Anda di halaman 1dari 7

Bearing Capacity of Roads, Railways and Airfields – Loizos et al.

(Eds)
© 2017 Taylor & Francis Group, London, ISBN 978-1-138-29595-7

Evaluation of a resilient modulus model for unsaturated soil conditions

P. Thirthar Palanivelu & C.E. Zapata


Arizona State University, Tempe, Arizona, USA

ABSTRACT:  The strength and the durability of pavement systems are directly related to the stiffness
of the unbound and subgrade soil materials. The 1993  AASHTO flexible pavement design guide and
more recently, the AASHTO ware software for the mechanistic-empirical design of pavements relies on
the resilient modulus as the primary input parameter to characterize the stiffness of the subgrade and
unbound base materials. Models that describe the modulus as a function of externally imposed stresses
has been adopted in design procedures, but models that consider the internal stress state of the material
at all levels of saturation have not been evaluated due to the lack of data available. This paper presents
an evaluation of the Cary and Zapata (2010) model parameters, which includes the matric suction stress
state of the material as a driver of the moisture soil retention capability. The evaluation was performed
with data collected from resilient modulus testing on six different soil types at different moisture contents.
Laboratory test performed include the soil-water characteristic curve, filter paper method and a compre-
hensive set of index properties. The resilient modulus tests were performed on soil specimens in accord-
ance with the NCHRP 1-28 A protocol at different compaction conditions.

1  INTRODUCTION approach. Incorporation of moisture content into


the design approach emphasizes the role of envi-
In mechanistic pavement design procedures, the ronmental factors into the pavement design guide
stiffness of pavement layers is characterized by procedures (Lekarp 2000). In the early to mid-2000
the elastic modulus and the Poisson’s Ratio. For period, major advances allowed for the incorpora-
unbound layer materials, this modulus is typi- tion of unsaturated soil principles into a pavement
cally referred to as the “resilient modulus”, which design methodology in United States in a more
reflects only the true elastic or recoverable portion mechanistic fashion. This was accomplished with
of the deformation phase. The variability of the the development of the American Association
resilient modulus depends on different factors such of State Highway and Transportation Officials
as moisture content, dry density, number of load (AASHTO) Mechanistic-Empirical Pavement
repetitions, the compaction energy involved, the Design Guide (M-EPDG).
compaction method used, and the state of stress Recently, several experimental studies have
in the field (Andrei 2003). Historically, pavement been performed to analyze the influence of mois-
design has been based, for the most part, on the ture content on the resilient response of the pave-
selection of material design properties that are at ment unbound material (Doucet 2004, Cary 2010,
the “worst possible environmental conditions”, Andrei 2003, Rahman 2012). Various models have
regardless of where the site is located. Thus, all been developed using these studies to predict the
unbound materials (base, subbase and subgrades) moisture content parameter and its variation due
are classically selected at fully saturated moisture to seasonal effects. Results from previous studies
conditions. However, in the past 30 years, there has shown that the values of MR in soils decrease
have been significant advancements in the field with an increasing in moisture content and increase
of unsaturated soil mechanics, that has allowed with a decrease in moisture content; which in turns
assessing pavement performance predictions using means that the seasonal variation of post compac-
real time environmental, seasonal and aging effects tion moisture content in base and subgrade soil
to predict actual changes in the unbound mate- materials has a determinant influence in the overall
rial strength and resilient behavior of the materi- structural performance of pavement systems.
als. One of the major factors playing a significant Unsaturated soil mechanics allow us to explain
role in the mechanical response of the unbound that the variations in the moisture content of
pavement materials is the moisture content of the fine‑grained subgrades due to seasonal variations
soil. Hence, proper consideration should be given can result in changes in the stress state of the soil.
to this factor in any sustainable pavement design This change can be explained in terms of changes

