tensile Strain @ Bottom of ACP = ____________ Vertical Comp. Strain @ Top of Subgrade = ____________

Calculate the number of cycles to failure: Rutting Nf = ____________ Fatigue Nf = ____________

6-64

6.4

D A T A FILE FORMAT

The following file descriptions are for data files generated by FWD's that are currently commercially available, pre sented in alphabetical order. 6.4.1 DynatestFW D There are various versions o f the Dynatest FWD field program available. The edition is identified by the last three characters in the first line o f the data sets e.g., F20 is edition 20 o f the FP. A complete description o f the FP20 contents is included on the DATA diskette in the file labeled FILE.DOC. For the purpose o f this course, the following is sufficient information to retrieve data from these files for use in any o f the backcalculation programs. Remember some o f the programs can access the FWD files directly. a) The two digits preceding the FP identifier indicate the number o f lines in the "header" i.e., the portion of the file that precedes the deflection data. For instance, 36F20 indicates a 36 line header. The header is used for recording information about the project (dates, locations, operator, etc.) and the FWD setup at the time o f test (serial number, sensor information, including load cell, measurement units, test sequences, etc.). For this course, the significant line in the header is the one that begins with: 150 0 200 ...... or 225 0 300 ...... which identifies load plate radius (mm) and active sensor locations in mm from the load plate center used for the test. It is line 3 for FP20. b) After the header, the first line o f data begins with the station indicator (S) and is for first test location e.g., S 1012.... identifies the test carried out at location 1012 (units are often ft. or m., but various other
6-65

indicators can be chosen and are identified in the header). The rest o f this line contains information such as temperature (air, surface or other), time o f day, cracking indicator (if used), wheelpath indicator (if used), etc. c) Immediately following the S line is the load and deflection peak value data on one line for each test load. Metric data is recorded in the first 32 columns The first value is contact pressure in kPa, followed by deflections DO through D7 (usually) in microns. Each entry field is four columns wide. Following the metric data, on the same line, the data may be repeated in U.S. units, with the first value being the total load (not pressure) in Ibf, followed by deflec tions in mils. Depending on FP version there may be up to 32 lines o f data at each station. Most project level data will be for two to four load applications.

6.4.2 JILSFW D The following description was provided by Foundation Mechanics, Inc:

6-66

FOUNDATION MECHANICS,

INC.

FACSIMILE MESSAGE

P A G E 1 O F _ 2 _____

4 2 1 S. EL SEGUJTDO BLVD. EL S E G i m D O , C A 9 0 2 4 5 F A X (310) 3 2 2 - 5 1 4 6 P H O N E (310) 3 2 2 - 1 9 2 0

TO: B o b B r i g g s Dynatest F A X KO. (805) 6 4 0 - 0 3 4 5 DATE: 07/16/93

D e a r Bob: I a m s e n d i n g a p r i n t o f a t y p i c a l A S C I f i l e w e u s e to s t o r e t h e s t a n d a r d J I L S t e s t dat a . T h i s is n o t t h e f o r m we u s e for s a v i n g w a v e f o r m s . M T h i s is t o i d e n t i f y t h e i i l e as a j i l s d a t a file. Date-Time T h i s is a u t o m a t i c f r o t h e c o m p u t e r c l o c k on o p e n i n g or a p p e n d i n g t o a d a t a file. Se n s o r s : 0 7 5 - 1 T h e s e are t o i n s t r u c t t h e c o m p u t e r t o t h e c a l i b r a t i o n f i l e s for e a c h v e l o c i t y sens o r . T h e y a r e n o . l t o n o . 7. W e i g h t / s p r i n g : 3 T h i s is t o t e l l t h e c o m p u t e r w h i c h s p r i n g - m a s s c o b i n a t i o n the o p e r a t o r h a s s e l e c t e d for t h i s t e s t file, w e u s e 1, 2, 3, or 4 for t h i s f u n c t i o n . It as to p r o v i d e t h e f o r c e r a n g e t o be used; Tom p : 65 T h i s is t h e a m b i e n t t e m p e r a t u r e as e n t e r e d b y t h e o p e r a t o r . O p e r a t o r : T h i s is t h e o p e r a t o r s name. Comments: demo T h i s line w a y be u s e d t o e n t e r a n y c o m m e n t . Location: I m i s s e d this l i n e above. It is f o r t h e o p e r a t o r t o i d e n t i f y t h e l o c a t i o n at w h i c h t h e t e s t f i l e is taken. P o rce: 1 1 . 0 0 8.00 T h i s is t h e f o r c e i n K I P S s e l e c t e d w h e n l a s t the s y s t e m w a c used. F o rce: 7.00 8.00 10.00 1 1 . 0 0 T h e s e are t h o t a r g e t f o r c e s I e n t e r e d

for this te3t sequence.

The col col col col col col c ol col col col

d a t a l i n e is a B f o l lows: 1 : t e s t s t a t i o n 1 to 1 0 0 0 2 : test lane 1 to 7 3 : a c t u a l p e a k f o r c e in k i p s 4 : p e a k d e f l e c t i o n at s e n s o r n u m b e r 1, in mils. 5 : sensor 2 6 : sensor 3 7 : Bensor 4 8 : sensor 5 9 : sensor 6 10: s e n s o r 7

In t h i c f i l e t h e s y s t e m r a n f o u r d r o p s at e a c h s t a t i o n as s e l e c t e d by the operator. S h o u l d t h e o p e r a t e r c h a n g e a n y f u n c t i o n , o r e x i t a n d r e - e n t e r t h e file, t he n e w c o n d i t i o n s e l e c t e d w i l l be d i s p l a y e d as w a c t h e f o r c e c h a n g e I d i d at the s t a r t of t h i s file a n d w h e n I c h a n g c d t o r u n four d r o p s , e a c h at 9 KIPS. P l e a s e c a l l m e f o r a n y h e l p y o u m a y n e e d w i t h t h i c file. Some of tbose u s i n g the JILS use other formats w h i c h they have requested a n d e i t h e r w e a r e t h e y r e v i s e d t h e p r o g r a m . W e l i k e to u s e t h e S t a t i o n a n d L a n e s e t - u p co t h a t w e c a n p u t t h e d a t a i n t o a m a t r i x f o r p l o t t i n g with some simple programs we provide. Best regards Bill Johnson

04-10-1993 08:00:48 sensors: 07 5 - 1 075 1 075-1 Weight/spring: 3 L o c a t i o n : shop Temp: 65 O p e r a t o r : jwj Comments: demo Fore: 1 1 . 0 0 8 . 0 0 Force: 7.00 8.00 10.00 11.00 6.39 7.14 5.11 11 7.51 7.87 7.77 7.00 11 8.48 8.70 5.57 1 1 1 0 . 3 8 10.90 7.03 9.78 8.79 1 1 11. 3 0 11.58 1 0 . 4 0 7.52 9.33 5.41 2 1 7.45 8.31 7.43 6.75 9.12 8 . 1 2 5.90 2 1 8.48 7.31 2 1 10. 3 6 10.95 7.11 9.86 8 . 8 3 2 1 1 1 . 3 8 11. 1 8 9.97 7.13 8.94 3 1 7.36 8.26 5.41 7.49 6.74 5.97 7.37 3 1 8.35 9.11 8 . 1 4 3 1 10.36 10.93 9 . 8 8 8 . 8 7 7.17 3 1 11. 2 8 11 . 4 8 1 0 . 3 1 9 . 2 8 7.48 5.39 4 1 7.31 8.26 6.73 7.44 4 1 9.39 9.09 8 . 1 1 7.33 5.97 4 1 10.24 10.70 9.64 8.61 7.02 4 1 11.23 11.50 1 0 . 3 3 7.49 9.19 5 1 7.25 8.15 7 . 3 7 6.60 5.35 5 1 8.28 5.98 7.36 9.06 8 . 1 6 5 1 10 . 1 6 10 . 6 3 9 . 6 4 8 . 5 2 6.92 5 1 11.11 11 . 7 3 1 0 . 6 2 7.61 9.44 6 1 7.30 8.24 5.3/ 7.40 6.68 6 1 8.31 9.08 8 . 2 0 5.95 7.34 6 1 10.22 10.87 9.83 8 . 7 8 7. 1 0 6 1 11.18 11 . 6 8 1 0 . 5 5 9.39 7.54 D a t e - T i m e : 04-10 - 1 9 9 3 1 8 : 3 5 :12 D 7 1 7.09 7.96 7. 1 6 5.24 6.49 7 1 8 . 1 8 8.94 5.84 8. 0 3 7.24 7 1 10.01 10 . 4 9 9.40 8 . 4 8 6.86 7 1 11.01 11 . 3 3 1 0 . 1 8 9.09 7.38 8 1 7.18 8.13 5.35 7.34 6.64 8 1 8.26 9.00 8 . 0 3 7.28 5.89 8 1 10.00 10 . 5 9 9.53 8 . 5 6 6. 9 2 8 1 11.05 1 1 . 3 2 1 0 . 1 6 9.17 7.35 9 1 7.21 8.14 7.38 6. 6 8 5.36 9 1 8.26 9.05 8 . 1 0 7.27 5.96 9 1 10.19 10 . 6 9 9. 6 6 8.68 7.01 9 1 11.04 11 . 3 2 1 0 . 2 1 9.16 7.37 Fore: 9. 00 9. 00 9.130 9.00 10 1 9.19 9.79 8 . 9 2 8 . 0 2 6.41 10 1 9.16 9.87 8.80 7 . 9 5 6.42 10 1 9.16 9.84 8 . 8 6 7 . 9 2 6.40 10 1 9.19 9.88 8 . 7 9 7.97 6.41 11 1 9.19 9.75 8.97 6.42 8.00 11 1 9.18 9.87 8.85 6.42 7.95 11 1 9.22 9.91 8.84 7.94 6.43 11 1 9.19 9.91 8 . 8 2 6.43 7.94 12 1 9.17 9.75 9.00 8 . 0 2 6.43 12 1 9.22 9.90 8.83 7. 98 6.44 12 1 9.22 9.89 8 . 8 3 7 . 9 9 6.43 12 1 9.19 9.87 8 . 8 9 7 . 9 5 6.42 13 1 9.16 9.70 7.96 9.02 6.39

Date-Time:

075 - 1

075-1

075-1

075-1

4.43 4.91 6.12 6.56 4.73 5.17 6.19 6.27 4.69 5.22 6.20 6. 59 4.65 5.24 6.11 6.57 4.58 5.24 5.93 6.68 4.61 5.25 6.16 6.60 4.52 5.09 5.98 6.45 4.64 5.20 6.01 6.47 4.65 5.13 6.15 6.46 5.71 5.65 5.54 5.66 5.56 5.58 5.65 5.64 5.58 5.68 5.63 5.59 5.61

2.87 3.11 3.92 4.12 3.05 3.33 4.00 3.97 3.05 3.36 3.98 4.14 3.03 3.38 3.89 4.15 3.04 3.38 3.90 4.27 3.01 3.40 3.97 4.18 2.95 3.32 3.81 4.12 3.01 3.36 3.86 4.10 3.02 3.39 3.92 4.13 3.67 3.61 3.55 3.61 3.59 3.61 3.61 3.58 3.60 3.61 3.58 3.61 3.60

1.17 1.27 1.56 1.79 1.23 1.39 1.58 1.66 1.25 1.41 1.62 1.81 1.21 1.38 1.67 1.81 1.20 1.41 1.64 1.82 1.24 1.40 1.65 1.70 1.21 1.37 1.60 1.76 1.21 1.36 1.63 1.76 1.16 1.37 1.64 1.76 1.54 1.54 1.51 1.54 1.48 1.56 1.56 1. 5 3 1. 5 1 1.53 1.54 1.57 1.52

6-68

6.4.3 KUABFWD
The following description was provided by Engineering Research Int'l (a copy o f the file is included on the data diskette as KUAB FWD):

6-69

IFWD DATA FILE

HProject HEngineer HContractor HOperator HOate HHighway HUork Order HSection HLane HSairple Unit HO irecti on
H

C:\SFUD\DATA\U15TL.FUD Pennsylvania MOT Testing PennOOT Engineering and Research, Int' Doug Steel* 5/13/93 S.R. 219
1

5 Traffic W15TL Northbound 10 .15 .45 05-13-1993 ICUAB FUD Model 150 4.08 4 (9+9 small buffers, 12 extra weights) 5.91 (in) 10918.00 130 1.00000 09-27-1991 24:05:00

HPvt Thickness hPoisson of PCC HPoisson of SC IDate Created HMachine Type HSoftware Version ILoad Mode : IPlate Radius : HPlate Cal Factor : : HPlate Cal Add HPlate Cain factor: HPlate Cal Date : HPlate Cal Tioie : HAir HAir HAir KAir Cal Cal Cal Cal Factor Add Date T io

=La j P o c l m T I oaJ

: 0.074065 : -273 : 05-04-1993 : 17:51:44

HOMI Cai Factor : -0.279876 HOMI Last Cal Date: 05-06-1993 HOMI Last Cal Tine: 14:23:04 lDrop Sequence IRecord Drop? : 123 : YT 4 6 7-31003 5-31007 6-30937 60.00 72.00 (in) 36.00 BEHIND BEHIND BEHIND 0.30878 0.12077 0.12118 1.05000 1.05000 1.05000 1.00000 1.00000 1.00000 05-04-93 05-11-93 17:03:56 D60 n ils 1.30 1.66 2.09 0.55 0.95 1.34 0.86 1.07 1.74 0.69 1.00 1.80 1.12 1.53 2.18 1.24 1.87 072
1 nils

1Channel 0 1 3 2 lSensor ID 0-30829 1-30890 2-30862 3-30864 IDistance 0.00 12.00 12.00 24.00 IPosition CENTER FRONT BEHIND I EHIND B HStatic Cal Factor 0.56417 0.54289 0. 53022 0.27020 HDyn Cal Factor 1.05000 1.05000 1. 05000 1.05000 HGain Factor 1.00000 1.00000 1. 00000 1.00000 HChamel Cal Date 05-10-93 06-19-92 06- 19-92 05-10-93 HCharrel Cal Time 19:28:20 15:55 :51 16:01:40 19:51:34 H J Station Imp Load DO D12 D12 D24 D36 ni Is J ft tui Ibf ni Is nils nils nils
J ...................................

Air
*F

Pav e Time *F h h : n rn :s s

P U TE

POS

D D D D D D D D D D D D D D D 0 D

85336 8SJ6 85386 85422 85422 85422 85446 85446 85446 85447 85447 85447 85479 85479 85479 85508 85508

1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2

9432 12209 16222 9437 122S1 16243 9424 12231 16222 9424 12225 16209 9429 12176 16243 9404 12212

3.99 4.94 6.55 2.57 3.24 4.38 2.85 3.64 4.97 2.80 3.64 4.90 2.96 3.75 4.97 5.01 6.39

3.52 4.62 6.08 2.36 3.01 4.53 2.54 3.34 4.38 2.47 3.19 4.22 2.76 3.57 4.83 4.78 6.19

3.64 4.65 6.05 2.30 3.05 3.95 2.45 3.27 4.23 2.45 3.20 4.32 2.67 3.42 4.54 4.43 5.61

3.52 4.64 5.86 1.99 2.50 3.38 1.99 2.70 3.52 2.01 2.65 3.48 2.31 2.97 3.99 3.71 4.74

2.90 3.80 4.74 1.56 2.06 2.74 1.62 2.12 2.80 1.56 2.06 2.77 1.90 2.43 3.22 3.03 3.83

1.31 1.59 ; 2.09 0.67 1 0.90 1 1.22 I 0.71 0.98 I 1.30 0.69 I I 0.91 1.20 0.84 I

1.11

1.48
1.2 0

67.6 67.6 67.6 68.3 68.3 68.3 69.2 69.2 69.2 69.5 69.5 69.5 67.2 67.2 67.2
68.1 68.1

1.55

86.9 12:04: 06 86.9 12:04: 14 86.9 12:04: 22 88.1 12:05: 23 88.1 12:05: 29 88.1 12:05: 37 88.3 12:06: 37 88.3 12:06: 43 88.3 12:06: 51 87.4 12:07: 43 87.4 12:07: 48 87.4 12:07: 54 88.6 12:08: 51 88.6 12:08: 57 88.6 12:09: 04 88.3 12 : 10 : 06 88.3 12 : 10 : 12

LEAVE LEAVE LEAVE CENTER CENTER CENTER APPROACH APPROACH APPROACH LEAVE LEAVE LEAVE CENTER CENTER CENTER APPROACH APPROACH

OP
G A T

6-70

3~C4 aJC4/Z'0 P/<T<LO ' < i T o H ^ ) T i e


r 5508 2.79 8.06 7.28 6.22 4.95 D 3 16088 8.21 1.80 4.29 D B5509 4.65 3.90 3.20 1 9347 4.85 2.49 85509 6.01 5.07 4.21 D 2 12173 6.25 5.59 85509 7.07 6.52 5.36 3 16089 8.09 7.76 3.21 D 2.30 2.08 1.74 85535 1 9450 2.54 2.56 D 2.73 1.16 85535 3.20 2.78 2.31 0 2 12235 3.54 3.23 1.68 85535 4.27 3.65 3.03 0 3 16250 4.71 4.33 85569 2.54 2.06 1.11 D 1 9404 3.24 3.03 2.83 3.84 1.46 D 85569 3.64 3.33 2.55 2 12199 4.10 1.86 85569 3.46 D 3 16217 5.57 5.16 4.80 4.26 1.17 1.94 2.89 2.40 D 85570 1 9450 3.08 2.76 1.51 3.68 2.43 D 85570 3.94 3.50 3.13 2 12212 1.98 3.24 85570 4.09 D 3 16133 5.29 4.78 4.91 1.02 2.94 2.14 85601 2.96 2.59 D 1 9462 3.20 1.25 2.68 85601 3.68 3.33 0 4.03 3.73 2 12241 2.36 3.55 85601 4.87 D 3 16241 5.12 4.31 5.25 0.83 2.87 2.27 85631 1 9404 3.38 0 3.75 3.46 1.40 85631 3.48 2.86 D 4.71 4.44 4.23 2 12157 2.30 4.66 3.80 85631 6.27 5.57 0 3 16185 5.75 2.34 1.05 85632 3.07 2.78 0 1 9417 3.52 3.31 1.45 3.63 2.96 D 85632 4.50 4.13 4.32 2 12231 2.39 3.83 85632 6.04 4.79 D 3 16198 5.70 5.66 1.30 2.16 85664 3.07 2.98 2.54 D 1 9488 3.31 2.68 1.66 85664 3.27 D 4.20 3.95 3.86 2 12218 2.18 3.52 85664 4.91 4.22 0 3 16247 5.46 5.12 1.30 2.64 85691 3.22 D 1 9409 4.10 3.90 3.90 1.89 3.36 85691 4.03 4.98 4.71 D 2 12173 5.18 2.47 4.28 85691 6.09 5.18 D 6.76 6.42 3 16204 2.74 1.53 3.19 D 85693 3.99 3.57 3.79 1 9432 2.09 3.40 85693 4.16 4.05 D 2 12141 5.01 4.62 3.10 6.20 5.26 4.43 85693 3 16247 6.51 5.99 0 2.34 1.43 3.64 85720 1 9424 3.87 2.86 0 3.31 1.72 2.83 85720 4.23 3.62 D 4.59 4.62 2 12225 3.74 2.29 4.64 85720 6.09 D 3 16262 6.24 5.39 3.98 2.46 85733 4.50 D 1 9442 5.53 4.98 5.02 3.00 4.80 5.56 0 85733 6.79 6.25 2 12164 6.19 4.35 7.09 6.14 85733 8.09 0 3 16222 8.77 8.12 2.44 3.70 85735 1 93S4 5.01 4.40 4.93 4.26 D 4.67 5.39 3.13 85735 6.37 5.98 0 2 12109 5.52 6.04 4.24 85735 8.30 6.96 D 3 16166 7.45 7.91 3.57 2.92 2.00 85764 1 9437 3.64 3.68 D 4.03 85764 3.87 1.62 0 2 12141 5.01 4.62 4.71 4.83 2.44 85764 4.77 D 3 16166 6.55 6.08 6.09 5.96 C Test ins Coment: BOTH EDGES OF THE JOINT AND THE CENTER SLAB C Testing Comment: CONCRETE PATCHES 85799 1 9327 1.90 0.90 0 3.41 2.38 2.96 2.72 85799 1.16 D 2 12109 4.27 2.46 3.84 3.05 3.53 85799 3.14 1.99 3 16099 5.64 4.01 D 4.94 4.65 0.72 85801 1.90 D 1 9417 3.20 2.69 2.35 2.78 85801 2.36 1.06 3.08 0 2 12193 3.99 3.64 3.75 1.90 85801 3.24 3 16076 4.87 4.05 D 5.36 4.51 2.58 1.76 D 85893 1 9417 2.97 3.64 3.34 3.31 2.39 85893 3.33 0 2 12176 3.90 4.62 4.40 4.32 3.18 D 85893 3 16206 4.45 6.09 5.88 5.13 5.83 2.57 1 9372 5.47 4.80 D 85922 6.83 6.62 6.25 3.29 85922 6.14 D 2 12164 7.96 7.01 8.65 8.53 7.91 4.86 9.10 0 85922 3 16125 11.22 11.20 10.37 3.26 85924 5.18 D 6.83 6.49 6.62 5.83 1 9461 6.55 4.42 85924 8.39 8.44 7.68 0 2 12109 8.82 5.66 85924 8.32 D 3 16114 11.33 11.18 10.92 9.76 1.84 0.88 85967 D 1 9521 2.36 2.45 1.95 2.73 1.39 85967 2.54 2.40 D 2 12238 3.47 3.19 3.20 1.69 85967 4.66 3.11 0 3 16232 4.33 4.23 3.82 85983 1.32 D 1 9412 3.87 3.64 3.35 2.92 2.51 1.59 85983 D 2 12176 4.34 3.74 3.11 4.85 4.71 4.00 85983 2.23 D 3 16093 5.72 4.96 6.55 6.19 85985 2.68 2.01 D 1 9482 3.08 5.18 3.23 3.42 1.33 0 85985 2 12228 4.09 3.33 6.25 4.22 4.60 3.96 2.02 85985 3 16174 D 7.98 5.59 5.32 5.98 86016 1 9404 2.40 1.29 D 3.24 2.92 2.69 2.85 1.47 66016 2 12180 3.11 D 3.88 4.03 3.41 3.75 1.89 86016 3 16133 3.93 0 5.41 4.89 4.83 4.98

*TKQ

Pi-CO~

2.13 1.34 1.74 2.30 0.78 1.03 1.39 0.84 1.05 1.42 0.79 1.04 1.43 0.99 1.25 1.67 1.04 1.32 1.82 1.03 1.34 1.83 0.94 1.19 1.59 1.21 1.55 2.05 1.26 1.56 2.05 1.00 1.27 1.67 2.26 2.76 3.61 2.14 2.64 3.41 1.33 1.57 2.15 ARE 0.82 1.05 1.38 0.81 1.03 1.44 1.43 1.83 2.50 2.36 3.02 3.85 2.63 3.29 4.22 0.96 1.20 1.60 1.05 1.28 1.69 1.12 1.36 1.80 1.24 1.54 1.86

68.1 68.5 68.5 68.5 70.5 70.5 70.5 70.5 70.5 70.5 72.1 72.1 72.1 71.5 71.5 71.5 72.4 72.4 72.4 72.4 72.4 72.4 72.4 72.4 72.4 72.1 72.1 72.1 73.1 73.1 73.1 68.9 68.9 68.9 68.3 68.3 68.3 70.1 70.1 70.1 73.7 73.7 73.7 74.7 74.7 74.7 74.7 74.7 74.7 78.9 78.9 78.9 78.3 78.3 78.3 75.3 75.3 75.3 71.7 71.7 71.7 74.1 74.1 74.1 78.7 78.7 78.7 78.0 78.0 78.0

88.3 86.7 86.7 86.7 88.1 88.1 88.1 89.2 89.2 89.2 87.4 87.4 87.4 89.0 89.0 89.0 91.3 91.3 91.3 88.8 88.8 88.8 88.6 88.6 88.6 89.4 89.4 89.4 67.8 87.8 87.8 89.2 89.2 89.2 86.5 86.5 86.5 87.4 87.4 87.4 89.4 89.4 89.4 86.9 86.9 86.9 88.3 88.3 88.3 68.0 88.0 88.0 90.8 90.8 90.8 88.8 88.8 88.8 91.7 91.7 91.7 89.0 69.0 69.0 69.7 89.7 89.7 90.8 90.8 90.8

12:10:18 12:11:10 12:11:16 12:11:23 12:12:23 12:12:29 12:12:35 12:13:48 12:13:56 12:14:04 12:14:56 12:15:02 12:15:08 12:15:58 12:16:05 12:16:11 12:17:12 12:17:18 12:17:25 12:18:12 12:18:21 12:18:28 12:19:23 12:19:29 12:19:36 12:20:32 12:20:39 12:20:47 12:21:38 12:21:44 12:21:51 12:22:53 12:23:13 12:23:21 12:24:34 12:24:39 12:24:46 12:25:40 12:25:46 12:25:53 12:26:54 12:27:01 12:27:08 12:29:03 12:29:09 12:29:16 12:30:49 12:30:55 12:31:03 12:36:03 12:36:09 12:36:15 12:37:13 12:37:19 12:37:26 12:38:17 12:38:23 12:38:31 12:39:44 12:39:50 12:39:57 12:41:38 12:41:44 12:41:51 12:42:49 12:42:58 12:43:06 12:43:57 12:44:09 12:44:17

APPROACH LEAVE LEAVE LEAVE CENTER CENTER CENTER APPROACH APPROACH APPROACH LEAVE LEAVE LEAVE CENTER CENTER CENTER APPROACH APPROACH APPROACH LEAVE LEAVE LEAVE CENTER CENTER CENTER APPROACH APPROACH APPROACH LEAVE LEAVE LEAVE CENTER CENTER CENTER APPROACH APPROACH APPROACH LEAVE LEAVE LEAVE CENTER CENTER CENTER APPROACH APPROACH APPROACH LEAVE LEAVE LEAVE CENTER CENTER CENTER APPROACH APPROACH APPROACH LEAVE LEAVE LEAVE CENTER CENTER CENTER APPROACH APPROACH APPROACH LEAVE LEAVE LEAVE CENTER CENTER CENTER

C Q A iA f -u
L ^ i x j . $

ozAJs sverno
A \> J
GO^O

6-71

6.4.4 PH0NIX FWD


The following description was provided by Resource International:

6-72

++

* *

* W*

1 1 1 1 1 I 1 1 1 1

Client *
3+C. OO.i

000J OMi

Link p o .i

RESOURCE | InLematiooal Inc. U a t xel.t sure ci Surtocc: Point 1S1/1


3*9/1 Roncxy 5- n _ _

D esign da*:

06-16-199 >3 0511

M i . dAZat

SOCTJIWEiT EKD_ F t U ____ O w n . A.load psl STAX__ 267. 9 264.3 264.3 29.C 288.3 273.3 280.4 TAX T_ 269.6 284.7 283.6 280.4 2B6.S 264.7 293. 293- 280.4 287.) 289.8 280.4 BE1432 289.2 Ari432 282.1 AP 1 262.S 267.0 259.0 END 281.2 Dfcf.l nils 27.21 33.31 24.57 37.C5 21.58 34 .89 28.78 37.60 20. 25.44 27.99 1C. 07 16.30 19.26 15.71 34.S3 22.01 21.19 >4.89 22.95 32.37 63.51 61.89 70.95 30.83 Dei .2 mil* 13.94 20.79 20.12 22.(4 15.94 22.80 18.90 23.01 16.30 19.09 18.15 11.97 10.35 13.83 11.61 23.58 16.34 15.94 23.54 13.86 22.95 41.89 41.34 41.65 21.77 Dci.l dla 9.02 14.49 15.91 16.18 13.50 16.30 13.70 17.05 14.21 16.22 13.86 10.35 14.65 13.35 9.9 18.03 13.62 13.46 17.44 9.80 17.60 29.41 29.13 31.77 17.36 Dei .4 >11 5.98 10.12 12.20 10.20 10.51 11.02 9.49 12.09 11.26 12.48 9.69 7.91 10.75 10.47 7.83 11.93 10.20 10 63 13.56 5.87 11.77 17.68 1890 20.63 13. IS Df .5 alia 4.29 7 .05 8.70 6.38 7.40 7.60 6.S4 7.56 7.95 8.27 6.46 5.47 7.28 7.40 5.47 7.24 6. 97 7.64 8.23 3.70 7.01 10.91 11.10 11. S4 10.24 Dei.6 Temp r die .2.72 J.98 S. 16 3.62 4.17 4.53 4.13 3.94 4.17 3.86 J .58 2.83 3.66 3.94 3.03 3.S4 3.74 4.37 4.37 2.40 3 .23 5.28 4.21 4.69 6.81 70 69 71 71 69 69 70 68 76 69 68 7 67 67 68 68 68 63 65 66 65, 66 65 66 66

553/i 155/1 ssiyi 1149/1 1350/1 1551/1 1751/1 1951/1 2150/1 2350/1 2551/1 2751/1 2951/1 31S0/1 33S1/1 355o/l 3750/1 3950/1 4169/1 4350/1 4551/1 4752/1 4950/1 1. 2. 3. 4.
5.

Colusa 1 - Point Column 2 - Remark

Column 3 - A.Load, psi Co 1 m m 4 to 9 - Dei. 1 to Def 6 , mil C o lu m n 10 - Temp.F Keasured temperature of paveoent at the time of test, F.

Location of FWD station in feet oeasi starting point noted ac the beginning of the table. Refer to additional remarlts entered by the operator In Che field. Applied load on the plate, pi These are the measured deflections in mils

6-73

6.5 ADDITIONAL PROJECT DATA


The following FWD deflection data files are included in directions on the on the DATA diskette, and were generated over the past few years for actual pavement evaluation pro jects. Each file contains numerous data points and they reflect the pavement variability typically encountered in these projects. Some thought should be given to how one should deal with the complete project data set. The data has been chosen to illustrate a variety o f pavement struc tures, some o f which may be considered "interesting". The raw FWD data files are identified by the extension .FWD. Normalized deflections for the test drop closest to 40 kN (9,000 lb.) have been included in ,DEF files. 6.5.1 Project Data a) AC Subdirectory (i) Medium to Thick AC Pavement Files: ACLRN.FWD, ACLRC.FWD, ACLRS.FWD for NB, center and SB lanes. Test Points: 1162 to 2432 - 100 mm (4") AC and 750 - 900 mm (30-36") base Test Points: 2432 to 25900 - 38 mm (l'/2") AC, 175 mm (7") ATB and 750-900 mm (30"-36") base [Equation: 3953 back = 23629 ahead] AC binder is probably an AC-5. MR values for AC from laboratory tests at 0.1 sec, 10 C (50 F) vary from 3,100 to 6,200 MPa (450 to 900 ksi) for five cores. "Base" consists o f two or three unbound granular layers o f moderate to good quality. The pavement is approximately seven to eight years old. Condition is

6-74

generally good. Design traffic is approximately 260,000 ESALs over 10 years. Center deflections for the ACLR files are shown in Figure 6.29. Project evaluation focused on test points 2432-25900. (ii) Medium AC Pavement File: AC281.FWD Test Points: 200.68 to 205.008 All Points: 138 mm (5.5") AC on 300 mm (12") granular base. Little additional information is available except that the AC binder may be an AC-5 material. Traffic is low (~ 140,000 ESALs over 20 years) and the pavement surface appears aged. There is some localized patching. b) PCC Subdirectory These projects include PCC joint tests. Typically slab centers and joints are tested in the morning, and joints retested in the afternoon. Slabs are numbered XX, with XX. 1 identifying center tests, XX.2 for AM tests and XX.22 for PM tests. D2 and D3 span the joints. (i) Files: PCI2WBAM.FWD morning tests PC12WBPM.FWD afternoon tests

All Points: 175 mm (7") PCC, 100 mm (4") LTB, 150 mm (6") LTSG. Slabs 1 through 13 have been undersealed. Design traffic is approximately 1.70 x 106 ESAL's. Figure 6.30 shows joint load transfer efficiency for AM and PM tests.

6-75

(ii)

Files: PC 12EBAM.FWD morning tests PC12EBPM.FWD afternoon tests All Points: 175 mm (7") PCC, 100 mm (4") CTB.

Slab 29 is cracked near the center. Design traffic is 1.75 x 106 ESAL's. c) THAW Subdirectory Thin AC spring thaw period tests.

Files: THAW1.FWD, THAW2.FWD, THAW3 .FWD for 4/22/91, 4/29/91 and 9/04/91 All Points: 50 mm (2") AC, 150-300 mm (6"-12") base AC binder is probably an AC-5. There was little information available regarding base material at the time o f the tests. Subsequent excavation revealed a schist material that had degraded very extensively. Pavement condition was very poor with extensive patching and surface distress (cracking and distor tion). FWD data was gathered during spring thaw on 4/22/91, 4/29/91 and 9/04/91 at the same test locations, based on DMI positioning. The 9/04/91 data was used for reference purposes. Center de flections for the THAW files are plotted in Figure 6.31. This data was not used for the project evaluation.

6-76

d)

COMPOSIT Subdirectory File: COMPW7.FWD (morning tests) All Points: 50 mm (2") AC, 150 mm (6") PCC This is an old overlay with most PCC joints reflected through the AC. Center and joint tests were carried out and are included in file COMPW7.FWD. As before XXX. 1 identifies slab centers and XXX.2 are the joint tests with sensors two and three on either side o f the joint. Traffic is 1 x 106 ESALs over 20 years. Analysis suggested that the pavement was essentially structurally adequate for the design traffic.

e)

RUBBLE Subdirectory File: RUBL75N.FWD All Points: 108 mm (4'A) AC, 250 mm (10") rubblized PCC and 75 mm (3") ATB

This is a fairly new section tested for research pur poses and the surface appearance is generally good. Design traffic is 3 x 106 ESALs over 20 years. The intent o f the project was to evaluate structural re sponse o f the rubblized material.

6-77

ACLR 9 KIP DEFLECTIONS

DEFLECTION

(M IL)

D IS T. FRO M S T A . 1 1 + 6 2

Figure 6 .2 9 - M edium A C Pavement Center Deflections

PC12WB JOINT LOAD TRANSFER EFFICIENCY

L E (% T

*----- AM TESTS ----- PM TESTS

LOCATION

Fiugre 6.30 - PC12W B Joint R esponse

SPRING THAW 9 KIP DEFLECTIONS

DEFLECTION (MIL)

----- ------ -----

4 /22/91 4/29/91 9/4/91

10

11
LOCATION

12

13

14

F igure 6.31 - Spring Thaw Deflections

SECTION 6.0 REFERENCES

Haiping, Z., Hicks, R.G ., and Bell, C.A., "BOUSDEF: A Backcalculation Program for Determining Moduli o f a Pavement Structure," Paper Presented to the Transportation Research Board, Washington, D.C ., 1990. "User's Guide to the BO U SD EF Program," Civil Engineering Department, Oregon State University, Corvallis. Ullidtz, P., Pavement Analysis, Elsevier, 1987. Lee, S.W ., "Backcalculation o f Pavement M oduli by U se o f Pavement Surface Deflections," Ph.D. Dissertation, University o f Washington, Seattle, 1988. "Users Guide to EVERCALC," Civil Engineering Department, University o f Washington, Seattle, 1991. M ichelow, J., "Analysis o f Stresses and D isplace ments in an n-Layered Elastic System Under a Load Uniformly Distributed on a Circular Area," Cali fornia Research Corporation, Richmond, California, September 1963. Southgate, H.F. and Deen, R.C., "Temperature D is tributions in Asphalt Concrete Pavements," Highway Research Record N o. 549, Highway Research Board, Washington, D .C ., 1975. Uzan, J., et al., "A Microcomputer Based Procedure for Backcalculating Layer Moduli from FW D Data," Research Report N o. 1123-1, Texas Transportation Institute, College Station, Texas, September 1988.

6-81

Lytton, R.L., Germann, F.P., Chou, Y.J., and Stoffels, S.M ., "Determining Asphaltic Concrete Pavement Structural Properties by Nondestructive Testing," Report N o. 327, National Cooperative Highway Research Program, Washington, D.C ., June 1990. Rohde, G.T., and Scullion, T., "MODULUS 4.0: Expansion and Validation o f the M odulus Backcalculation System", Research Report 1123-3, Texas Transportation Institute, Novem ber 1990.

