Tenth U.S.

National Conference on Earthquake Engineering

Frontiers of Earthquake Engineering
July 21-25, 2014
10NCEE Anchorage, Alaska

COMPREHENSIVE EVALUATION OF
POST-TENSIONING ANCHORAGE
SYSTEMS FOR SEISMIC RESILIENT
ROCKING WALL STRUCTURES
E.S. Musselman1, M. Fournier2, D. Abramson2, and S. Sritharan3

ABSTRACT

The motivation to develop seismic resilient structures stems from the substantial economic losses
that can be caused by earthquake damage. One method of achieving seismic resiliency is to
design buildings with self-centering structural systems capable of resisting lateral earthquake
loads, dissipating energy, and minimizing significant structural damage. The growing popularity
of unbonded post-tensioned (PT) precast concrete rocking walls designed for seismic resiliency
in conjunction with recent research indicating the inability of PT anchorage systems to meet
current industry certification standards has prompted the need to further test and better
understand the behavior of PT anchorages in these systems. To investigate these issues research
was conducted to evaluate the fracture and ultimate strength of two types of systems: single
strand and multistrand post tensioned anchorages. The testing program encompasses two
anchorage manufactures, half inch and 0.6 inch strand, and two wedge geometries under both
monotonic and cyclic loading. In addition, couplers were tested in the single strand system and
anchor head misalignment was evaluated for the multistrand system. Results from the research
indicate that for both single strand and multi-strand systems, high intensity-short duration loads
do not adversely affect the performance of the system. However, even under monotonic loading
a significant number of the systems do not meet published standards. Therefore, a modified
wedge geometry was evaluated, and showed significant improvement in the ultimate strength of
the system.

1
Assistant Professor, Dept. of Civil and Environmental Engineering, Villanova University, Villanova, PA 19085
2
Graduate Student Researcher, Dept. of Civil Engineering, University of Minnesota Duluth, Duluth, MN 16823
3
Wilson Engineering Professor, Dept. of Civil, Construction & Environmental Engineering, Iowa State University,
Ames, IA 50011

Musselman ES, Fournier M, Abramson D, Sritharan S. Comprehensive evaluation of posttensioning anchorage

systems for seismic resilient rocking wall structures. Proceedings of the 10th National Conference in Earthquake
Engineering, Earthquake Engineering Research Institute, Anchorage, AK, 2014.
 Tenth U.S. National Conference on Earthquake Engineering
Frontiers of Earthquake Engineering
July 21-25, 2014
10NCEE Anchorage, Alaska

COMPREHENSIVE EVALUATION OF POST-TENSIONING

ANCHORAGE SYSTEMS FOR SEISMIC RESILIENT ROCKING
WALL STRUCTURES
E.S. Musselman1, M. Fournier2, D. Abramson2, and S. Sritharan3

ABSTRACT

The motivation to develop seismic resilient structures stems from the substantial economic losses
that can be caused by earthquake damage. One method of achieving seismic resiliency is to design
buildings with self-centering structural systems capable of resisting lateral earthquake loads,
dissipating energy, and minimizing significant structural damage. The growing popularity of
unbonded post-tensioned (PT) precast concrete rocking walls designed for seismic resiliency in
conjunction with recent research indicating the inability of PT anchorage systems to meet current
industry certification standards has prompted the need to further test and better understand the
behavior of PT anchorages in these systems. To investigate these issues research was conducted
to evaluate the fracture and ultimate strength of two types of systems: single strand and
multistrand post tensioned anchorages. The testing program encompasses two anchorage
manufactures, half inch and 0.6 inch strand, and two wedge geometries under both monotonic and
cyclic loading. In addition, couplers were tested in the single strand system and anchor head
misalignment was evaluated for the multistrand system. Results from the research indicate that for
both single strand and multi-strand systems, high intensity-short duration loads do not adversely
affect the performance of the system. However, even under monotonic loading a significant
number of the systems do not meet published standards. Therefore, a modified wedge geometry
was evaluated, and showed significant improvement in the ultimate strength of the system.

Introduction

The growing popularity of unbonded post-tensioned (PT) precast concrete rocking walls
designed for seismic resiliency in conjunction with recent research indicating the inability of PT
anchorage systems to meet current industry certification standards has prompted the need to
further test and better understand the behavior of PT anchorage systems. Recent research has
focused on single strand PT anchorages primarily under monotonic loads in an attempt to better
understand the failure mechanisms and ultimate stress and strain capacities of these systems.
However, virtually no published literature is available regarding the behavior of these systems
under the short duration, high intensity cyclical loads to which the systems would be exposed in
a rocking wall. In addition, virtually no published literature is available regarding the behavior
of multistrand anchorages under either monotonic or cyclical loading due to the Post-Tensioning
Institutes recommendation to test only one strand at a time in multistrand anchorages [1].