3
in the soil matric suction, which is the difference Cary and Zapata (2010) proposed a model that
between the pore-air and the pore-water pres- incorporates the influence of seasonal environ-
sures in the soil matrix. Therefore, a more rational mental variations by including the matric suction
approach to incorporate seasonal effects in the as stress state variable into the model developed by
resilient modulus predictive models would be to Witczak and Uzan. They studied the effects of the
consider matric suction as a fundamental vari- moisture content in the soil in terms of suction and
able in the stress state of the subgrade soil, rather obtained sets of regression constants from triaxial
than considering moisture content as a driving experiments on granular base and subgrade mate-
mechanism. rials. The proposed model is shown in Eqn (1):
The overall objective of this study is to inves-
tigate the influence of moisture content on the   2
k k3
mechanical response of the unbound pavement θ − 3∆uw − sat 
 
τ oct + 1
M R = k p net 
1 a 

 pa 
materials using the model developed by Cary and  pa 

Zapata (2010) and the evaluation of the Cary and  (ψ


k
m0 − ∆ψ m )
 4
Zapata parameters, which includes the suction 
+ 1 (1)

 p 
stress of the material that represents the saturation a

state.
In this study, data from six different subgrade where pa = atmospheric pressure; θ net = θ − 3ua = the
soil were used in the analysis. The six soil materi- net bulk stress (θ = bulk stress = σ 1 + σ 2 + σ 3 and ua
als were tested at the Arizona State University = pore‑air pressure); ∆uw − sat = pore‑water pressure
(ASU) Geotechnical Laboratory using a custom build up under saturated condition; τoct = octahedral
built advanced triaxial cells and control units. The shear stress = 13 (σ 1 − σ 2 )2 + (σ 1 − σ 3 )2 + (σ 2 − σ 3 )2 ;
system enables full control/measurement of pore- ψ m0 = initial matric soil suction; ∆ψm = relative
water and pore-air pressure of the cylindrical test change in soil matric suction with respect to ψ m0
specimen during the Repeated Load Triaxial (RLT) due to pore-water pressure build up under unsatu-
test. The load sequences used for this purpose were rated condition; k1 ≥ 0, k2 ≥ 0, k3 ≤ 0 and k4 ≥ 0
in accordance to the NCHRP 1-28  A (2004) pro- are regression constants.
tocol “Harmonized Test Methods for Laboratory The Cary and Zapata model was validated and
Determination of Resilient Modulus for Flex- the regression constants were determined and cali-
ible Pavement design” after some modifications to brated for all six soils.
the procedure that would allow for testing soil in
unsaturated conditions with matric suction control.
More details on the modification to the NCHRP 3  SUBGRADE SOIL PROPERTIES
1-28 A loading procedure and triaxial cell control
unit can be found in Cary and Zapata (2010). 3.1  Soil index properties
Six soils were used in this study. Data for two of
the soils used for this study was collected and ana-
2  MODELING OF SOIL RESILIENT lyzed by Cary (2010). These soils were named IOSG
MODULUS and GESG, after the location of the soils found in
Arizona, USA. The other two soils were obtained
Resilient modulus of the soil is widely used to from two different sites in Sweden named Torpsbruk
characterize the mechanical stiffness of the pave- and Luleå. Data for these two soils was obtained and
ment unbound materials and stiffness of unbound analyzed by Salour (Salour 2014). The remaining two
pavement materials. The resilient modulus of the soils (County and Dupont clays) were obtained from
soil is measured under the conditions representa- the Federal Aviation Administration (FAA) National
tive of the external stress state and the environment Airfield Pavement Test Facility located in New Jersey,
conditions experienced by the unbound materials USA. All the six soils were analyzed in ASU labora-
in the pavement system. Various researches have tory. Wet sieve analysis, specific gravity, maximum dry
been studied the influence of different parameters density, optimum moisture content and Soil‑Water
and to develop mathematical model to predict the Characteristic Curve (SWCC) tests were performed
stress dependence of the resilient modulus of the on the selected materials. A summary of the soil index
soil using a total stress approach. A model devel- properties is presented in Table 1. The test results of
oped by Witczak and Uzan (1992), also known as Proctor Compaction tests are presented in Table 2.
the Universal model, clearly explain the influence
of the externally applied stress parameters on the
3.2  Soil Water Characteristic Curve (SWCC)
resilient modulus, but it does not directly account
for the influence of the environmental factors into The SWCC for the six subgrade materials were
the stress sate conditions. determined using the Fredlund SWCC device, an

4
Table 1.  Summary of the subgrade soil index properties.

Subgrade USCS Passing Specific Plasticity


soil class* no.200 (%) gravity index

Luleå SM 42.2 2.68 NP


Torpsbruk SM 27.4 2.67 NP
IOSG SC 37.5 2.72   7.2
GESG SM 47.7 2.72   3.3
County ML 58.0 2.84 17
Dupont MH 80.0 2.79 29

*USCS—Unified Soil Classification System.