6-82

R esu lts from

BOUSDEF EVERCALC MODULUS MODCOMP for SHRP Section C

6-83

SHRP Pavement Section C (GPS-2: Asphalt Concrete Pavement with Bound Base Section Located in North Carolina)

6-84

BOUSDEF Program SHRP Section C Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6616 lb. 667.4 (4601) 532.4 (3671) 36.9 (254) 2 9522 lb. 673.3 (4642) 429.9 (2964) 38.2 (263) 3 12,958 lb. 675.5 (4658) 451.4 (3112) 37.4 (258) 4 16,696 lb. 703.4 (4850) 496.4 (3423) 39.1 (270)

Soil Cement Base

Silty Sand Subgrade

Deflection, mils Load 0" (0 mm) 3.39 3.38 0.4 4.87 4.86 0.3 6.64 6.62 0.3 8.12 8.09 0.3 8" (203 mm) 2.72 2.62 3.6 3.89 3.77 3.1 5.30 5.16 2.7 6.48 6.30 2.8 12" (305 mm) 2.37 2.41 -1.9 3.40 3.47 -2.0 4.69 4.76 -1.4 5.73 5.82 -1.6 18" (457 mm) 2.03 2.11 -3.8 2.95 3.04 -2.9 4.07 4.18 -2.6 4.98 5.12 -2.8 24" (610 mm) 1.77 1.82 -2.8 2.57 2.63 -2.5 3.57 3.64 -1.9 4.39 4.46 -1.7 36" (914 mm) 1.38 1.35 1.9 2.03 1.99 2.0 2.83 2.76 2.6 3.47 3.39 2.3 60" (1514 mm) 0.89 0.83 6.2 1.31 1.25 4.3 1.79 1.74 2.7 2.21 2.14 3.0 ARS = 14.6 RMS = 2.3 ARS = 14.4 RMS = 2.2 ARS = 17.3 RMS = 2.8 ARS = 21.3 RMS = 3.6 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-85

BOUSDEF, SHRP Section C, Run No. 1


Layer Number 1 2 3 Seed Moduli, ksi (MPa) 1000.0 (6895) 200.0 (1379) 15.0 (103) Range of Modulus, ksi (MPa) Minimum 100.0 (690) 50.0 (345) 1.0 (6.9) Maximum 3000.0 (20685) 2000.0 (13790) 50.0 (345) Poisson's Ratio 0.35

0.25 0.40

Deflection Convergence Criterion: ARS < 10% Maximum Iterations Criterion: 9

6-8 6

BOUSDEF Program SHRP Section C Run No. 2


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6616 lb. 553.3 (3815) 989.6 (6823) 28.4 (196) 1000.0 (6895) 2 9522 lb. 540.5 (3727) 1029.9 (7101) 27.4 (189) 1000.0 (6895) 3 12,958 lb. 543.9 (3750) 1073.1 (7399) 26.8 (185) 1000.0 (6895) 4 16,696 lb. 601.3 (4146) 1207.0 (8322) 27.8 (192) 1000.0 (6895)

Soil Cement Base

Silty Sand Subgrade

Rigid Base @ 300" (7.6 m)

Deflection, mils Load 0" (0 mm) 3.39 3.36 0.8 4.87 4.83 0.8 6.64 6.59 0.7 8.12 7.92 2.5 8" (203 mm) 2.72 2.57 5.6 3.89 3.67 5.6 5.30 5.04 5.0 6.48 6.11 5.7 12" (305 mm) 2.37 2.41 -1.5 3.40 3.45 -1.6 4.69 4.74 -1.1 5.73 5.77 -0.7 18" (457 mm) 2.03 2.15 -5.7 2.95 3.10 -5.2 4.07 4.27 -5.0 4.98 5.22 -4.8 24" (610 mm) 1.77 1.88 -5.9 2.57 2.73 -6.2 3.57 3.77 -5.6 4.39 4.63 -5.4 36" (914 mm) 1.38 1.39 -0.4 2.03 2.04 -0.5 2.83 2.83 0 3.47 3.50 -0.9 60" (1514 mm) 0.89 0.77 13.8 1.31 1.41 12.9 1.79 1.59 11.0 2.21 1.99 10.1 ARS = 30.0 RMS = 5.2 ARS = 28.4 RMS = 5.4 ARS = 27.4 RMS = 4.8 ARS = 34.4 RMS = 6.4 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-87

BOUSDEF, SHRP Section C, Run No. 2


Layer Number 1 2 3 4 Seed Moduli, ksi (MPa) 1000.0 (6895) 200.0 (1379) 15.0 (103) 1000.0 (6895) Range of Modulus, ksi (MPa) Minimum 100.0 (690) 50.0 (345) 1.0 (6.9) 1000.0 (6895) Maximum 3000.0 (20685) 2000.0 (13790) 50.0 (345) 1000.0 (6895) Poisson's Ratio 0.35 0.25 0.40 0.50

Deflection Convergence Criterion: ARS < 10% Maximum Iterations Criterion: 9

6-88

EVERCALC Program SHRP Section C Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6616 lb. 588.3 (4056) 383.9 (2647) 37.3 (257) 2 9522 lb. 567.2 (3911) 438.5 (3024) 36.5 (252) 3 12,958 lb. 589.5 (4065) 428.1 (2952) 36.0 (248) 4 16,696 lb. 612.0 (4220) 471.0 (3248) 37.6 (259)

Soil Cement Base

Silty Sand Subgrade

Deflection, mils Load 0" (0 mm) 3.39 3.39 0.0 4.87 4.87 0 6.64 6.65 -0.1 8.12 8.13 -0.1 8" (203 mm) 2.72 2.65 2.7 3.89 3.79 2.5 5.30 5.21 1.7 6.48 6.37 + 1.8 12" (305 mm) 2.37 2.37 0 3.40 3.41 -0.3 4.69 4.70 -0.1 5.73 5.74 -0.2 18" (457 mm) 2.03 2.06 -1.6 2.95 2.99 -1.2 4.07 4.12 -1.1 4.98 5.04 -1.2 24" (610 mm) 1.77 1.81 -2.1 2.57 2.63 -2.3 3.57 3.62 -1.5 4.39 4.45 -1.3 36" (914 mm) 1.38 1.39 -0.9 2.03 2.04 -0.6 2.83 2.82 0.4 3.47 3.47 0 60" (1514 mm) 0.89 0.87 2.4 1.31 1.28 2.1 1.79 1.77 0.9 2.21 2.19 +1.1 ARS = 5.7 RMS = 1.0 ARS = 5.8 RMS = 1.0 ARS = 9.1 RMS = 1.6 ARS = 10.0 RMS = 1.7 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-89

EVERCALC, SHRP Section C, Run No. 1


Layer Number 1 2 3 Seed Moduli, ksi (MPa) 1000.0 (6895) 200.0 (1379) 15.0 (103) Range of Modulus, ksi (MPa) Minimum Maximum Poisson's Ratio 0.35 0.25 0.40

Deflection Convergence Criterion: RMS < 1% Modulus Convergence Criterion: < 1% Maximum Iteration Criterion: 10

EVERCALC Program SHRP Section C Run No. 2


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6616 lb. 755.1 (5207) 552.0 (3806) 28.9 (199) 1000.0 (6895) 2 9522 lb. 745.6 (5142) 617.3 (4257) 28.2 (194) 1000.0 (6895) 3 12,958 lb. 792.1 (5462) 588.3 (4057) 27.7 (191) 1000.0 (6895) 4 16,696 lb. 828.3 (5712) 642.5 (4431) 29.0 (200) 1000.0 (6895)

Soil Cement Base

Silty Sand Subgrade

Rigid Base @ 300" (7.6 m)

Deflection, mils Load 0" (0 mm) 3.39 3.37 0.6 4.87 4.84 0.6 6.64 6.61 0.4 8.12 8.08 0.4 8" (203 mm) 2.72 2.68 1.6 3.89 3.84 1.2 5.30 5.29 0.3 6.48 6.46 0.3 12" (305 mm) 2.37 2.33 1.8 3.40 3.35 1.5 4.69 4.61 1.6 5.73 5.64 1.5 18" (457 mm) 2.03 2.03 0 2.95 2.94 0.3 4.07 4.05 0.4 4.98 4.96 0.3 24" (610 mm) 1.77 1.82 -3.0 2.57 2.65 -3.1 3.57 3.65 -2.3 4.39 4.48 -2.0 36" (914 mm) 1.38 1.44 -4.3 2.03 2.11 -3.8 2.83 2.91 -2.7 3.47 3.57 -3.0 60" (1514 mm) 0.89 0.85 4.2 1.31 1.26 3.7 1.79 1.75 2.5 2.21 2.15 2.6 ARS = 10.2 RMS = 1.8 ARS = 10.2 RMS = 1.8 ARS = 14.1 RMS = 2.4 ARS = 15.6 RMS = 2.7 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-91

EVERCALC, SHRP Section C, Run No. 2


Layer Number 1 2 3 4 Seed Moduli, ksi (MPa) 1000.0 (6895) 200.0 (1379) 15.0 (103) 1000.0 (6895) Range of Modulus, ksi (MPa) Minimum 1000.0 (6895) Maximum 1000.0 (6895) Poisson's Ratio 0.35 0.25 0.40 0.50

Deflection Convergence Criterion: RMS < 1% Modulus Convergence Criterion: < 1% Maximum Iteration Criterion: 10

6-92

EVERCALC Program SHRP Section C Run No. 3


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6616 lb. 515.3 (3554) 587.2 (4050) 35.1 (242) 1000.0 (6895) 2 9522 lb. 497.6 (3431) 684.0 (4717) 34.3 (236) 1000.0 (6895) 3 12,958 lb. 513.8 (3543) 668.2 (4608) 33.7 (232) 1000.0 (6895) 4 16,696 lb. 534.4 (3686) 738.5 (5093) 35.3 (243) 1000.0 (6895)

Soil Cement Base

Silty Sand Subgrade

Rigid Base @ 999" (25.4 m)

Dcflection, mils Load 0" (0 mm) 3.39 3.38 0.2 4.87 4.86 0.2 6.64 6.63 0.1 8.12 8.11 0.1 8" (203 mm) 2.72 2.61 4.0 3.89 3.74 3.9 5.30 5.14 3.0 6.48 6.28 3.1 12" (305 mm) 2.37 2.36 0.2 3.40 3.40 0.0 4.69 4.68 0.1 5.73 5.73 0 18" (457 mm) 2.03 2.09 -2.8 2.95 3.02 -2.4 4.07 4.16 -2.2 4.98 5.10 -2.4 24" (610 mm) 1.77 1.83 -3.6 2.57 2.67 -3.7 3.57 3.68 -3.0 4.39 4.51 -2.7 36" (914 mm) 1.38 1.40 -1.5 2.03 2.05 -1.1 2.83 2.83 0 3.47 3.49 -0.4 60" (1514 mm) 0.89 0.85 4.0 1.31 1.26 3.6 1.79 1.75 2.4 2.21 2.16 2.5 ARS = 11.4 RMS = 2.1 A R S= 11.1 RMS = 2.0 ARS = 15.0 RMS = 2.7 ARS = 16.4 RMS = 2.8 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-93

EVERCALC, SHRP Section C, Run No. 3


Layer Number 1 2 3 4 Seed Moduli, ksi (MPa) 1000.0 (6895) 200.0 (1379) 15.0 (103) 1000.0 (6895) Range of Modulus, ksi (MPa) Minimum 1000.0 (6895) Maximum 1000.0 (6895) Poisson's Ratio 0.35 0.25 0.40 0.50

Deflection Convergence Criterion: RMS < 1% Modulus Convergence Criterion: < 1% Maximum Iteration Criterion: 10

6-94

MODULUS Program SHRP Section C Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6616 lb. 925.0 (6378) 293.4 (2023) 30.8 (212) 2 9522 lb. 904.0 (6233) 325.8 (2246) 30.1 (208) 3 12,958 lb. 886.0 (6109) 340.7 (2349) 29.5 (203) 4 16,696 lb. 927.0 (6392) 370.1 (2552) 30.9 (213)

Soil Cement Base

Silty Sand Subgrade

Rigid Base @300" (7.6 m) (selected by program)

Deflection, mils Load 0" (0 mm) 3.39 3.24 4.6 4.87 4.66 8" (203 mm) 2.72 2.69 1.1 3.89 3.86 12" (305 mm) 2.37 2.43 -2.5 3.40 3.49 18" (457 mm) 2.03 2.11 -3.3 2.95 3.06 24" (610 mm) 1.77 1.84 -3.8 2.57 2.67 60" 36" (914 mm) (1514 mm) 1.38 1.39 -0.7 2.03 2.42 0.89 0.80 11.2 1.31 1.19 ARS = 43.7 RMS = 8.5 6.64 6.38 5.30 5.28 4.69 4.79 4.07 4.20 3.57 3.69 2.83 2.83 1.79 1.66 ARS = 20.2 RMS = 3.7 8.12 7.79 6.48 6.46 5.73 5.86 4.98 5.15 4.39 4.52 3.47 3.47 2.21 2.04 ARS = 20.7 RMS = 3.8 ARS = 26.8 RMS = 4.8 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-95

MODULUS, SHRP Section C, Run No. 1


Layer Number 1 2 3 4 *Selected by program Seed* Moduli, ksi (MPa) ? ? Range of Modulus,* ksi (MPa) Minimum 636.0 (4385) 100.0 (690) Maximum 1750.4 (12069) 3000.0 (20685) Poisson's Ratio* 0.35 0.35 0.40

15.0 (103) ?

6-96

MODCOMP Program SHRP Section C Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 66161b. 661.2 (4560) 308.9 (2130) 38.6 (266) 2 9522 lb. 616.3 (4250) 349.5 (2410) 38.3 (264) 3 12,958 lb. 636.6 (4065) 375.6 (2590) 36.7 (253) 4 16,696 lb. 687.3 (4740) 400.2 (2760) 38.3 (264)

Soil Cement Base

Silty Sand Subgrade

Deflection, mils Load 0" (0 mm) 3.39 3.39 0.0 4.87 4.87 0 6.64 6.65 0 8.12 8.11 0 8" (203 mm) 2.72 2.68 -2.1 3.89 3.82 -2.4 5.30 5.24 -1.5 6.48 6.42 -1.3 12" (305 mm) 2.37 2.36 0.2 3.40 3.39 0.1 4.69 4.69 0.1 5.73 5.75 0.1 18" (457 mm) 2.03 2.05 -0.3 2.95 2.95 -0.6 4.07 4.09 0.8 4.98 5.04 0.5 24" (610 mm) 1.77 1.77 0 2.57 2.56 0 3.57 3.58 0 4.39 4.4 0 36" (914 mm) 1.38 1.38 0 2.03 1.97 -4.1 2.83 2.76 -2.3 3.47 3.43 -1.4 60" (1514 mm) 0.89 0.91 +2.2 1.31 1.22 -6.4 1.79 1.73 -4.4 2.21 2.52 -2.9 ARS = 6.2 RMS = 1.3 ARS = 9.1 RMS = 2.0 ARS = 13.6 RMS = 3.0 ARS = 4.8 RMS = 3.2 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-97

MODCOMP, SHRP Section C, Run No. 1


Layer Number " f '- v 2 3 Seed Moduli, ksi (MPa) 1000.0 (6895) 200.0 (1379) 15.0 (103) Range of Modulus, ksi (MPa) Minimum Maximum Poisson's Ratio

0.35 0.25 0.40

Deflection Convergence Criterion: RMS < 1% Modulus Convergence Criterion: < 1% Maximum Iteration Criterion: 10

MODCOMP Program SHRP Section C Run No. 2


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6616 1b. 630.8 (4350) 368.3 (2540) 32.5 (224) 1000.0 (6895) 2 9522 lb. 594.5 (4100) v 410.4 . (2830) 32.2 (222) 1000.0 (6895) 3 12,958 lb. 614.8 (4240) 437.9 (3020) 30.7 (212) 1000.0 (6895) 4 16,696 lb. 662.7 (4570) 468.4 (3230) 32.0 (221) 1000.0 (6895)

Soil Cement Base

Silty Sand Subgrade

Rigid Base @ 300" (7.6 m)

Deflection, mils Load 0" (0 mm) 3.39 3.39 0 4.87 4.88 0 6.64 6.65 0 8.12 8.11 0 8" (203 mm) 2.72 2.64 -2.4 3.89 3.78 -2.5 5.30 5.24 -1.6 6.48 6.42 -1.4 12" (305 mm) 2.37 2.36 .1 3.40 3.30 .1 4.69 4.69 .1 5.73 5.75 .1 18" (457 mm) 2.03 2.05 -.1 2.95 2.95 -.3 4.07 4.09 1.1 4.98 5.04 .7 24" (610 mm) 1.77 1.77 0 2.57 2.56 0 3.57 3.58 0 4.39 4.41 0 36" 60" (914 mm) (1514 mm) 1.38 1.34 -3.8 2.03 1.93 -5.9 2.83 2.72 -4.1 3.47 3.35 -3.1 0.89 0.75 -18.0 1.31 1.10 -16.2 1.79 1.57 -14.1 2.21 1.93 -12.6 ARS = 17.9 RMS = 4.9 ARS = 21.0 RMS = 5.6 ARS = 25.0 RMS = 6.6 ARS = 24.4 RMS = 6.9 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-99

MODCOMP, SHRP Section C, Run No. 2


Layer Number 1 2 3 4 Seed Moduli, ksi (MPa) 1000.0 (6895) 200.0 (1379) 15.0 (103) 1000.0 (6895) Range of Modulus, ksi (MPa) Minimum Maximum 1000.0 (6895) Poisson's Ratio 0.35 0.25 0.40 0.50

1000.0 (6895)

Deflection Convergence Criterion: RMS < 1% Modulus Convergence Criterion: < 1% Maximum Iteration Criterion: 10

6-100

Results from

BOUSDEF EVERCALC MODULUS MODCOMP for SHRP Section F

6-101

Asphalt Concrete 7.65 in. (194 mm) Crushed Limestone Base 14.47 in. (368 mm)

Silty Sand Subgrade 175.88 in. (4,467 mm) or<

Possible Shale Rock Layer @ 16.5 ft (198 in.) (5 m)

SH R P Pavement Section F (GPS-6A: AC Overlay of AC Pavement Section Located in Kentucky)

6-102

BOUSDEF Program SHRP Section F Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6534 lb. 979.5 (6754) 99.9 (689) 45.4 (313) 2 95121b. 1065.8 (7349) 66.2 (456) 45.3 (312) 3 12,662 lb. 1053.4 (7263) 55.6 (383) 43.9 (303) 4 16,8121b. 1180.3 (8138) 48.6 (335) 42.7 (294)

Crushed Limestone Base

Silty Sand Subgrade

Deflection, mils Load 0" (0 mm) 3.28 3.30 -0.8 5.07 5.15 -1.5 7.28 7.26 0.3 9.71 9.70 0.2 8" (203 mm) 2.69 2.63 2.3 4.32 4.19 3.0 5.97 5.95 0.3 8.17 8.10 0.9 12 (305 mm) 2.33 2.32 0.3 3.67 3.71 -1.1 5.17 5.27 -1.9 7.11 7.22 -1.6 18" (457 mm) 1.88 1.90 -1.0 2.99 3.01 -0.6 4.26 4.25 0.1 5.88 5.88 0 24" (610 mm) 1.56 1.56 0 2.40 2.43 -1.4 3.49 3.42 2.0 4.81 4.74 1.5 36" (914 mm) 1.09 1.10 -1.4 1.69 1.68 0.4 2.37 2.34 1.4 3.29 3.23 2.0 60" (1514 mm) 0.68 0.67 2.0 1.01 1.00 0.8 1.33 1.38 -3.4 1.83 1.88 -3.0 ARS = 8.5 RMS = 1.5 ARS = 9.8 RMS = 1.8 ARS = 9.2 RMS = 1.5 ARS = 6.7 RMS = 1.2 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-103

BOUSDEF, SHRP Section F, Run No. 1


Layer Number 1 2 3 Seed Moduli, ksi (MPa) 350.0 (2413) 30.0 (207) 15.0 (103) Range of Modulus, ksi (MPa) Minimum 100.0 (690) 5.0 (34.5) 1.0 (6.9) Maximum 2000 (13790) 150.0 (1034) 50.0 (345) Poisson's Ratio 0.35 0.35 0.40

Deflection Convergence Criterion: ARS < 10% Maximum Iterations Criterion: 9

6-104

BOUSDEF Program SHRP Section F Run No. 2


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6534 lb. 921.0 (6350) 5.0 (34.5) 30.5 (210) 1000.0 (6895) 2 95121b. 940.9 (6488) 117.1 (807) 28.8 (199) 1000.0 (6895) 3 12,662 lb. 849.6 (5858) 110.1 (759) 27.4 (189) 1000.0 (6895) 4 16,812 lb. 956.3 (6594) 99.7 (687) 26.2 (181) 1000.0 (6895)

Crushed Limestone Base

Silty Sand Subgrade

Rigid Base @ 198" (5.0 m)

Deflection, mils Load 0" (0 mm) 3.28 7.54 -129.9 5.07 5.13 -1.2 7.28 7.31 -0.4 9.71 9.77 -0.7 8" (203 mm) 2.69 6.61 -145.7 4.32 4.12 4.7 5.97 5.83 2.4 8.17 7.96 2.5 12" (305 mm) 2.33 5.99 -157.2 3.67 3.68 -0.2 5.17 5.19 -0.4 7.11 7.13 -0.3 18" (457 mm) 1.88 4.93 -162.1 2.99 3.06 -2.3 4.26 4.31 -1.1 5.88 5.94 -1.0 24" (610 mm) 1.56 3.89 -149.3 2.40 2.54 -5.8 3.49 3.56 -2.1 4.81 4.92 -2.4 60" 36" (914 mm) (1514 mm) 1.09 2.28 -109.2 1.69 1.75 -3.8 2.37 2.45 -3.5 3.29 3.40 -3.2 0.68 0.77 -13.3 1.01 0.87 13.8 1.33 1.21 8.9 1.83 1.68 8.0 ARS = 18.3 RMS = 3.6 ARS = 18.7 RMS = 3.9 ARS = 31.7 RMS = 6.2 ARS = 866.7 RMS = 132.8 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-105

BOUSDEF, SHRP Section F, Run No. 2


Layer Number 1 2 3 4 Seed Moduli, ksi (MPa) 350.0 (2413) 30.0 (207) 15.0 (103) 1000.0 (6895) Range of Modulus, ksi (MPa) Minimum 100.0 (690) 5.0 (34.5) 1.0 (6.9) 1000.0 (6895) Maximum 2000 (13790) 150.0 (1034) 50.0 (345) 1000.0 (6895) Poisson's Ratio 0.35 0.35 0.40 0.50

Deflection Convergence Criterion: ARS < 10% Maximum Iterations Criterion: 9

6-106

EVERCALC Program SHRP Section F Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6534 lb. 980.0 (6759) 91.5 (631) 45.7 (315) 2 95121b. 1028.8 (7095) 69.7 (481) 44.3 (305) 3 12,662 lb. 1149.2 (7926) 50.5 (348) 43.8 (302) 4 16,812 lb. 1237.6 (8535) 45.4 (313) 42.5 (293)

Crushed Limestone Base

Silty Sand Subgrade

Deflection, mils Load 0" (0 mm) 3.28 3.32 -1.1 5.07 5.17 -2.0 7.28 7.24 0.6 9.71 9.72 -0.1 8" (203 mm) 2.69 2.66 1.3 4.32 4.19 2.9 5.97 5.98 -0.1 8.17 8.12 0.6 12" (305 mm) 2.33 2.31 0.9 3.67 3.65 0.5 5.17 5.24 -1.3 7.11 7.16 -0.7 18" (457 mm) 1.88 1.89 -0.4 2.99 2.98 0.4 4.26 4.27 -0.3 5.88 5.88 0 24" (610 mm) 1.56 1.56 0 2.40 2.44 -1.8 3.49 3.49 0 4.81 4.82 -0.2 36" 60" (914 mm) (1514 mm) 1.09 1.12 -2.4 1.69 1.71 -1.2 2.37 2.39 -0.6 3.29 3.30 -0.3 0.68 0.67 1.7 1.01 ARS = 7.8 RMS = 1.3 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

1.00
1.3 1.33 1.34 -0.6 1.83 1.84 -0.3

ARS = 10.0 RMS = 1.6

ARS = 3.7 RMS = 0.6

ARS = 2.2 RMS = 0.4

6-107

EVERCALC, SHRP Section F, Run No. 1


Layer Number 1 2 3 Seed Moduli, ksi (MPa) Used Default Moduli Equation Used Default Moduli Equation Used Default Moduli Equation Range of Modulus, ksi (MPa) Minimum Maximum Poisson's Ratio 0.35 0.35 0.40

Deflection Convergence Criterion: RMS < 1% Modulus Convergence Criterion < 1% Maximum Iterations Criterion: 10

6-108

EVERCALC Program SHRP Section F Run No. 2


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6534 lb. 846.3 (5837) 158.4 (1092) 28.9 (199) 1000.0 (6895) 2 9512 1b. 859.9 (5930) 128.8 (888) 27.9 (192) 1000.0 (6895) 3 12,662 lb. 872.1 (6014) 105.7 (729) 27.3 (188) 1000.0 (6895) 4 16,812 lb. 975.3 (6726) 97.0 (669) 26.2 (181) 1000.0 (6895)

Crushed Limestone Base

Silty Sand Subgrade

Rigid Base @ 198" (5.0 m)

Deflection, mils Load 0" (0 mm) 3.28 3.39 -3.3 5.07 5.29 -4.4 7.28 7.49 -2.9 9.71 10.05 -3.5 8" (203 mm) 2.69 2.54 5.7 4.32 4.01 7.1 5.97 5.75 3.7 8.17 7.83 4.1 12" (305 mm) 2.33 2.21 5.4 3.67 3.49 4.9 5.17 4.99 3.4 7.11 6.85 3.7 18" (457 mm) 1.88 1.91 -1.4 2.99 2.99 0 4.26 4.25 0.1 5.88 5.87 0.2 24" (610 mm) 1.56 1.63 -4.5 2.40 2.54 -5.8 3.49 3.58 -2.5 4.81 4.95 -2.9 36" 60" (914 mm) (1514 mm) 1.09 1.18 -7.8 1.69 1.80 -6.7 2.37 2.50 -5.5 3.29 3.46 -5.2 0.68 0.62 8.2 1.01 0.94 7.4 1.33 1.27 4.7 1.83 1.75 4.5 ARS = 24.1 RMS = 3.7 ARS = 23.0 RMS = 3.6 ARS = 36.3 RMS = 5.6 ARS = 36.3 RMS = 5.8 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-109

EVERCALC, SHRP Section F, Run No. 2


Layer Number 1 2 3 4 Seed Moduli, ksi (MPa) 700.0 (4826) 35.0 (241) 20.0 (138) 1000.0 (6895) Range of Modulus, ksi (MPa) Minimum 1000.0 (6895) Maximum 1000.0 (6895) Poisson's Ratio 0.35 0.40 0.45 0.50

Deflection Convergence Criterion: RMS < 1% Modulus Convergence Criterion < 1% Maximum Iterations Criterion: 10

6-110

MODULUS Program SHRP Section F Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6534 lb. 840.0 (5792) 129.9 (896) 31.8 (219) 300" (7.6 m) 2 9512 lb. 903.0 (6226) 99.1 (683) 30.9 (213) 209" (5.3 m) 3 12,662 lb. 859.0 (5923) 87.2 (601) 29.8 (205) 170" (4.3 m) 4 16,812 lb. 936.0 (6454) 80.9 (558) 28.7 (198) 165" (0.6 m)

Crushed Limestone Base

Silty Sand Subgrade

Rigid Base @

Deflection, mils Load 0" (0 mm) 3.28 3.31 -0.9 5.07 5.15 -1.6 7.28 7.31 -0.5 9.71 9.82 -1.2 8" (203 mm) 2.69 2.63 2.3 4.32 4.16 3.8 5.97 5.91 1.1 8.17 8.04 1.6 12" (305 mm) 2.33 2.29 1.5 3.67 3.64 0.9 5.17 5.16 0.2 7.11 7.07 0.6 18" (457 mm) 1.88 1.90 -1.2 2.99 3.00 -0.4 4.26 4.24 0.4 5.88 5.85 0.5 24 (610 mm) 1.56 1.60 -2.3 2.40 2.50 -4.0 3.49 3.51 -0.6 4.81 4.85 -0.9 36" (914 mm) 1.09 1.15 -5.2 1.69 1.76 -4.1 2.37 2.45 -3.5 3.29 3.40 -3.3 60" (1514 mm) 0.68 0.62 8.3 1.01 0.93 8.0 1.33 1.28 4.0 1.83 1.76 3.8 ARS = 11.8 RMS = 2.1 ARS = 9.8 RMS = 2.0 ARS = 22.7 RMS = 4.0 ARS = 22.8 RMS = 4.2 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-111

MODULUS, SHRP Section F, Run No. 1


Layer Number 1 2 3 Seed* Moduli, ksi (MPa) ? ? Range of Modulus, ksi (MPa) Minimum 490.0 (3379) 10.0 (69) Maximum 1481.3 (10214) 150.0 (1034) Poisson's Ratio 0.35 0.35 0.40

15.0 (103) ?

4 Selected by program

6-112

MODCOMP Program SHRP Section F Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6534 lb. 1139.7 (7860) 78.0 (538) 47.0 (324) 2 9512 1b. 1189.0 (8200) 59.6 (411) 45.7 (315) 3 12,662 lb. 1009.2 (6960) 60.2 (415) 43.2 (298) 4 16,812 lb. 1194.8 (8240) 49.6 (342) 41.9 (289)

Crushed Limestone Base

Silty Sand Subgrade

Deflection, mils Load 0" (0 mm) 3.28 3.27 0 5.07 5.07 0 7.28 7.28 0 9.71 9.72 0 8" (203 mm) 2.69 2.68 -.7 4.32 4.17 -3.5 5.97 5.91 -1.1 8.17 8.11 -1.2 12" (305 mm) 2.33 2.32 -.1 3.67 3.66 -.1 5.17 5.16 -.1 7.11 7.11 -.1 18" (457 mm) 1.88 1.89 .6 2.99 2.99 0 4.26 4.17 -1.5 5.88 5.87 -.3 24" (610 mm) 1.56 1.57 .7 2.40 2.44 1.9 3.49 3.43 -2.5 4.81 4.80 -.1 36" 60" (914 mm) (1514 mm) 1.09 1.10 .1 1.69 1.69 .1 2.36 2.39 .1 3.29 3.31 .1 0.68 .66 -3.3 1.01 .98 -5.7 1.33 1.38 1.6 1.83 1.85 .8 ARS = 2.6 RMS = 0.5 ARS = 3.9 RMS = 1.3 ARS = 11.2 RMS = 2.6 ARS = 5.5 RMS = 1.3 Convergence Error (%)

Measured 1 Predicted
% Error

Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-113

MODCOMP, SHRP Section F, Run No. 1


Layer Number 1 Seed Moduli, ksi (MPa) 350.0 (2413) 2 30.0 (207) 3 15.0 (103) 0.40 0.35 Range of Modulus, ksi (MPa) Minimum Maximum Poisson's Ratio 0.35

Deflection Convergence Criterion: RMS < 1% Modulus Convergence Criterion < 1% Maximum Iterations Criterion: 10

6-114

MODCOMP Program SHRP Section F Run No. 2


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 6534 lb. 1067.2 (7360) 110.5 (762) 32.3 (223) 1000.0 (6895) 2 9512 lb. 1113.6 (7680) 87.4 (603) 31.2 (215) 1000.0 (6895) 3 12,662 lb. 949.8 (6550) 85.7 (591) 29.7 (205) 1000.0 (6895) 4 16,812 lb. 1123.8 (7750) 74.1 (511) 28.4 (196) 1000.0 (6895)

Crushed Limestone Base

Silty Sand Subgrade

Rigid Base @ 198" (5.0 m)

Deflection, mils Load 0" (0 mm) 3.28 3.27 0 5.07 5.08 0 7.28 7.28 0 9.71 9.72 8" (203 mm) 2.69 2.64 -1.0 4.32 4.17 -3.8 5.97 5.91 -1.3 8.17 8.07 -1.4 12" (305 mm) 2.33 2.32 -.1 3.67 3.66 -.1 5.17 5.16 -.1 7.11 7.13 -.1 18" (457 mm) 1.88 1.93 1.4 2.99 2.99 0 4.26 4.21 -.7 5.88 5.87 -.3 24" (610 mm) 1.56 1.57 .6 2.40 2.48 3.1 3.49 3.46 -1.2 4.81 4.84 .9 36" (914 mm) 1.09 1.10 .1 1.69 1.69 .1 2.37 2.36 0 3.29 3.31 .1 60" (1514 mm) 0.68 0.55 -17.9 1.01 0.83 -19.8 1.33 1.14 -14.4 1.83 1.57 -14.1 ARS = 16.9 RMS = 5.4 ARS = 17.7 RMS = 5.5 ARS = 26.9 RMS = 7.7 ARS = 21.1 RMS = 6.8 Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-115

MODCOMP, SHRP Section F, Run No. 2


Layer Number 1 2 3 4 Seed Moduli, ksi (MPa) 700.0 (4826) 35.0 (241) 20.0 (138) 1000.0 (6895) Range of Modulus, ksi (MPa) Minimum Maximum 1000.0 (6895) Poisson's Ratio 0.35 0.40 0.45 0.50

1000.0 (6895)

Deflection Convergence Criterion: RMS < 1% Modulus Convergence Criterion < 1%

6-116

Results from

BOUSDEF EVERCALC MODULUS MODCOMP for SHRP Section G

6-117

SHRP PAVEMENT SECTION G (GPS-7A: AC OVERLAY OF CONCRETE PAVEMENT SECTION LOCATED IN NEBRASKA)

6-118

BOUSDEF Program SHRP Section G Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 2 9398 lb. 815.6 (5624) 9000.0 (62055) 30.7 (212) 3 12,256 lb. 944.5 (6512) 5594.8 (38576) 30.0 (207) 4 16,350 lb. 899.8 (6204) 5603.4 (38635) 32.6 (225)

PCC Combined Sand Base and Silty Sand Subgrade

Deflection, mils Load 0" (0 mm) 2.81 2.77 1.3 3.90 3.89 0.2 5.03 4.98 1.1 8" (203 mm) 2.41 2.35 2.4 3.40 3.36 1.1 4.37 4.24 2.9 12" (305 mm) 2.33 2.33 0 3.32 3.32 0 4.22 4.19 0.7 18" (457 mm) 2.24 2.25 -0.5 3.17 3.19 -0.7 4.04 4.02 0.4 24" (610 mm) 2.14 2.14 0.2 3.02 3.03 -0.2 3.84 3.80 1.0 36" (914 mm) 1.90 1.86 2.0 2.67 2.64 1.0 3.38 3.30 2.3 60" (1514 mm) 1.40 1.36 2.8 1.96 1.95 0.4 2.45 2.42 1.4 ARS = 9.8 RMS = 1.6 ARS = 4.0 RMS = 0.7 ARS = 9.3 RMS = 1.7 ARS = RMS = ___ Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

6-119

BOUSDEF, SHRP Section G, Run No. 1


Layer Number 1 2 3 Seed Moduli, ksi (MPa) 350.0 (2414) 3500.0 (24132) 15.0 (103) Range of Modulus, ksi (MPa) Minimum 100.0 (690) 2500 (17238) 1.0 (6.9) Maximum 2000.0 (13790) 9000.0 (62055) 50.0 (345) Poisson's Ratio 0.35 0.15 0.40

Deflection Convergence Criterion: ARS < 10% Maximum Iterations Criterion: 9

6-120

BOUSDEF Program SHRP Section G Run No. 2


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 2 9398 lb. 743.0 (5123) 7108.2 (49011) 150.0 (1034) 19.5 (134) 1000.0 (6895) 3 12,256 lb. 803.4 (5539) 8416.5 (58045) 150.0 (1034) 16.8 (116) 1000.0 (6895) 4 16,350 lb. 2000.0 (13790) 2500.0 (17238) 150.0 (1034) 25.1 (173) 1000.0 (6895)

PCC

Sand Base

Silty Sand Subgrade

Rigid Base @ 300" (7.6 m)

Deflection, mils Load 0" (0 mm) 2.81 3.09 -10.0 3.90 3.92 -0.5 5.03 5.15 -2.3 8" (203 mm) 2.41 2.62 -8.5 3.40 3.37 0.7 4.37 4.59 -5.0 12" (305 mm) 2.33 2.59 -11.0 3.32 3.35 -1.0 4.22 4.43 -5.0 18" (457 mm) 24" (610 mm) 2.14 2.35 -9.9 3.02 3.11 -3.1 3.84 3.84 0 36" (914 mm) 1.90 2.04 -7.4 2.67 2.77 -3.7 3.38 3.19 5.8 60" (1514 mm) 1.40 1.43 -2.0 1.96 2.04 -4.0 2.45 2.06 15.8 ARS = 36.9 RMS = 7.1 ARS = 15.6 RMS = 2.6 ARS = 60.3 RMS = 9.1 ARS = ___ RMS = ___ Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

2.24 2.49 -11.0 3.17 3.25 -2.7 4.04 4.16 -2.9

6-121

BOUSDEF, SHRJP Section G, Run No. 2


Layer Number 1 2 3 4 5 Seed Moduli, ksi (MPa) 350.0 (2413) 3500.0 (24132) 50.0 (345) 25.0 (172) 1000.0 (6895) Range of Modulus, ksi (MPa) Minimum 200.0 (1379) 1000.0 (6895) 10.0 (69) 1.0 (6.9) 1000.0 (6895) Maximum 2000.0 (13790) 9000.0 (62055) 150.0 (1034) 50.0 (345) 1000.0 (6895) Poisson's Ratio 0.35 0.15 0.40 0.40 0.50

Deflection Convergence Criterion: ARS < 10% Maximum Iterations Criterion: 9

6-122

EVERCALC Program SHRP Section G Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 2 9398 lb. 1538.4 (10610) 5643.9 (38923) 31.7 (224) 3 12,256 lb. 2105.7 (14522) 3603.2 (24850) 29.4 (203) 4 16,350 lb. 1578.2 (10884) 4296.3 (29630) 31.8 (219)

PCC

Combined Sand Base and Silty Sand Subgrade

Deflection, mils Load 0" (0 mm) 8" (203 mm) 12" (305 mm) 2.33 2.33 0 3.32 3.32 0 4.22 4.23 -0.3 18" (457 mm) 24" (610 mm) 36" (914 mm) 60" (1514 mm) Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

2.81 2.81 0 3.90 3.93 1

2.41 2.40 0.3 3.40 3.43 -0.7 4.37 4.36 0.3

2.24 2.24 0 3.17 3.18 -0.2 4.04 4.04 0

2.14 2.12 0.9 3.02 3.00 0.8 3.84 3.81 0.9

1.90 1.89 1.7 2.67 2.63 1.4 3.38 3.32 1.7

1.40 1.41 -0.6 1.96 1.98 -1.0 2.45 2.47 -0.7 ARS = 4.5 RMS = 0.8 ARS = 5.0 RMS = 0.8 ARS = 3.7 RMS = 0.8 ARS = ___ RMS = ___

5.03 5.06 1

6-123

EVERCALC, SHRP Section G, Run No. 1


Layer Number 1 Seed Moduli, ksi (MPa) 250.0 1724) 4000.0 (27580) 25.0 (172) Range of Modulus, ksi (MPa) Minimum Maximum Poisson's Ratio 0.35

0.20

0.45

Deflection Convergence Criterion: RMS < 1% Modulus Convergence Criterion: <1% Maximum Iterations Criterion: 10

6-124

MODULUS Program SHRP Section G Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1

2 9398 lb. 1046.0 (7212) 7000.0 (48265) 72.8 (502) 32.1 (221) 300 (7.6 m)

3 12,256 lb. 1153.0 (7950) 5743.5 (39601) 48.7 (336) 29.8 (205) 300" (7.6 m)

4 16,350 lb. 1146.0 (7902) 5412.3 (37318) 69.2 (477) 32.0 (221) 300" (7.6 m)

PCC

Sand Base

Silty Sand Subgrade

Rigid Base @

Deflection, mils Load 0" (0 mm) 8" (203 mm) 12" (305 mm) 2.33 2.35 -0.6 3.32 3.32 0 4.22 4.24 -0.5 18" (457 mm) 24" (610 mm) 36 (914 mm) 60 (1514 mm) Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

2.81 2.81 0 3.90 3.90 0 5.03 5.03 0

2.41 2.43 -0.8 3.40 3.44 -1.0 4.37 4.40 -0.7

2.24 2.24 0 3.17 3.16 0.2 4.04 4.03 0.2

2.14 2.12 0.8 3.02 2.99 0.9 3.84 3.81 0.9

1.90 1.87 1.4 2.67 2.64 1.2 3.38 3.33 1.4

1.40 1.42 -1.1 1.96 1.99 -1.4 2.45 2.49 -1.5 ARS = 5.3 RMS = 0.9 ARS = 5.1 RMS = 0.9 ARS = 5.6 RMS = 1.0 ARS = ___ RMS = ___

6-125

MODULUS, SHRP Section G, Run No. 1


Layer Number 1 Seed Moduli, ksi (MPa) ? Range o f Modulus, ksi (MPa) Minimum 200.0 (1379) 1000.0 (6895) 10.0 (69) Maximum 2000.0 (13790) 7000.0 (48265) 150.0 (1034) Poisson's Ratio 0.35

0.15

? 25.0 (172) ?

0.40

0.40

Note: user-specified imits

6-126

MODCOMP Program SHRP Section G Run No. 1


Layer Moduli, ksi (MPa) Layer Number 1 Material Type Asphalt Concrete 1 2 9398 lb. 973.0 (6710) 9019 (62200) 31.0 (214) 3 12,256 lb. 986.6 (6680) 9729.5 (67100) 27.0 (186) 4 16,350 lb. 1015.0 (7000) 8105.5 (55900) 29.9 (206)

PCC Combined Sand Base and Silty Sand Subgrade

Deflection, mils Load 0" (0 mm) 8" (203 mm) 12" (305 mm) 2.33 2.32 0 3.32 3.27 -.6 4.22 4.21 .4 18" (457 mm) 24" (610 mm) 36" (914 mm) 60" (1514 mm) Convergence Error (%)

Measured 1 Predicted % Error Measured 2 Predicted % Error Measured 3 Predicted % Error Measured 4 Predicted % Error

2.81 2.80 0 3.90 3.90 0 5.03 5.04 0

2.41 2.40 0.3 3.40 3.39 0 4.37 4.37 0

2.24 2.24 -.6 3.17 3.15 -.7 4.04 4.06 0

2.14 2.13 -.1 3.02 3.03 -.1 3.84 3.86 -.1

1.90 1.89 -.2 2.67 2.72 1.8 3.38 3.43 1.4

1.40 1.46 2.8 1.96 2.13 8.6 2.45 2.64 8.0 ARS = 9.9 RMS = 3.1 ARS = 11.8 RMS = 3.3 ARS = 3.7 RMS = 1.1 ARS = __ RMS = _ _

6-127

MODCOMP, SHRJP Section G, Run No. 1


Layer Number 1 Seed Moduli, ksi (MPa) 250.0 1724) 4000.0 (27580) 25.0 (172) Range of Modulus, ksi (MPa) Minimum Maximum Poisson's Ratio 0.35

0.20

0.45

Deflection Convergence Criterion: RMS < 1% Modulus Convergence Criterion: <1% Maximum Iterations Criterion: 10

6-128

SECTION 7.0 COURSE WRAP-UP

7.1

SUMMARY AND REVIEW Items covered in some detail in the course include: Section 2 Materials Characterization

The intent o f this section was to ensure that any information that is available regarding the pavement layer materials at a specific deflection test location can be used to evaluate whether the backcalculated moduli are reasonable. Section 3 Fundamentals of Mechanistic-Empirical Design

This section provided background information on the mechanistic-empirical design approach. Some existing design procedures were briefly described. The typical mechanistic approach was described and the use o f current analysis methods demonstrated using ELSYM5. Analysis results were used in typical fatigue cracking and rutting evaluation approaches. Section 4 Pavement NDT Devices

This section was used for a brief presentation o f existing surface deflection measurement equipment, including comparisons and correlations between different types o f measuring devices. Section 5 Deflection Analysis Techniques

Some o f the empirical deflection basin parameter uses were discussed, followed by a demonstration o f "manual" backcalculation by using forward analysis with ELSYM5 and adjusting moduli based on simple observation o f differences between calculated and measured deflections. Automated backcalculation procedures were described and discussed. Critical issues were addressed and recommendations for
7-1

dealing with these issues are made. Reliability and errors related to deflection measurements were presented in some detail. Section 6 Use of Backcalculation Programs

This is the major "hands-on" workshop portion o f the course, and involved use o f three backcalculation programs viz. EVERCALC, MODULUS and MODCOMP, although BOUSDEF is also included. Various pavement structures are evaluated in the demonstration mode, while a number of special interest FWD deflection files are also included for additional work.

7.2 FUTURE TRENDS IN PAVEMENT DEFLECTION ANALYSIS Improved backcalculation approaches speed accuracy simplicity analysis techniques (e.g. FE, Dynamic Analysis) material modeling

Improved Pavement Design Approaches analysis techniques response evaluation performance evaluation

Equipment Effects high speed moving wheel deflections improved thickness data complementary NDT tools (e.g. SASW) dynamic analysis of FWD data

7-2

7.3 QUESTIONS AND ANSWERS (a) Backup (805) 646-2230 Nick Coetzee Dick Stubstad Bob Briggs Starr Kohn Chuck Gemayel Rohan Perera Lynne Irwin

(904) 964-3777 (313) 454-9900

(607) 255-2805 (b) General questions?

7.4 COURSE EVALUATION

7-3

APPENDIX A BOUSDEF Users Guide

APPENDIX A BOUSDEF USERS GUIDE

A-1 Introduction BOUSDEF was developed at Oregon State University to determine in-srtu pavement layer moduli using deflection data through backcalculation technique. The program is developed to: conventional flexible pavements and assumes that the pavements in analysis consisting of a fine grained subgrade and a coarse grained aggregate base/subbase on the top of subgrade. The program is based on the method of equivalent thicknesses and Boussinesq theory. The backcalculated moduli may be used for evaluating the existing pavement structural strength and/or for use in mechanistic overlay design. BOUSDEF can be operated on any IBM or compatible microcomputers with a DOS version 3.1 or higher. BOUSDEF is an integrated program which includes creating, editing, and analyzing a data file functions. A menu screen of the program is shown in Figure A-1. Three selections can be made and each of them is discussed in the following.

A-2 Create Data File This option allows the user to create a data file for later analysis. By pressing key C (Create) or 2 in the main menu, the program will ask for a file name and display a data input screen as shown in Figure A-2. The data input screen provides a friendly environment for data entry. The user may use cursor keys to move the cursor to any fields and enter required data. If the [F1] key is pressed, the program will display a brief explanation of what information is required for the field. After entering all necessary information, the user may press function key [F8] to run the data right away. The analysis results will be displayed and the program will return to the data input screen for possible edit. Function key [F10] allows the user to save a data file under same filename or under a new filename. If the data is not going be saved, press Esc key.

A -l

This program allows user to b a c k c a 1cu 1ate pavement layer moduli from deflection basin data. The program was developed for use with Falling Weight Deflectometer (FWD) data. However, other NDT data may also be used with some modification of the data.

(1). (2). (3). A-2

Ed it a Data File Create a Data File Analyze a Data File

Enter your selection

--- Press Esc to Exit ---

Figure A-1

BOUSDEF Menu Screen

Pavement Structure Data Number of Layers: 0 Layer No. 1. 2. 3.


A .

File Name: Po isson Rat io 0.00 0 .00 0 .00 0 .00 0 .00 Minmum M o d u 1us 0 0 0 0 0

EXAMPLE Maximum M o d u 1us 0 0 0 0 0 Initial M o d u 1us 0 0 0 0 0 Dens ity (pe f ) 0 .0 0 .0 0 .0 0 .0 0 .0

5.

Layer for M 1 1 1 0 0

Thickness (in e h .) 0.00 0 .00 0. 00 0. 00 0 .00

Deflection Measurement Data Load Plate Radius: 0.00 Number of Sensors: 0 Sensor Locations 0 .0 0 .0 0 .0 0 .0 0 .c 0 .0 0 .0 Load (Ib) Deflection Readings at Corresponding Sensor Locations Test 1 0 0.00 0 . 00 0 . 00 0.00 0.00 0 . 00 0 . 00 Test 2 0 0 . 00 0 . 00 0 . 00 0.00 0.00 0 . 00 0 . 00 Test 3 0 0. 00 0 . 00 0.00 0.00 0.00 0 . 00 0.00 0 . 00 Test 4 0 0 . 00 0.00 0.00 0.00 0 . 00 0.00 Tolerance Fl-Help (%): 10 F8-Run Number of Iterations: 3 Esc-Exit (No save)

FlO-Save

Figure A-2 BOUSDEF Data Input/Edit Screen

The following information are needed to create a file: 1. Number of layer (required) - total pavement layers, including subgrade. 2. 3. Number of layer for modulus (required). Tolerance (required) - deflection error tolerance to stop program execution. Usually set at 5-10 percent. 4. 5. Number of iterations (required) - usually set at three iterations. Layer for modulus (required) - 1 for calculating modulus for the layer. In this case, minimum, maximum, and initial modulus must be provided. 0 for not calculating modulus for the layer. In this case, minimum and maximum moduli are not required. Initial modulus must be given and is treated as fixed value for the layer. 6. 7. Layer thickness (required, except subgrade) - in inches. Poissons ratio (required) - for asphalt concrete, Poisson's ratio = 0.35. 8. Minimum and maximum modulus (required if modulus for the layer needs to be calculated) - in psi. These values are used to set up the range of possible modulus. 9. Initial modulus (required) - if layer for modulus is set at 0, this value will be used as a fixed modulus for the layer. 10. 11. Number of sensors (required) - maximum 7. Sensor spacings (required) - in inches. Starts from load center line. 12. Test (required) - Four tests can be entered. Load in pounds, deflection in mils. Maximum 7 deflections allowed. The input data are saved in a text file in ASCII form and can be accessed and edited by the program or by other word processor software.

A-3 Edit Data File This option allows the user to edit a data file that has been created
A4

previously. By pressing key E (Edit) or 1 in the menu screen (Figure A-1) and providing a file name to be edited, the same screen used to create the data file will be displayed. The information saved in the existing data file will be shown in corresponding fields. The user may use cursor keys to move to each filed and edit. After editing all necessary information, the user may press function key [F8] to run the data right away. The analysis results will be displayed and the program will return to the data edit screen for further edit. Function key [F10] allows the user to save a data file under same filename or under a now filename. If the edited data is not going be saved, press Esc key.

A-4 Analyze Data File This option allows the user to analyze a data file created previously. By pressing key A (Analyze) or 3 and giving the file name to be analyzed, the calculation will begin.

A-5 Output The output will be displayed on the screen. The output include pavement modulus for each layer, bulk stress (BSTRS) and deviator stress (DSTRS), NDT load, and material coefficient k 1 and k2 for base/subbase and subgrade.

A-5

APPENDIX B User's Guide for EVERGALC

EVEIRCALC 3.3

T A B L E OF C O N T E N T S INTRODUCTION.................................................................................................................................B-l C H A R A C TE R IST IC S OF EVERCALC.................................................................................... B-l H A R D W A R E REQ U IR EM EN TS................................................................................................ B-6 ST R U C T U R E OF THE PROG RAM ............................................................................... I......... B-6 IN PU T D A T A FOR BACK CALCULATION...........................................................................B-8 G en eral Data F ile ................................................................................................................ B-8 Deflection Data File...............................................................................................................B-10 OUTPUT................................................................................................................................................B -l 1 REFERENCES......................................................................... ........................................................... B-12

B-i

E'VIEKCALC 3.3

ii|

L IST OF FIG U R E S Figure 1. T itle S c r ee n ................................................................................................................... B-13 Figure 2. Menu S creen .................................................................................................................... B-14 Figure 3(a). General Data Screen (US U nits)........................................................................... B-15 Figure 3(b). General Data Screen (Metric U nits)...................................................................... B-16 Figure 4(a). Deflection Data Screen (US U nits)....................................................................... B-17 Figure 4(b). D eflection Data Screen (Metric Units).............................................................B-18 Figure 5. Screen Output During Backcalculation....................................................................... B-19 Figure 6(a). ECALC1.GEN (US U n its)..................................................................................... B-20 Figure 6(b ). EC A LC 1.G EN (M etric U nits)........................................................................... B-20 Figure 7(a). ECALC1.DEF (US U nits).................................................. ................................... B-21 Figure 7(b). ECALC1.DEF (Metric U nits)................................................................................. B-21 Figure 8(a). B ackcalculation Output (US Units).................................................................. B-22 Figure 8(b). Backcalculation Output (Metric Units)................................................................... B-24 Figure 9(a). Summary Output File (US U nits).....................................................................B-26 Figure 9(b). Summary Output File (Metric Units).................................................................... B-26 Figure 10(a). Bar Chart Plot of Normalized Moduli (US U n its)......................................... B-27 Figure 10(b). Bar Chart Plot o f Normalized Moduli (Metric U n its)................................B-27 Figure 11. General Data Screen Using Stiff Layer Option (US Units)......................................... B-28 Figure 12. Deflection Data Screen Using Stiff Layer Option (US U nits).....................................B-29 Figure 13: Backcalculation Output Using Stiff Layer Option (US Units).....................................B-30 Figure 14: Summary Output Using Stiff Layer Option (US Units)............................................... B-31 Figure 15: Bar Chart Plot of Normalized Moduli Using Stiff Layer Option (US Units)........... B-31

B-ii

KVEIRCALC 3.3

IN TRO D U CTIO N EVERCALC is pavement analysis computer program which estimates the elastic moduli of pavement layers. The program is based on multilayered elastic theory using an inverse solution technique. The program runs on personal computers and is primarily for the evaluation of flexible pavement using Falling Weight Deflectometer (FWD) surface deflection measurements. It estimates the elastic modulus for each pavement layer, determines the coefficients of stress sensitivity for unstabilized materials, stresses and strains at various depths, and optionally normalizes asphalt concrete modulus to a standard laboratory condition (temperature). This user guide explains the characteristics of the program, hardware requirements, input and output, and the execution of the program. CHARACTERISTICS OF EVERCALC
General

EVERCALC is a pavement analysis computer program, based on the multilayered elastic pavement analysis program, CHEVNL, which was developed by the Chevron Research Corporation [1]. CHEVNL makes the following assumptions: a) Layers are infinitely long in the horizontal directions b) Layers have uniform thickness c) Bottom layer is semi infinite in the vertical direction d) Layers are composed of homogeneous, isotropic, linearly elastic materials, characterized by elastic modulus and Poisson's ratio. An inverse solution technique is used to determine elastic moduli from FWD pavement surface deflection measurements. The program is capable of evaluating a flexible pavement structure containing up to five layers. From an initial, rough estimate of layer moduli (seed moduli), the program iteratively searches for the "final" modulus for each pavement layer. The deflections calculated using CHEVNL are compared with the measured ones at each iteration. When the discrepancies in the calculated and measured deflections as characterized by root mean square (RMS) error (Equation 1), or the changes in modulus (Equation 2) falls within the allowable tolerance, or the number of iteration has reached a limit, the program terminates. Using the final set of moduli, the stresses and strains at the bottom of the AC layer, middle of the other layers except the subgrade, and at the top of the subgrade are calculated. When deflection data for more than one load level is available at a given point, coefficients of stress sensitivity for unstabilized

B-l

EVEKCALC 3 .3
materials are also computed. Optionally, AC layer modulus is normalized to a standard laboratory condition.
Seed Moduli

Two options for estimating the seed moduli are available. When a pavement structure containing up to three layers is being analyzed, a set of internal regression equations can be used [2]. These regression equations determines a set of seed moduli from the relationships between the layer modulus, surface deflections, applied load, and layer thicknesses. Alternatively, the user can provide these values. When more than one deflection data set at a given location is analyzed, the final moduli from the previous deflection data set is used as seed moduli for the next one in order to improve the performance of the program.
Termination

The program terminates when one or more of the following conditions are satisfied: i) Deflection Tolerance:
rms

= io o y i =1

D - D0!2 1 ------M < deflection tolerance

where D, and D, the measured and calculated deflection at i-th sensor and n- is the number of sensors. Normally a deflection tolerance of one percent is considered adequate, ii) Moduli Tolerance: < modulus tolerance, for all i = 1 to m
Eik)

where, Ei

and Ej

are the i-th layer modulus at the k-th and (k+l)-th iteration,

respectively, and m- is the number of layers with unknown modulus. Again a modulus tolerance of one percent is considered adequate. iii) Number of iterations has reached the Maximum Number o f Iteration. At every iteration a minimum of (m+1) calls to CHEVNL is made, where m- is the number of layers with unknown moduli. Normally, a maximum number of JU iterations is adequate.

B-2

EVERCALC 3.3
Qxfllcieni o f Stress Sensitivity'

The stress sensitivity characteristics of unstabilized material moduli are usually formulated as follows: Eb = kj0^2 for coarse-grained soils E, = k 30 d^i for fine-grained soils where EB = resilient modulus of coarse grained soil, E, = resilient modulus of fine grained soil, 8 = Bulk Stress, o d = Deviator Stress, and kj, k2, k3, and k4 = regression coefficients The program determines the stress sensitivity coefficients using a linear regression method when FWD deflection data for two or more load levels at a given point are available.
Load Rate and Temperature Correction

The stiffness of asphalt concrete is primarily affected by temperature and loading rate. While FWD loads occur over a 25 to 35 millisecond loading time (approximately) and at ambient temperature, the standard laboratory condition is taken to be a 77*F (25C) temperature and 100 millisecond loading time. Thus, the stiffness of asphalt concrete is normalized to view the backcalculated modulus in terms of the traditional" laboratory values (at least this is what is being attempted). Temperature normalization of asphalt concrete is accomplished using the relationship between the modulus and temperature. The relationship for WSDOT Class B asphalt concrete was found as follows [3]: log E k = 6.4721 - 1.47362 x 10^(Tp)2 where = modulus of asphalt concrete (psi), and Tp = pavement temperature (*F) From the above modulus-temperature relationship, the backcalculated AC modulus at the insitu field temperature is multiplied by an adjustment factor in order to obtain a "standard" modulus at a temperature of 77*F (25*C):

B-3

EVEKCALC

3 .3

J A F = 1 0 000147362(Tp2 -772)

where

TAF = temperature adjustment factor, and Tp = pavement temperature (F)

Pavement temperature is determined either by direct measurement or Southgate's method which uses pavement surface temperature, the previous five day mean temperature, and pavement thickness. Both measurement methods are incorporated in the program. The effect of different loading rates can be normalized using the Asphalt Institute relationship 13,4], which predicts asphalt concrete modulus when various material parameters and testing conditions are known. The material parameters of WSDOT Class B asphalt concrete were substituted into the Asphalt Institute equation and the moduli were predicted for various temperatures and two loading duration conditions (33 and 100 milliseconds). The ratio of the modulus for a 33 millisecond load pulse to that for a 100 millisecond load pulse for various temperatures was regressed as follows: R = 0.791 + 0.00813Tp where R = loading adjustment factor of FWD to laboratory conditions, and Tp = pavement temperature (*F) This relationship can be used to adjust the "field" backcalculated asphalt concrete modulus by dividing the backcalculated modulus by R. This specific relationship was used in prior versions of EVERCALC but was deactivated for Version 3.3. It should be noted that these corrections (temperature and loading rate) are based on regression equations that were developed for the WSDOT Class B asphalt concrete and their validity to other classes o f asphalt concrete are not known. However, Class B is a traditional dense asphalt concrete mixture.
Depth to Stiff La\er

A depth to stiff layer is estimated using the scheme reported by Rohde and Scullion [5]. The basic assumption is that surface deflection will occur beyond the offset (measured from the load plate) which corresponds to the intercept of the applied stress zone and a stiff layer (the stiff layer modulus being 100 times larger than the subgrade modulus). Thus, the method for estimating the

B-4

'EVERCALC 3.3
depth to stiff layer assumes that the depth at which zero deflection occurs (presumably due to a stiff layer) is related to the offset at which zero deflection occurs. This feature in EVERCALC is optional within the General Data File (to be discussed in later sections). It is common to expect a stiff layer condition within a 30 ft (9 m) depth. Accounting for
a

stiff layer condition generally reduces the subgrade modulus (layer above the stiff layer) and

increases the base coarse modulus. When the backcalculation analysis is done for all layers at a specific station, only one depth to stiff layer is used; however, depths to stiff layer are calculated for each FWD load level (drop). This stems from the original development of the depth to stiff layer scheme [5] in which all equations and assumptions were based on one load level (9,000 lb (80 kN)). Thus, load levels closer to 9,000 lb (80 kN) should be more reliable. The single depth to a stiff layer is calculated as follows [5]: D= n
1 -

where

D = average depth to a stiff layer (ft), B, = depth to stiff layer for the i-th deflection basin Goad level) (ft), and n = number of deflection basins within one standard deviation of the mean of l/Bj

The original development of this scheme by Rhode and Scullion [5] requires the use of one of four separate regression equations for estimating 1/B. These equations are categorized by the thickness of the AC surfacing as follows:
AC thickness less than 2 in. (50 mm) AC thickness between 2 to 4 in. (50 to 100 mm) AC thickness between 4 to 6 in. (100 to 150 mm) AC thickness greater than 6 in. (150 mm)

Within each of these equations are various independent variables, one of which is "r0" (1/r intercept, where 1/r is the inverse of the deflection offset). For some deflection basins, r0 can be negative (which is outside the range of the regression equations) and can cause 1/B to be negative as well. When this occurs, the depth to stiff layer is automatically set to a default value of 600 in. (15.2 m).

B-5

[BVEKCASLC 3.3

HARDW ARE REQUIREMENTS The EVERCALC program is coded in Microsoft Quick Basic version 4.5 and Fortran version 5.0 and designed to run on IBM or compatible personal computers with PC-DOS or MS-DOS version 2.0 or later operating system. The program has been compiled and linked to produce 80286 processor instructions and 80287 coprocessor floating point instructions. Therefore a machine with 80286 or higher processor and a coprocessor is required to run the compiled programs. However, the source files are included in the disks and can be recompiled to produce executable code for other machines. The program contains numerous lengthy numerical computations and an AT (or compatible) or better machine with a coprocessor is advisable.

Since ANSI commands are used for screen control, the ANSI.SYS driver must be loaded before running the program (Include a line DEVICE = drive:Npath\ANSI.SYS in the CONFIG.SYS file). STRUCTURE OF THE PROGRAM The disk contains four executable files, EVERCALC.EXE, MAIN.EXE, FWDCNVRT.EXE and PLOT.EXE. EVERCALC.EXE file is a menu program which first displays the title screen (Figure 1) and then provides the following seven menu options (Figure 2): 1) Edit General Data File - This menu option is used to create/edit the general data file (to be described later) which is used in the backcalculation program. An editor, with named fields and brief messages for each field, is provided for easy data entry (Figure 3). When this menu option is selected a name for general data file is prompted and if that file already exists it is loaded in to the editor, otherwise a fresh editor with blank fields except the File Name field is loaded. 2) Edit Deflection Data File - This menu option is used to create/edit the deflection data file. Only four deflection data sets can be inputted. An editor, with named fields and brief messages for each field, is provided for easy data entry (Figure 4). When this menu option is selected the name of the general data file and a name for deflection data file are prompted. If the deflection data file already exists it is loaded into the editor, otherwise a fresh editor with blank fields except the File Name field is loaded. The editor has space for a maximum of four deflection data sets and if there are more than four deflection data sets a text editor has to be used.

B-6

| P VIEIRCALC 3.3 S

3) Perform B ackcalculation - This menu option loads the program MAIN.EXE and

performs the backcalculation analysis. Two data files, the general data file and the deflection data file are the input files. It generates two output files, a backcalculation output file and a summary file. The backcalculation output file is lengthy and contains backcalculated modulus, deflections, stresses and strains, and normalized modulus among other relevant information. The summary file reports the normalized modulus, bulk stresses and stress sensitivity in a concise form. While performing backcalculation, plots of the measured and calculated basins, layer moduli, and RMS error at each iteration are displayed on the screen enabling the user to monitor the performance of the program (Figure 5).
4) Convert FW D Data File - This menu option loads the program FWDCNVRT.EXE

and produces a data file for the backcalculation analysis from the FWD raw data file. This is specially formatted for the Washington State Department of Transportation FWD data file format and may not work correctly with other formats. Two data files, the FWD raw data file and a file containing layer information (deflection file created using menu option 2 with or without deflection data sets) are required. It interactively asks for the number of layers and thicknesses of the layers. Terminate the thickness input by typing a non-numeric character at the milepost prompt. This is intended for inputting mileposts where changes in. layer thicknesses occurs (e.g., if the thicknesses of a three layer system at Milepost 10.000 is 4.0 in. and 13.0 in. and at Milepost 20.000 is 3.0 in. and 15.0 in, then these Mileposts and the thicknesses are inputted in that order). At the third Milepost prompt terminate the data input by typing a non-numeric character. The program uses the thickness provided at milepost 10.000 (4.0 in. and 13.0 in.) for stations between Mileposts 10.000 and 20.000 and uses the thickness provided at Milepost 20.000 stations on and after Milepost 20.000). The program extracts the data sets and other pertinent information from the raw data file and appends this to the deflection data file.
5) Plot Bar Chart - This menu option loads the program PLOT.EXE and generates a bar

chart of the normalized modulus from the summary data file created in backcalculation. This plot is written to a file. This file can be later printed or viewed on the screen using DOS PRINT or TYPE commands. (Note: Since extended ASCII characters are used to create the bar chart, the DOS command GRAPHICS should be executed before printing)
6) Execute DOS Commands * This menu option lets you execute DOS commands

without exiting the program. The program remains in the memory and opens a second command shell to execute DOS commands.

B-7

|IVIIRCALC 3.3
7) Exit to DOS. The menu program (EVERCALC.EXE) is intended for ease of use and informational messages are displayed dynamically about various data fields. Even though the programs in menu options 3 (MAIN.EXE), 4 (FWDCNVRT.EXE), and 5 (PLOT.EXE) arc stand alone and can be executed separately, it is advised that the menu program be used at least until a familiarity of the various data required and their order in the general data file and the deflection data file is achieved. The disk also contains the source files. MAIN.FOR, FWDCNVRT.FOR and PLOT.FOR are Fortran program s and contains some features which are extensions provided in MS FORTRAN 5.0. Therefore MS FORTRAN 5.0 or later version is required to recompile these programs. EVERCALC.BAS is written in QUICK BASIC 4.5 and requires QUICK BASIC 4.5 or later version for recompiling. INPUT DATA FOR BACKCALCULATION The backcalculation program MAIN.EXE (menu option 3) requires two files, a general data file generated in menu option 1, and a deflection data file generated in menu option 2. These files are in free format and can also be created externally using a text editor. G eneral D ata File The general data file contains input data which do not change frequently, in free format (i.e., different data on the same line are separated either by at least one space or a comma). This file should include the following information (refer to Figures 3 and 6 (the item # and line # in brackets refers to Figure 6)): Rome: No of Layers: Units: Load Plate Radius: No of Sensors: Radial Offsets: Temp Correction: Route name or the title of the analysis (hem 1, line 1) Number of layers in the pavement structure (item 2, linel) Units of measurements ( for English units or M for Metric units) (item 1, line 2) Load plate radius (in inches if English units or in cm if Metric units) (maximum of 10 allowable) (item 1, line 3) Number of sensors(maximum of 10 allowable) (item 2, line 3) Radial offsets of deflection sensors (in inches if English units or in cm if Metric units) (item 1-10, line 4) Temperature correction required? (X for Yes and N for No) (item 1, line 5)

B-8

|SVIIRCALC 3.3

Temp Measurement: Seed Modulus Option: Stiff Layer Option: Maximum Iteration: RMS Tolerance: Modulus Tolerance:

Method of temperature measurement (D for Direct method or 5 for Southgate method) (item 1, line 6) Seed modulus option (/ for Internal equations or for Engineering judgment) (item 1, line 7) Stiff layer to be calculated and used in backcalculation? (V for Yes and N for No) (item 1, line 8) Maximum number of iterations allowed (item 1, line 8) Deflection tolerance criteria for convergence (RMS in percent) (item 2, line 8) Modulus tolerance criteria for convergence (in percent) (item3, line 8)

It should be noted that the internal equations for the seed moduli can only be used when there are three or less layers in the pavement structure. Further, if you have selected the Stiff Layer Option , do not include this layer in the No of Layers or attempt to include its information in the Deflection Data File. Also, since the inclusion of an stiff layer will increase the number of layers during the backcalculation phase (Menu Option 3), the internal equations for the seed moduli can only be used when there are two or less layers in the pavement structure (excluding stiff layer) under this condition. You will be prompted for the stiff layer modulus and Poisson's ratio during backcalculation (Menu Option 3) and will be provided with the calculated depth to stiff layer at each section, a maximum of 21 stations at a time (a plot of the depth to stiff layer along the road will be displayed when there at least more than two stations), and you will have the option to change any of the depths to stiff layer. A sample general data file ECALC1.GEN is included in the disk and contains default general data. This file can be edited in menu option 1 or using a text editor. Alternatively, a new general data file can also be created. The ECALC1.GEN file as loaded in menu option 1 is shown in Figure 3. When loading a general data file, the File Name field in menu option 1 (Figure 3) will contain this file name. This can be changed to include a valid file name and the data can be saved to a that file effectively keeping the original file unchanged. A print out of ECALC1.GEN is shown Figure 6.

B-9

[IEVIS IRCALC $.3

D eflection Data File

The deflection data file contains input data which change frequently, in free format (i.e., different data on the same line are separated either by at least one space or a comma). This file should include the following information (refer to Figure 4): Route: Known/Unknown:
Poisson's Ratio: Seed Moduli:

Route name or the title of the analysis Whether the modulus of this layer is known or unknown (0 for known moduli or 1 for unknown modulus) Poisson's ratio of this layer. True modulus if the modulus is known or seed modulus if it is unknown (if engineering judgment was selected in General Data File) (in psi if English units or in kPa if Metric units). Otherwise enter 0.0. Lower limit (minimum) for the backcalculated modulus of this layer (in psi if English units or in kPa if Metric units). For no lower limit on the moduli enter 0.0. Upper limit (maximum) for the backcalculated modulus of this layer (in psi if English units or in kPa if Metric units). For no upper limit on the moduli enter 0.0. Station number or milepost (in real numbers). Thickness of each layer, excluding the subgrade (first layer thickness first; in inches if English units or in cm if Metric units) Temperature (pavement tem perature if direct method or surface and 5-day mean tem perature if Southgate method; in *F if English units or in *C if Metric units) Number of data sets at this station (number of drops; maximum of 10) Load applied (in lbs if English units or in kg if Metric units) Deflection measurements (in mils if English units or in microns if Metric units)

Min Moduli:

Max Moduli:

Station:
Thickness:

Temperature:

Mo o f Data Sets: Load: Sensor No:

The layer information must be repeated for all the layers. If you are using a text editor to create/modify a Deflection Data File, the following order must be maintained (each line of data in a separate line) (refer to Figure 7):

B-10

EVERCALC 3.3
Layer No, Known/Unknown, Poissons Ratio, Seed Modulus, Min Modulus, Max Modulus

3
1 , for ^ > as you want j slations

.... (repeat for all the layers)..... Station, Thickness, No of Data Sets, Temperature Load, Deflections .... (one line per drop, maximum 10 drops per station).....

A sample deflection data file ECALC1.DEF is included in the disk. The ECALC1.DEF file as loaded in menu option 2 is shown in Figure 4. When loading a deflection data File, the File N am e field in menu option 2 (Figure 4) will contain this file name. This can be changed to include a valid file name and the data can be saved to that file effectively keeping the original file unchanged. A print out of ECALC1.DEF is shown Figure 7. OUTPUT The backcalculation program (menu option 3) produces two output files, backcalculation output file and summary file. The backcalculation output file contains detailed backcalculation results such as seed moduli, iterations, errors, and stresses and strains for the pavement system with the estimated moduli as shown in Figure 8. The summary file (Figure 9) contains the normalized moduli in a concise form that can be used in menu option 5 to generate a bar chart of the normalized moduli along the road (Figure 10). Figures 11 through 13 show the same data but with the stiff layer option.

B -ll

KVEKCALC 3.3

23
R EFER EN C ES
c

1. Michclow, J., Analysis of Stress and Displacement in an n-Layered Elastic System Under a Load Uniformly Distributed on a Circular Area, Chevron Research Corporation, Richmond, California. 2. Newcomb, D. E., "Development and Evaluation of Regression Method to Interpret Dynamic Pavement Deflections," Ph.D. Dissertation, Department of Civil Engineering, University of Washington, Seattle, 1986. 3. Bu-Bushait, A. A., Development of a Flexible Pavement Fatigue Model for Washington State," Ph.D. Dissertation, Department of Civil Engineering, University of Washington, Seattle, 1985. 4. The Asphalt Institute, Research and Development of the Asphalt Institutes Thickness Design Manual (MS-1), Ninth Edition, Research Report No. 82-2, The Asphalt Institute, College Park, Maryland, August 1982. 5. Rohde, G. T. and Scullion, T., "MODULUS 4.0: Expansion and Validation of the Modulus Backcalculation System," Research Report 1123-3, Texas Transportation Institute, Texas A&M University System, College Station, Texas, November 1990

B-12

1EVERCALC 33________ _________________ _________

13|

Pavesant valuation Program

rVERCXLC 3.3

Developed
by

The Washington Stata Transportation Cantar Univarsity of Washington and the Washington Stata Department of Transportation February 1992 Press any key to continue....

Figure 1: Title Screcn

B-13

|c. v u k u m a , j j

EVERCA1C VERSION 3.3 - February 1992 Washington Stata Department of Tranaportation - University of Washington

THE MENU Use ARROWS> to select <ENTER> to Choose 1 2 3 4 5 6 7 Edit Cenerai Data File Enter Deflection Data Xnteractively Perforo Backealculataion Convert FVD Rav Data File Plot Dar Chart of Noroalited Hodulus Excut DOS Conaands Exit to DOS

Edit file containing information such as I of layers, I of sensors etc..

Figure 2: Menu Screen

B-14

[EVERCALC~33

CEWERAL DATA Rout*: EXAMPLE PROBLEM No of Layers: 3 Load Plata Radius: Sensor No: Radial Offsets: T u p Correction: Y Seed Modulus Option: I Kaxiaux Iteration: 5 3 0 5. 9 2
8

INSERT OFF

File Nae: ECALC1.GEN Units: E No of Sensors: 3 12 4 24 5 36 6 7 5 8 9 10

Tenp Measurement: D Stiff Layer Option: N RMS Tolerance(): 1. - Modulus Tolerance(1): 1.

Please Enter Route Nane or Title of Analysis, 25 Characters Kax Use Arrows,TAB Or ENTER To Hove Highlight Bar, <F1> When Done Editing Screen

Fipure 3(a: General Data Screen (US Units)

B-15

CENERAL DATA Route: EXAMPLE PROBLEM No of Layers: 3 Load Plate Radius: Sensor No: Radial Offsets: Teop Correction: Y Seed Modulus Option: I Haxisua Iteration: 5 1 0 15. 0 2 20.3 File Nane: ECALC1M.CEN Units: H No of Sensors: 3 4 5 6 30.5 61.0 91.4 7 5 8

INSERT OFF

10

Tenp Measurement: D Stiff Layer Option: N RMS Tolerance(%): 1. Modulus Tolerance(%) : 1.

Please Enter Route Nane or Title of Analysis, 25 Characters Max Use Arrows,TAB Or ENTER To Move Highlight Ear, <F1> When Done Editing Screen

Figure 3(b): Genera) Data Screen (Metrie Units !

B-16

|EVERCALC33

DEFLECTION DATA Route: EXAMPLE PROBLEM Layer Mo Known/Unknown 1 1 2 1 3 1 Poisson's Ratio .35 .40 .45

INSERT orr

Pile Name: SCALCI.DEF Seed Moduli Min Moduli Max Moduli 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Station: 600.050 No of Data Sets: 2 Data Data Data Data Set Set Set Set 1 2 3 4 load 6864. 12251.

Thickness: 6.60 Tespereture: 76.00 Sensor No 1 2 3 4 7. 953 6.220 5.116 3.110 13. 307 10.669 8.896 5.669

21.00 5 2.008 3.898

Please Enter Route Mane or Title of Analysis, 25 Characters Max Use Arrows,TAB Or ENTER To Move Highlight Bar, <P1> When Done Editing Screen

fig u re 4(a): Deflection Data Screen fUS Units)

B-17

{EVERCALC 33

18J

DEFLECTION DATA Rout: EXAMPLE PROBLEM Layer No Known/Unknown 1 1 2 1 3 1 Poisson's Ratio .35 .40 .45

INSERT OFF

Pile Naae: ECALC1M.DEF Seed Moduli Min Moduli Max Moduli 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Station: 600.050 No of Data 8ets: 2 Data Data Data Data Set Set 6et Set 1 2 3 4 Load 3120.0 5567.0

17.3 Thickness: Temperature: 21.1 Sensor No 4 1 2 3 202 158 130 79 338 271 226 144

53.3 5 51 9

Please Enter Route Naae or Title of Analysis, 25 Characters Max Use Arrows,TAB Or ENTER To Move Highlight Ear, <fl> When Done Editing Screen

Figure 4(b): D eflation Date Screen (Metric Units)

B-18

[e v e r c a l c T J

-: . - **-' V "

-' i '

> - '.: :* '- m d f e i K . ii-a:rjff.;iijVlar- t -4 " '"/* '* * - * ... 1 *' ^ -r -


v " '' '~ ' ' - 'T WXt. s'40!H. -3K71. PAW.