Table 2.  Proctor compaction test results. Figure 2.  SWCC for County and Dupont soils.

Max. dry Opt.


Subgrade unit weight moisture
soil (kN/m³)* content. (%)

Luleå 19.6 10.1


Torpsbruk 20.3   7.6
IOSG 18.7 12.1
GESG 19.0 12.5
County 16.3 21.0
Dupont 14.8 26.3

*Maximum dry unit weight is based on standard Proctor


compaction test (ASTM D698).

Figure 3.  SWCC for IOSB soils (Cary 2010).

4  RESILIENT MODULUS TESTING

4.1  Equipment and system set-up


The triaxial system used to perform the resilient
modulus testing of the soil for this testing is a
custom-built system, which is capable of applying
Figure 1.  SWCC for Lulea and Torpsbruk soils (Salour, repeated cyclic load of a haversine-shaped load
2014). pulse. The unsaturated soil testing system is an
electro-hydraulic system with closed-loop digital
oedometer-type apparatus for applying matric suc- servo control. A direct measurement or control
tions from near zero values up to 1500 kPa under of the pore water pressure (uw) and pore air pres-
various stress paths. Saturated ceramic stones sure (ua) can be achieved in this system. The pore
with variable High Air Entry Values (HAEV), air pressure (ua) is applied through the top platen
were used for this test. The soils specimens were though a porous stone. The bottom plate consists
compacted in the brass ring of the apparatus to of a HAEV ceramic disc similar to the one used
reach the target dry density based on the standard in the SWCC device. The pore water pressure (uw)
Proctor test results, and then they were completely is applied through the bottom platen. The HAEV
saturated. The drying branch of the SWCCs was disc restricts the air and only allows the water to
obtained from this test as air pressure was applied pass through the disc, provided the water pressure
to a saturated specimen. For each soil, the equilib- does not exceeds the HAEV.
rium moisture contents at different matric suctions All the components mentioned above can be
were measured; and the SWCC for the whole range seen in Figure 4. GCTS CATS v1.6 is the software
of suction values was predicted. The results can be available for the operation of the testing systems
seen in Figures 1 to 3 for the Lulea and Torpsbruk used in this study. The software allows a testing
soils, the County and Dupont clays, and the IOSB load sequence according to the NCHRP 1–28  A
soil, respectively. protocol, which was previously modified Cary and

5
Table  3.  Target degrees of saturation, moisture con-
tents and matric suctions for resilient modulus testing of
the subgrade soils.