S t u r tiie d*U s e t
Ser.c JIntuli: . Oirrert
- T* - . t .- .

1 at'S !*tian iiW.'jj * t 1L:*J:3M

. F:rc.Tl BJfc-liiiiw: I n iir *1 i h Si.-t B-ic^iikilysis


2TAu.

e r c e r t B f e o t lue Error * t t i c Lu o it r t ion

1 '

Figure 5; Screen Output During Backcalculaiion

B-19

| EVERCALC33
EXAMPLE PROBLEM

20]

3
E

5 .9 5 0 8 12 24 36
Y D

1 N 5 1.

1.

Figure 6(a): ECALC1.GEN (US Units)

EXAMPLE PROBLEM

3
M

15.0 5 0 20.3 30.5 y


D

61.0

91.4

1
N

5 1.

1.

Figure 6(b): EC ALCI GEN (Metric Units)

B-20

1EVERCALC33
EXAMPLE P ROBLEM

ID
0 .0 0 .0 0 .0 2 76.00 5.118 3.110 2.008 8.898 5.669 3.898

1 .3 5 1 .4 0 1 .4 5 6864. 12251.

0 .0 0 .0 0 .0

0 .0 0 .0 0 .0

600.050

6.80 21.00 6.220 13.307

7.953

10.669

Figure 7(a): ECALC1.DEF (US Units)

EXAMPLE PROBLEM

1 .3 5 1 .4 0 1 .4 5 3120.0 5567.0

0 .0 0 .0 0 .0

0 .0 0 .0 0 .0

0 .0 0 .0 0 .0 2 21.1 1 3 0 79 51 2 2 6 144 99

600.050

17.3 53.3 20 2 1 5 8 338 271

Figure 7(b): ECALC1.DEF (Metric Units)

B-21

|EVERCALC~33
a t * c a l c u l a t i o n
sy

rrncAU

vulsiok

.)

KXAW LC PROSULM
mi

i* p o t

WuMltmT of Layera TMcfcnaaaaa(In) Pav* nt T M p . 4F )

00.050 3

i.l

11 .0

>4.0
I M M . I 310(7. 4*4744. 41537. >0534. 2041. !M(&.

Loafl (Ita) Id Moduli Uaad Moduli Calculated

(pal) (pal)

Dafloctiona IK)

(ila)

Aftor

Iteration

2 DlfFEREHCE -.002 .041 -.057 .020

OfTSET Iin)

CALCULATED 7. 55 . 17f S . 175 3.090 2.014

MEASURED 7.S3

.0
12. 0
24.0 34.0

.0

.220
S.Ill 3.110

- 1.

2.001
A B SO L. A R IT M . SUM: SUM:

-.0
.1252

MS
Strasaea (pal) traina

two:

<10*- in/in)

.00
.00

I
.0
17.30 STRESS : STRAIN: STRESS : STRAIN: STRESS: STRAIN:

VERTICAL TANGENTIAL
-12.7 - 151.19S2 -S.0414 -141.7453 -2.355 -3.5105 .7454 130.434 -.1533 3 .224 -.1*42 41.2040

RADIAL

ULK

1.7454 13C.434 -.1533 3 224 -.1442 41.2040

10.0I
-5.3410 -2.47

.00

27.10

Error tolaranc# criteria it satisfied. Load (lba) Seed Moduli Uaed Moduli Calculated (pal) (pal)

12251. : t 44744. 513230. 30534. 372S. 24545. 23177.

Deflectiona 90 1 2 3 4 5

(ails)

After Iteration

1 DirFEHEKCE -.123 .073 -.110 -.011 -.005 DIFF(I)

orrsETdn) .0 .0 12.0 24 .0 3 .0

CALCULATED 13.430 10.594 9.00 S.17 3 .C3

MEASURED 13.307 10 44 . 5. 44 3.

-1.2 -. 3 -. 1

ABSOL. SUM: WITH. SUM: RMS ERROR:


(pal) 4 Strains (10*- in/in)

I I
. 00 to .00 17.30

X
STRESS: STRAIN: STRESS: STRAIN: STRESS: STRAIN:

VERTICAL TANCENTIAL
-23.921 -250. *772 -S.(142 -243.1575 -4 .2430 -170.453 150.10*0 204.2403 .1424 4.3415 -.2144 754010

RADIAL
27.533 -5314

150.1010 204.2403 .1424 .341b -.2144 75.4010

.00

27.0

-4.4922

Figure 8(a): Backcalcuiation Output (US Units)

B-22

EVER CALC 3.3


l i w u r y of fceckcalculation at itation 00.OSO

LOAD *4. 122J1.

K (11 444744. 113230.

KAD 441233. 447244. 4442S0. 4S9444.

112) 30S34 >4724. >34)0 >2944 444

M T9 .14 11.71 10.12 U O .>40

C*T* 1.04 .74 .91 .94 1.00

KOI 24S4S. 23477. 2S221. 499. >4241 .

DSTPt 2.04 4.44 > . S 3.49 -.2427

CST* 1.93 1.97 1.9 J 1.94 1.00

MAH OfcK. 4IS949. to 000 lba 11. *2 4 U O

Voto 1

1(1) Mo d u l ua of i- th layer (pal) * A D Adj. moduli of aaphtlt layer for 7? * 9 P (pal) fcSTft Bulk i t r t n Ips 11 CSTR * Confining ilrtii (pal) M T U Deviator atresa ipsl) 11,112 - Iirei aenaltlvlty coefficlenta ILSQ - Coefficient of ielermlnation

Figure 8(a): Backcalculation Output fUS Units) (continued)

B-23

|EVER C ALC 33
BACK CALCULATION IT VUICALC V M S 10* 3.3

U A W U In roit unbe r of U y i n Thickne(cm ) f r M n t To*p. LMd (kg) |C) I UP) UP) 00.OSO

PROOLEX

17.3 I) .1 3120.

eed Moduli Utd Moduli Calculated

t
:

2139(15. 320(421.

334990. 209100.

203359. 113507.

Deflection# HO 1 2 3 4 S

(micron) Aftar Iteration CALCULATED 202 .009 157.034 1 3 1 .43( 7*.470 1 .155

2 DIFFERENCE -.009 .9(( -1 .43( .530 -.155 .1219 irr d ) .0 .( -i.i .7 -.3 2.7

orrsET(c) .0 20.3 30.5 1.0 1 .4

MEASURED 202.000 151.000 130.000 79.000 SI.000 A B 5 0 L . UM: ARITM. UM: MS IHIO:

-.1
.7

tresaea

(kPa)

traina

<10*-( c n / c n )

.00 17.30 .00 3.5

t
STRESS : STRAIN: STRESS : STRAIN: STRESS : STRAIN:

VERTICAL TANCENTIAL -7.M I C -157.9513 0 4 JS7| -1(1.9944 -1.194 2 -93 4774 $97.9199 130.170 1 Oil) (3.3445 -1.15(0 41 .1514 597.9199 1101.0920 130.1170 -1.0119 (3.344 5 -1.15(0 41 .1514 -37.0354

.00

70.CO

- 2 C .50(3

Error tolaranca criteria la aatiafiao. Load (kg) (kPa) (kPa) < 55(7. 320(421. 3531(5. 209100. 252443. 113507. 1(4145.

Seed Moduli Used Moduli Calculated

Deflection*

(micron) After Iteration CALCULATED 341.020 2(9 277 221 SCO 144 .407 9.1(0

1 DIFFERENCE -3 020 1.723 -2. IOC -.407 -.1(0 .3193 D1FF(%) -.9 .( -1.2 -.3 -.2 3.2 -1.9 .

M orrsET<oo O
1 2 3 4 5 .0 20.3 30.5 (1.0 91 .4

MEASURED 331.000 271.000 2 2 ( .000 144.000 99.000 ABSOL. SUM: ARIT*. SUM: MS ERROR:

Streaaea 00

(kPa) t

Straina

(10*-( AD1AL ULK

VERTICAL TANGENTIAL STRESS: STRAIN: STRESS: STRAIN: STRESS: STRAIN:

17.30

-1(2 (404 1034.1570 1034.1570 1905.(740 20(04(1 20(04(1 -250.5331 -(0.7093 -243.4447 -29.4247 -170.3C4 34 ( 1.41(1 -1.4904 75 3 5 K 934( 91.41(1 -1.4904 7S.3SK -51.9400

00 00

43.95

TO .(0

-32.4055

Error tolerance criteria ia aatltfled.

Figure 8(b); Backcalculation Output (Metric Units)

B-24

IEVER CALC 3 3
l i m i r y of ftackcaleulatlon at Station 400.030

LOAD >110. >>47.

(1) >2044 21. >331431.

SAD 2239294. 24 93394. 2)U>44. 24 >91>2.

1(2) 209100. 232443. 231122. 242342. 411.

u n > *4 0.74 9.94 7 > .49 .>171

CSTS 1. 32 4.34 >79 1.00

1(3) 103307. 144143. 174174. 149170. 211144.

OSTS 19.73 >0.43 2111 21.11 -.2440

CSTS 13.29 13.43 13.44 13.>3 1.00

MSA* >372340. O A K .Xiips. to >000 11, 12 so

ota:

1(1) Modulua of 1-th layar (If) AD - Adj. oduli of asphalt layar for 23 dg C U fa) M T U - Bulk atroaa UPa) CSTS - Confining atraaa (kPa) OSTk - Daviator atrai (kPa) El ,Si - Strait aanaltlvlty coafficianta MO Coofflclont of datanilnation

Figure 8(b); Backcalculation Output (Metric Units) (continued)

B-25

[e v e r c a l c T ?
AC CALCULAT IOK OY RVIRCALC VLAJIO* 3.2 out: I X M C L E PROtLKM

26)

Modulus ar Roramlltod to 000 lb* Mil 00.OSO BAD 45*.J til) 4*4.0 K<2) 3S.0 MTII2I t.* 1(21 ***. K2 (2 ) .S* KO) 25. OSTRO) 2.S RIO) 24211. R20) -.24 ARMS

.11

Voto:

K( i ) - Modulus of 1-th lyr (ksi) RAD - Aoj. moduli of asphalt layar for 77 dg F (ksi) OST* Ouik strass (psi) K1.K2 - Strass nsitivity Coafflclonta ARMS - Avara R W

Figure 9(a); Summary Output (US Units)

OACK CALCULATION OY CVtRCALC VtftSlON 3.3 RXAMFLt FkOBLtH

Rout:

Modulus ara Koranllsad to S000 kg km 00.050 CAD 243*.2 til) 3441.7 K (2) 242. OSTM2) 75. 7 *1(2) 417. K2(2l .5 ID) 1**2 OSTRO) 2*1 RIO) 2111*. R20) -.24 ARMS .70

Mot:

H i ) - Modulus of 1-th layr (KFa) CAT/ - Adj. moduli of sphalt layar for 25 dg C (HFa) OST* - Owlk strss (kfa) R1.K2 - trass nsltlvlty Coefficients ARKS - Average RKS Ralativ Krror

Figure 9(b): Summary Output (Metric Units)

B-26

|EVERCALC~33

1!)

BACKCALCULATION BY XVZRCXtC VERSION 3.3 Rout*: EXAMPLE PROBLEM

Modulus sr* Morsaslitsd to 9000 lbs Psvsasnt Modulus Adjusted for Tsapsrsturs Mil* o P t v u i n t Modulu 0 to 1000 ksi 0 2 4 6 10 < &* Modulus 0 to 50 ksi 0 10 20 30 40 50 <-Subgr*d* Modulus- 0 to 50 ksi 0 10 20 30 40 50

, I I I I I
460 ^ 6 2 4 6

600.050 0

^ 10

----1---- I-----1 33 , , , , ----1---- 1


0 10 20 30 40 50

, ----1---- 1---- 1---- | ---- 1 26


10 20 30 40

50

Figure 10(a): Bar Chan Plot of Normalized Moduli (US Units)

BACK CALCULATION BY EVER CALC VERSION 3.3 ' Bouts: EXAMPLE PROBLEM

Modulus arc Morsaslitsd to 5000 kg Psvsasnt Modulus Adjusted for Tsapsrsturs k* <-Pav*aent Modulus* 0 to 10000 kPa 0 2 4 6 10 < Bast Modulus - 0 to 500 kPa 0 100 200 300 400 500 <-Subgrsds Modulus- 0 to 500 kPa 0 100 200 300 400 500

----- 1 I-I-----I---- | ---- |


2439 4 6 2

| ---- | ---- | | ---- | ---- |


243 ^

, ---- | ---- 1---- 1---- 1---- 1


169

600.050 y H M 0

^ 0

10

100 200 300 400 500

100 200 300 400 500

Figure 10(b): Bar Chan Plot of Normalized Moduli (Metric Units)

IEVER CALC 3 3

28J

CEHXRAL DATA Rout: EXAMPLE PROBLEM No of Layera: 3 Load Plat* Radius: Sansor Mo: Radial Offacta: 1 0 5.9 i Fi1 Mas: ECALC1R.CEN Onita: E Mo of Sanaora: 5 7

INSERT OFF

12

I 24

5 36

10

Teap Correction: Y Sd Modulua Option: E Maxiaua Iteration: 5

Tanp Measurement: D Stiff Layer Option: Y RMS Tolaranca(I): I. Modulus Toirane(I): 1.

Please Entar Pout Man or Titl* of Analyaia, 25 Charactara Max Use Arrows,TAB Or ENTER To Mova Highlight Bar, <F1> When Dona Editing Screen

Figure 11: General Data Screen Using Stiff Laver Option (US Units)

B-28

fcVmCALC 3 3

2V|

DEFLECTION DATA Routa: EXAMPLE PROBLEM Layer Mo Rnown/UnXnown 1 1 2 1 3 1 Poisson' Ratio .35 .40 .45

INSERT OFF

Fila Ma: ECALC IR.DEF Sd Moduli Min Moduli Max Moduli


200000.0 0.0 0.0

50000.0
20000.0

0.0
0.0

0.0
0.0

Station: 00.050 Mo of Data Sats: 2 Data Data Data Data Set St Set Set 1 2 3 4 Load 864. 12251.

Thlcknass: 6.80 Teprature: 76.00 Snsor No 1 2 3 4 7.953 6.220 5.118 3.110 13.307 10.669 8.898 5.669

21.00 5 2.008 3.898

Plaasc Enter Eoute Mas or Title of Analysis, 25 Characters Max Use Arrows,TAB Or ENTER To Mova Highlight Ear, <F1> Vhan Done Editing Screen

Figure 12: Deflection Data Screen Using Stiff Laver Option 0JS Units)

|E V E R C A L C 33
SAC* CALCULATION BY EVERCALC VERSION 3.3

EXAMPLE PROBLEM Mllepoat


bu*Uc>er o f L a y tra

Thickneaaaa(In) tava*ent Tep. (F)

00.OO S 4
II
h o

Load

(lbs)

i (pii) (pal) i i

1(4. 200000. 41335?. 50000. 35522. 20000. 231?*. 1000000. 1000000.

Seed Moduli Uaed Moduli Calculated

Deflect Iona bO

(11)

After Iteration

4 DIFFERENCE

Orr*tT(in)

CALCULATED

MEASURED ?. S3

oirr( )
.3
.4 -1.5 * 1.5

0.0
12. 0
24 .0 )4.0

.0

7.121 .1? . It?


3.045
2 . 01

.220
I.ill
3.110 2.001

. 02$
.023

-.0 ?

-.011
.1131

.045

-.
4.3 .1

ABSOL. SUM: AA1TM. SUM: EMS ERROR:


Streaaea E .00 . 0 (pal) I Straina (10*- ln/in

1.0

VERTICAL TANCENTIAL

STRESS : TRAIN : STRESS : STRAIN: STRESS : STRAIN : STRESS : STRAIN:

-13.750 -152575? -5.0412 -144 .4423 -.0421 -2 1472 -.0045 -.0053

SI .33 124.4?2
.0172 51.4443 .0013 1.0270 .0015 .0022

SI .33 124 4?2


.0172 51.4443 .0013 1 .0270 .0015 .0 02 2 .0022

.00 o O Load bO 1 2 3 4 5 E 00 00

17..30

-.?
-.0242

313 .0

.00 00 .00

- . 001

Modulua tolerance criteria la aatlafled.

(lba)

: (pal) (pal)

12251. 4335?. 51111. 35522. 4319. 231?S. 20413. 1000000. 1000000.

Seed Moduli Used Moduli Calculated

Deflectiona

(mila)

After Iteration

orrsETdn)
.0

CALCULATED
13.317 10.432 .054 5.431 3.14

MEASURED
13.30? 10.44

DIFFERENCE
-.010 .03? -.154 .031 -.014

DIFF (%)
-. .3

s.o
12.0 24 0 34.0

S.ll
5 44 3.1!

-l.S
.1 -.4 -l.S

ABSOL. SUM: AR1TM. SUM: EMS ERROR:


Streaaea (pal) t .10 1?,.30 4 Stralna (10*- in/in)

3.1 .

VERTICAL TANCENT1AL

STRESS : STRAIN : STRESS : STRAIN. STRESS : STRAIN: STRESS : STRAIN :

-25.?34 -239.5112 -1.1214 -213.0931 -.0751 -4.2? -.0010 -.003

131.3301 1*3.31?0 .71

131.3301 1*3.3170 .71

250.Il1 -7.4715

S .51I?
.014 2.0217 .002 .003

S .51 I?
.014 2.021? .002 .003 -.045

00 313 .0

00 00 .00

-.002

Error tolerance criteria la aatlafled.

Figure 13: Backcalculation Output Using Stiff Laver Option (US Units)

B-30

Ie v e r c a lc T J
t a w r y i f ta o lo tlw litlw it l U ll w 00.916

ID
BAD 4)91)4. 40192, Bll) 3)122. 4)919. MT 4.44 10.41 4.24 .49 .7411 CBTB .42 . .1) .19 BO) 2)17. 041). 21094. 221&9. MTI 4.20 4.24 4.22 4.22 e*n 10.47 10.47 10.47 10.07 1.40 BID 1400000. 1400000. 1400000. 1000000. 1000000. MTU 29 02 29.02 29.02 29.42 .4004 CtT* 14.41 14.40 14.40 14.40 .40

LOAD 144. U t i l.

Bll) 40))?. 111110.

MBA* 407117. 442920, 9721. MO*.* 402171. 9112. 410149 to 400 lba 7244. Bl. M I M O BufcfraOe ie o Pine-rained Boll Motoi

} , -12.0004

t (1) - Modulus of 1-th layer (pal) BAD - Ad] M u l l of eaphalt layer for 77 # P (psi) IITl - Kill i t r m Ipai)

CST - Confining o tr o aa (pal) MTU D ov lator i t r m (pal)


K l . U I t i i M oenaltlvity coefficients M O - Coofflclont of Mtormirvatlon

Figure 13; Backcalculation Output Using Stiff Laver Option (US Units) (continued)

M C H C X L C U L A T !OM T BVKBCALC V I U 1 0 M ) .)

M uto:

U A K P L l PBOOLL* 4000 lba

Modulus oro B o r a a l ii e d t o Milo


I00.0SD

MS
410.2

KID
402.0

K ill
>0.9

M T Rlt )
t.l

K) (2)
724.

M il)
.74

K ill
22.2

MTU) I
4.2

BIO)

MO)

ABHS .94

to:

K i l l - Modulua o f 1 - t h l a y e r (Mai)
M E - Adj. Moduli of aaphalt layer for 7 M T i - Bull atreaa (pl) B 1 0K2 - tree* Sensitivity Coefficient Mhs - Averae M S Relative Krror

Figure 14; Summary Output Using Stiff Laver Option (US Units)

BACKCALCULATION BY EVER CALC VERSION 3.3


a w w i

Bout*:

EXAMPLE PROBLEM

Modulua ara Moraaalitad to 9000 lbs Bavaaant Modulua Adjusted for Taaparstura Mila <aPavaaant Modulus- 0 to 1000 kai 0 2 4 t 10 5 <
0

----- 1 1-1---- 1---- 1---- 1


2 4 10

| ---- 1---- 1---- 1---- 1---- 1


3 10 20 30 40 50

Bast Modulus 0 to 50 kai 10 20 30 40

50

oSubgrad* Modulus 0 to 50 ksi 0 10 20 30 40 50

| -----, ---- , ---- 1---- 1---- ,


22 10 20 30 40

00.050

, ----, -----, ---- , -----,-----1


0

, , , , I ,
0

| | ---- 1---- 1---- 1---- 1


0

50

Figure 15: B ar Chan Plot of Normalized Moduli Using Stiff Laver Option (US Units) B-31

APPENDIX C User's Guide for MODULUS 4.0

PRELIMINARY USER'S MANUAL - VERSION 4.0

The MODULUS program, a modulus backcalculatlon system described In this report has been developed by the Texas Transportation Institute for the Texas Department of Highways and Public Transportation. See TTI Research Reports Any technical questions 1123-1 and 1123-3 for background and technical details. address: Pavement Systems Program Texas Transportation Institute Texas A&M University College Station Texas 77843 (409) 845-9913 This system is intended to be used when analyzing data collected with only the Falling Weight Deflectometer. thi.s version of the software, 1) 2) 3) 4) Automatic calculation of a depth to a stiff layer (H4)t which can be overwritten by the user. Automatic calculation of weighting factors for each sensor. Detection of non-linearity 1n the subgrade and automatic selection of the optimum numbers of sensors to use 1n the backcalculatlon process. The use of the corps of Engineers WES5 linear elastic program which 1s considerably 5) faster than existing programs end has no copyright restrictions. Inclusion of a routine to permit manual Input of deflection bowl data. The following enhancements have been made in

regarding'this software should be directed to Tom Scullion at the following

One minor restriction on this version of HODULUS Is that four of the deflection sensors must be located at offsets of 0, 12, 24, 36 Inches from the center of the load plate. It does not matter which four, for example other sensors could be Included at 8 and 18 Inches. The MODULUS system has the three following subsystems. Subsystem 1 This Data Input either the user can anually Input subsystem has two options:

deflection bowl data or 1t can be automatically read from the field diskette

C-l
i

(currently Dynatest versions 9, 10 and 20 data formats only). This subsystem creates an Input file for the modulus backcalculatlon procedure. The field diskette file must have a .FWD extension end the created file is given the extension .OUT. This subsystem may be skipped if the .OUT file already exists. Subsystem 2 Modulus Backcalculatlon Three options are available depending Reads the .OUT file, performs the backcalculatlon and creates a .DAT file, which contains the calculated t values. Subsystem 3 subsectioning. Plot Deflection the data stored In the .DAT file and performs on the level of familiarity the user has with backcalculatlon schemes. Graphically displays

C-2

1. INTRODUCTION

A. Getting Started The TTI MODULUS Analysis System program is distributed 1n a 5 1/4" or 3 1/2" high density disk. To make backup copies of this diskette use the DISKCOPY command from DOS to insure that all the files are copied to the backup diskette. It 1s recommended that the user create a subdirectory called MODULUS and then copy the entire distribution diskette into that subdirectory. 8. System Requirements Minimum system requirements to run the program are: IBM AT or compatible microcomputer 640 Kb of RAM DOS (version 3.00 or later) operating system Math coprocessor chip (80287, 80387, or similar) A hard disk with 1MB of available storage space A EGA or VGA graphics card with 256 Kb of screen memory and a compatible RGB or monochrome monitor Printer

It is recommended that an advanced microcomputer, a 286 or even a 386 based machine, be used in order to minimize program execution time. C. File Naming Conventions The TTI MODULUS Analysis Systems program uses several types of files. filename: .LBR: .WES: .DAI: .DA2: Input/output screen display library. A file produced by the MODBAC program. for the WES5 program. These files contain the final results. The PRMODRES program uses these files to produce summary and detailed output tables which can be sent to a printer. It contains Input data The type of each of these files 1s identified by the three letter extension to the

TMP.DEF:

This file contains Input Information provided by the user when selecting programs. the full inalysls backcalculatlon option. The Information 1s later used by the HODBAC, WES5, end SEARCH

WES.RES:

This file stores the normalized deflection data base that are calculated by the WES5 program. programs as TMP.DEF. The file 1s used by the same

TMP1.DEF:

This files contain default information for 24 fixed pavement designs. If any of the fixed TMP1.DEF - TMP24.DEF designs are selected, the corresponding file 1s renamed to TMP.DEF and used as Input to the search routine.

to TMP24.DEF

WES1.RES: .DAT:

These files contain the default deflection data base, they are renamed to WES.RES and used as Input to the search routine. Files with this extension store deflection data is used by the DEL IN IAT program. The DES1GN.DAT file contains the default names for the fixed analysis option of the modulus backcalculatlon subsystem. See the section on Customizing Fixed Designs" for Instructions on how the user can create Its own fixed designs. The DEFAULT.DAT file stores default values for options two and three of the modulus backcalculatlon subsystem. readings and corresponding backcalculated moduli for each test point. This

DEFAULT.DAT:

This file stores values used 1n the previous run of modulus backcalculatlon program. The values are displayed as default Inputs during current execution of the program.

DESIGNS.DAT: .EXE: .FWD: .OUT:

This file contains the name descriptions for the 24 existing fixed designs. Identifies executable files. The field Input file as obtained from the Falling Weight Deflectometer. This files are produced by the FWDREAD program. They contain deflection Information extracted from the FWD files (.FWD) or manual input Into the system.

DEPTHS.OUT:

Contains the depth to bedrock as calculated by HODBAC.FOR program. The vilues are printed 1n the summary report.

C-4

2 A B S 0 L U T .O U T :

Stores the ABSOLUTE error vilues of the fit is cilculited by the SEARCH.FOR program. summary report. These vilues ire also printed 1n the

STATS.OUT:

Stores the mein ind standard devlitlon for the depths to bedrock numbers. These vilues ire presented In the summary for each report. A speciil

.VAL:

file contilnlng Poisson Ratio vilues

pavement liyer. .BAT:

This pirtlculir file 1s only used for output

purposes by the PRMODRES progrim. Batch files used for Instil 1at Ion of the system 1n a hard disk ind for setting up and starting the program.

C-5

2.

r u n n i n g the p r o g r a m

Starting the Program To run the T7I MODULUS Analysis System programs, make the MODULUS subdirectory active by typing CD\MODULUS after the DOS prompt. If another drive 1s active, type the letter of the drive where the system has been Installed and press <ENTER>; then type CD\HODULUS. Once 1n the MODULUS directory type MODULUS followed by <ENTER> to start the program. After a few seconds, the Introductory screen will be displayed. At this time the main program should appear on the screen, see Figure 1. Main Program Menu Options The Main Program Menu screen allows the selection of any of the four available programs. the same way. The following programs are available: Convert FWD data to INPUT data: options are available: This program builds the .OUT file Two which contains the FWD deflection bowls for later processing. To execute any of the programs, use the up/down arrow keys All menus 1n this package work in to highlight the selection and press <EN7ER>.

h 1. 2.

Manual Input of load and deflection data. Automatic read of Dynatest FWD field diskette. This program

Run Modulus Backcalculation Program: This option allows the user to execute the Modulus Backcalculation (MODULUS) program. uses INPUT files (files with the .OUT extension) that have been converted from FWD files using option one above, or It can also process files that have been custom-made using a text editor or similar program. Plot Deflection and/or Moduli values: Select this option to produce plots of deflection data or backcalculated moduli values, as a function of project length. The program uses a cumulative difference algorithm to achieve unit delineation for either deflection and moduli data. The delineation approach 1s useful for identifying units of sections or stations that present similar structural behavior. This option will also plot the absolute error and depth to bedrock data.

C-6

>

V4.0 MODULUS MAIN MENU PROGRAM

*]) *2) *3) *4)

Input Data Conversion Options

Run Modulus Backcalculation program* Plot Deflection and/or Moduli values* Print results of latest analysis* *5) Exit to DOS*

Use the t or i keys or enter the option NUMBER and press <ENTER> (C) Copyright 1989, Texas Transportation Institute. All Rights Reserved

Figure 1.

Main Program Menu Screen.

C-7

Print results of latest analysis: This option permits the user to skip directly to the Print Menu In order to obtain sunmary nd/or detailed printouts of the last analysis performed by the Modulus Back Calculation program. To finish a session just select option five to exit to DOS.

C-8

3. 2.

OPTION 1 INPUT DATA CONVERSION OPTION The first screen within the Input data conversion option 1s shown 1n Figure To get data Into the system the user has two options. The first 1s the

automatic conversion of the Dynatest field diskette file Into the Input data format for MODULUS. The second 1s a screen to manually Input defleclton data. Both are describe below. 3.1. Convert .FWD to .OUT Data The format used in the Dynatest FWD files 1s highly elaborate and most of the information that they contain 1s not relevant to the programs contained 1n the TTI MODULUS Analysis System. the specific data was developed. Consequently, a program capable of extracting This program extracts the following variables

from a FWD file: district number, county number, highway prefix and number, mile point position of the station (to 3 decimal places), and load and deflection readings (up to seven) for a pre-spec1fied drop along the length of a project. The program then stores this information in a new file and appends to U s name the extension .OUT. These files form the actual Input to the Modulus In This Backcalculation program and are hereon referred to as INPUT or .OUT files. general during FWD testing one or more drops are made at one location. program can handle from one to eight drops per location. requested to select one of the eight drops for processing. To start the FWD conversion program, select option one from the menu and press <ENTER>. A window will appear In the lower part of the screen asking you to verify your choice. Enter <Y> 1f the choice Is correct. After a few moments, the program Input screen (Figure 3) 1s displayed. There are five fields of required Information that the user needs to Input before the program can run. These re: Enter the letter Identifier of the If the FWD file 1s In the If Finish this - DRIVE WHERE THE FWD FILE RESIDES:

The user will be

drive where the FWD file to be converted 1s stored.

hard disk, enter the U t t e r of the drive from which the program Is running. the file resides 1n a floppy disk, enter the letter of that drive. Input by pressing <ENTER>. - FWD DATA FILENAME: converted.

In this field enter the name of the FWD file to be

Enter the name of the file, up to eight alphanumeric characters,

without entering the extension name (It will be automatically appended to the

C-9

V4.0 MODULUS INPUT DATA OPTIONS MENU ]. 2. 3. Convert .FWD to .OUT Data Enter .OUT Data Manually Return to MAIN MENU

Use the T and i keys or enter the option NUMBER nd press <ENTER>

Figure 2.

Data Input Screen.

C-10

V4.0 MODULUS FALLING WEIGHT DEFLECTOMETER DATA CONVERSION PROGRAM INPUT SCREEN

DRIVE WHERE FWD FILE RESIDES ..................... X FWD DATA FILENAME OUTPUT FILE NAME ................... XXXXXXXX.FWD ..................... XXXXXXXX.OUT

NUMBER OF DROPS RECORDED AT EACH POINT ........... X NUMBER OF FWD DROP TO USE FOR CONVERSION.........X PROCESS ANOTHER FILE? X

(C) Copyright 1989, Texas Transportation Institute.

All Rights Reserved

Figure 3.

Data Conversion Screen.

C-ll

name you entered) and press <ENTER>. county number and highway name. the selected drive press <F1>. - OUTPUT FILE NAME:

In Texas this filename Is a combination of

To see a listing of all FWD files residing 1n To select a file, use the up/down arrows until It can also be up

the desired file 1s highlighted then press <ENTER>. Supply the name of the output file. to eight characters long and the .OUT extension will be automatically appended. Again press <ENTR> to finish this Input. - NUMBER OF DROPS RECORDED AT EACH POINT: automatically displayed. - NUMBER OF FWD DROP TO USE AT EACH POINT: At this point, enter the number This option permits the of the drop to be analyzed. Frequently four drops are recorded at different load levels, e.g., 5,000, 8,000, 12,000, and 15,000 pounds. user to select the load level of interest. Check the input carefully. If a mistake has been Bade, press the <ESC> key and the cursor will be set back to the beginning of the input process, at the position of the drive letter designator. Press <ENTER> to validate the entries until the Incorrect one is reached. name and press <ENTER> to validate 1t. that entry. Pressing this last <ENTER> will start the conversion process, which should take about 5 to 10 seconds depending on the disk access speed and the length of the file being converted. When the program has successfully executed, a window with the following message will be displayed: FILE XXXXXXXX.OUT CONTAINS or stations stored In the file. As a last option, the program will prompt to detemlne If another file 1s to be processed. Enter <Y> to extract another FWD file. To quit, enter <N> and press <ENTER> 1n order to go back to the Main Program Menu. To change 1t Just enter the new value or Keep on pressing <ENTER> until the last In this field the numbe- of drops (up to eight) performed at each point or station during the test will be

field is reached. If it Is also correct, press <ENTER> once gain to validate

III

POINTS,

where XXXXXXXX corresponds to the .OUT file name and

III

to the number of points

C-12

3.2

Manual Input of Deflection low! Dati To start the manual deflection bowl data Input program, select option two A window will appear asking you to

from the Input data menu and press <ENTER>. verify your choice.

Enter <Y> 1f the choice 1s correct.

The manual deflection bowl data Input screen (Figure 4) will be displayed. The user will be asked to supply the following Information to create the file: - OUTPUT FILE NAME: characters. Enter the name of the file, up to right alphanumeric that the extension .OUT will be appended Note

automatically to the name entered by the user. - NUMBER OF BOWLS TO BE ENTERED: - DISTRICT: - COUNTY: - HIGHWAY: In this field enter the total number of deflection bowls you Intend to enter manually. Enter the district number in this field. (2 digits) Enter the SDHPT county number 1n the field. (3 digits) In this field enter the SDHPT highway prefix and number. (6 In this field the program will display

alphanumerics) - ENTER INFORMATION FOR BOWL NO.: deflection data. - STATION OR MILEPOST: - LANE: - LOAD: In this field enter the milepost (to three decimal places) where the deflection bowl was measured. Enter the lane (R or L) In which the deflection bowl was measured. In this field enter the load 1n pounds used for the deflection Enter the deflection In mils measured it each of the FWD Carefully check the data 1n fields. If everything Is correct the number of the deflection bowls for which the user 1s entering

test drop. - W1 to W7: - VALIDATE:

sensors 1 through 7 for this deflection test. enter <Y> to validate the entries and the data will be read by the program and written to the .OUT file. To change any of the entries enter <N> and the cursor will be repositioned at the STATION OR HILEPOINT field. go to the next field. Press enter until the cursor Is at the field you want to change, enter the new data and press enter to When all the data entries are correct enter <Y> at the VALIDATE field and the data will be read and written to the .OUT file. Note that each deflection bowl is entered and validated, the ENTER INFORMATION FOR BOWL s No. field 1s automatically updated and the cursor is positioned at the STATION OR MILEPOST field for each subsequent deflection bowl. After the user

C-13

Print Screen BOWLS.AID Input Order: Horizont*! V4.0 .MODULUS/ FALLING WEIGHT DEFLECTOMETER MANUAL INPUT SCREEN OUTPUT FILE NAME NUMBER OF BOWLS TO BE ENTERED DISTRICT -- XX COUNTY - XXX HIGHWAY XXXXXXXX.OUT XXX XXXXXX

ENTER INFORMATION FOR BOWL NO. STATION OR MILEPOST --- XXXXXXXX W1 --- XXXXXX W5 --- XXXXXX SDHPT version. W2 XXXXXX W6 --- XXXXXX LANE -- X W3

XXX: LOAD --- XXXXXXXX W4 --- XXXXXX VALIDATE ? - X

XXXXXX

W7 --- XXXXXX

Developed by the Texas Transprotetlon Institute.

Figure 4.

Deflection Bowl Manual Entry Form.

C-14

has entered and validated the last deflection bowl the program will return to the Main Program Menu screen nd the user can select one of the five options available. To abort this program, press the <ESC> key if the cursor 1s positioned 1n the first Input field; otherwise press 1t twice. These <ESC> key sequences will quit the program and return to the Main Program Menu. Alternatively, 1f the user has t text editor he or she can create a .OUT file containing as few or as many deflection bowls as desired by using the data input format of the .OUT file listed below. Data Item
(a) (b)

Inpvt Fcririi 13 14 IX A8 F8.3 IX A1 16 F6.2 F6.2 F6.2 F6.2 F6.2 F6.2 F6.2

Column 1-3 4-7 8 9-16 17-24 25 26 27-32 33-38 39-44 45-50 51-56 57-62 63-68 69-74

(c) (d) (e) (f) (9) (h) (i) U) (k) 0) (m)

State County Blankfield Highway Milepost Blankfield Lane Load(lbs) Sensor ](miIs) Sensor 2(mils) Sensor 3 (mils) Sensor 4(mils) Sensor S(miIs) Sensor 6 (miIs) Sensor 7 (mi Is)

C-15

4.

OPTION 2 MODULUS BACKCALCULATION PROCEDURE Option two of the Main Program Menu allows the user to run the Modulus Inputs to the program consist of a series of default They are created and

Backcalculatlon program. read automatically. After

and temporary files, which are transparent to the user.