Degree of Moisture Matric


Subgrade saturation, content, suction,
soil S (%) w (%) ψ m0 (kPa)
Luleå 30.0   3.8 443.9
50.0   6.4   81.0
79.3 10.1   35.0
94.3 12.0   6.7
Torpsbruk 30.0   3.6 316.0
50.0   5.4   51.6
70.1   7.6   23.3
97.1 10.4   0.0
GESG 71.0 10.5 236.0
84.0 12.2   52.0
98.0 14.2   1.0
IOSB 78.0 12.1 157
Figure 4.  Equipment used to conduct resilient modulus 80.0 12.4   30
test. 83.0 13.0   0.0
County 79.1 22.3 343.3
Zapata (2010) to allow the testing of unsaturated 82.3 25.3 163.3
soil conditions. A double-layered glass chamber 83.6 26.5   70.0
was used for confinement and air was used as Dupont 82.3 26.0 333.3
the confining material in this study. The vertical 85.5 30.9 236.6
88.4 32.3 160.0
deformations on the test specimen were measured
using two Linear Variable Differential Transform-
ers (LVDTs). The LVDTs are mounted directly to retain the homogeneity in moisture distribution.
onto studs buried during the preparation of the For the County and Dupont soils, the soil was
soil specimen. A schematic overview of the triaxial compacted to the target moisture content and dry
system is depicted in Figure 4. density directly and the preconditioning process
was not performed.
After the specimens were at equilibrium, they
4.2  Resilient modulus triaxial testing
were mounted on the triaxial cell pedestal. The air
The resilient modulus soil specimens were pre- entrapped under the ceramic stone and connec-
pared in accordance to the NCHRP1-28  A test tion lines was first flushed out. The confinement
protocol. The soil samples were thoroughly mixed chamber was assembled and a confining pressure
at the required moisture content and stored in a was applied. Then the target matric suction of the
sealed container in a controlled temperature room soils was imposed. Under the applied matric suc-
for approximately 48  hours in order to obtain a tion, the specimen absorbed or released water until
homogenous mix. The materials were then com- moisture equilibrium was reached. This was the
pacted to the targeted dry density based upon the last phase of the moisture equilibration process.
standard Proctor test results. These specimens were Once the equilibration process was completed, the
cylindrical with 102 mm in diameter and 203 mm resilient modulus test was started.
in height (4 × 8 inches). The soils were tested at dif-
ferent moisture contents as shown in Table 3.
The target moisture content of the soil to run 5  TEST RESULTS AND MODEL
the resilient modulus tests was achieved in the PARAMETER ANALYSIS
initial preconditioning of the soil. The specimens
were either saturated or dehydrated to the target The Resilient modulus tests on the soil samples were
moisture content. This was controlled by regular performed according to the NCHRP 1-28  A test
measurements of the weight of the specimens. protocol. The test results showed that an increase
Once the specimen was near to the target moisture in the bulk stress resulted in increase in the resilient
content, the latex membrane was assembled and modulus. Also, an increase in the moisture content,
the specimen was stored in a moisture‑sealed bag or decrease in suction, resulted in decrease in the
for about 24 hours. During this period, the speci- resilient modulus. It was also observed that at lower
men was intermittently turned over upside down degrees of saturation, the resilient modulus of the

6
subgrade increased at higher rate with increase in the Table  4.  Resilient modulus model regression
bulk stress. The subgrade materials generally exhib- parameters for the subgrade soils.
ited a softening behavior with respect to increase
in the deviator stress. For a constant confinement Model Goodness
parameters of fit
stress, an increase in the deviator stress resulted in Subgrade
decrease in the resilient modulus. However, the sof- soil k1 k2 k3 k4 R2
tening behavior of the materials diminished as the
moisture content was increased. At high degrees Luleå 1489 0.746 -2.633   0.434 0.761
of saturation, the magnitude of the deviator stress Torpsbruk 1524 0.774 -1.470   0.475 0.784
had no significant influence on the resilient modulus IOSG 1480.3 0.420 -2.932   1.650 0.534
results. GESG   993.8 0.288 -1.384 12.711 0.454
The regression parameters for the Cary and County 2402.7 0.136 -5.074   0.665 0.714
Zapata model presented in Eq. (1) were calculated by Dupont   332.5 0.220 -3.140   2.044 0.798
the least square curve fitting on the resilient modu-
lus test data. In total, 123 data points for the Luleå,
150 data points for the Torpsbruk subgrade, 162 data
points for GESG soil, 108 data points for ISOB soil,
115 data points for County soil and 128 data points
for Dupont soil, that correspond to different combi-
nations of stress invariants and matric suctions, were
used in the statistical analysis. The Solver function in
Microsoft Excel was used for the parametric optimi-
zation. The coefficient of determination (R2) was cal-
culated to evaluate the goodness of fit of the model
used. The calibrated model parameters and the
goodness of fit statistics are summarized in Table 4.
Figure 5 to 9 shows the measured versus predicted
results for Lulea, Torpsbruk, IOSG, GESG, County
Figure 5.  Measured and predicted resilient modulus for
and Dupont clay respectively. GESG soil (after Cary 2010).
In Table 4, it can be seen that the parameters of
k1, k2, k3 and k4 varies by soil type. It was noted
that the k4 parameter increases with an increase in
the plasticty of the soil with Dupont clay having
the highest value of 2.0445. The k4 parameter is
influenced by the suction of the soil which can be
in turn related to the PI or P200 of the soil. Also, it
can be noticed that the value of k2, which is related
to the influence of the bulk stress on the magnitude
of the resilient modulus is lower for the soils with
higher plasticity. Both County and Dupont clay are
plastic soils which has a low k2 parameter. In fine
grained soils, the octahedral stress state parameter
plays a dominant role on the resilient modulus of
the soil and in a granular base soil the influence
of the bulk stress more. Considering that the both
County and Dupont are fine grained soils, the
value of k2 parameter was found to be low when
compared with the other soils. Figure 10 shows the
influence of P200% on the k2 and k4 parameters of
the model. Figure 6.  Measured and predicted resilient modulus for
The R2 value obtained from the regression analy- Torpsbruk soil (Salour, 2014).
sis are 0.76 and 0.78 for Luleå and Torpsbruk sub-
grades, respectively. This is considered to be good
for unbound subgrade materials which generally the proposed Cary and Zapata model that includes
show high data scatters in experimental tests such matric suction as a fundamental stress variable
as the resilient modulus. From the graphs plotted performs well for all kinds of soil. The model pro-
in Figure 6 and given the goodness of fit param- vides a very good fit for both granular and fine
eters presented in Table 4, it can be concluded that grained soils.

7
Figure 7.  Measured and predicted resilient modulus for
Luleå soil (Salour, 2014).

Figure 10.  Influence of P200 on k2 and k4 parameters.

6  CONCLUSIONS

Changes in suction of unbound materials is identi-


fied as one of the major driving factors for sea-
sonal changes in the resilient modulus. The matric
suction, that can be considered an independent
stress state variable, directly affects the soil mois-
ture content, which in turn affects the resilient
modulus prediction of subgrade soil materials.
Results from several resilient modulus tests, con-
ducted on six different subgrade soils and at various
degrees of saturation, were presented. The test data
Figure 8.  Measured and predicted resilient modulus for
were used to determine regression parameters of the
Dupont soil. Cary and Zapata model that takes into account mat-
ric suction as a stress state variable. The prediction
model showed a relatively good correlation with the
experimental data, for soils with low plasticity but
deteriorated for soils with high plasticity. The bulk
stress state and suction stress state parameters were
found to be related to the P200 value of the soil.
Given the goodness of fit of the prediction model, it
can be concluded that considering matric suction as
a stress state variable might be a rational approach,
but a more extensive database that includes high
plastic soils is necessary to validate the approach
and improve the model presented in this study.

ACKNOWLEDGMENTS

This study was partially supported by Federal Avi-


Figure 9.  Measured and predicted resilient modulus for ation Administration (FAA). The authors would
County Soil. like to acknowledge the contribution of the data