The only file that 1s user-supplled Is the .OUT file, which and validating option two from the Main Menu, the

was created using option one of the Main Program Menu as explained above. selecting Input/Output Information screen (Figure 5) 1s displayed. In this screen the user

1s requested to enter the name of the .OUT file (the file created by option one). From the .OUT file supplied, MODULUS will generate a file which will store the deflection Information and the corresponding backcalculated moduli values for each pavement layer. This file 1s referred hereon as the OUTPUT file and 1s given the extension .DAT. Enter the name of the INPUT (.OUT) file, up to eight characters long. Before pressing <ENTER> be sure the input file is correct. If Incorrect, then An use the backspace key to make the correction and press <ENTER> to continue. prefix as the INPUT (.OUT) file. file and press <ENTER>. name field.

output (.DAT) file will be automatically created by the system using the same Press <F1> to see a listing of all the current .OUT files 1n the current directory. Use the Up/Down arrow keys to select a .OUT The .OUT file name will be displayed 1n the Input file Specify which deflection bowls are to be used 1n the modulus

backcalculatlon procedure by entering the beginning mile point and the ending mile point, or enter 0.0 (in both) to use all the deflection bowls 1n the .OUT file in the analysis. screen. After entering the INPUT filename, the program displays the Modulus Backcalculatlon Menu screen (Figure 6) which allows the user to select any of three alternative ways of running the program or to return to the Main Program Menu. There are three options for performing backcalculatlon 1n MODULUS. are: USE AN EXISTING FIXED DESIGN: This option was Intended for those cases The data base for the This option lets where the agency runs the same pavement type many times. They The number of deflection bowls to be used 1n the analysis will be displayed, then press <ENTER> to go to the modulus backcalculatlon menu

fixed designs are stored and only the search routine 1s run.

C-16

MODULUS INPUT/OUTPUT FILE INFORMATION NAME OF THE INPUT F I L E .......... XXXXXXXX.OUT (PRESS *F1FOR A LIST OF ALL .OUT FILES) BEGINNING MILE POINT ............ ENDING MILE POINT .............. XXXXXXXXX XXXXXXXX

V4.0

(NOTE - - LEAVE BLANK TO USE ALL DEFLECTIONS) NUMBER OF BOWLS FOR THIS ANALYSIS - - - XXX PRESS ENTER" TO CONTINUE SDHPT version. Figure 5. Developed by the Texas Transportat1jw_Inst1tute.

Input/Output File Information.

V4.0 MODULUS MODULUS BACKCALCULATION MENU *1) *2) *3) *4) Use in existing fixed design * Input material types * Run a full analysis * Return to Main Menu *

Use the t and 1 keys or enter the option NUMBER and press <ENTER> (C) Copyright 1989, Texas Transportation Institute. Figure 6. Modulus Backcalculatlon Menu. All Rights Reserved

C-17

the user select between 24 designs (8 pavement designs X 3 depth to bedrock) for which 11 Input parameters, except for the deflection and load values, have been already calculated and stored 1n disk files. This option provides the fastest analysis since 1t only has to perform the Search algorithm 1n the program. Note: The current data bases were built assuming seven sensors at one foot spaclngs and either a clay (240"), sandy/day (180") or sand (120") subgrade type, with the depth to bedrock as shown 1n parentethlsls. In the section "Customizing Fixed Designs", you will find Instructions on how to create alternative fixed designs that comply with particular characteristics which are applicable to your needs. - INPUT MATERIAL TYPES: For this option the user selects the material types, thicknesses for the pavement layers, and test temperature, and the program assigns the range of acceptable moduli and poisson values to be used 1n the analysis. This was Intended for the user who is unfamiliar with moduli values. In this option the user supplies all of the input It 1s Intended that once the user is This option selects acceptable ranges for each material type. - RUN A FULL ANALYSIS: parameters needed to perform the analysis. option.

familiar with modulus backcalculation that this will be the most frequently used The user has full control over all of the inputs to the analysis. To quit the program while 1n the menu screen, select option 4 to return to the Main Program Menu. Using the Modulus Backcalculation Menu Options: The principal difference in the three options of the Modulus Backcalculation menu is that in Option 1, default databases are used and no runs of the WES5 program are required. Each of the three options is discussed 1n detail. Option 1 - Fixed Designs: Select this option and the program will display the Existing Fixed Designs screen (Figure 7). These layer thicknesses are common in Texas but can be modified to fit a particular user's need (see section "Customizing Fixed Designs"). The moduli values used to build these default databases are shown In Table 1. The screen presents the user with two prompts: First, select the type of subgrade for the analysis, sand, sandy/day or clay. Pressing <ENTER> after the selection validates the choice and moves to the next prompt. Select one of the 8 available designs by entering the appropriate number and pressing <ENTER>. If no suitable designs are available for the pavement under analysis, press <ESC> twice to return to the previous enu which will permit selection of an alternate backcalculation option.

C-18

V4.0
MODULUS

TYRE OF SUBGRADE, (l)SAND (2)SAND/CLAY (3)CLAY 1) I" SURFACE TREATMENT, 2) I" SURFACE TREATMENT 3) 2" HMAC 4) 2" HMAC 5) 4" HMAC 6) 4" HMAC 7) 4" HMAC 8) 6" HMAC FIXED DESIGN NUMBER 6FLEXIBLE BASE 8" FLEXIBLE BASE 8" FLEXIBLE BASE 10FLEXIBLE BASE 8" FLEXIBLE BASE 10" FLEXIBLE BASE 12" FLEXIBLE BASE 12* FLEXIBLE BASE XX

.
Figure 7. Existing Fixed Designs.

C-19

Table 1.

Modulus Defaults for the Eight Fixed Designs (ks1). - Asphalt Min. Max. 500 500 500 500 200 200 200 200 500 500 500 500 1200 1200 1200 1200

Design Number (Figure A5) 1 2 3 4 5 6 7 8

Min. 5 5 5 5 5 5 5 5

llii.

Max. 100 100 100 100 100 100 100 100

Min.

-.Siifefriit

Max.

Subgrade

Pressing <ENTER> after selecting the desired fixed design choice verifies the selection and starts execution of the program. The screen will clear and the message "The Search Program 1s Running..." will be displayed. displayed. When done, the Refer to the messages "Program terminated normally", and "Press any key to continue..." are Press any key to display the Print Results Menu screen. section "Print Results Program" later 1n this manual for instructions on how to obtain printed output of the analysis. When the program has completed printing the analysis results, 1t automatically returns to the Main Program Menu. Option 2 - Input Material Types: When this option 1s selected, the program prompts for the required information using two separate input screens. The first screen (Figure 8), displays default settings for the FWD test. the FWD default setting need to De changed then press <FK2>. to the HMAC surface layer thickness input field. layer thickness in inches and press <ENTER>. The cursor is However, if automatically positioned at the HMAC surface layer thickness Input.

Press <ENTER> to

move through the fields and make the necessary changes, until the cursor returns Otherwise enter the surface Enter <l> for The cursor moves to the second

field where the program requests the surface layer material. <ENTER> to continue to the next input field. field deserve a brief explanation.

crushed limestone aggregate or <G> for crushed river gravel aggregate, and press The options to be selected in this

The program has built into it equations for stiffness versus temperature for typical mixes found in Texas (crushed limestone or river gravel mixes). Also equations which represent the reasonable range of stiffnesses are also available. These were generated by analyzing stiffness results and obtained on rutted mixes (low stiffnesses) and badly cracked mixes (high stiffness). In the backcalculation procedure, if the user wishes to use a fixed default asphalt modulus, which is often the case on these pavements, then a single value is calculated based on the coarse aggregate type and FWD test temperature. However, If an asphalt modulus is to be backcalculated, then an acceptable range of moduli values 1s generated using the equation for rutted and cracked mixes, and the FWD test temperature. with materials This option was Intended for field personnel who are familiar information but who have limited experience with modulus In this field select whether you want the program

backcalculation techniques.

to backcalculate a fixed value or a range of values for the asphalt moduli. Enter <F> for a fixed value or <R> for a range and press <ENTER> or press <FK1>

to see the formulas used for each of these two options. temperature value and press <ENTER>.

The last field in this Enter the

screen prompts for the pavement temperature 1n degrees Fahrenheit. field and make changes, as explained previously.

Use the <ESC> key to return to the first

After validating the pavement temperature with a <ENTER>, the program displays a second screen (Figure 9). sections to be analyzed. In this screen the user selects the vaterial to be used for the base, subbase 1f any, and subgrade of the pavement The input sequence is organized In five fields. In the The second If a subbase 1s present, Input its first field enter any of the nine available base material options. field takes the base thickness in Inches. type and thickness as for the base. <ENTER> if there is no subbase. section as per the option 11st. <ESC> key as described previously. run the program.

Leave the subbasse type blank by pressing

Enter the predominant type of subgrade for the Changes to the screen can be made using the Press <ENTER> to validate the Input and to When WES5 is complete, Completion of

This time the message "The WES5 Program is running..." appears

in the screen to indicate that the program is executing.

the data base is generated, and the Path Search algorithm program takes over; the respective message is displayed to indicate that it is executing. "Press any key to continue" messages. Results Menu. Option 3 - Pun a Full Analysis: This option of the Modulus Back Calculation Menu l ^ r = G i e u s e r specify the thickness, moduli ranges, and polsson ratios for up to four layers wvthin a pavement section. input re 10) 1s displayed. When you request this option, the To run the The values that are displayed on the The existing values can be edited at the search phase is confirmed by the "Search program terminated normally!" and Pressing any key leads you to the Print

screen are the values used in the most recent run of the program. program with these values press <ENTER>.

three levels which are accessible through function keys two to four. The editing levels correspond to the degree of likelihood 1n which the user would change the values, from less to more likely. For all practical purposes, Information such as plate radius, number of sensors and sensor distance to the plate are prone to remain the same throughout the length of a project since these values reflect the characteristics of the FWD. abort the program. At this point press the <ESC> key 1f you want to Editing 1s done 1n the same way as for the previous programs;

that is, you enter the desired value and validate 1t by pressing <ENTER>.

V4.0 MODULUS <FK2> PLATE RADIUS (IN) - - - - XXXXXX SENSOR NUMBER WEIGHTING FACTOR 1 XXXX 2 XXXXXX XXXX 3 XXXX 4 XXXX 5 XXXX NUMBER OF SENSORS - - X 6 XXXX 7 XXXX XXXXX DISTANCE FROM PLATE - XXXXXX XXXXXX XXXXXX XXXXXX XXXXXX :XXXXXX

HMAC SURFACE LAYER THICKNESS (IN) USE A (F)1XED VALUE OR A (R)ANGE OF VALUES FOR THE ASPHALT MODULUS BASED ON TEMPERATURE .......... INPUT ASPHALT TEMPERATURE (?F) - figure 8. Input Material Types.

HMAC WITH CRUSHED (L)IMESTONE OR CRUSHED RIVER (G)RAVEL AGGREGATE - - X

XXXX

V4.0 MODULUS BASE AND SUBBASE TYPES 1) CRUSHED LIMESTONE 2) ASPHALT BASE 3) CEMENT TREATED BASE 4) LIME TREATED BASE 5) IRON ORE GRAVEL 6) IRON ORE TOPSOIL 7) RIVER GRAVEL 8) CALICHE GRAVEL 9) CALICHE THICKNESS BASE TYPE - - - X SUBBASE TYPE - X > XXXX SUBGRADE T Y P E .........X > XXXX PREDOMINANT SUBGRADE TYPE 1) GRAVELLY SOILS 2) SANDY SOILS 3) SILTS 4) CLAYS, LL < 50 5) CLAYS, LL > 50

igure 9 . Input Base and Subgride Types.

C-23

V4.0 MODULUS <FK2> PLATE RADIUS ( I N ) .........XXXXXX SENSOR No. 1 2 XXXX 3 XXXX HI XXXXX MINIMUM (KSI) <FK4> SURFACE LAYER - - BASE LAYER ......... XXXXXXXXX XXXXXXXXX 4 XXXX H2 XXXXX MAXIMUM (KSI) XXXXXXXXX XXXXXXXXX XXXXXXXXX (KSI) SUBGRADE MODULUS (MOST PROBABLE VALUE) - -XXXXXXXXX Figure 10. Full Analysis. 01 S. FROM PLATE - XXXXXXX WEIGHTING FACTOR XXXX <FK3> LAYER THICKNESS (IN) ........... MODULUS RANGES FOR: NUMBER OF SENSORS - X 5 XXXX H3 XXXXX 6 XXXX 7 XXXX H4 XXXXX

XXXXXXX XXXXXXX XXXXXXX XXXXXXX XXXXXXX XXXXXXX

POISSON'S RATIO XXXX XXXX XXXX POISSON'S RATIO XXXX

SUBBASE LAYER - - - - XXXXXXXXX

C-24

If you want to change *11 the values on the screen, press the <FK2> key. The cursor will be positioned 1n the plate radius field. 1n Inches and the number of deflection sensors. Enter the plate radius for the Then enter the spacing of the

sensors 1n inches and the weighing factor to be used for each sensor, 60 and 72 Inches. 4.0.

FWD, a typical plate radius 1s 5.91 Inches with spaclngs t 0, 12, 24, 36, 48, Note sensors must be placed at 0, 12, 24 and 36 In MODULUS

The user may wish to assign weight factors to each geophone pr let the Leave all the weighting factors at zero for

program assign them automatically. automatic assignment.

However, if the user wishes to overlde this he simply Weighting Factors typically range from 0

Inputs the desired weighting factors.

to 1.0, a weighting factor of 0 will remove that sensor from the calculation process. To change layer thicknesses and modulus ranges press the <FK3> key. In these four fields labelled HI to H3, enter the pavement thicknesses in inches. HI represents the surface layer, H2 the base layer and H3 the subbase layer. The H4 value is automatically calculated and represents the depth in the pavement at which a stiff layer is calculated to occur. (See Report 1123-3 for detailj) This may be bedrock or a stiff clay layer, for example. number if needed. three The user can overwrite this For a For a

Use 300 inches as the maximum possible value of H4.

four layer pavement, enter the thicknesses 1n their respective fields. layer system with no subbase, enter the surface thickness, thickness, then zero <0> to Indicate the absence of subbase. the base layer. To change the modulus ranges only, press the <FK4> key.

the base

For a two layer

pavement, the procedure is the same except that a thickness of <0> 1s entered for Enter the lower

modulus boundary value, the upper boundary value, and the polsson value for the surface layer. Then, depending on whether the pavement Is a two, three, or four

layer system, the cursor will move to the corresponding field allowing you to edit the values for the particular layer. Next, enter the most probable modulus After

value 1n ksl and the corresponding polsson ratio value for the subgrade.

entering the value for the subgrade layer polsson ratio, check all the Input values and 1f necessary, change any values using the appropriate function key and repeat the above process. the program. screen. If satisfied with the Input, press <ENTER> to execute The "WES5 Program 1s Running...* message should now appear on the

C-25

When the program 1s complete, 1t displays the appropriate message and asks the user to press any key. The Print Results Menu is then displayed and the user can obtain a printout of the analysis results.

C-26

5.

OPTION 3 PLOT DIRECTION AND/OR MODULI VALUES PROGRAM This program allows the user to analyze pavement response variables, mainly

deflection readings, calculated moduli values, depth to bedrock and average errors, from a graphical point of view, along the entire project length. It will also perform a unit delineation analysis using the cumulative difference approach in order to Identify units of sections having similar characteristics. Report 1123-1) To run this program select option three from the Main Program Henu and press <ENTER>. After validating the choice, the Pavement Response Variable Graphic Here Representation and Delineation Analysis screen Is displayed (Figure 11). calculated moduli values for each of the pavement layers. (See

enter the name of the data file containing both the deflection readings and the These files are

characterized by the extension .DAT 1n their file names; 1t Is automatically appended to the name of the file that was specified 1n the backcalculatlon phase. Enter the file name, up to eight characters long and press <ENTER>. Next, select the response variables to be plotted. There are a maximum of Deflections seven deflection readings, and four moduli values for each station.

are identified by a number from 1 to 7, 1 corresponding to the sensor closest to the loading plate, 2 to the second closest, and so on. Moduli values have labels from 8 to 11 where 8 Identifies the modulus of the surface layer, 9 the base, 10 the subbase, and 11 the subgrade. shows the depth to bedrock. Lable 12 displays the absolute error and 13 Enter the number corresponding to the response

variable required, 1 through 7 for deflections or 8 through 11 for moduli values, 12 for absolute error or 13 for depth to bedrock, and press <ENTER>. The last item of Information requested is the minimum section length that will be used by the delineation subroutine to perform the unit delineation of the chosen response variable. If consecutive Inflection points 1n the cumulative This feature 1s provided to avoid the

difference curve occur at Intervals that are less than the minimum section length entered, the program will Ignore them. response variable variability. clutter of unit delineations that might occur 1n projects with unusually high Enter this value In miles including fractions To make any changes,

cf a mile, that Is, as a decimal value, and press <ENTER>. use the <ESC> sequence as in the other programs.

As soon as the <ENTER> key 1s pressed, the program starts txecutlng and In a few seconds the screen 1s cleared and a plot of the selected response variable

C-27

V4.0 M ODULUS

PAVEMENT RESPONSE VARIABLE GRAPHIC REPRESENTATION AND DELINEATION ANALYSIS

THE FOLLOWING INFORMATION IS REQUIRED:

NAME OF THE D A T A FI LE...................XXXXXXXX.DAT RESPONSE VARIABLE ............................... XX

MINIMUM SECTION LENGTH - - - - ...............XXXXX

(C) Copyright 1989, Texas Transportation Institute. All rights Reserved

Figure 11.

Setup for Graphics.

C-28

as a function of distance along the project 1s produced (Figure 12). displayed. the message appears.

At the

bottom of the screen a table of statistics for each of the unit delineations 1s If there are more than three delineated sections 1n the plot, press Press any key until (C/Q): The "Would you like to combine sections manually.or Quit? any key to see the statistics for the remaining sections.

To quit the program at this point enter <Q>, otherwise enter <C>. Enter the number and press <ENTER>.

manual combination routine then prompts for the number of sections the user would like the combination to have. plot. To arbine all sections, enter the total number of sections that were delineated In the If the user did not elect to combine all sections, a prompt 1s displayed asking for the number of the last section to be Included as the new section one. The subroutine repeats the last prompt until all of the sections have been accounted for. Then 1t recalculates and replots the curve showing the new Repeat the above sequence for

delineations and their respective statistics.

manual combination 1f there are sections left to combine or quit the program. When the user answers <Q> to the prompt, the user Is given the choice of printing the statistics displayed: <R> Program Menu:". redisplays for the latest delineation. Then, the following prompt 1s "Enter <R> to analyze other Responses or <Q> to quit to the Main Selecting <Q> returns to the Main Program Menu while entering the Pavement screen, Response Variable Graphic Representation and

Delineation

Analysis

allowing the user to select

another response

variable for graphical analysis.

C-29

Rond: F12818

Pavewent Response V ariab le:

E4 (Moduli values in KSI)

C-30

S ectio n H 1 2 3

Fro To Mean 0.041 0.940 14900.56 0.948 1.600 27429.71 1.600 2.359 24673.33 P ress any hey to continue

S.Dev. 4065.28 5520.06 3237.58

Figure 12.

Plot for Subgrade Moduli Values for a Section of FN 2818.

e.

PRINT RESULTS PROGRAM The Print Results of Latest Analysis option 1n the Main Program Menu gives (Figure 13) that 1s

the user direct access to the same Print Results Menu

displayed after any of the three options 1n the Modulus Back Calculation Program terminate execution, and allows the user to print a results summary table or a detailed estimated deflection report, or both for the analysis that was performed the last time the Back Calculation program was used. The options 1n this menu are: - PRINT DEFLECTION & MODULI SUMMARY TABLE: This option lets the user print a table listing the deflection readings, the calculated moduli values, and the estimated absolute percent error per sensor for each station 1n the project and the estimated depth to a stiff layer for each deflection bowl. At the bottom of the figure, statistics are printed for all of the above variables (Figure 14). The asterisk printed at the end of the line denotes moduli values which hit upper or lower limits of either, the range of allowable input moduli values or modular ratios used within the program to generate the deflection data base. If more than 10% of deflection bowls hit a limit then the user should rerun the program with wider limits to eliminate this problem. - PRINT ESTIMATED DEFLECTION TABLE: Option <2> of the Print Results Menu produces a detailed station by station result report which includes the back calculated deflection values, absolute error and squared error values, force and pressure at the loading plate, and a 11st of checks indicating 1f the moduli values are close to the given limits, If the convexity test falls, or 1f the solution to the particular station was Infeasible (Figure 15). - PRINT BOTH OF THE ABOVE TABLES: This option prints the summary table

first, advances the paper to the beginning of a new page, and then prints the detailed section by section report. RETURN TO MAIN MENU: It does just that.

C-31

V4.0
MODULUS PRINT RESULTS MENU *1) Print Deflection I Moduli summary table *2) Print Estimated Deflection table * *3) Print both of the above tables * *4) Return to Main Menu * Use the 1 and 1 keys or enter the option Number and press <ENTER>

(C) Copyright 1989, Texas Transportation Institute.


------- ' "

All Rights Reserved


J . J M - M l arm

'

hf

' v

l .............................

- !------ --------------------------------------------------------- '

Figure 13.

Print Results Menu.

C-32

Til its m um usui isunin itfoin


Il l ir ic i : It tl Citili: I l i l u i / t i W : Fllt Tl l d i t i t ( l i ) Ml SMI Ml 141.11

(liriitt 4. 1) lOICll lilCIIfii) lim n 111,1) 111, 1)1 11,111 11,111 1 1 It,III li ft! It

lutili

fiititil: li l t : Stlltit: Sit(ri<t:

fittili 1.141 1.114 1.114 1. 1 411 .SOI M il .Til l.tll 1.141

( 111) I I , 4M 11, 11] 11, 11? 11,411 11,411 11, 1(7 11,111 11,111 11,111 11,111

l i m i l i Itll tcllti (l il t) : 11 14 II 11 11.11 41.11 11.11 11.14 11.11 Ml

it Ml Ml S.ll 1.14

S.ll l.tt Ml Ml 4.11 Ml Ml Ml Ml MI l.lt Ml

11 Ml Ml Ml Ml Ml M4 1.11 l.lt Ml Ml

CtlttltltMWiJl f i l i t i (III): Mitili SllfUl) im iti) stmt)) SIICIt ) MIDI'S III. SII. IIS. III. III. 111. 411. III. III. 411. SM IM 11.4 11.1 11.1 11. S IS.I SI.S SM 1S.S 11.1 11.1 IM M M M M M M II M M M M M M 1. 11. 1.1 M 1.1 1.1 1.1 l.l S.l 1.1 lt. 1

ti. It<rtl Ki ll 111.11


111 II 10.11 U4.ll It?.II m

11.11
1.41 Ml lt.lt 11.11 11.11 11.11 Ml Ml 11.11 Ml SMI

1.11
Ml Ml Ml 11.11 Ml Ml I.IS 1.41 Mi Ml 14.11

11.11
11.11 41.11 11.11 14.11 11.11 11.11 14.11 11.11 ll.lt 11.11

11.11
St.!! 11.1? 11.41 11.11 11.41 11.14 11.41 11.41 Ml 14.41

11.
14. 1. S.

1.11
Ml 1.14 1.14 Ml 1.11 4.11 1.41

1.
1. 11.

ii:.ii i? * in .
IMI

11. 11. 1.
SS.

1.1
4.1 1.1

leu: 514. It: ?n Citfl(l):

s.n
Ml Sl.ll

III.
111. SS.

I t i . II

mi
41.11

11.11 11.11

Hit

Figure 14.

Summary Listing.

C-33

Fui: I
ITI I l l i r i c i : il Cult): li SUI 1.III II.Ili l.lll III.Ili ICCUI JIU1SIS SlSltl fSICTlOI UIMT) ( l u m i (.li l. lll 11.Ili S1.III II . Il i II.Ili Il III 11.111

rimi
t i f i t i ( l u t i l i f l i t i ( l i ) : S uf i c i t i i c l i i i i l l i l : l i t i t l I c l i t i M II): Sibili ttlckuii(li): Stillili l l l t l i m d i l :

l U t n c i (li) ( t u m i n i l l u t i l i f l i t l i u n i r : Il i: ICI 550 u n o m v t s li II: I H K li 11: D I 11: I Il II: 1.41 Il

Stilili: l. lll Il lim ili Militili: 11.11 Cilulitil liflictlu: 11.11 III lii|lt Fnlir: l mot 1.41 StlfiCt(ft) Ujir: (itili f i l m dii): IMI (l ui li liuti* US Stilili: 1.114 11 l i m i l i li flict lu: 41.11 Cilulittl Dtdttliu: 41.41 lii[lt fictu: l .t l 1 tlIOR 4.11 lijir: sememi) l i d l l fi lit i (lil): 110.4 (l ui li lilil? US Stiliti: 1.214 11 l i m r t i li fltct In : 11.1? Cilcilitil l i f 1ict iil : 11.41 l t 11kt Fic1ir : l.tl 1 tuoi l.l? lijtr: StlflCE(U) 1*0)1 filiti (hi): HS.t (l ui U Usiti? IO Stilili: l . ll l 11 l i m i l i lif Itti In: 11. tt Cilciliti! Dtfltdlii: 14.41 lii|lt ficlii: 1.40 t tUOR 4.11 Ujir: SllFlCMl) li lil f i l m di i ) : tilt (l ui II lilil? IO Stilili: 1.411 l i m r i l lifltctlii: Cilcililil Itdictlii: Iti|kt ridi r: 1 Q I 01 Ujtr: I*) 1 f i l m (Iti): ( l u i li Usiti? II 14.11 44.14 1.11 *1.11

1! 11.11 II. Ili 1.11

li II ti. 1.11 1.11 12.11 MI MI l.tl MI USUO) Sl)lliSl(tl) 11.1 M IO l/l

II li MI 1.11 MI 1.11 MI t.ll MI l.tl svienila M li li MI Mi MI MI MI MI l.tl III SBICliDtdd 11.4 * li II l.ll MI MI MI t.ll l.tl MI MI stianta ii.i it II 1.14 MI Mi l.tl MI MI l.tl MI suemtao IM

11 MI 1.11 MI t.ll

ridi l u i 11,41: Ih Ulti f i u t i t i 144.ili pii

Unt ili Su il 1 11I0I 14.110 Sflirt triir 4.tll f ilili ( u t i l l l ) lui? IO

11 II 11 MI I I. 11.11 MI III IMI 1.14 1.41 MI 1.11 11. 11.11 sti li st a usta tu it IO i/i 14 li 11 in i III 4.11 11. tl 1.11 1.11 t.ll 1.11 MI MI MI 1.14 USUO) smistiti) t.t 11.4 IO i/i 11 IMI 11.14 MI MI 11 MI MI t.ll *M1 li
mi

11 MI l.ll MI MI

fidi lui * 11,111 Ih fi lli f u t u r i 111.111 pii

Unt ili Sii it 1 EU I U lti O l i u ti trnr 4.1 Filili Cintili) tilt? IO

11 1.11 MI MI MI

Plltl l u i 11,111 1 Filli Fttiiirt 141.141 rii

Unt i li Su ! 1 UIOI > M l l Sturi trnr l.ll! Filli! Ciimlt} fui? IO

u su o
11.1 IO

4.11 1.11 t.ll insta t.t i/i 14 1.11 MI MI MI

11 MI MI MI MI

Filli Ini * 11.110 Di Filli Futuri * 1.111 rii Unt i ti Su it 1 UIOI l.lll Sfuri trnr Mll Fallii Cmtiil) lui? IO

11 11.11 11.11 MI MI

11 IMI 11.11 1.11 1.11

li 1.11 MI MI Mt M

II l. ll MI MI t.ll

r
MI MI MI Mt

Pilli l u i 11,101 111 fi l li Futuri 114.111 fi! U nt i l i Su Ialiti (tur

siuiccdi ) in.! ns

linai) ii.i li

smistai) mainai
t .t i/i

1 Q10!

Mll l.lll

Filili Cimili) Int? IO

Figure 15.

Detailed Bowl by Bowl Listing.

C-34

7.

CUSTOMIZING FIXED D E S I G N S The first option 1n the Modulus Backcalculation menu, "Use an Existing Fixed allows the user to access 24 different design types that ire

Design*,

characteristic of Texas.

These are divided Into three groups, fixed designs one

through eight, nine through sixteen, and seventeen through twenty-four that differ from each other 1n the depth to bedrock and type of subgrade assumed. Designs one through eight assume a SAND subgrade and depth to bedrock of 120 Inches, designs nine through sixteen are for a sandy/clay with a 180 Inch depth to bedrock and designs seventeen through twenty-four assume a CLAY subgrade and a 240 Inch depth to bedrock. The process for creating fixed designs 1s straight forward and Includes three steps:

Running the "Full Design" option from the Modulus Backcalculation Menu using the parameters for the new fixed design;

Renaming the TMP.DEF and WES.RES files produced In the previous step to TMPI.DEF and WES#.RES respectively, where I stands for the number of the existing fixed design being replaced; and

Modifying the DESIGNS.DAT file which stores the fixed design enu definitions to reflect the change. The above procedure has to be duplicated every time a new fixed design 1s

to be Incorporated Into the system. To replace a new design 1t Is recommended that the new design be new for all three depths to bedrocks (240", 180", and 120") to Include the clay, sandy/day and sand subgrade types. For example

TMPI.DEF and WES1.RES (120"), TMP9.DEF and WES9.RES(180"), and TMP17.DEF and VES17.RES (240") contain Information for pavement type 1 of the fixed design option. Replace all of these files with the output from the modulus backcalculation and change the appropriate entry 1n the DESIGNS.DAT file to reflect the layer thicknesses and liyer aterlal types of the new fixed design you have created. Note: The system should be backed up regularly.

APPENDIX D MODCOMP User's Guide

DIRECTIONS FOR BATCH PROCESSING Of KOOCOMP3 DATA FILES..

1.

These b a tc h f i l e s a its t be r i n tn d e r DOS 3 .5 or hig her.

2 . For s i n p l i c i t y . th e MOOCONP3.EXE pro cessor should be in the sane s u b d i r e c t o r y with th e K001.BAT aixi th e K002.BAT batch f i l e s and the input d a ta file s. I t i s p o s s i b l e w ith DOS t o have th e processor and batch f i l e s in o th e r s u b d i r e c t o r i e s , s e p a r a t e froai th e d a t a f i l e s , but th e PATH w ill need to be s e t 143 t o f i n d them, o r th e b a tc h f i l e s w ill r e q u i re o d i f i c a t i o n to include the p a th . 3 . All d a t a f i l e s w ith t h e e x te n s io n .DAT in th e su bd ire ctory will be s e q u e n t i a l l y p ro c e s s e d through N00C0HP3. Each f i l e should have been cre ate d c o r r e c t l y , and no o t h e r f i l e s with th e .DAT f i l e e xtension shoutd be present i n th e s u b d i r e c t o r y . I f t h e r e a r e in c o r r e c t d a ta f i l e s in the subdirectory, HOOCOHP3 w ill iw re th a n l i k e l y be h a l t e d due to a read e r r o r . 4 . There sh o u ld be a d e q u a te space on th e disk (o r hard d riv e ) to accept the o u tp u t f i l e s . Two o u tp u t f i l e s w i l l be c re a t e d fo r each .DAT input f i l e . In a d d i t i o n a tem porary in p u t f i l e i s c r e a t e d , which is erased a f t e r the p ro c e s s in g i s co m pleted. I f th e d is k does not have enough space, NOOCCM P3 w ill more th a n l i k e l y be h a l t e d due t o a DOS w rite e r r o r . 5 . If t h e r e a r e any f i l e s on th e svixJirectory which have no f i l e extension th e y w i l l be t e m p o r a r i l y re r a n e d w ith a .THP f i l e ex tensio n. The su bd irec tory w i l l be scanned and a warning p o s te d on th e screen before t h i s is done. A f te r th e .DAT d a t a f i l e s have a l l been processed, the .TMP f i l e extension w i l l be removed. 6 . To s t a r t th e b a tc h p r o c e s s o r p ly ty pe M001 and a l l of the .DAT d ata f i l e s in t h e s u b d i r e c t o r y w i l l be ri*i through M00C0MP3. The output wilt have th e same filen am e a s t h e ir^xjt d a t a f i l e . The r e g u la r output f i l e w ill have th e f i l e e x t e n s i o n A ST, and th e N00C0KP3.0UT surmary f i l e w ill have the f i l e e x te n s io n .OUT. 7 . K001.BAT c a l l s M002.BAT, so I t i s n e cess ary t h a t both a r e present in the sifcdi r e c t o r y . S. The b a tc h p r o c e s s in g c a n be i n t e r r i p t e d by typing CTRl-BREAK or CTRL-C. The p ro c e s s in g w i l l c o n t i n u e i n t i l t h e next read or w rite operation is e n co un te re d, which may t a k e i n u t e or s o . Some of th e temporary f i l e s needed f o r batch p r o c e s s i n g w i l l p ro b a b ly be l e f t in th e su b d ire c to r y . These can be c le a n e d out aianua lly u s i n g DOS, o r a u to m a ti c a ll y by typing H003.

Background .................................................... Whats N e w ? ................................................. The Strategy of Back-Calculation................ A Limitation in Back-Calculation................ Method of A p p ro a ch ................................. .. Deflection Assignments to Unknown Layers Bedrock and Other Sources of Variability . Tolerances .................................................. Deflection Fit Tolerance................ Modulus Precision Tolerance Number of Ite ra tio n s................................. Assessing the Accuracy of the Results . . . Input F o rm a t............................................... Layer Types and M o d e ls .............. Material Density ............................ Lateral Earth Pressure Ratio . . . . Output Format O ptions.............................. Acknowledgem ents.................................... Disclaimer ..................................................

C opyright....................................................

Appendix 1 - Program Test Data

Background

The computer program MODCOMP 3 was developed to interpret the moduli of elasticity of pavement layers from surface deflection data. This process is commonly called "back-calculation". It involves an iterative approach in which the layer moduli are systematically varied until the desired fit of the surface deflection data is achieved. MODCOMP 3 is a successor to two previous versions. MODCOMP I was originally released in December 1981. It treated all pavement layers as being linearly elastic (e.g., E = constant). MODCOMP 2 was first released as Version 2.1 in November 1983, with numerous minor improvements added through Version 2.44 in August 1987. It added the capability to assign known, nonlinear constitutive models to the pavement layers. MODCOMP 3 incorporates all of the features of versions 1 and 2, plus it adds the capability to determine the constants in nonlinear constitutive models. A test version 3.1 was first released in March 1991. MODCOMP I and MODCOMP 2 were developed by Lynne H. Irwin and Daniel P. T. Speck at Cornell University under the sponsorship of the U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory. MODCOMP 3 was developed by Lynne Irwin and Thomas Szebenyi at Cornell University. MODCOMP 3 utilizes the Chevron elastic layer computer code for determining the stresses, strains and deformations in the pavement system. MODCOMP is written in the FORTRAN language. The first two versions of MODCOMP were written to run on mainframe computers. MODCOMP 2 was subsequently ported to the IBM-PC. MODCOMP 3 has been developed specifically to run on desktop computers. The Chevron code has been extensively modified to improve its accuracy.

W hats New? MODCOMP 3 offers many improvements over its predecessors. Data can be processed for up to ten (10) deflection sensors at up to six (6) load levels. The pavement system can have up to twelve (12) layers, although it is recommended that no more than five or six of the layers have unknown moduli (eg., moduli that are to be back-calculated by the program). The remaining layers should be known, which means that either their moduli or the constants in their constitutive models must be specified.- This admonishment can be violated, if necessary, but as the number of unknown layers is increased beyond five, it is possible that the resultant solutions may not be unique, eg., the specified fit might be accomplished by more than one set of layer moduli. It is also possible that with too many unknown layers the program may fail to come to closure on a specific set of moduli. Having
1

too few unknown layers may also pose a difficulty for obtaining a good deflection basin fit. There will be more discussion on this subject later in this manual. Through the use of improved numerical techniques MODCOMP 3 comes to a solution more quickly than its predecessors. It utilizes the goodness of fit data at the end of each iteration as a means of adjusting the moduli that will be used to seed the next iteration. This accelerates the closure on a solution. The user can specify the precision of fit of the deflection basin as either "low," "medium," or "high." The interpretation of this, in terms of the allowable percent difference between measured and calculated deflections, is somewhat different for systems composed of all linear layers, versus systems with one or more nonlinear layers. The user can also specify the minimum percent change in layer moduli (from one iteration to the next). Backcalculation will cease when either the basin fit tolerance or the modulus precision tolerance, or both, are satisfied. It will also cease when the user-specified number of iterations is exhausted. Improvements have also been made in the Chevron elastic layer system code which is utilized by MODCOMP 3. The original CHEVLAY code incorporated four-part quadrature in its integration procedure. This was adequately precise for layered systems having relatively low modular ratios. However, when there is a thick, stiff (e.g., high modulus) layer present, it is desirable to use eight-part and even sixteen-part quadrature on the first few intervals. MODCOMP 3 contains the improved CHEVLAY 2 computer code, which has the more sophisticated quadrature procedure. Finally, and perhaps most importantly, MODCOMP 3 has the capacity to treat any combination of pavement layers as being either linearly elastic or nonlinearly stress dependent. The program contains a total of eight models, including three new models that were not available in MODCOMP 2. Some of the nonlinear layers use a log-log constitutive model of the form E = k, S kj while other models use a newer semilogarithmic form E = k, exp(S * kj) (2) (1)

where E is the modulus of elasticity, S is a user-selected stress-strain parameter, and k, and k2 are constants that are determined using regression methods. The constants k, and k2 for the models may be either known or unknown. If they are known, they are input by the user. If they are unknown, they are back-calculated by the program if deflection data for at least three load levels are provided. For best results, it is recommended that deflection data for four or more load levels be utilized when the unknown nonlinear feature is used. Where k, and k2 are known, these parameters would usually be determined using laboratory repeatedload triaxial testing.
2

The constitutive models, particularly the ones involving a semilogarithmic form, may represent a significant improvement in the state-of-the-art of back-calculation. Until now nearly all of the constitutive models have taken the log-log form of equation 1. Depending upon whether the exponent k2 was positive or negative, the log-log models suggest that the modulus E goes either to zero or infinity as the stress parameter S approaches zero. The log-log models also imply that the stress parameter cannot be negative, as the exponential would be undefined. In reality, both of these aspects of the log-log models can be incorrect. For instance, as the stress approaches zero, the modulus approaches the initial tangent value. And at shallow depths in a pavement system, near the load the stresses do become tensile (ie., "negative"), and thus the bulk stress and some other stress parameters can be negative, which results in the modulus being undefined in the log-log models. The semilogarithmic form of equation 2 remedies both of these problems. When the stress is zero, then E = k,, and thus k! is the initial tangent modulus. Furthermore, the stress parameter can be negative and the modulus is still defined.

The Strategy of Back-Calculalion The surface deflection data (Figure 1) will ordinarily come from dynamic, nondestructive testing where loads are applied on the pavement surface (Figure 2) and vertical deflections are measured by electronic transducers (geophones or seismometers). MODCOMP 3 is designed to accept the data that is typical of devices such as the falling weight deflectometer, the Road Rater, and the Dynaflect. Because Benkelman beam test and plate-load test loadings ordinarily involve plastic creep as well as elastic deformation, they are not recommended as a source of surface deflection data, since the analytical model presumes that the measured deflections are purely elastic without any creep. Pavement surface deflections are relatively insensitive to minor variations in pavement layer moduli. However, this means that in back-calculation, pavement layer moduli will be highly sensitive to minor variations (or errors) in measured surface deflections. Deflation errors on the order of only 1 or 2 microns (0.04 to 0.08 mils) will have a dramatic effect on the resultant moduli (reference ASTM paper). Therefore, for the best results with backcalculation, it is vitally important to use data from an NDT device that has been calibratedIn the United States four regional centers have been established for the purpose of calibration of FWDs. These are located in Pennsylvania, Texas, Nevada and Minnesota, and they are administered by the respective state Departments of Transportation. In addition, a commercial calibration center is administered by Dynatest, Inc. in Starke, Florida. In order to be able to determine the unknown layer moduli, surface deflections are associated with the pavement layers. MODCOMP 3 can use its own logic to make the assignments, or the assignments can be made directly by the user. In the latter case the deflections further from the load should be associated with the deeper unknown layers. There is a limit, however, regarding how deep a layer with an unknown modulus can be
3

Distonce fro m lood

D e fle c tio n

I
F igure 1. D e f l e c t i o n B a s i n S howi ng Me a su re d D e f l e c t i o n s .

deflection sensors

loading plates
1 L *
q

*
^ \

r, * 0

P l a n V i e w - D y n a f l e c t and Road R a t e r L o a d i n g

P rofile

- S u r f a c e M e a s u r e m e n t s and P a v e m e n t L a y e r s

Figure

2.

S c h e m a t i c Geometry o f D e f l e c t i o n M e a s u r e m e n t s .

positioned. As a general rule-of-thumb, MODCOMP 3 will be able to determine the moduli for layers which lie at a depth that is no more than the radial distance from the center of load to the outermost deflection sensor. This rule can be exceeded to some degree, but if the layer is so deep that it has little or no influence on the outermost deflection, then it wi not be possible to accurately determine the layers modulus. Layers deeper than 20 to 30 feet (7 to 10 meters) generally fall into this category. The modulus of layers below that depth should be assigned as "known." In addition to the need to know the deflections and the loading conditions, it is also necessary to know the pavement layer thicknesses and a value of Poisson s ratio for each layer. Although the latter typically will not be known precisely, the calculations are not highly sensitive to small variations in Poissons ratio for values less than 0.40. Thus it is possible to make informed assumptions of Poissons ratio, based up>on a knowledge of the material typ>e, plasticity, and moisture content. An accurate knowledge of the layer thicknesses is important, as the computational results are sensitive to their values. The layer thicknesses should be known to a degree o precision of five percent or better (e.g., to the nearest quarter-inch in five inches, etc.). Several pavement corings should be made to determine the surface course, base, and su ase thicknesses. The recovered materials can be taken to the laboratory for classification testing, which may be helpful for estimating moduli. Asphalt concrete cores and other cores of cemented materials can easily be tested in the laboratory using a repeated-load splitting tensile test to determine their moduli, thus reducing the number of unknown layers in the back-calculation. Auger borings should be made to a depth of ten to twenty-five feet when it is susp>ected that bedrock or a stiff layer may underlie the surface materials. Alternatively, seismic refraction techniques may be used to detect moderately deep, stiff layers. When the water table exists at a shallow depth (e.g., less than 25 feet), it may be useful to knov. its depth also. Such layers may not influence the measured surface deflections to a large degree, and yet failure to account for their presence when the pavement is configured in the computer model as a system of elastic layers may lead to computational errors and to incorrect backcalculated moduli. The more you know about what lies beneath the pavement surface, the more likely you are to come up with a correct back-calculated set of layer moduli.

A Limitation in Back-Calculation It is important for the program user to understand that there are certain features of the pavement that affect the surface deflections and others that have little influence. In general, back-calcularion can only determine the moduli of pavement layers thai significantly influence the surface deflections. As an extreme example of this principle, suppose that in the pavement profile shown in 6

Figure 2c there is a thin layer of aluminum foil at the interface between layer 2 and layer 3. Although aluminum has a very high modulus of elasticity (in comparison to the adjacent soil moduli), it would jtqI be possible to accurately back-calculate the modulus of the foil layer if it were to be treated as a separate, unknown layer. The simple reason is because the foil is too thin to have any influence on the measured surface deflections. Therefore, it is a basic principle that if a pavement feature has little or no effect on the deflections then it will not be possible to use measured deflections to gain information about the pavement feature. Thus it is difficult to back-calculate the moduli of very thin surface layers. A bituminous surface treatment on top of a granular base would fall into this category. Asphalt concrete surfaces thinner than 2 to 3 inches (5 to 7 cm.) also pose problems. Other "problem" layers include: thin asphalt overlays over portland cement concrete - the concrete dominates the deflections and thus it is difficult to back-calculate the modulus of the overlay. thick asphalt overlays over portland cement concrete - the asphalt dominates the deflections and thus it is difficult to back-calculate the modulus of the concrete. asphalt surfaces composed of several layers of different ages or compositions - these will usually have to be treated as a single layer. thin subbase layers or lenses in the subgrade - often these will have to be combined with adjacent layers. The term thin" has a different meaning the deeper you are in the pavement system. At the depth of the upper subgrade a thin layer might be several feet thick. a lens of frozen material under thawed, soft layers - if it is "thin" it may not have any effect on the deflections, and if it is "thick" it may have the same influence as bedrock. black (i.e., asphalt concrete) bases under portland cement concrete or thick, stiff asphalt concrete surfaces - sometimes in the latter instance the base and surface layers can be treated as a single layer.

Usually if a "problem" layer exists, it will show up in back-calculation as having either a very high or a very low modulus. The program has both a maximum and a minimum value that it will allow. When one of the limits is invoked MODCOMP 3 will print a warning message. These limits are well beyond what would be normal values of moduli for pavement materials. If these limits are hit too many times, the program will stop execution. The sensitivity of a given deflection to changes in the modulus of the layer to which it is assigned is shown in the MODCOMP 3 output. The sensitivity is the inverse of the slope of the log modulus - log deflection relationship. Its magnitude is not influenced by the system of units, metric or English, that is being used. As the sensitivity approaches zero this 7

is an indication that the layer modulus has very little influence on the deflection that is being used to determine the modulus. If the layer is too insensitive it may help to designate the problem layer as "known", assigning it a reasonable modulus value. If the modulus of the layer cannot be determined because it has little effect on the deflections, then assigning a modulus to the layer will generally have only a small effect on the overall accuracy of the solution. Another strategy is to "ignore" the layer by merging its thickness in with an adjacent layer. Very little information is lost when the adjacent layers have similar moduli, such as an asphalt surface joined with a black base, or a dirty gravel base joined with a subgrade. In such cases the modulus that is back-calculated represents a "composite" value for the layer. Sometimes it will help to assign a different deflection sensor to the layer. The center deflection (at R = 0) is sensitive to the moduli of all of the layers, while the outer deflections are only sensitive to the moduli of the deeper layers. If the layer is too deep, none of the deflections are very sensitive to its modulus.

Method of Approach There is no closed-form solution for determining layer moduli from surface deflection data. Thus an iterative approach is used in the computations. The basic principle is to start off with a user-supplied set of "seed" moduli from which surface deflections are computed using the Chevron code. The computed deflections are compared to the measured deflections and the seed moduli are adjusted as a function of the magnitude of the difference in deflections. Then a modulus for the layer is interpolated which agrees with the measured deflection (Figure 3). This process is repeated for each layer until the agreement between the calculated and the measured deflections is within the specified tolerance (i.e., "low," "medium," or "high* accuracy), or until the allowed number of iterations has been exhausted. If the rate of change of the layer moduli falls below the user-specified rate, then the iterative calculations will cease. This outcome may not lead to a true "solution" of the problem, however. The deflection basin fit information should be checked to verify whether or not an acceptable solution has been obtained. If the fit is poor, then the rate of change of moduli (e.g., the modulus precision tolerance) may have been satisfied because one or more of the deflections were too insensitive to the layer moduli. The modulus for each layer can be either "known" or "unknown". If it is unknown, it will be determined by back-calculation. If it is known, then the information provided by the user will be held fixed. If the layer model is linearly elastic (eg., E = constant), the sod modulus is held fixed. If the layer model is nonlinear (ie., given by equation 1 or 2), then the k, and k2 parameters are held fixed. Where a nonlinear model has been specified for a
8

Modulus (log scale)

Deflection (log scale)

F ig u re

3.

I n t e r p o l a t i o n o f Modulus U s i n g C a l c u l a t e d a n d M e a s u r e d D eflections.

know'll layer, a subroutine in the program utilizes the NELAPAV (Nonlinear ELastic Layer Analysis for PAVements) code to determine a stress-compatible modulus for the layer. Where unknown, nonlinear models are to be determined, the program evaluates a modulus for the layer for all of the load levels. A preprocessor fine tunes the seed moduli for up to six iterations by treating the layers as if they were linearly elastic. Thereafter, in the MAIN portion of the program, the coefficients for the nonlinear models are determined. After two iterations in MAIN, a second deflection, usually located at an inner radius, is included. The moduli and the associated stresses at the mid-depth of the layer are passed to a subroutine which performs a regression analysis to determine the kj and k2 parameters. A hypothesis test is performed to assure that the nonlinear model is significant. If the model is significant it is used for the remainder of the calculations in the iteration. If the model is not found to be significant the layer is treated as being linearly elastic for the rest of the iteration. In the latter case, if the range of the moduli calculated for the layer is sufficiently narrow, then an average E value is used. The program first evaluates the modulus of the deepest unknown layer, and then it works upward to the surface layer. Measured deflections at larger radii are assigned to the deeper layers. Thus as the program works upward from the deeper layers, it also works inward toward the center of the deflection basin. On each subsequent iteration the moduli of the upper layers are known with increasing precision, and therefore the solution for the system of unknown layers converges steadU) toward the shape of the measured deflection basin. This progression will generally proceed regardless of the initial magnitude of the seed moduli, however, fewer iterations will be required if the seed moduli are close to the final solution. The user-specified tolerance regarding modulus precision should be set small enough (normally 2 percent or less), and the deflection basin fit criterion should normally be set to "high." If this is done, and if the root-mean-square (RMS) error of the basin fit is small (typically less than one percent), then the solution is probably both precise and accurate.

D eflection A ssignm ents to Unknown Layers

In order to back-calculate the modulus of any unknown layer, a surface deflection must be identified which can be used for adjusting the layer modulus until the calculated surface deflection closely matches the observed value. MODCOMP 3 provides three ways the user can assign deflections to the unknown layers. 1. The user can utilize a special routine in MODCOMP 3 which uses default criteria to make the assignments of measured deflections. The user can directly specify the assignments by indicating which deflection number is associated with a given layer.
10

2.

3.

The user can utilize a curve fitting routine which combines with the default criteria to make interpolations (and extrapolations) of the surface deflections.

The best initial strategy for most problems would be to use method #1, allowing the computer to make the assignments. Thereafter, if the solution seems to warrant it due to marginal basin fit, the user can use method f/2 and manually adjust the sensor assignments. The interpolation scheme would seldom be used unless there was an unusual juxtaposition of sensor spacings and layer thicknesses. The surface layer is considered to be Layer 1, with the layer numbers increasing downward. The deflection sensor nearest the center of load is Deflection 1, with the numbers increasing radially. The program does not presume that Deflection 1 is at the center of load, and the radial distance from the center of load to each sensor must be entered. Thus the program can work with data from the Dynaflect and the Road Rater, as well as the falling weight deflectometer (FWD). It is important to have an understanding of how the deflection is distributed beneath the pavement surface in order to be able to make reasonable deflection assignments. Figure 4 shows that the depth beneath which 95 percent of the surface deflection occurs gradually moves downward at increasing radial distance from the loading plate. The actual shape and position of this line is a function of the moduli and thicknesses of the pavement layers. Most of the registered surface deflection is attributable to compression that occurs in the layers that are below the line. Only a very small proportion (viz., 5 percent) of the surface deflection occurs in the materials above the line. Thus at a large distance from the load, most of the deflection that is measured at the surface is due to compression that occurs in Layer 4 and below. If a deflection sensor at this radius were assigned for determining the modulus of Layer 1, the result would be very inaccurate. This is because the modulus of Layer 1 has very little influence on the deflection at large radii, the sensitivity parameter would be very small, and hence this deflection cannot be used to determine the surface layer modulus. Therefore the outermost deflections would usually be assigned to the deepest unknown layers. The center deflection would be assigned to the uppermost unknown layer, since it is the deflection that has the greatest sensitivity to the modulus of the surface layer. The user can use these criteria when making deflection assignments. These criteria also form the basis for the default assignment procedure that is offered by MODCOMP 3. While the actual location of the 95 percent line is unknown for a particular problem, it is presumed in MODCOMP 3 that the line can be approximated by a 34-degree line (Fig. 4). It has been determined that in some cases a line flatter than 34 degrees might be preferred, while in other cases a steeper line would be better. Since the 95 percent deflection criterion is arbitrary (some might prefer to use 96 percent, or 94 percent, etc.) the arbitrariness of the 34-degree line matters little. If the line of approximation is reasonable, convergence will ultimately be obtained through iteration.
11

L oad in g plate
V77//J/ Z 7 L {77
V

\3 4

Loyer t ^ - 9 5 % of surfoce S ' deflection occurs \ below this line

Layer 2

XV
\

Loyer 3

Loyer 4

F igure

4.

P e v e p e n t S y s t e a Sh ov in g L i n e o f 95 P e r c e n t D e f l e c t i o n .

12

If the user specifies that the default assignment procedure be used, then the deflections are assigned to given layers from the set of input data using the 34-degree line. The deflection that falls closest to the intersection between the upper layer interface and the 34-degree line will generally be used. In a few instances where the layers are tfiin or the measured deflections are spread far apart (and hence they are assigned to an adjacent layer), an interpolation scheme is automatically invoked to obtain a deflection. The list of deflection assignments on the output will note if interpolation or extrapolation has been used. Based upon a special third-degree spline, the interpolation method has been found to be fairly accurate outside the radius of the load plate. Unfortunately it does a very poor job of extrapolation. The program issues a warning whenever the spline subroutine must be used to extrapolate the basin beyond the outermost measured deflection. The modulus of an unknown layer obtained using an extrapolated deflection is likely to be very questionable. On some occasions the user may wish to specify that the curve fitting spline be used. This might be used, for instance, when it is desired to make the deflection assignments in the input file, but it is not clear which deflections should go where. The ideal sensor positions could be identified during an initial run (perhaps specifying zero iterations), and then the user could pick the "best choice" measured deflections. Regardless of whether the deflections are assigned by the user, assigned by the program, or interpolated by the program, MODCOMP 3 provides a list of the deflection assignments, indicating which deflection is associated with each layer. It is also noted whether the deflection was assigned by the user or by the computer. Generally speaking, the greatest allowable depth to the upper interface of an unknown layer cannot be much more than the radial distance from the center of load to the outermost deflection sensor. Thus if the outer sensor is at a radius of 6 feet (2 meters), the deepest unknown layer should be approximately 6 feet deep. This is a genera^ rule of thumb, an it may of course be too conservative for some situations. However, as the depth to the deepest unknown layer increases, the sensitivity of the outermost deflection to the modulus o the layer gradually diminishes. Thus in MODCOMPs iterations, the modulus of the layer may head toward the maximum allowable value (MAXMOD) or the minimum allowable value (MINMOD). If this happens, the modulus of the deep layer will probably have to be treated as a "known" layer.

Bedrock and O ther Sources of Variability One of the major difficulties in back-calculation involves dealing with the unseen differences between the way the pavement profile is modeled in the computer, and the way it really exists in the field. Using elastic layer theory, the bottom layer is assumed to be semi infinite in depth, with a constant elastic modulus. However, in a real pavement, bedrock can exist at any depth, which may lead to an unaccounted for high-modulus layer. When bedrock occurs at a depth of less than 50 feet, its presence will need to be considered, or the
13

back-calculated moduli may be greatly in error (reference Yangs thesis). Probably the worst situation occurs when bedrock is very shallow, but it is not included at all in coding the problem into the computer. Many times this will result in exhausting the iterations without coming to a solution, or an implausible solution may be obtained. If bedrock is known to exist, it can be handled as a known" layer and assigned an arbitrarily high modulus value. The magnitude of the modulus matters much less than the depth. Typically a modulus of 3500 MPa (500,000 psi) is assigned to a bedrock layer. Where shallow bedrock is not a problem, a more subtle difficulty may arise due to a gradient in the subgrade. Since the overburden pressure increases with depth, it is reasonable to expect that the modulus of the subgrade will also increase with depth. However, elastic theory presupposes that the modulus is constant with depth. To account in part for this vertical variability, one can create an "artificial layer", 30 to 60 inches (0.7 to 1.5 meters) deep at the top of the subgTade. Doing this offers two advantages. First it results in a more correct subgrade modulus immediately beneath the base course, which yields a more accurate base course modulus. Second, it involves an additional, outer deflection in the calculations, which usually results in a lower RMS error. Other features, such as a shallow depth to the water table, and nonlinear stressdependent properties of the subgrade, may also have an effect on the subgrade modulus. However, the use of a nonlinear model for the subgrade layer should not be substituted for accurate modelling of the actual layering of the pavement system. The nonlinear model can properly account for the horizontal change in the layer modulus. But it should not be used to account for the vertical change in modulus, whether the change is a gradient due to increasing overburden pressure, or abrupt, due to bedrock, water, materials, etc. The latter features should be handled by introducing additional known or unknown layers in the system.

T olerances

Tolerances are specified in back-calculation as a way of indicating when the solution (eg., the set of layer moduli) is "close enough." In MODCOMP 3 two tolerances are designated: a deflection fit tolerance and a modulus precision tolerance. The deflection fit is the percent difference between the calculated and measured deflection at a particular sensor location. The modulus precision is the percent difference in the modulus of a given layer from one MODCOMP iteration to the next. By using deflection basins as input data to MODCOMP 3 which were calculated from hypothetical problems run through the revised Chevron elastic layer system program, it has been possible to assess the accuracy of MODCOMP 3 and the tolerance level that must be required to obtain an accurate solution. In this way it has been determined that a very tight deflection fit tolerance must be specified in order to get accurate results. Deflection tolerances on the order of 0.02 percent or less were found to be necessary to get the same moduli from back-calculation that were used in the hypothetical problems.
14

As a practical matter, it is not possible to solve real, field-generated deflection basins to the same degree of precision that can be achieved with hypothetical problems. The reasons for this are numerous. In the field situation there is always an amount of uncertainty regarding the pavement and subgrade layering. The presence of bedrock at an unknown depth, and/or the horizontal variability of layer thicknesses and materials properties combine to make "real" data more difficult for back-calculation. It is vitally important that the FWD or other deflection measuring device be calibrated, so that the data does not include systematic errors in load or deflections. The effect of B fldgm measurement errors can be reduced by making multiple drops with the FWD at a single point, and then computing the average deflections. The standard error of the mean decreases in proportion to the square root of the number of observations in the average. Thus the standard error of the average of four readings is half as big as the standard error of a single reading. Averaging multiple readings does not reduce the systematic error, however, so the device must be calibrated to assure the most accurate moduli will come from back-calculation. The "practical realities" listed above have been considered jointly in arriving at the following recommendations regarding the choice of tolerances for MODCOMP 3.

Deflection Fit Toleranp-p MODCOMP 3 provides the user with three levels of deflection fit tolerance: low, medium, and high. The actual percent tolerance that is set differs depending upon whether the system contains any unknown, nonlinear layers. Deflection Fit Tolerance Lcvgl (L)ow (M)edium (H)igh Tolerances in percent All Linear One or More Nonlinear Layers Layers 0.50 0.30 0.15 1.00 0.60 0.30

The user-assigned tolerance level is noted on the first page of the MODCOMP 3 output. There is a tradeoff between the tolerance level that is chosen and the number of iterations that will be required. Life is seldom so rushed that it would not be possible to specify the "high" level of fit, and thus hopefully assure the best, most accurate backcalculated moduli. However, there are some problems that simply will not be able to satisfy the "high" tolerance, even when they are allowed a very large number of iterations. This is 15

usually an indication that the pavement layering has not been correctly modeled. Some feature, such as a hard or a soft layer, may be present in the field that is not being accounted for in MODCOMP. Sometimes, by trial and error, a plausible solution can be found by adjusting the number of layers or their thicknesses. The solution may be plausible, but it is not necessarily right! When the high level of deflection fit tolerance is specified for a pavement system made up of all linear layers (E = constant), a tolerance of 0.15 percent will be used. For a center deflection on a basin that is 500 to 1000 microns (20 to 40 mils) or so, this would require an agreement between the calculated and measured deflection of about 0 .7 5 to 1 .5 microns ( 0.03 to 0.06 mils). In the outer reaches of the basin, where the deflections might be less than 50 microns (2 mils), the agreement would have to be 0.075 microns (0.003 mils). Statistically-based studies have shown the accuracy of the falling weight deflectometer deflections is about 2 microns (0.08 mils). While a calculation tolerance of 0.15 percent may seem to be somewhat stringent, it is required to avoid compounding measurement uncertainties with calculation uncertainties. Such a small tolerance is also required in order to assure that the back-calculated moduli are independent of the seed moduli. A general check on the accuracy of an) backcalculation program can be made by changing the seed moduli. If a substantially different set of back-calculated moduli occur when the seed moduli are changed, then something is wrong, and the accuracy of the solution is doubtful. When the specified tolerance has been satisfied for each of the deflections thoi are assigned to unknown layers, then the iterative calculations will stop. The program provides a message that this is the reason why the calculations terminated.

Modulus Precision Tolerance MODCOMP 3 also allows the user to specify a tolerance on the minimum rate of change of layer moduli from one iteration to the next. This provides some control on the precision (not the accuracy) of the moduli. The iterations will not go on and on if the moduli are changing less than the tolerance, even if the deflection fit tolerance is not yet satisfied. The user may wish to start with a precision tolerance on the order of 1.5 percent. The tolerance may need to be made larger or smaller, depending on the particular circumstances of the problem that is being solved. The calculations will stop if the moduli of all unknown layers change from one iteration to the next by an amount less than the specified tolerance. Again, the program provides a message if this is the reason why the calculations terminated. If both the deflection fit tolerance and the modulus precision tolerance are satisfied
16

simultaneously, a message to this effect will be given. In order to have the deflection fit be the only criterion that is used, one can simply set the modulus precision tolerance close to zero.

Number of Iterations Commonly, as the number of unknown layers increases and as the departure of the seed moduli from the final solution increases, the number of iterations to amve at a solution also increases. Simple problems with all linear layers and two or three unknowns will usually achieve tolerance within five to ten iterations. More complex problems with four or five unknowns, perhaps with one or two of them nonlinear, may require ten or more iterations. Normally a solution will be reached in no more than twenty iterations. Information is displayed on the computer monitor to indicate the start of each iteration. Near the end of the iteration the elapsed time and the RMS (root mean square) error is displayed. If the program is heading toward a solution, the RMS error will usual!) decrease with each iteration, although the rate of change typically slows with each iteration, and sometimes the RMS error even increases slightly during the last several iterations. The user can watch the progress on the monitor, and if it appears that MODCOMP 3 is moving away from a solution (because the RMS error is steadily increasing) the execution of the program can be halted by pressing the CTRL and the BREAX keys down at the same time. [It may require a period of time before this has the desired effect, as an output statement must be encountered before the halt occurs.] MODCOMP 3 will allow you to specify zero iterations. This can be helpful as a quick way to check on the deflection assignments and the reasonableness of the seed moduli. No modulus adjustment computations are made, but the tolerance check for each sensor will provide an indication of whether or not the seed moduli are in the ballpark. The user can compare the calculated and measured deflections, and the spatial distribution of deflection assignments, and adjust accordingly. This is not an essential step, since the end solution is more or less independent of the starting seed moduli. However, this refinement will often reduce the required number of iterations, and hence it will reduce the computation time. On a slow computer, the effort may be worthwhile.

Assessing the Accuracy of the Results One of the most difficult things to do in back-calculation is to determine whether or not the results (ie., the moduli) are believable. To do this requires both a good deal of experience in the use of the back-calculation computer program (to recognize the signs of whether or not the results might be valid), and as much information as possible about the materials in the pavement layers (to know what modulus values to expect). If you have finly the deflection test data, combined with imperfect knowledge about the layer thicknesses, little
17

or no information about the subsurface materials properties, and no information about the depth to bedrock and/or the water table, then you probably have no way of assessing whether the back-calculated moduli are accurate or not. Under such circumstances, there is a good chance the moduli will be inaccurate. As explained previously, three levels of deflection fit tolerance are available in the program. By requiring high level of fit the calculated deflections for the assigned sensors will closely match the measured deflections. Thus some assurance is provided that the results will be accurate, and independent of the seed moduli. However, for a pavement with only four unknown layers, only four of seven or more measured deflections are actually being worked with. The data for the unused sensors provides a good indication of the real accuracy of the moduli. This is reflected in the RMS error of the deflection fit. A small RMS error (less than 1 percent) is usually a good indication that the moduli are accurate. The user can specify the modulus precision fit. If this is set too high (perhaps 3 percent or more), the computer may stop iterating too soon. If the modulus tolerance is set too low (perhaps less than 1 percent), it may take a large number of iterations to achieve a solution. At a certain level both the deflection fit tolerance and the modulus precision tolerance may be satisfied on the same iteration. Can you say that you have both a precise and an accurate solution if both the modulus tolerance and the deflection fit tolerance are satisfied on the same iteration? Maybe, but this is not a certainty. The answer depends on the size of the RMS error. If the RMS error steadily goes down with each iteration, and if it is a number less than one percent on the last iteration, this is a good sign that the results are believable. On the other hand, if the RMS error ranges up and down erratically from one iteration to the next, the results should be suspect. There may be an unaccounted for feature of the pavement profile (bedrock perhaps), or there could be one layer whose modulus has little effect on the deflections. In these cases, the RMS error may wander, or fail to close rapidly. However, the iteration with the smallest RMS error does not necessarily have the most accurate moduli. Probably the best way to assess the accuracy of the results is to have a good knowledge of the pavement materials. Asphalt concrete that was very old or very cold at the time of test will have a very high modulus of elasticity. Badly broken asphalt or portland cement concrete should have a modulus near that of crushed stone. Due to the seasonal effects of weather, the upper few feet of a subgrade will often have a lower modulus than the deeper portion. If you know something about the condition and the properties of the pavement materials, you can anticipate what to expect in terms of back-calculated moduli. One aspect o f back-calculation is universally true. The modulus of the deeper layers affects all of the surface deflections about equally, while the modulus of the surface layer mainly affects the deflection nearest to the center of load. Thus the calculations are fairly sensitive to minor variations in the subgrade modulus, while a small variation in the surface
18

course modulus has little or no effect. It is therefore reasonable to expect that the moduli for subgrade materials will be correct to perhaps 3 percent, while the surface course moduli will be correct to about 20 percent. The uncertainty of the intermediate layer moduli will usually fall somewhere between 20 and 3 percent.

In p u t Form at

Requirements for the input file are given in Table 1. The program is set up to run in the batch mode. The user therefore creates a file which is subsequently read as an input file by the program. An example input file is included in Appendix 1. For the most part the program is set up for format-free input. The variable types are specified in Table 2. Real variables generally should have a decimal point, while integgr variables do not permit the use of a decimal point, and alpha(numeric) variables must be entered in single quotation marks (e.g., H). Where more than one entry is required on a given line, the items may be separated by a comma and/or one or more blanks. The user may specify whether the input data is in SI metric or U.S. customary units. The expected units of each input variable are specified in Table 2. Internally the program converts the input data to a consistent set of units (either pounds, inches and p.s.i., or Irilonewtons, meters and kilopascals). To protect against inadvertent output scale errors, the user should carefully adhere to the input dimension requirements. All of the input data is echoed in the printout before unit conversions are made, making it easy and convenient to check for input errors. The program does a modest amount of error checking of the input data, and if no units are specified metric input is the default.

Laver Types and Models

Each layer can be either "known" or "unknown". If it is known, the user must specify either a linear modulus or the kj and k2 parameters for the layer. If it is unknown, the program will determine these values. If the layer is known, specify TYPE = K , and if the layer is unknown, specify TYPE = U \ The program is set up to accommodate eight different constitutive relationships, as outlined in Table 3. A linear layer (E = constant) is denoted by MODEL = 0. All other constitutive relationships are log-log or semilogarithmic models of the form given in equations 1 and 2.

Material Density

For nonlinear, stress-dependent materials, the stress-strain parameter S in equation 1 or


19

Table 1. MODCOMP 3 Input Variable Definitions

Line 1 2

Variable Title UNITS OUTPUT

Comments Descriptive title for problem (80 characters max.) Either METR for SI metric input, or ENGL for U.S. customary units (see Table 2 for further information). Default units are metric. Either BRIE for brief (i.e., only the final iteration results are tabulated), or LONG if the results of each iteration are desired, or ALL if intermediate calculations are also of interest. Default output is brief. (Format: four characters in columns 1-4, followed by a space or a comma, followed by four characters in columns 6-9) Deflection fit tolerance. Enter L for low, M for medium, or H for high. (The use of (H)igh accuracy fit is suggested.) Modulus precision tolerance specifies minimum difference between layer moduli in consecutive iterations, given in percent. (A decimal number on the order of 1.5 percent is suggested.) The maximum allowed number of iterations in the main program loop. (An integer number in the range of 10 to 15 is suggested.) Number of layers in the pavement system, counting the subgrade as one layer, including both the known and the unknown layers (maximum =12). (One line for each layer, I) Enter U if the layer modulus or constitutive model is unknown, or K if the modulus or model is known (i.e., fixed). (see Table 3 for description) The deflection number assigned to this layer, counting the sensor closest to the load as No. 1. If the integer number zero is input, the computer will make the selection of the "best" deflection. If a -1 is input, the computer will interpolate a deflection at the "optimum" radius by fitting a curve through the deflection data. If a positive integer is input, the program will assign this deflection number to the layer. Assigned deflection numbers should be in ascending order for increasing layer numbers. A seed modulus (i.e., a starting point for the computations) for nonlinear, known layers and for all unknown layers. For linear, known layers this modulus will be used for all calculations. Poissons ratio for the layer.

TOL2 TOL3

MAXITN

NS

5a, 5b, 5c, etc. TYPE(I) MODEL(I) DEFNUM

E(I)

V(I)

20

Table 1 contd. MODCOMP 3 Input Variable Definitions


Ling Variable Comments

DEN(I) K0(I) THKNES(I): K1(I):

K2(I):

Unit weight of layer material (can be input as 0.0 if all layers in the system are linear. At-rest lateral earth pressure coefficient (can be input as 0.0 for each linear layer). Layer thickness (should be input as 0.0 for the bottom layer). Coefficient in nonlinear model. (See Table 3 for further information.) (Required only for known, nonlinear layers; can be input as 0.0 for all others.) Exponent in nonlinear model. (See Table 3 for further information.) (Required only for known, nonlinear layers; can be input as 0.0 for all others.) Number of load levels (maximum = 6).

NLOADS:

7a, 7b, etc. (One line for each load level, J) LOAD(J): Load force on plate PRESUR(J): Plate pressure LDRAD(J): Plate radius (Only two of these three parameters must be input. If the third is input as 0.0 it will be computed, assuming a circular plate.) 8 9 NDEF: INPRAD(K): Number of surface deflections (maximum = 10). Radial distance measured from the center of the load plate to each deflection sensor.

10a, 10b, etc. (One line for each load level.) INPDEF(J,K): Surface deflection data.

Except on line 1 and line 2 where an alphanumeric input format is specified, all input is format-free. Each item of format-free input may be separated by a comma or one or more blank spaces. See Table 2 for further information regarding variable types and required units.

21

Table 2. MODCOMP 3 Input Variable Types and Required Units

Variable Name DEN(I) E(I) INPDEF(J,K) INPRAD(K) KO(D K1(D K2(D LOAD(J) LDRAD(D MAXITN MODEL(D NDEF NLOADS NS OUTPUT PRESUR(J) THKNES(T) TITLE TOL2 TOD TYPE(D UNITS V(I)

Parameter Density Elastic Modulus Measured Deflections Deflection Radii Earth Pressure Coefficient Nonlinear Model Coefficient Nonlinear Model Exponent Load Force Plate Radius Iteration Limit Layer Model Number of Deflections Number of Load Levels Number of Layers Output Format Plate Pressure Layer Thickness Problem Title Deflection Fit Tolerance Modulus Precision Tolerance Layer Type Input Units Poissons Ratio

Variable Type Real Real Real Real Real Real Real Real Real Integer Integer Integer Integer Integer Alphanumeric Real Real Alphanumeric Alphanumeric Real Integer Alphanumeric Real

Required Units Metric System U.S. System pounds/cu. ft. pounds/sq. in. mils (lO-3 in.) inches dimensionless pounds/sq. in. dimensionless pounds inches pounds/sq. in. inches percent ENGL dimensionless kilograms/cu. m. megapascal sCMN/m7 ) microns (10^ m.) meters (m) dimensionless megapascals (MN/m1 ) dimensionless kilonewtons (KN) meters kilopaseals (KN/mi) meters percent METR dimensionless

I = layer number; J = load level; K = deflection or sensor number

Table 3. Layer MODEL Descriptions M ODEL DESCRIPTION

Linear irodel of the form: E = constant. In general, most asphalt concrete, portland cement concrete, and treated (i.e., stabilized) soil materials conform to this constitutive relationship.

Bulk stress model of the form: E = ki exp(S * kj), where S = 6 = ct + at + o,

(at = vertical stress, at tangential stress, and a, radial stress at the mid-depth of the layer). The parameter 6 is known as the bulk stress or as the first stress
invariant. Some coarse-grained untreated base and subbase gravels have been found to conform to this relationship.

Deviatoric stress model of the form: E = k, exp(S kj), where S = O - o} i the difference between the major principal stress and the minor principal stress at the mid-depth of the layer. Note that this model differs from the customary linear model (E = k,S + kj). It can be used for fine-grained, untreated materials.

Second Stress Invariant model of the form: E = kj exp(S * kj), where S = V(1 + O The parameter J2 is the second stress invariant, which has units of stress squared. J2 = atot + ota, + atal + rnJ Where oz, at and a, are as defined for Model 1, and ra is the shear stress in the rz plane, is the octahedral shear stress, which is the root mean square deviatoric stress. t, = 1/3[(a, - o f + (a, - a,)7 + (a, - o f + 6 rnI],' This constitutive relationship has been found to be useful for both base course and subgrade materials, and the coefficient k, appears to be primarily a function of material density and soil moisture tension.

23

Table 3 contd. Layer MODEL Descriptions

Octahedral shear stress model of the form: E = k, exp(S * k j , where S = (1 + O This relationship has been found to be useful for fine grained materials and for frozen soils, and the coefficient k, appears to be a function of soil temperature and moisture content.

Vertical stress model of the form: E = k, S 2, where S = at The parameter at is the vertical stress, which is always compressive (ie., positive). This model is very similar in some respects to MODEL 6, and it has a log-log form.

Major principal stress model of the form: E = k, S 2, where S = ax The parameter is the major principal stress, which is generally always compressive. This model is used by Dynatest in the ELMOD back-calculation program. It is included here for completeness, and to permit comparisons between the two programs. As used in ELMOD, the overburden stress is not included in the major principal stress, so the overburden stress is not included here either.

k k Cornell constitutive model of the form: E = k, S 2(kJ 3 , where

S = (eo + 6t2 022(1 *Pm) (1 + o ' 034 P 2 )


The parameter 60 is the initial bulk stress, due only to overburden stress. The parameter 8p is the bulk stress at peak load, including both the overburden stress and the load stress. The parameter is the octahedral shear stress, due only to load stress. The parameter P* is the percentage of the materials gradation that passes the number 200 sieve, and it is a required input for both known and unknown models. It is entered on datafile line 5 as the K2(I) parameter. The parameter k, is the lateral earth pressure coefficient, which represents the anisotropy of the overburden stress. The coefficient k, has been found to be a function of material density and soil moisture content, along with other materials properties. The exponent k3 = -0.69, is defined internally in the program.

24

2 must include both the effects due to load as well as those due to pverfrurfen. The effects due to both sources are added together before the S parameter is evaluated. The vertical overburden pressure, W is determined in the program according to Wv = E 7 ih, i= l where 7 ,h, is the product of the unit weight of the material, 7 ,, times the layer depth, hj. This product is summed over n layers, down to the actual depth of the point of calculation. It is important that the appropriate value of unit weight be input, in order to assure that the correct overburden pressure will be calculated. Abt>vE the bulk unit weight (not the "dry density") should be used, since the weight of water in the partially saturated soil res contributes to the intergranular pressure. Belpw ihg wgtgr taMg the pore water supports itself and adds nothing to the intergranular pressure. Thus the submerged or buoyed unit weight should be used. The submerged unit weight, 7 , is determined by subtracting the unit weight of water, 7 . (62.4 pounds per cubic foot, or 1,000 kilograms per cubic meter), from the saturated bulk unit weight of the aggregate, 7 .
v* 7 =
Tial

(3)

- 7v <*

(^)

The uplift or buoyant effect of the water, which reduces the intergranular overburden pressure, is thus properly taken into consideration. For problems where all of the layers beneath a given depth are linear (MODEL = 0), the overburden pressure does not enter into the calculations, and therefore it is sufficient to input a value of 0 . 0 for the unit weight of the materials.