8
obtained by Dr. Salour and Dr. Cary to the conclu- Tests. Transportation Research Record: Journal of
sions of this study. the Transportation Research Board, No. 2335, Trans-
portation Research Board of the National Academies,
pp. 121–129.
Salour, F., Erlingsson. S. & Zapata, C. 2014. Resilient
REFERENCES Modulus Modeling of Unsaturated Subgrade Soils
with Matric Suction 12th International Conference on
Andrei, D. 2003. Development of a Predictive Model for Asphalt Pavements, ISAP 2014 - Raleigh, NC, United
the Resilient Modulus of Unbound Materials, Doc- States, volume number 2, 1145–1154.
toral Dissertation, Arizona State University, Tempe, Santha, B. L. 1994. Resilient Modulus of Subgrade Soils:
Arizona. Comparison of Two Constitutive Equations, Trans-
ARA, Inc., 2004. ERES Consultants Division. Guide portation Research Record: Journal of the Trans-
for Mechanistic–Empirical Design of New and Reha- portation Research Board, Vol. 1462, Transportation
bilitated Pavement Structures. Final report, NCHRP Research Board of the National Academies, pp. 79–90.
Project 1-37 A. Transportation Research Board of the SWC-150 Fredlund Soil Water Characteristic Device,
National Academies, Washington, D.C. (Available at: 2007. User’s Guide and Reference Manual, Version
http:// www.trb.org/mepdg/) 1.1, GCTS Testing Systems. (Available at: www.gcts.
Cary, C. E. & Zapata, C. E. 2010. Enhancement of the com).
Model for Resilient Response of Soils due to Sea- Uzan. J, Witczak M.W., Scullion. T, & Lytton R. L. 1992.
sonal Environmental Changes Implemented in the Development and validation of realistic pavement
M-EPDG. Transportation Research Record: Journal response models. Proceedings of the 7th International
of the Transportation Research Board. Vol. 2170, Conference on Asphalt Pavements, Nottingham.
pp. 36–44. International Society for Asphalt Pavements (ISAP),
Cary, C. E. & Zapata, C. E. 2011. Resilient Modulus Philadelphia, PA, USA, vol. 1, pp. 334–350.
Testing for Unsaturated Unbound Materials, Inter- Witczak, M.W., & C. E. Zapata. 2000. Development
national Journal of Road Materials and Pavement of Bulk and Octahedral Shear Stress Relationships
Design, Vol. 12, No. 3, pp. 615–638. between Equivalent Pavement Depths. Development
Doucet, F. & Doré, G. 2004. Resilient Modulus and of the 2002 Guide for the Development of New and
Resilient Poisson Coefficient of the C‑LTTP Granu- Rehabilitated Pavement Structures, NCHRP 1-37  A,
lar Materials. Proceedings of the Annual Conference Inter Team Technical Report, Arizona State Univer-
of the Canadian Geotechnical Society, Canadian sity, Tempe, Arizona.
Geotechnical Society, Quebec City, Canada. Witczak M.W., X. Qi, & Mirza. W.M. 1995. Use of Non-
Khoury, N. N. & Zaman, M. 2004. Correlation among linear Subgrade Modulus in AASHTO Design Pro-
Resilient Modulus, Moisture Variation and Soil Suc- cedure. ASCE Journal of Transportation Engineering,
tion for Subgrade Soils, Transportation Research Vol. 123, No. 3, 1995, pp. 273–282.
Record: Journal of the Transportation Research Witczak M.W., W.N. Houston, & Andrei, D. 2000. Resil-
Board, Vol. 1874, Transportation Research Board of ient Modulus as Function of Soil Moisture—A Study
the National Academies, pp. 99–107. of the Expected Changes in Resilient Modulus of the
Lekarp, F., Isacsson, U. & Dawson, A. 2000. State of Unbound Layers with Changes in Moisture for 10 LTPP
the Art. I: Resilient Response of Unbound Aggre- Sites. Development of the 2002 Guide for the Devel-
gates. Journal of Transportation Engineering, 126(1), opment of New and Rehabilitated Pavement Struc-
pp. 66–75. tures, NCHRP 1-37  A, Inter Team Technical Report
National Cooperative Highway Research Program, 2004. (Seasonal 2). Arizona State University, Tempe, Arizona.
Laboratory Determination of Resilient Modulus Yang, S. R., Huang, W. H. & Tai, Y. T. 2005. Variation of
for Flexible Pavement Design, In Research Results Resilient Modulus with Soil Suction for Compacted
Digest, No. 285, Transportation Research Board of Subgrade Soils. Transportation Research Record:
the National Academies, Washington, D.C., 52 p. Journal of the Transportation Research Board, Vol.
Rahman, M. S. & Erlingsson, S. 2012. Moisture Sensitiv- 1913, Transportation Research Board of the National
ity of Unbound Granular Materials. Proceedings of Academies, pp. 99–106.
the 4th European Pavement and Asset Management Zapata, C. E., Andrei, D., Witczak, M. W. & Houston,
Conference (EPAM 4). CD‑ROM., Malmö, Sweden. W. N. 2007. Incorporation of Environmental Effects in
Salour, F. & Erlingsson, S. 2013. Moisture Sensitive and Pavement Design, International Journal of Road Mate-
Stress Dependent Behavior of Pavement Unbound rials and Pavement Design, Vol. 8, No 4, pp. 667–693.
Materials from Insitu Falling Weight Deflectometer

Anda mungkin juga menyukai