Lateral Earth Pressure Ratio In order to take into consideration the effect of overburden pressure on the horizontal stresses, the "at-rest" lateral earth pressure coefficient, ke, is used. The horizontal overburden pressure, Wh , is computed according to

wb- wv* k

(5)

For normally consolidated subgrade materials ko may be expected to be in the range 0.4 < k,, < 0.6, with a typical value of about 0.5. However, materials that have been previously subjected to high pressure, such as due to glaciation, desiccation, or in the case of pavement materials, due to compaction, k has been found to be a function of the overconsolidation ratio (OCR). Such materials may be expected to have a k* in the range 0.8 < k. < 2.0, with a typical value of 1 . 0 to 1 .5 .

25

Since the lateral earth pressure ratio will generally be unknown for all materials in the pavement, it is recommended for use in MODCOMP 3 that k* be entered as 1.0 to 1.25 for compacted pavement layers. However, for materials that are subjected to a seasonal loss of density due to frost heaving, may be somewhat less than 1.0, perhaps 0.7 to 0.9.

Output Form at Options Output can be specified to be either BRIEF, LONG or ALL . If this field h left blank, BRIEF output will be provided. In all three cases the input data will be reiterated. With BRIEF output only the table of results for the last iteration will be printed. For nonlinear systems, specifying LONG output does not add greatly to the total processing *ime, and it may be helpful to look at the results for each iteration in order to assess the rate of convergence. The ALL output option will probably be of interest only to those who want to assess how the program goes about the process of obtaining a solution, and it incorporates the LONG output also. Regardless of the choice of output format, a standard output summary file, MODCOMP3.0UT, is provided. This lists only the modulus values for each layer and the RMS error for each iteration.

Acknowledgements Financial support for the development of the MODCOMP I and MODCOMP 2 programs was provided by the Cornell University Local Roads Program, the Cornell University Agricultural Experiment Station, and the U.S. Army Corps of Engineers Cold Regions Research and Engineering Laboratory. Support for the development of MODCOMP 3 was provided by the Cornell Local Roads Program and by the Department of Agricultural and Biological Engineering at Cornell University.

Disclaimer Use of the MODCOMP 3 computer code is subject to the judgment of the user and no express or implied warranty of the accuracy of the results or any interpretations made therefrom is made by the author or the sponsoring agencies. Citation of brand names does not constitute an endorsement, and the contents of this report are not to be used for advertising or promotional purposes.

26

Copyright Copyright for the MODCOMP 3 computer program is asserted by the Cornell University Local Roads Program and by Cornell University. All rights are reserved.

27

Appendix 1 - Program Test Data

28

Test Problem # 1 - Problem with some nonlinear layers. The following input and corresponding output may be used to verify that MODCOMP 3 is working properly. The results reported below were obtained using an IBM type desktop computer, with a 33 MHz 80386 processor, and an 80387 math coprocessor.

Input Line 1: Trial Problem Line 2: ENGL,BRIEF Line 3: *H\ 1.5, 10 Line 4: 4 Line 5a: *U 0 0 200000. 0.35 145. 1.0 5.0 0.0 0.0 Line 5b: U 1 0 15000. 0.35 135. 1.0 4.5 0.0 0.0 Line 5c: U 1 0 12000. 0.45 130. 1.0 4.5 0.0 0.0 Line 5d: K 0 0 7500. 0.45 130. 1.0 0.0 0.0 0.0 Line 6: 4 Line 7a: 0.0 50.0 6.0 Line 7b: 0.0 72.0 6.0 Line 7c: 0.0 100.0 6.0 Line 7d: 0.0 145.0 6.0 Line 8: 7 Line 9: 0.00 7.87 11.81 17.72 25.59 35.43 47.24 Line 10a: 21.20 17.2914.72 11.54 8.457 5.925 4.242 Line 10b: 30.09 24.55 20.93 16.46 12.13 8.545 6.118 Line 10c: 41.27 33.6828.73 22.65 16.79 11.84 8.514 Line lOd: 59.0248.17 41.12 32.49 24.24 17.28 12.38

| Layer j Data 1 1 | Load j Data 1 1 | Deflec| tion | Data j j

Results (Iteration No. 5) Layer 1 2 3 4 Type Unknown Unknown Unknown Known Model (psi) = 256246.74 (psi) = 16256.9 S 0.177 (psi) = 2420.06 S 0.817 (psi) = 7500.0 CedT 0.999 0.999 Cpmmgnls = constant (linear layer) = bulk stress model = bulk stress model = constant (linear layer)

E E E E

E S S E

"This is the correlation coefficient, r, which denotes the statistical goodness o f fit o f the constitutive model. A coefficient of 1.00 (the sign of r will depend on the sign of ka) denotes perfect fit and a strong model, while a coefficient near zero denotes a lack of fit. In the latter case k will not be statistically significantly different from zero, and MODCOMP 3
29

wiU consider the layer to be linearly elastic, of the form E constant.

Calculated/Supplied Deflection Agreement

Radius 0.00 7.87 11.81 17.72 25.59 35.43 47.24

Calculated Deflection 59.019 48.187 41.167 32.541 24.222 17.339 12.354

Measured Deflection 59.020 48.170 41.120 32.490 24.240 17.280 12.380

Percent Difference +0.00 +0.04 +0.12 +0.16 -0.07 +0.34 -0.21

Mean Square Error = 0.171 percent Computation terminated because tolerance was satisfied after 5 iterations.

The above computation required approximately 50 minutes on an 80386 IBM-PC equipped with a numeric co-processor.

30

APPENDIX E THE SPECTRAL-ANALYSIS-OF-SURFACE-WAVES METHOD

INTRODUCTION Spectral-analysis-of-surface-waves (SASW) method is a testing procedure for determining shear wave velocity, shear modulus, and Young's modulus profiles of pavement systems in situ. The test is performed from the ground surface and thus requires no boreholes. Measurements are made at strain levels below 0.001 percent, where elastic properties are considered independent of strain amplitude. The key elements in SASW testing are the generation and measurement of Rayleigh waves. The SASW method has proven to be a valuable tool for determining shear wave velocity profiles. The ability to determine a detailed shear wave velocity profile entirely from surface measurements results in substantial time and cost savings compared to other seismic methods, e.g., crosshole, downhole. SASW is of potential interest to the highway engineer for a number of reasons: 1. It can provide information to determine the approximate thickness of individual pavement layers without coring. 2. It can provide a starting point for estimating moduli, and modular ratios of the pavement layers during the backcalculation process. 3. It can detect rigid layers and provide an estimate of depth. 4. It can more accurately determine & quantify modular values of thin ACP layers on the surface, as compared to the FWD. SASW is a natural complement to the FWD. It cannot replace the FWD however, as it cannot predict moduli of paving layers under traffic loads, as most materials behave in a non-linear fashion. A number of publications in recent years have described in detail the SASW method (Hiltunen [1988], Nazarian [1984], Nazarian and Stokoe [1989]). A schematic of the experimental arrangement for SASW tests is presented in Figure 1. Current practice calls for E-l

locating two (or more) vertical receivers on the pavement surface a known distance apart and a transient wave containing a large range of frequencies is generated in the pavement by means of a hammer. The surface waves are detected by the receivers and are recorded using a Fourier spectrum analyzer. The analyzer is used to transform the waveforms from the time to the frequency domain and then to perform spectral analyses on them. The spectral analysis functions of interest here are the phase information of the cross power spectrum and the coherence function. Knowing the distance and the relative phase shift between the receivers for each frequency, the velocity of the surface wave (phase velocity) and wavelength associated with that frequency are calculated. The final step is application of an inversion process that constructs the shear wave velocity profile from the phase velocity-wavelength information (dispersion curve). To date, use of the SASW method for the in situ characterization of pavement materials has included: Determination of the small-strain shear wave velocity, shear modulus, or Young's modulus as a function of depth. The stiffness profile provides a valuable tool for tracking the structural integrity or condition of the pavement system with time (Drnevich, et al. [1990], Nazarian and Stokoe [1989]). Determination of the thicknesses of the pavement layers (Roesset, et al. [1990]). The estimation of the layer thicknesses can often provide substantial savings in lieu of destructive coring of the pavement. The as-constructed thicknesses are used in the design of rehabilitation strategies such as surface overlays, and can also be used when the SASW method is used in conjunction with other pavement NDT methods such as the falling weight deflectometer (FWD). Determination of the depth to bedrock below the pavement surface (Nazarian and Stokoe [1989]).
E-2

Determination of the rate of strength increase (curing) in portland cement concrete (Rix, Bay, and Stokoe [1990]). Use of the SASW method for the measurement of the rate of strength development in portland cement concrete has been demonstrated. Widespread use of the technique for routine construction is not anticipated; rather, use can be made under special or critical circumstances. Detection and delineation of voids, cracks, and delaminations (Nazarian [1990]). Use of the SASW method for the detection and delineation of voids, cracks, and delaminations has been limited to date. Demonstration of the possibilities have been made, however, and further research and development of the technique are underway.

Investigation of a site using the SASW method consists of the following three phases: 1) collection of data in the field, 2) determination of the Rayleigh wave dispersion curve, and 3) inversion of the Rayleigh wave dispersion curve. Each phase is individually discussed in the following sections. COLLECTION OF DATA IN THE FIELD A schematic of the current test procedure is shown in Figure 1. The procedure is quite simple. Two vertical receivers (geophones or accelerometers) are located on the surface at the site. A surface wave is generated by an appropriate source. The generated wave front is detected by the receivers as it propagates past them and is recorded on the appropriate device. A more thorough discussion of the equipment required and of how the measurements are made follows.

Source The source should be able to generate Rayleigh waves over a wide range of frequencies with adequate amplitude so that they can be
E-3

detected by the receivers. Simultaneously, the source should generate minimal compression (P) and shear (S) wave energy. Heisey et al. (1982) identified two important factors in source selection. First, the magnitude of the input energy does not seem to be a critical factor as long as it exceeds the background noise level. Rather, source selection should be based on the range of frequencies that can be sufficiently excited to adequately sample the site. Further, the energy of excitation should not be focused on a few frequencies but should be distributed over all frequencies in the measurement bandwidth. The actual frequency values required for testing a site are dependent upon the stiffness of the materials to be tested and on the desired depth of investigation. Nazarian (1984) has indicated that the highest frequency required for testing soil sites is typically 200-800 Hz. The more important value for testing soil sites, however, is usually the lowest frequency to be excited. This is because it is usually desirable to measure the shear wave velocity profile as deep as possible and it is the low frequency (long wave length) surface waves that sample the deep materials. Nazarian (1984) indicated that one of the most important additions to the SASW method would be a source capable of generating frequencies below 5 Hz so that deeper deposits can be sampled. The author has sometimes found it difficult to obtain acceptable data below 25 Hz at sites where strong background noise is present. A source capable of generating adequate energy at these low frequencies would indeed be a very important addition to the SASW method. For testing pavement sites, on the other hand, it is the high fre quencies that are typically most important and more difficult to generate and resolve. Since it is normally not required to determine the stiffness profile below about 3 meters (10ft.) from the surface, the lowest frequency to be generated is usually on the order of 30 Hz, which is usually not difficult to produce. However, for sampling the thin, stiff layers near the surface of pavement systems, it is necessary to generate very high frequencies, occasionally as high as 100 kHz. For impact testing methods this requires a source capable of producing a very short impulse, such as a ball peen hammer, spring loaded center punch, etc.

E-4

A second important factor identified is the coupling of the source with the test surface, which influences the transfer of energy. Heisey et al. (1982) found that the use of a plate between the hammer and impact surface should be avoided if the generation of low frequencies (long wavelengths) is desired. Two excitation techniques have been employed to date, namely, impact and random vibration. Impact techniques are more popular at the present time, although Dmevich, et al. (1985) have used random noise excitation with some success. For testing the stiff layers near the surface of pavements using an impact source, a light hammer producing a short impulse is much better than a large weight which produces a relatively cushioned impulse. The opposite is true for testing soil sites. Examples of sources used to date include: claw hammer, range of sizes o f ball peen and sledge hammers, drop hammer (compaction hammer), hammer and chisel, Standard Penetration Test (SPT) hammer, and various size shakers excited with a random signal. Again, the light-weight sources capable of generating high fre quencies are used for investigations where the properties of shallow layers are of primary concern. The size and weight of the source (for impact testing) is increased, and thus the upper frequency bound decreased, as the required depth of investigation increases. An SPT hammer has been successfully employed in one case study by the author to sample materials as deep as approximately 50 feet.

Receivers
Vertically oriented transducers (velocity or acceleration, depending on required frequency range) are used as receivers in the field. Based on tests performed on pavements, Heisey et al. (1982) found that vertical transducers provide a more accurate velocity profile than do horizontal. They found that velocities obtained from measurements using horizontal transducers were generally too high, and suggested that this was probably due to the greater sensitivity of horizontal transducers to higher-velocity P-waves that contaminate the measured signal to some degree.

E-5

The range of frequencies over which the receivers should function depends on the site being tested. To sample deep materials (15 to 30 meters (50 to 100 ft)), velocity transducers with a low natural frequency, typically within the range of 1 to 2 Hz, are appropriate. In contrast, for sampling the shallow layers more common to pavement sites, the receivers should be able to respond to frequencies in the kHz range, i.e., the use of accelerometers is more appropriate. Velocity transducers are typically useful in the range 1-1,000 Hz, while accelerometers can have a useful range from 1 to 20,000 Hz. However, accelerometers will generally produce less output at lower frequencies than velocity transducers. It is recom mended, therefore, that velocity transducers be used if the fre quencies to be measured are less than 1000 Hz. The result of this recommendation is that velocity transducers are used almost exclusively for investigating soil sites. For investigating pavement systems, on the other hand, a combination of velocity transducers and accelerometers are used. Accelerometers are typically used for receiver spacings below 1.2 meters (4 ft.), while velocity transduc ers are used for spacings of 1.2 meters (4 ft.) and above. An issue of some concern when employing velocity transducers in the SASW method is that of phase-angle distortion. It is usually desirable to use velocity transducers that display a horizontal response curve over the widest range of frequencies possible. This is achieved by adding a certain amount of damping to the velocity transducer. However, damped velocity transducers always introduce distortion of complex or superimposed waveforms. This is due to a phase shift or time delay between the mechanical input and the electrical output signal of the instrumentation system. Therefore, velocity transducers that are intended for use in measuring either transient disturbances or vibrations containing several frequency components simultaneously, such as in the SASW method, should be calibrated to determine that the phase-angle distortion is acceptably small. Alternatively, because it is the relative phase difference between two velocity transducer signals that is desired in the SASW method, some phase-angle distortion can be tolerated as long as the distortion caused by each transducer is approximately the same. Phase-angle distortion is also possible with accelerometers. It is highly recommended that a matched pair of receivers be used at all times and that periodic checks be made of
E-6

transducer pairs to ascertain the range of frequencies where the phase-angle distortion is below a tolerable level. The transducers should be used only within this frequency range or the measured results will be questionable. Another important aspect related to receivers is that of attaching them to the test surface. Positive coupling between the receivers and the test surface is essential. Velocity transducers used at soil sites are typically coupled through detachable spikes. For sites with loose materials near the surface, partially burying the velocity transducers often provides more positive coupling. Both velocity transducers and accelerometers have been successfully mounted on pavement surfaces with materials such as bees-wax (Nazarian [1984]), potter's clay, epoxy, and Super Glue. This technique, however, would not be practical for an automated testing procedure. More research is required to develop an efficient means to positively couple receivers to a pavement surface. Location of Source and Receivers The factors that affect appropriate spacing of the receivers have been studied by Heisey et al. (1982). These factors include: 1) velocity of the material to be tested, 2) desired depth of investi gation, 3) range of frequencies, 4) attenuation properties of the me dium, and 5) sensitivity of the instrumentation. On the basis of studies conducted at several soil sites, Heisey et al. (1982) found that a spacing arrangement in which the first receiver (receiver nearest the source) is located at an increasing distance from the source is more favorable than an arrangement in which the first receiver is fixed at a reference location close to the source. In addition, Heisey et al. (1982) found that much of the scatter in the velocity profile can be reduced by filtering out data for wavelengths that are inappropriate for the spacing of the receivers. Wavelengths that are too short for a given spacing may attenuate excessively, whereas wavelengths that are too long may not have traveled a sufficient distance to sample adequately the depth proportional to the wavelength. Furthermore, since frequency analyzers possess a potential relative error, i.e., percentage error, in measuring phase relationships, it is desirable to set a lower limit on the measured
E-7

phase angles one is willing to accept in order to keep the absolute error below a specified level. This transforms in the present case to setting an upper limit on the wavelengths one is willing to accept at each receiver spacing. Based on these considerations, an appropriate range of wavelengths (Lr ) for a given receiver spacing (x) was found to be: x -< L r -< 3x 2 R Eq. E -l is used at each receiver spacing to eliminate those data points not satisfying the wavelength criterion. Theoretically, one test at one receiver spacing is sufficient to determine the properties of the medium. However, due to the factors discussed above, several tests with different receiver spacings are typically required to obtain data over the complete range of frequencies desired. In each test the distance between the receivers is generally doubled. A typical setup is shown in Figure 1(b). The receivers are always placed symmetrically about the selected, imaginary centerline. This pattern of testing is called the common receivers midpoint (CRMP) geometry. Nazarian and Stokoe (1983) have shown that use of this setup reduces scatter in the data due to the fact that the distances covered in the previous tests are always included in the next tests. Small receiver spacings are used to obtain short wavelengths (high frequencies) and thus shallow depths, while large spacings are used to obtain long wavelengths (low frequencies) and thus large depths. Typical receiver spacings range from 0.15 meters (0.5 ft.) to 2.4 meters (8 ft.) for pavement sites and from 0.6 to 19.5 meters (2 to 64 ft.) for soil sites. These values may change depending on the range of depths to be sampled at a particular site. In addition, at each receiver spacing, two series of experiments are performed. First, the test is carried out from one direction (forward profile). Then, without relocating the receivers, the same test is performed with the source on the opposite side of the receivers and the receiver cables switched on the recording device (reverse profile). By running forward and reverse profiles and by averaging the results of these two tests, the effect of any internal phase shift
E-8

(Eq. E -l)

between receivers is minimized, and the effect of dipping layers along the distance between the receivers is averaged.
Recording Device

The records of wave arrivals at different receivers can be easily recorded on an oscilloscope or other data acquisition hardware and then stored on magnetic tape or disk. These measurements are thus made in the time domain. By means of the fast Fourier transform algorithm, the results can then be converted to the frequency domain, and the data can be reduced to develop the Rayleigh wave dispersion curve. A more convenient device is a Fourier spectrum analyzer. A Fourier spectrum analyzer is a digital oscilloscope that has the ability to operate in either the time or frequency domain. The analyzer can directly calculate all of the required time and frequency domain functions. Thus, the data can be easily checked in the field to determine its suitability for future analysis. In addition, the type and number of averages, frequency span, trigger conditions, input voltage ranges, and the type of measurement window function can all be specified by the operator. The analyzer can also be easily interfaced with a magnetic tape recorder or disk drive for permanent storage and recall of data. The major drawback of Fourier spectrum analyzers, however, is that they are relatively expensive (approximately $25,000).

Measurement Parameters

The measurement parameters of concern here are the setup parameters for the recording equipment. The discussion is pri marily directed towards the use of a Fourier spectrum analyzer as the recording device. The pertinent parameters are the type and number of signal averages, frequency span, input ranges, triggering, and measurement window function. As discussed above, SASW testing is conducted at each site for a series of receiver spacings. In addition, to help decrease the effect
E-9

of background noise, more than one record is obtained at each receiver spacing and source location and the records are ensemble averaged in the frequency domain. Averaging of signal pairs in the frequency domain is also necessary to provide a meaningful coherence function. Heisey et al. (1982) have found that results obtained with five averages will provide reliable data. They have shown that additional averages are not warranted because they do not significantly improve the end results. The second setup parameter of concern is the frequency span, i.e., the maximum frequency to be included in the measurements (assuming baseband or zero start frequency measurements). The primary objective in choosing the frequency span is to obtain the best resolution in the frequency domain (and thus in the dispersion curve) as possible while at the same time obtaining the complete range of frequencies for the particular test setup. A trade off must be made because a spectrum analyzer digitizes a fixed number of points independent of any of the setup parameters. Thus, as the frequency span increases the frequency resolution decreases. The choice of frequency span depends upon two factors: the receiver spacing and the stiffness of the material under test. For small receiver spacings, large frequency spans are required because it is the high frequencies that are of interest, and vice versa for large receiver spacings. For a given receiver spacing, as the stiffness of the material under test increases, the required frequency span must increase to fully describe the dispersion curve. The phase of the cross power spectrum and the coherence function are used as guides for selecting the appropriate frequency span for a given receiver spacing and material profile. Nazarian (1984) suggests a rule of thumb for selecting the frequency span: the optimum bandwidth is obtained when more than 3/4 of the frequency range contains data of high quality, identified by coherence values greater than 0.9. Use of eq. 1 can also be made in selecting the appropriate frequency span. Eq. 1 suggests that the data should be filtered for wavelengths smaller than 1/2 the receiver spacing and larger than 3 times the receiver spacing. This transforms into accepting continuous or unwrapped phase values from the cross power spectrum between 120 and 720 degrees.
E-10

Thus, frequencies with phase values larger than 720 degrees should not be recorded in the field since they will subsequently be filtered out of the data. The optimum frequency span, then, is obtained as the lesser of the two frequencies from the above rules. The third input parameter to be chosen are the voltage ranges for the recording channels. The voltage ranges should be set as low as possible without overloading to obtain the best resolution in the analog to digital conversion. The value for the channel connected to the receiver nearer the source (usually channel 1) will typically be higher than the value for the farther receiver; this does not create any problems and is preferred to obtain the best resolution. Also, the ranges will likely need adjustment for each receiver spacing. The fourth consideration is triggering of the instrument to begin recording data. The primary objective is to obtain the complete signal for all channels and not to introduce any internal phase difference between the channels. For the impact testing technique the recorder is typically triggered off the signal from the near receiver about half way up (down) the first major pulse. A pre-trigger delay is then set to ensure that the initial portion of the signal is obtained. It is imperative that the pre-trigger delay be an equal amount for all recording channels, otherwise an internal phase difference will be introduced. Nazarian (1984) suggests that a pre-trigger delay of 10 percent of the total time record is sufficient. The final measurement parameter to select is the measurement window function. The measurement window function is the function applied to the time domain records to ensure that the signals have zero value at each end of the record, and thus be periodic with respect to the measurement window. This is neces sary to prevent leakage. For impact testing the signal usually decays to zero before the end of the record and a rectangular or uniform window function is used. This, in effect, does nothing to the measured time signal. For testing techniques where leakage may be a problem, e.g., pure random, an appropriate window should be applied, a Hanning window for example.

E -ll

DETERMINATION OF THE RAYLEIGH WAVE DISPERSION CURVE

The in-house data reduction associated with the SASW method consists of construction of the dispersion curve and then inversion of the dispersion curve. It was noted earlier that the variation of wave velocity with frequency (or wavelength) is known as disper sion. A plot of velocity versus wavelength is called a dispersion curve. The dispersion curve is constructed from the spectral analysis functions discussed above. The information of major interest is obtained from the coherence function and from the phase of the cross power spectrum. A typical coherence function and phase of the cross power spectrum are shown in Figure 2. From the coherence function, the range of frequencies that should be considered in each record is selected. Nazarian and Stokoe (1986) have found that frequencies with a coherence value greater than 0.90 provide useful data. The phase information of the cross power spectrum provides the relative phase between two signals at each frequency in the range of frequencies excited in the SASW test. Phase information should only be accepted for those frequencies satisfying the coherence criterion above. As seen in Figure 2, regions with low coherence exist, and phase data in these regions would be deleted in constructing the dispersion curve. In the end, however, it is the phase angle of the cross power spectrum that is used in the calculation of the dispersion curve. Using the coherence function as the sole criterion can lead to questionable results if correlated "noise" is measured at each receiver. The coherence function will indicate "good" data in this instance, yet the phase angle may yield incorrect dispersion data. The experienced user will consider both the phase angle of the cross power spectrum and the coherence function to determine the ranges of good data. For a travel time equal to the period of the wave, the phase dif ference is 360 degrees. Thus, for each frequency the travel time between receivers can be calculated by:

>-r4n
E-12

(E q.E -2)

where: t = travel time associated with the given frequency, < = |) phase shift (continuous or unwrapped) for that frequency, T = \ / f = period of the wave for the given frequency, and / = the given frequency. The distance between the receivers, x, is known. Therefore, the phase velocity, Vph, at the given frequency is calculated by:

V h~X p t and the corresponding wavelength is equal to:

(Eq. E-3)

LR ~ 3 f* l By repeating this procedure for each frequency in the acceptable range of frequencies, a dispersion curve for the given receiver spacing and source location (forward or reverse) is obtained. The procedure is repeated for the data from all receiver spacings and source locations. Each dispersion curve is then filtered for wave lengths not meeting the requirements of eq. 1. Typically, the dispersion data from one receiver spacing will cover a different range of wavelengths than the other spacings. However, some overlap in the data will occur between spacings so that the com plete range in wavelengths is covered. The filtered dispersion curve data from all receiver spacings and source locations are then sta tistically combined to provide an average curve to represent the site. A typical dispersion curve for a pavement site is shown in Figure 3.

(Eq. E-4)

INVERSION OF THE RAYLEIGH WAVE DISPERSION CURVE

The velocities from the dispersion curve are not actual Rayleigh wave velocities but are apparent or phase velocities. The existence of a layer with a higher or lower velocity at the surface of the medium affects the measurements of the velocities for the underlying layers. Thus, a method of determining actual wave velocities from the apparent velocities is necessary. E-13

Inversion of the Rayleigh wave dispersion curve consists of determining the shear wave velocity profile from the phase velocity versus wavelength data. Stokoe and Nazarian (1985) suggest that the simplest method of inversion is to assume that the shear wave velocity is approximately equal to 110 percent of the phase velocity, and that the effective sampling depth (the depth to material having the calculated shear wave velocity) for each wavelength is equal to 1/3 to 1/2 of that wavelength. Use of this simple inversion method inevitably results in some error due to the assumptions made. As noted above, existence of a layer with a relatively high or low velocity near the surface causes a shift in the measured velocities of the underlying layers towards higher or lower velocities. If the contrast in velocities is small, then the simple inversion method may work reasonably well. However, Nazarian and Stokoe (1983) have found that use of the simple method normally results in shear wave velocity profiles that are doubtful. Therefore, a refined inversion process has been devel oped by Nazarian (1984). The refined inversion process used is to obtain a theoretical dispersion curve that matches reasonably well with the experimental dispersion curve. The theoretical dispersion curve is constructed using a modified version of the Thomson (1950)-Haskell (1953) matrix formulation for multi-layered elastic media, as developed by Thrower (1965). The mathematics have been thoroughly described by Nazarian (1984) and will not be repeated here. To begin the process, the medium is divided into a number of layers and a shear wave velocity, Poisson's ratio (or compression wave velocity), and mass density is assigned to each layer. The theoretical dispersion curve for this assumed profile is then calcu lated and compared with the experimental dispersion curve. If the theoretical and experimental curves match, the desired profile is obtained. However, if the two curves do not match, the shear wave velocity profile is modified and another theoretical curve is constructed. This trial-and-error procedure is continued until the two curves match to within a reasonable tolerance. The final velocity profile obtained, then, is the postulated variation of shear E-14

wave velocity with depth. The shear or Young's modulus profile can then be constructed using equations from the theory of elas ticity. The shear wave velocity and Young's modulus profiles constructed from the dispersion curve in Figure 3 are shown in Figure 4. It should be noted that the inversion process also requires that the mass density and Poisson's ratio of each layer in the profile be known. However, reasonably assumed values for these parameters can usually be used, since it has been shown by Nazarian (1984) that the effect o f these parameters on the final outcome is small, especially for civil engineering materials in which Poisson's ratio and mass density fall into fairly narrow ranges. At the same time, care should be exercised in choosing the value of Poisson's ratio. Nazarian and Stokoe (1983) indicate that several studies to evaluate Poisson's ratio of different materials show that the values are quite small in the low-strain range. This should be considered for dynamic tests such as the wave propagation methods.

E-15

SUMMARY OF CASE STUDIES

Over the past decade, the SASW method has been employed at hundreds of sites nationwide. The author has used the method to date at over 30 sites. Together, the sites have included both flexible and rigid pavements, soil sites, and a concrete dam. In a number of cases the results have been compared with crosshole seismic tests, a well established method for measuring shear wave velocity in situ. The procedure for performing crosshole tests has been discussed in detail in a number of articles, and is not repeated here. In general, the moduli obtained from the two methods compare favorably with one another. One comparison between the two methods for a flexible pavement site is shown in Figure 4. The profile consists of 5 in of asphaltic concrete, 8 in of lime rock base, and subgrade. The shear wave velocity profile determined from SASW tests, along with the results from crosshole tests, are shown in Figure 4(a). It is observed that the results from the two methods compare well. Young's modulus profiles calculated from the shear wave velocity profiles are shown in Figure 4(b). These results also agree well with one another, which is not surprising, since the two shear wave velocity profiles are nearly alike.
DISADVANTAGES

The primary disadvantage of the SASW method at this time is that the testing and data reduction are not performed rapidly. The SASW method currently requires that several impact tests be performed at each location of interest. For each impact test, two transducers are placed on the surface of the ground or pavement at a known distance apart. To fully characterize the layered system, the series of tests at each location requires increasing the distance between the two transducers and generating impacts with different frequency contents to correspond to the transducer spacings. This technique must currently be performed in a stationary mode, and it requires a good deal of time at each location. However, with some development, it seems possible to convert this technique to a rapid, on-the-move test.

E-16

A further limitation of the SASW method for pavement evaluation and other applications may be the fact that the moduli obtained are for low strain levels, i.e., in the range of strains where the moduli are not strain dependent. The loadings encountered in real pavements are often large enough to cause the pavement materials to exhibit strain-dependent behavior with respect to modulus. However, the SASW method does give a starting point from which strain-dependent moduli can be determined from correlations with laboratory results, such as resonant column or cyclic shear tests. This technique is used regularly in earthquake engineering applications.
SUMMARY

This chapter has described the SASW test method in detail. The three phases o f SASW testing, namely, collection of data in the field, determination of the Rayleigh wave dispersion curve, and inversion of the Rayleigh wave dispersion curve, were individually described. Many case studies have shown that the results obtained from SASW tests compare well with the results from crosshole tests, a well established testing technique. The major limitations of the SASW method at this time are that the testing and data re duction are time consuming and the modulus measured with the SASW test is for small strains.

E-17

BIBLIOGRAPHY

Aouad, M. F., Stokoe, K. H. II, and Briggs, R. C. (1993), "Stiffness of the Asphalt Concrete Surface Layer from Stress Wave Measurements," Paper Presented to Annual Meeting, Transportation Research Board, Washington, D. C., January. Bay, J. A. and Stokoe, K. H. II (1990), "Field Determination of Stiffness and Integrity of PCC Slabs Using the SASW Method," Proceedings, Nondestructive Evaluation of Civil Structures and Materials, University of Colorado, Boulder, Colorado, October, pp. 71-85. Bay, J. A. and Stokoe, K. H. II (1992), "Field and Laboratory Determination of Elastic Properties of PCC Using Seismic Tech niques," Paper Presented to Annual Meeting, Transportation Research Board, Washington, D. C., January. Dmevich, V. P., Hossain, M. M., Wang, J., and Graves, R. C. (1990), "Determination of Layer Moduli in Pavement Systems by Nondestructive Testing," Paper Presented to Annual Meeting, Transportation Research Board, Washington, D. C., January. Dmevich, V. P., Kim, S.-I., Alexander, D. R., and Kohn, S. (1985), "Spectral Analysis of Surface Waves in Pavement Systems with Random Noise Excitation," Expanded Abstracts with Biographies, 55th Annual International Society of Exploration Geophysicists Meeting, Washington, D. C., October 6-10, pp. 143-145. Haskell, N. A. (1953), "The Dispersion of Surface Waves on Multilayered Media," Bulletin of the Seismological Society of America, Vol. 43, No. 1, January, pp. 17-34. Heisey, J. S., Stokoe, K. H. II, Hudson, W. R., and Meyer, A. H. (1982), "Determination of In Situ Shear Wave Velocities from Spectral Analysis of Surface Waves," Research Report No. 256-2, Center for Transportation Research, The University of Texas at Austin, December, 277 pp.

E-18

Heisey, J. S., Stokoe, K. H. II, and Meyer, A. H. (1982), "Moduli of Pavement Systems From Spectral Analysis of Surface Waves," Research Record No. 852, Transportation Research Board, pp. 22-31. Hiltunen, D. R. (1988), "Experimental Evaluation of Variables Affecting the Testing of Pavements by the Spectral-Analysisof-Surface-Waves Method," Technical Report GL-88-12, U. S. Army Engineer Waterways Experiment Station, Vicksburg, Mis sissippi, August, 303 pp. Hiltunen, D. R. (1991), "Nondestructive Evaluation of Pavement Systems by the SASW Method," Geotechnical News, BiTech Publishers Ltd., Vancouver, B. C., September. Hiltunen, D. R. and Woods, R. D. (1989), "Influence of Source and Receiver Geometry on the Testing of Pavements by the Surface Waves Method," Nondestructive Testing of Pavements and Backcalculation o f Moduli, ASTM STP 1026, A. J. Bush III and G. Y. Baladi, eds., American Society for Testing and Materials, pp. 138-153. Hiltunen, D. R. and Woods, R. D. (1990), "Variables Affecting the Testing of Pavements by the Surface Waves Method," Research Record 1260, Transportation Research Board, Washington, D. C., pp. 42-52. Nazarian, S. (1984), "In Situ Determination of Elastic Moduli of Soil Deposits and Pavement Systems by Spectral-Analysis-of-Surface-Waves Method," Ph.D. Dissertation, The University of Texas at Austin, 453 pp. Nazarian, S. (1990), "Detection of Deterioration Within and Be neath Concrete Pavements with Sonic and Ultrasonic Surface Waves," Proceedings, Nondestructive Evaluation of Civil Struc tures and Materials, University of Colorado, Boulder, Colorado, October, pp. 391-406. Nazarian, S. and Stokoe, K. H. II (1983), "Evaluation of Moduli and Thicknesses of Pavement Systems by Spec
E-19

tral-Analysis-of-Surface-Waves Method," Research Report No. 256-4, Center for Transportation Research, The University of Texas at Austin, December, 123 pp. Nazarian, S. and Stokoe, K. H. II (1984), "Nondestructive Testing of Pavements Using Surface Waves," Research Record No. 993, Transportation Research Board, pp. 67-79. Nazarian, S. and Stokoe, K. H. II (1986), "Use of Surface Waves in Pavement Evaluation," Research Record No. 1070, Transportation Research Board, Washington, D. C., pp. 132-144. Nazarian, S. and Stokoe, K. H. II (1989), "Nondestructive Evaluation of Pavements by Surface Wave Method," Nondestruc tive Testing o f Pavements and Backcalculation of Moduli, ASTM STP 1026, A. J. Bush III and G. Y. Baladi, eds., American Society for Testing and Materials, pp. 119-137. Nazarian, S., Stokoe, K. H. II, and Briggs, R. C. (1987), "Nondestructively Delineating Changes in Modulus Profiles of Secondary Roads," Research Record 1136, Transportation Re search Board, Washington, D. C., pp. 96-107. Nazarian, S., Stokoe, K. H. II, Briggs, R. C., and Rogers, R. (1988), "Determination of Pavement Thicknesses by SASW Method," Paper Presented to Annual Meeting, Transportation Research Board, Washington, D. C., January. Nazarian, S., Stokoe, K. H. II, and Hudson, W. R. (1983), "Use of Spectral Analysis of Surface Waves Method for Determination of Moduli and Thicknesses of Pavement Systems," Research Record No. 930, Transportation Research Board, Washington, D. C., pp. 38-45. Rix, G. J., Bay, J. A., and Stokoe, K. H. II (1990), "Assessing In Situ Stiffness of Curing Portland Cement Concrete with Seismic Tests," Paper Presented to Annual Meeting, Transportation Research Board, Washington, D. C., January.

E-20

Roesset, J. M., Chang, D. W., Stokoe, K. H. II, Aouad, M. (1990), "Modulus and Thickness of the Pavement Surface Layer from SASW Tests," Research Record 1260, Transportation Research Board, Washington, D. C., pp. 53-63. Sanchez-Salinero, I., Roesset, J. M., Shao, K. Y., Stokoe, K. H. II, and Rix, G. J. (1987), "Analytical Evaluation of Variables Affecting Surface Wave Testing of Pavements," Research Record 1136, Transportation Research Board, Washington, D. C., pp. 86-95. Stokoe, K. H. II and Nazarian, S. (1985), "Use of Rayleigh Waves in Liquefaction Studies," Measurement and Use of Shear Wave Velocity fo r Evaluating Dynamic Soil Properties, Proceedings of a Geotechnical Engineering Division Session at ASCE Convention, Denver, Colorado, May 1, pp. 1-17. Thomson, W. T. (1950), "Transmission of Elastic Waves Through a Stratified Soil Medium," Journal o f Applied Physics, Vol. 21, No. 2, February, pp. 89-93. Thrower, E. N. (1965), "The Computation of the Dispersion of Elastic Waves in Layered Media," Journal of Sound and Vibration, Academic Press, London, England, Vol. 2, No. 3, July, pp. 210-226.
FIGURES

Figure 1. Schematic of Experimental Arrangement for SASW Tests (after Nazarian [1984]) Figure 2. Typical Cross Power Spectrum and Coherence Function Figure 3. Dispersion Curve Constructed from SASW Tests on a Flexible Pavement Site (from Nazarian and Stokoe [1986]) Figure 4. Shear Wave Velocity and Young's Modulus Profiles from Flexible Pavement Site (from Nazarian and Stokoe [1986])

E-21

Spectrum Anolyzer

Veri icol Rece 1 ver


% V W w / / \\T X /2 *- ---------XI v o r 1 o b 1 e 1 f

l o ) G e n e r o 1 ConFigurot 1 on oF S A S V T e e t s

Figure 1. (Appendix E) E-22

aoo

(a) Cross Power Spectrum

Phase (degrees)

Frequency (Hz)

Magnitude

Frequency (Hz)

Figure 2. (Appendix E) E-23

Phase Velocity (fps)

Figure 3. (Appendix E)

-------- '

.........-*

. . . - j - '

4
1
teptfc, ft

T ut Mthod |UI CmrtW

!
.
0 1000 2003 5000 S M i r M m V e l o c i t y , fp () S h e *r M t*t V ilo c U j P r v ft U

'
4000 T o u n j '* J t e c u lt r t , p i i i i O * (k ) T o u n 't Modulus e f l i t r

Figure 4. (Appendix E) J E-25

Appendix F

SHRPs Layer Moduli Backcalculation Procedure

SHRP-P-655

SHRPs Layer Moduli Backcalculation Procedure

Strategic Highway Research Program National Research Council

Strategie Highway Research Program Executive Committee

John R. Tabb, Chairman Mississippi Highway Department William G. Agnew General Motors Research (retired) E. Dean Carlson, ex officio Federal Highway Administration A. Ray Chamberlain Colorado Department o f Highways Michael J. Cuddy New York Department o f Transportation Raymond F. Decker University Science Partners Inc. Thomas B. Deen, ex officio Transportation Research Board Thomas M. Downs New Jersey Department o f Transportation Francis B. Francois, ex officio American Association o f State Highway and Transportation Officials William L. Giles Ruan Transportation Management System Jack S. Hodge Virginia Department o f Transportation Donald W. Lucas Indiana Department o f Transportation Harold L. Michael Purdue University Wayne Muri Missouri Highway and Transportation Department M. Lee Powell, III Ballenger Paving Company, Inc. Henry A. Thomason, Jr. Texas Department o f Highways and Public Transportation Stanley I. Warshaw National Institute o f Standards and Technology Roger L. Yarbrough Apcon Corporation Damian J. Kulash Executive Director Guy W. Hager Implementation Manager Edward T. Harrigan Asphalt Program Manager Kathryn Harrington-Hughes Communications Director Don M. Harriott Concrete & Structures/Highway Operations Program Manager Harry Jones Finance & Administration Director

Key SHRP Staff

SHRP-P-655

SHRPs Layer Moduli Backcalculation Procedure

PCS/Law Engineering

Strategic Highway Research Program National Research Council Washington, DC 1993

SHRP-P-655 Contract P-001 Program Manager: Neil F. Hawks Project Manager: Cheryl Allen Richter Production Editor: Marsha Barrett Program Area Secretary: Cynthia Baker August 1993 key words: backcalculation deflection testing moduli pavement evaluation

Strategic Highway Research Program National Academy of Sciences 2101 Constitution Avenue N.W. Washington, DC 20418 (202) 334-3774

The publication o f this report does not necessarily indicate approval or endorsement of the findings, opinions, conclusions, or recommendations either inferred or specifically expressed herein by the National Academy of Sciences, the United States Government, or the American Association of State Highway and Transportation Officials or its member states.

1993 National Academy o f Sciences

350/NAJV893

Acknowledgments

The research described herein was supported by the Strategic Highway Research Program (SHRP). SHRP is a unit of the National Research Council that was authorized by section 128 o f the Surface Transportation and Uniform Relocation Assistance Act of 1987.

TABLE OF CONTENTS Page

ABSTRACT

.............................................................................. ....................................................i x 1 1

INTRODUCTION............................................................................................................................ BACKCALCULATION SO FTW A RE.........................................................................................

BACKCALCULATION R U L E S .................................................................................................. 3 Definition of Layer Moduli R a n g e s................................................................................ 3 Asphalt Concrete L a y e rs ...................................................................................... 3 Portland Cement Concrete Layers ..................................................................... 9 Base and Subbase L ayers...................................................................................... 10 Subgrade L a y e r s .................................................................................................... 12 Modeling of Pavement Structure...................................................................................... 16 Subgrade L a y e rs .................................................................................................... 17 Thin L a y e rs............................................................................................................. 17 Pavement S tru c tu re ...............................................................................................17 Poissons Ratio . . . ...............................................................................................18 Evaluation of Analysis R e su lts......................................................................................... 20 Other Considerations.......................................................................................................... 20 SUMMARY AND CONCLUSIONS............................................................................................21 REFERENCES .............................................................................................................................. 22

LIST OF FIGURES
Figure Page

1 2 3

Typical Viscosity - Temperature R elationships............................................................. Schematic o f Stress Zone Within Pavement Structure Under FWD L o a d ...............14 Composite Modulus versus Radial Distance P l o t ..........................................................15
i T

LIST OF TABLES Table 1 2 Page Asphalt Viscosity at 70F Based on G ra d e ..................................................................... 7

Initial Modulus and Moduli Range for Unbound Base and Subbase Materials ............................................................................................................. 11 Initial Modulus and Moduli Range for Stabilized Base and Subbase Materials ............................................................................................................. 13 Poissons Ratio as a Function of Material Type ..........................................................19

vii

ABSTRACT
Deflection basin measurements for the purpose of structural capacity evaluation are a key component of the SHRPs Long-Term Pavement Performance monitoring program. In the near term, SHRP will apply a backcalculation procedure to these deflection measurements in order to estimate the in situ elastic moduli of the pavement layer materials. Because a standard method for evaluating the structural capacity of flexible pavements from deflection data does not presently exist, SHRP has undertaken a study to develop a layer moduli backcalculation procedure for use in the initial analysis of the SHRP deflection data. This procedure covers not only the software but also the rules and guidelines used in applying the program. This report focuses on the standard procedure used to ensure that the LTPP deflection data analysis is as consistent, productive, and straightforward as possible. The procedure consists of a rigorous set of application rules used to generate data files for direct input into the backcalculation program modeling of pavement structure and layer moduli ranges or initial moduli. Additional rules address the subsequent evaluation of the backcalculation results.

ix

INTRODUCTION
Since the Spring of 1988, SHRP has completed an initial round of deflection testing on nearly 800 in-service pavement test sections, and has begun a second round. Although the raw deflection data is the primary data to be stored for use by pavement researchers, the initial Long Term Pavement Performance (LTPP) data analyses require that SHRP derive estimates of the in-situ elastic moduli of the pavement layers from the deflection data. In order to do so, SHRP has developed a backcalculation procedure, consisting of an existing backcalculation program and a series of application "rules". The development process for the SHRP backcalculation procedure involved four phases: (1) a literature review to identify backcalculation programs which might be used in the procedure; (2) the selection of a limited number of programs for detailed evaluation; (3) a detailed evaluation of those programs; and (4) the development of a procedure around the selected program. The first three stages of this endeavor are discussed in detail elsewhere (1). This report focuses on the standard backcalculation procedure developed around the selected program. In general, backcalculation is a laborious process, requiring a high degree of skill, and the results are known to be moderately to highly dependent on the individual doing the backcalculation. This comes about for a number of reasons, including the lack of a consensus standard addressing all aspects of the backcalculation process. In order to ensure that the backcalculation process applied in the SHRP data analysis is as consistent, productive, and straight forward as possible, the SHRP backcalculation procedure combines an existing backcalculation program with a rigorous set of application rules. In addition, the initial backcalculation has been automated to a high degree, thus reducing opportunities for "operator" error, and between user inconsistencies. The SHRP backcalculation rules rely on information stored in the LTPP data base to generate the input modeling of pavement structure, layer moduli ranges, Poissons ratios, etc. -- for the backcalculation program. Data base queries are used to generate data files for input into the backcalculation program, and additional rules address the subsequent evaluation of the backcalculation results. It is anticipated that both the application rules and the evaluation rules will be refined as more is learned about the strengths, weaknesses, and requirements of the backcalculation procedure that SHRP has developed.

BACKCALCULATION SOFTWARE
The process by which SHRP pursued the selection of a backcalculation program for use in the LTPP data analysis involved the following steps: 1. Software identification;
1

2. 4. 5.

Preliminary software selection; Software evaluation; and Final software selection.

A brief summary of these is presented next, while a more detailed description of the process is presented in Reference (1). The first three steps in the process outlined above were quite straightforward. Software identification involved a review of the literature to identify a number of the programs available, and their pertinent features. The second step was accomplished through discussions at a meeting of SHRPs LTPP Expert Task Group (ETG) for Deflection Testing and Backcalculation in November, 1990. Based on ETG recommended criteria, six programs were selected for further evaluation. ELCON and ILLI-BACK were selected for rigid pavements, and ISSEM4, MODCOMP3, MODULUS, and WESDEF for flexible pavements. The purpose of SHRPs backcalculation software evaluation exercise was twofold: (1) to provide a basis for selecting a program for use in the SHRP backcalculation; and (2) to provide a basis for development of the procedures to be used with that software. For this endeavor a group composed of ETG members, the software developers, and SHRP contractors was assembled. Backcalculation results were evaluated on the basis of reasonableness, robustness and stability, goodness of fit, and general suitability for SHRPs purposes. Based on the results of the evaluation exercise, it was concluded that MODCOMP3, MODULUS, and WESDEF are useful tools for backcalculation, which can produce good results. The programs, however, were found to have different strengths and weaknesses. MODCOMP3 produced results which match the measured deflection basins quite well, were reasonably independent of the user, and were generally "reasonable". In addition, it was the most flexible of the programs evaluated. MODULUS did a slightly better job of matching the measured deflections basins, was slightly more independent of the user, and also produced results which were generally "reasonable". However, the lower degree of user dependence of MODULUS, as compared to MODCOMP3, comes about as a result of fewer options with respect to modelling o f the pavement structure (i.e., less flexibility). The performance of WESDEF was similar to that of MODCOMP3 (i.e., not quite as good as MODULUS), with respect to the ability to match measured deflection basins. However, the results were somewhat less independent of the user, and were subjectively judged to be slightly less "reasonable" for the sections evaluated. Overall, it was concluded from the results of the evaluation exercise that the performance of MODULUS was somewhat superior to that of the other programs, although one or both of the other programs may be better for an individual section. Thus, MODULUS was selected as the primary backcalculation program to be used in the initial analysis of the SHRP deflection data.

BACKCALCULATION RULES
As indicated earlier, backcalculation is a laborious process, requiring a high degree of skill, and the results are known to be moderately to highly dependent on the individual doing the backcalculation. In order to ensure that backcalculation process applied in the SHRP data analysis is as consistent, productive, and straightforward as possible, a standard procedure (i.e., rigorous set of applications rules) was developed around the MODULUS program. This procedure relies on the wealth of information stored in the LTPP data base deflection, pavement structure and materials, and surface layer temperature data -- to generate the input for MODULUS. In addition, the procedure has been automated to a high degree, thus reducing opportunities for operator error or inconsistency. The SHRP backcalculation rules address three major areas: definition of layer moduli ranges, modeling of the pavement structure, and evaluation of the analysis results. The first group of rules focuses on the definition of the moduli ranges required to run the MODULUS program, the second set o f rules addresses the modeling of the pavement structure for purposes of backcalculation, and the third and final set of rules focuses on the evaluation of the backcalculation results. A step-by-step discussion of these rules is presented next. In addition, new rules or modifications to the existing ones based on preliminary LTPP data analysis results are discussed in a later section.

Definition o f Layer Moduli Ranges


The MODULUS program requires that an estimate of the "expected" range of moduli be specified for each pavement layer, except the subgrade where only an estimate of the initial modulus is required. In the SHRP backcalculation procedure, predictive equations that rely on material property and field temperature data stored in the LTPP data base are used to establish the moduli range for asphaltic concrete (AC) layers - the specific algorithm used depends on the available information. Moduli ranges for portland cement concrete (PCC) layers and other stabilized materials are determined based on available laboratory test results, or assumed. Similarly, moduli ranges for unbound granular base and subbase layers are estimated on the basis of material type. Outer deflection readings and Boussinesqs one-layer deflection equation are used to estimate the initial subgrade modulus.

Asphalt Concrete Layers


The following rules are used to arrive at the modulus range for asphalt concrete layers:

1.

Determine Mid-Depth Temperature of AC Layer(s) Using the surface layer temperature gradient versus time data stored in the LTPP data base, the mid-depth temperature for each AC layer in the pavement structure at the time o f testing is determined (extrapolated or interpolated).

2.

Compute Initial Modulus of AC Layer(s) If mix data aggregate grading, maximum and bulk specific gravity of mix, and asphalt content are available from the LTPP data base, the following equation is used to estimate the initial modulus of AC layers (2): log10[E*] = 2.250053 - 0.091756*%* - 0.027949*V, - 0.096881^00 + 0.250094*p> - 0.006447% + 0.060612*f - 0.00007404%2 + bl 0.00191539*Vbe2 + 0.0082813*p2oo2 - 0.0010225 *pj/42 + 0.0001909W - 0.0801155*plb,2 + 0.0148592*7>7()1(r62 - 0.0024159'f2 + 0.00094015*p3/g*Vbe + 0.00084534*p3,4*Vbe + 0.0004965*p3/4*p4 0.00034328*p3/g*p4 - 0.00316297*p3/8*pib, (1)

where E = AC modulus, in 10s psi; = effective asphalt content, by volume percentage; V, = percent air voids in mix; P200 = percent aggregate weight passing the No. 200 sieve; p ^ = percent asphalt absorption, by weight of aggregate; f = test frequency of load wave, in Hz (assume 16 Hz in all cases); tp = test temperature, in Fahrenheit (from Step No. 1); p4, p3/g, and p3/4 = percent aggregate weight retained in the No. 4, 3/8" and 3/4" sieves, respectively; and tj701(r6 = asphalt viscosity at 70F, in 106 Poises. The effective asphalt content (V^). by volume percentage, is determined by means of the following equation (3): Vbc = [(p* - P.bs - Pbs*P.c/100)*G11 / Gb 1b] (2)

where p*. = percent asphalt content by weight of mix; p ^ = percent asphalt absorption by weight of aggregate; G ^ = maximum specific gravity of mix; and Gb = specific gravity of bitumen. If the specific gravity of the bitumen is not stored in the LTPP data base, a value of 1.010 is assumed. If aggregate (effective and bulk) and bitumen specific gravities are stored in the LTPP data base, the following equation is use to determine the percent asphalt absorption (p^J by weight of aggregate (3): P.bs
=

lOO'tiG* - GJ/(GA*GJ]*Gh

(3)

where G*. = effective specific gravity of aggregate; G^ = bulk specific gravity of aggregate; and, Gb = specific gravity of asphalt. Otherwise, it is assumed that p ^ = 0.5% for crushed stone, gravel and sand mixtures and 1.5% for slag. The percentage of voids in the mix, V,, is determined from the following relationship (2): V. = [100*(Gm -Gm m b)]/Gm m (4)

where G ^ = maximum specific gravity of compacted mix and G ^ = bulk specific gravity of compacted mix. The asphalt viscosity at 70F (^o.kts) can be determined in one of three ways. If measured absolute (140F, in poises) and kinematic (275F, in centistoke) viscosities are stored in the LTPP data base, a logOog(viscosity)) versus log(temperature) correlation is first established and then extrapolated to 70F (2). Figure 1 graphically illustrates the computation of rj10 l(r6 from known viscosity and temperature data. When using this procedure, special care must be taken to ensure that viscosity data have been converted into centipoise and temperatures into degrees Rankine. prior to the development of the correlation. If viscosity data is not available but a penetration value at 77F (Pen?;) is known, the following relationship between asphalt viscosity at 70F and penetration at 77F is recommended (3): 77o.icr6 = 475,300*Pen77'293 (5)

Finally, if the only information known about the asphalt consistency is the general grade, either viscosity or penetration grade, the values shown in Table 1 are used. In the event that asphalt consistency data are not available, viscosity values are assumed on a state-by-state basis; e.g., 2.5*106 poises (AC-20) for the State of Maryland. It has been assumed that a certain minimum amount of data -- aggregate grading, maximum and bulk specific gravity of mix, and asphalt content are available for the computation of the initial modulus for each AC layer in the pavement structure. In those cases where this information is not available from the LTPP data base, the initial modulus is computed using the following equation (3):

VISCOSITY, POISES

TEMPERATURE, F

FIGURE 1.

Typical Viscosity - Temperature Relationships.

Table 1 - Asphalt Viscosity at 70F Based on Grade

Basis for Grade Viscosity

Grade AC-5 AC-10 AC-20 AC-40

Viscosity (70F, 106 poises) 0.3 1.0 2.5 5.0 2.5 1.0 0.5 0.25 0.08 0.3 1.0 2.5 5.0

Penetration

60-70 85-100 100-120 150-200

After Residue (AR)1

10 20 40 80 160

Note: Viscosity values were established for both viscosity and penetration grade asphalts based on recommendations provided in Reference 2; AASHTO M-226-80 correlations were used to establish viscosity values for the AR grades of asphalt.

lg10[E*] =

0.553833 + 0.028829I(tp2oo*f^17 3 - 0.03476*V. + 0.070377^ 701(r6 + 03 0.000005*[tp(13 + O 4*25iog(f))*pic0.5]. o.00189[tp 3 + -49825lo(f))* pK05* f 1'] + 0.931757^ 02774 (6)

where E* = AC modulus, in 105 psi; V, = percent air voids in mix; f = test frequency of load wave, in Hz (assume 16 Hz in all cases); tp = test temperature; in Fahrenheit (from Step No. 1); p ^ = percent aggregate weight passing the No. 200 sieve; p. = percent asphalt content by weight of mix; and, v-ro.icre asphalt viscosity at 70 F, in 106 Poises. If the information contained in the LTPP data base is not sufficient to define one or more of the variables in Equation 6, the following default values are recommended: Percent air voids in mix, V,: 4% for surface courses, 5% for binder courses, and 7% for base courses. Percent asphalt content by weight of mix, p,c: 6% for surface courses, 5% for binder courses, and 4% for base courses; 8% for all sand asphalt mixtures. Percent aggregate weight passing the No. 200 sieve, p ^ : 6% for surface courses, 5% for binder courses, and 4% for base courses; 6% for all sand asphalt mixtures.

The asphalt viscosity at 70F (t70 1(r6) can be determined using any of the three 7 procedures described earlier for the definition of this variable in Equation 1. If none of the required grade information is present in the data base, viscosity values are assumed on a state-by-state basis. Combine AC layers of Same Construction Age In general, backcalculation procedures are unable to handle individual AC construction lifts separately. As a consequence, in the SHRP backcalculation procedure, AC layers having the same construction age are combined into a single layer e.g., binder and surface course for an overlay or original surface layer are combined into one layer. The specific rules for combining AC layers are as follows: Add thicknesses of all AC layers having the same construction age, including any surface treatments: = h(surface treatment) + h(surface course) + h(binder) + ... (7)

hcompost

Find initial composite modulus for the combination of AC layers having the same construction age: . = ^ / h ^ J * E ^ ]3 (8)

^com posite

where hj = thickness of the "i"th layer; Ej = modulus of the "i"th layer (from Step No. 2); and i = 1 to n, where n is the number of AC layers having the same construction age. For example, if during construction, a 2 inch AC surface course (modulus of 1,000,000 psi) is placed over a 3 inch AC binder course (modulus of 500,000 psi), the composite modulus for the combined 5 inch AC layer is 673,000 psi. When surface treatments are present, their thickness should be included in the total surface thickness, but their presence should be ignored when determining the composite modulus value. 4. Define Modulus Range for AC Layer(s) Once the initial or composite modulus of each AC layer has been defined, the range of moduli is determined as follows: Range = 0.25*E(initial or composite) to 3.00*E(initial or composite) The upper limit of the AC layer modulus range defined by the above relationship is not to exceed 3,000,000 psi. (9)

Portland Cement Concrete Layers


The procedure to define the modulus range for portland cement concrete layers is considerably simpler than that for asphalt concrete layers. One reason for this is that more strength tests are being performed on PCC layer materials (static modulus, compressive strength, and splitting tensile strength). Most of this testing has been completed and is now stored in the LTPP data base. The other reason is that PCC moduli are not as temperature dependent as those for AC materials, thus the anticipated range of values can be more easily approximated in the absence of any information. The specific set o f rules used to define the range of moduli for PCC layers is as follows: 1. Determine Initial Modulus of PCC Layer(s) Depending on the type of laboratory strength data available, the initial PCC modulus is determined in the following priority order: If static modulus (E) test results are available, these values are used directly in the definition of the layer moduli range. If static modulus data are not available but compressive strength results are, the following equation is used to determine the initial modulus value (5): E = 57,000 * (fc)-5 (10) 9

Table 2 - Initial Modulus and Moduli Range for Unbound Base and Subbase Materials

M aterial Type Crushed Stone, Gravel or Slag Bases Subbases Gravel or Soil-Agg. Mix, Coarse Bases Subbases Sand Bases Subbases Gravel or Soil-Agg. Mix, Fine Bases Subbases

Initial M odulus (ksi) 50.0 30.0 30.0 20.0 20,0 15.0 20.0 15.0

Moduli Range (ksi)

10.0 to 150.0 10.0 to 100.0 10.0 to 100.0 5.0 to 80.0 5.0 to 80.0 5.0 to 60.0 5.0 to 80.0 5.0 to 60.0

11

For stabilized base and subbase layers, estimates of the initial modulus and range of moduli are based on unconfined compressive strength data, which are generally available from the LTPP data base. The recommended values are summarized in Table 3, according to the stabilizing agent used. If unconfined compressive strength data is lacking, a value o f 400 psi is assumed for lime stabilized layers, 700 psi for asphalt stabilized layers, and 1000 psi for cement stabilized materials for input into Table 3.

Subgrade Layers
The concept used to estimate the initial subgrade modulus from the measured deflections is illustrated in Figure 2, which shows a pavement structure being deflected under a load. As the test is conducted, the load applied to the surface is distributed through the depth of the pavement system. The distribution o f stresses, represented in this figure by the "Zone of Stress", is obviously dependent upon the stiffness or modulus of the material within each layer. As the stiffness of the material increases, the stress is spread over a much larger area. Figure 2 also shows a radial distance (r = a^J in which the stress zone intersects the interface of the subbase and subgrade layers. When the deflection basin is measured, any surface deflection obtained at or beyond the a3e distance are due only to stresses, and hence deformations, within the subgrade itself. Thus, the outer readings of the deflection basin reflect the in-situ modulus of the subgrade soil. Using this concept, the initial subgrade modulus is estimated from the composite moduli predicted for radial distances greater than the effective radius, a3e, of the stress bulb at the pavement-subgrade interface; as indicated by the horizontal dashed line in Figure 3 for linearly elastic subgrades or by the upward trend for non-linear (stress dependent) subgrades. The composite modulus is a single value representation of the overall pavement stiffness, at a given radial distance, that combines the modulus of elasticity of all layers present in the pavement. The specific set of rules used in the SHRP backcalculation procedure involves the following steps (4):

12

Table 3 - Initial Modulus and Moduli Range for Stabilized Base and Subbase Materials

M aterial Type Lime Stabilized

Unconf. Comp. Strength (psi) < 250 250-500 > 500 < 300 300-800 > 800 < 750 750-1250 > 1250

Initial M odulus (ksi) 30.0 50.0 70.0 100.0 150.0 200.0 400.0 1000.0 1500.0 500.0 50.0

Modulus Range (ksi) 5.0 to 100.0 10.0 to 150.0 15.0 to 200.0 10.0 to 300.0 25.0 to 800.0 50.0 to 1500.0 50.0 to 1500.0 100.0 to 3000.0 150.0 to 4000.0 100.0 to 3000.0 10.0 to 150.0

Asphalt Stabilized

Cement Stabilized

Fractured PCC Others

13

FIGURE 2.

Schematic of stress zone within pavement structure under FWD load.

(Composite Pavement Modulus)

FIGURE 3. Composite modulus vs. radial distance plot

15

Calculate the composite modulus of the pavement at each radial distance beyond 5.91 in. (i.e., 8, 12, 18, 24, 36, and 60 in.) using the measured deflection data as input into Boussinesqs one-layer deflection equation: (13) Eicomp) = ------ e j --------------- def * r
Pr * C
* (l

~ U2) * C

where E(comp) = pavement composite modulus; pc = contact pressure applied by FWD, from data base; a,. = load plate radius, from data base (5.91 in.); u = Poissons ratio of subgrade, assume to be 0.4; def = measured deflection at given radial distance "r", from data base; r = radial distance for deflection in question; and C = deflection constant equal to: ( C = l.Uog \
+ 1.15

\ ac,

Assume that the initial subgrade modulus is equal to the minimum composite pavement modulus: E(subgrade) = E ic o m p )^ ^ (15)

Note that the MODULUS program requires only an initial modulus value for the subgrade, not a range of modulus.

Modeling o f Pavement Structure


Along with known layer thicknesses, the layer moduli derived from the SHRP backcalculation rules will provide much of the information required to run the MODULUS program, but not all. Because the MODULUS program is limited to a maximum of 4 unknown layers, prioritized guidelines are required for combining two or more layers in pavement structures with more than 4 layers. Likewise, rules for fixing layer moduli in complex pavement structures, thin asphalt concrete layers, and treated subgrade soils are also required. Another item that must be covered by these rules is the assignment of a Poissons ratio for each pavement layer. The specific set of rules used by the SHRP procedure for modeling of pavement structures in backcalculation analyses is as follows:

16

Subgrade Layers
Lime, asphalt (mixed in place), or cement stabilized subgrade is treated as a subbase layer. If shoulder boring data or other similar information indicates that a rigid layer is present within 20 feet of the surface, then the subgrade thickness is defined in accordance with this information. Otherwise, the MODULUS option to calculate the depth to an effective rigid layer is used; i.e., to look for rigid layer effects at depths of up to 50 feet. If no rigid layer is found within this range, then the depth to rigid layer defaults to 50 feet. When analyzing a three-layer pavement system, the subgrade is modeled as two layers and the thickness of the top subgrade layer is assumed to be equal to 36 inches. This is done to account for possible changes in subgrade modulus with depth due to such factors as the stress sensitivity of the subgrade soil, varying moisture conditions, etc. However, if the total subgrade thickness is less than 72 inches (due to presence of rigid layer) a single subgrade layer is used.

Thin Layers
If the total thickness of the AC layer is less than 3 inches, fix the modulus of this layer equal to that derived from Equation 1 or 6. If a thin layer (<_ 2 inches) exists beneath portland cement concrete, neglect the modulus of this layer and combine its thickness with that of the underlying layer.

Pavement Structure
As indicated earlier, the maximum number of layers (with known or unknown modulus) that can be modeled in the MODULUS program is 4, exclusive of the effective rigid layer. If more than 4 layers are present, the prioritized list of rules given below are used to reduce the number of layers included in the backcalculation analysis. 1. Combine adjacent granular base and subbase layers, if more than one is present and material types are similar (e.g., crushed stone base and crushed gravel subbase not crushed stone base and sand subbase). The total thickness for the composite layer is equal to the sum of the thicknesses for the adjacent layers, while the modulus range is defined by the combined range of the layers (i.e., largest maximum, smallest minimum).

17

2.

Combine adjacent AC layers of different construction dates, if more than one is present (e.g., overlay plus original surface). Use Equations 7, 8 and 9 to determine the total thickness and moduli range for the composite layer. If the total thickness is still less than 3 inches, fix the modulus of this composite layer as that generated from Equation 8. Combine adjacent stabilized base and subbase layers, if more than one is present and material types are similar (e.g., cement stabilized base and subbase, not cement stabilized base and lime stabilized subgrade, which is treated as a subbase). The total thickness and modulus range for the composite layer is determined in the same fashion as in Item 1 above. Combine adjacent granular base and subbase layers, if more than one is present; material types do not have to be similar; e.g., crushed stone base and sand subbase. The total thickness and moduli range for the composite layer is determined in the same fashion as in Item 1 above. Combine adjacent subbase and subgrade layers, if material types are similar; e.g., sand subbase over sandy subgrade. If this done, the initial subgrade modulus should be used in the backcalculation analysis. The thickness of this combined layer will depend on whether or not a rigid layer (actual or effective) exists below the subgrade. Combine adjacent AC and asphalt treated layers, if more than one is present; e.g., original surface plus asphalt treated base. Use Equations 7, 8 and 9 to determine the total thickness and moduli range for the composite layer. Combine adjacent cement-stabilized and lime-stabilized base/subbase layers, if more than one is present; e.g., cement stabilized base and lime stabilized subgrade (treated as subbase). The total thickness and moduli range for the composite layer is determined in the same fashion as in Item 1 above.

3.

4.

5.

6.

7.

Poissons Ratio
Recommendations for assigning Poissons ratios as a function of material type abound in the literature. Based on this information and recommendations by the Deflection Testing and Backcalculation ETG, the values shown in Table 4 have been selected for use in the SHRP backcalculation procedure.

Table 4 - Poissons Ratio as a Function o f Material Type

Material Type Asphalt Concrete E > 500 ksi E < 500 ksi Portland Cement Concrete Stabilized Base/Subbase Lime Cement Asphalt Other (stabilized subgrade) Other (fractured PCC) Granular Base/Subbase Cohesive Subgrade Cohesionless Subgrade

Poissons Ratio 0.30 0.35 0.15 0.20 0.20 0.35 0.35 0.30 0.35 0.45 0.35

19

Evaluation of Analysis Results


The third and final set of rules focus on the evaluation of the backcalculation results. Maximum allowable deflection matching error limits are established, both for the individual sensors as well as all sensors combined. Guidelines for checking the reasonableness of the results are also provided in these rules, along with procedures to be followed in case of bad or questionable data. The specific rules for the evaluation of the results are as follows: All backcalculation results must be carefully reviewed by an engineer familiar with the backcalculation process. If the results fail the convexity test, the range of moduli must be widened (reduce lower bound by 50% and increase upper bound by 100%), and the backcalculation rerun. If the results are similar to those from the first run, they should be accepted whether they pass the convexity test or not. If the results from the second run differ from those from the first run, but pass the convexity test, they should be accepted. Otherwise, they are not considered valid. Results having an average absolute arithmetic error in excess of 2% are not valid. This corresponds to a total sum of absolute error of 14% when all seven sensors are used in the back-calculation; 10% when only five sensors are used, and so on. Predicted moduli which hit the boundaries provided as input into the backcalculation are not considered valid. When the deflection errors fail to meet the 2% tolerance, the modulus results hit an upper or lower bound, or the results are considered "unreasonable" in the judgement of the reviewer, the engineer must look for obvious problems, by verifying the input data, comparing the results with laboratory data, and checking the distress film. In the absence of obvious errors, unacceptable results will be set aside for further evaluation.

Other Considerations
Despite all of the above rules, the evolving nature of the science (or art) of backcalculation makes it likely that early experience with this procedure will bring to light areas where further refinement is needed. Hence, it is anticipated that the initial release of the SHRP backcalculation procedure will be followed up, as we learn more about the strengths, weaknesses, and requirements of the process.

20

While the initial analysis of the LTPP deflection data has not been completed, it is anticipated that new rules will likely be added to the existing SHRP backcalculation procedure and/or that existing ones may be modified. For example, based on preliminary analysis results, it is possible that the following rules will be implemented in the SHRP backcalculation procedure: Using the same rules described earlier in the report, fix the modulus of AC layers having thicknesses of 6 inches or less and constructed on portland cement concrete or other stiff materials (e.g., cement treated bases). In the case of portland cement concrete pavements, combine adjacent unbound base and subbase layers underneath the PCC slab. The total thickness for the composite layer is equal to the sum of the thicknesses for the adjacent layers, while the moduli range is defined by the combined range of the layers. In addition to AC layers, combine other "thin" material layers placed below PCC slabs with other adjacent base, subbase or subgrade layer. As more laboratory modulus test results become available (i.e., stored in the LTPP data base), these data will be used to estimate the initial value and range of moduli for the various material types.

SUMMARY AND CONCLUSIONS


This report focused on a standard backcalculation procedure, developed around the MODULUS program , to ensure that the backcalculation process applied in the LTPP deflection data analysis is as consistent, productive, and straightforward as possible. The procedure consists of a rigorous set of application rules which rely on the wealth of information stored in the LTPP data base to generate the input -- modeling of pavement structure, layer moduli ranges, Poissons ratios, etc. -- for the backcalculation program. In conjunction with data base queries, the SHRP backcalculation rules are used to generate data files for direct input into the backcalculation program. As detailed in the report, these rules are used to model the pavement structure and to establish initial moduli or moduli ranges for use in conjunction with measured deflections and loads in the backcalculation analysis. Additional rules address the subsequent evaluation of the backcalculation results. Despite these rules, the evolving nature of the science (or art) of backcalculation makes it likely that early experience with this procedure will bring to light areas where further refinement is needed. Hence, it is anticipated that the initial release of the SHRP backcalculation procedure will be followed up, as we learn more about the strengths, weaknesses, and requirements of the process. Already, preliminary analysis results seem to

21

indicate that new rules will need to be added to the SHRP backcalculation procedure and/or that some of the existing ones may have to be modified.

REFERENCES
1. Strategic Highway Research Program: Layer Moduli Backcalculation Procedure Software Selection, Strategic Highway Research Program, July 1991. M.W. Witczak, "The Universal Airport Pavement Design System - Report II: Asphaltic Mixture Material Characterization", Department of Civil Engineering, University of Maryland, College Park, Maryland, May 1989. "Mix Design Methods for Asphalt Concrete and Other Hot-Mix Types", Manual Series No. 2, The Asphalt Institute, College Park, Maryland, 1988. Rada, G.R., Witczak, M.W. and Rabinow, S.D., "A Comparison of AASHTO Structural Evaluation Techniques using NDT Deflection Testing", TRB, Transportation Research Record 1207, Washington, D .C ., 1988. ACI Committee 318, "Building Code Requirements for Reinforced Concrete (ACI318-77)", American Concrete Institute, Detroit, Michigan, 1977. Hammitt, G.A., "Concrete Strength Relationships", Miscellaneous Paper S-74-30, U.S. Army Engineer Waterways Experiment Station, Vicksburg, Mississippi, December 1974.

2.

3.

4.

5.

6.

Long-Term Pavement Performance Advisory Committee


C hairm an William J. MacCreery W.J. MacCreery, Inc. David Albright Alliance fo r Transportation Research Richard Barksdale Georgia Institute o f Technology James L. Brown Pavement Consultant Robert L. Clevenger Colorado Department o f Highways Ronald Collins Georgia Department o f Transportation Guy Dore Ministere des Transports de Quebec Charles E. Dougan Connecticut Department o f Transportation McRaney Fulmer South Carolina Department o f Highways and Public Transportation Marlin J. Knutson American Concrete Pavement Association Hans Jorgen Ertman Larsen Danish R oad Institute, Road Directorate Kenneth H. McGhee Consultant Civil Engineer Raymond K. Moore University o f Kansas Richard D. Morgan National Asphalt Pavement Association William R. Moyer Pennsylvania Department o f Transportation David E. Newcomb University o f Minnesota Charles A. Pryor National Stone Association Cesar A.V. Queiroz The World Bank Roland L. Rizenbergs Kentucky Transportation Cabinet Gary K. Robinson Arizona Department o f Transportation Frederic R. Ross Wisconsin Department o f Transportation Ted M. Scott American Trucking Association Marshall R. Thompson University o f Illinois Kenneth R. Wardlaw Exxon Chemical Corporation Marcus Williams H.B. Zachry Company Liaisons Albert J. Bush, III USAE Waterways Experiment Station Louis M. Papet Federal Highway Administration John P. Hallin Federal Highway Administration Ted Ferragut Federal Highway Administration Frank R. McCullagh Transportation Research Board

Expert Task Group


Paul D. Anderson Mountairrview Geotechnical Ltd. Robert C. Briggs Texas Department o f Transportation Albert J. Bush, III USAE Waterways Experimental Station Billy G. Connor Alaska Department o f Transportation William Edwards Ohio Department o f Transportation John P. Hallin Federal Highway Administration Frank L. Holman, Jr. Alabama Highway Department William J. Kenis Federal Highway Administration Joe P. Mahoney University o f Washington Larry A. Scofield Arizona Transportation Research Center Richard N. Stubstad Dynatest Consulting, Inc. Marshall R. Thompson University o f Illinois Per Ullidtz Technical University o f Denmark Jacob Uzan Texas A& M University Wes Yang New York State Department o f Transportation

Appendix G

List of Backcalculation Programs

List of Backcalculation Programs


Program
MODULUS 4.2
Author Contact

Cost
No charge for program. D ocum entation - $20 Also see TRR # 1 2 6 0 pp 180-191. N o charge.

T exas Transportation Institute T exas A&M University College Station, TX 77843

S andy Tucker (409) 845-1636 Tom Scullion (Technical Assistance) (409) 845-9913 Lynne Irwin (607) 255-8033

M ODCOM P3

Cornell University Local R oads Program 4 1 6 Riley-Robb Hall Ithaca, NY 14S53 T ransportation Research Institute M enyfield Hall 100 O regon S tate University Corvallis, O R 97331-4304 W ashington S tate DOT M aterials L aboratory P.O . B ox 167 O lym pia, WA 98507-0167

BOUSDEF

Chris Bell (503) 737-4981

N o charge.

EVERCALC

Linda Pierce (206) 753-4661

N o charge for program. T here will be a charge for docum entation but unknow n at this time.

SHRP Calibration Programs

FWDREFCL FWDCAL

Federal Highway Administration Turner-Fairbanks Highway R esearch C enter - HNR40 6 3 0 0 G eorgetow n Pike M cLean, VA 22101

Cheryl Richter (703) 285-2183

No charge.

Other Programs

BISAR

Shell Oil, H ouston, Texas

Ken Spalding (713) 241-6903 Ken Spalding (713) 241-6903

$300

SH ELL PAVEMENT DESIGN MANUAL

Shell Oil, H ouston, Texas

$600

U.S. GOVERNMENT PRINTING OFFICE: 1994-301-719/15738

M 3(00W fr6-Z/IZ0-i76-IH-VM HJ

Anda mungkin juga menyukai