1
Assistant Professor, Dept. of Civil and Environmental Engineering, Villanova University, Villanova, PA 19085
2
Graduate Student Researcher, Dept. of Civil Engineering, University of Minnesota Duluth, Duluth, MN 16823
3
Wilson Engineering Professor, Dept. of Civil, Construction & Environmental Engineering, Iowa State University,
Ames, IA 50011

Musselman ES, Fournier M, Abramson D, Sritharan S. Comprehensive evaluation of posttensioning anchorage

systems for seismic resilient rocking wall structures. Proceedings of the 10th National Conference in Earthquake
Engineering, Earthquake Engineering Research Institute, Anchorage, AK, 2014.
Furthermore, results from the anchorage validation tests, conducted by anchor manufacturers, are
typically not available for public access. The lack of information in these critical areas calls for a
comprehensive evaluation of post-tensioned anchorage systems under both monotonic and short
duration, high intensity cyclical loads.

Background Information

Post-Tensioned Rocking Wall Systems

Prompted by the Chilean earthquakes of May, 1960, the study of structures being controlled by a
rocking mode was first investigated through the free vibration of rigid rocking blocks [2]. Based
largely on the principles quantified by Housner, since the 1990s several structural rocking wall
systems have been developed for the purpose of dissipating energy and minimizing damage
during seismic events [3-6]. In each of the structural rocking systems investigated, unbonded
post-tensioning was used to join precast concrete members to achieve the lateral resistance
required under seismic loads as well to provide self-centering capability. Additional benefits of
unbonded PT over monolithic cast-in-place reinforced concrete in this application include better
quality control, construction efficiency, and an ability to undergo large nonlinear lateral
displacements without sustaining significant damage [7].
From 1989 to 1999, researchers from the University of California at San Diego worked
on the Precast Seismic Structural Systems (PRESSS) research program. The testing program
culminated with a large-scale five-story precast concrete building constructed to 60 percent scale
being tested under simulated seismic loading conditions. The building included several seismic
resilient components, including an unbonded post-tensioned jointed rocking wall system. The
primary objectives of this program were: (1) to develop comprehensive and rational design
recommendations needed for a broader acceptance in seismic zones; and (2) to develop new
materials andd technologies for precast concrete construction in different seismic zones [4].
While the PRESSS buildings performance was touted as extremely satisfactory, several
deficiencies of the jointed rocking wall system were identified through this testing program that
limited its implementation in practice. Reduced moment capacity and expensive stainless steel
energy dissipating shear connectors made the cost of the jointed wall system prohibitive in
comparison to monolithic concrete wall construction [8].
Another self-centering wall system concept is the hybrid wall [9] [10]. This system
combines a single post-tensioned precast concrete wall with mild steel energy dissipating bars
placed across the wall-foundation interface. The hybrid wall system addressed the reduced
moment capacity deficiency while still providing sufficient energy dissipation. However, a new
resiliency deficiency arose with the use of mild steel reinforcing bars as an energy dissipation
element. The bars are likely to undergo low-cycle fatigue fracture during a large seismic event
and cannot be cost-effectively replaced [8].
To address the deficiencies identified above, a new structural wall system was developed
at Iowa State University [11]. The system consists of a Precast Wall with two End Columns
(PreWEC). It utilizes inexpensive, easily replaceable mild steel oval-shaped flexural plate
connectors (O-connectors) to dissipate energy and lightly prestressed unbonded post-tensioning
to self-center after a seismic event [12].
The system consists of a single precast concrete wall with two end columns, all with
unbonded post-tensioning. The end columns can be made up of steel columns, concrete filled
steel tubes, or precast concrete columns. Walls and columns are anchored to the foundation via
unbonded post-tensioning and joined together with energy dissipating O-connectors. Unbonded
post-tensioning strands or rods can be placed evenly along the wall or centered within the wall.
Using unbonded PT ensures a constant strain distribution over the length of the strand and
prevents localized strains at critical sections compared to bonded systems [8]. This allows the
wall to withstand significant levels of drift without sustaining substantial structural damage.
Conceptually, the PreWEC system allows the wall and end columns to rock
independently during a seismic event. Unlike most post-tensioning applications, the tendons will
only be initially stressed to about fifty percent of ultimate design strength (fpu) instead of the
typical initial stress of about seventy percent of fpu. This reduction in initial stressing is
implemented so that the tendons will remain elastic for the duration of the design seismic forces.
The energy absorbed within the tendons is transferred back into the wall system and, ultimately,
energy dissipating O-connectors.
Following a seismic event, the wall will re-center due to the restoring force of the
elongated PT strands. The primary source of hysteretic damping in the system is forced to the
mild steel O-connectors as they transfer energy laterally between precast segments and dissipate
energy through inelastic flexural yielding. After a significant seismic event, these connectors can
be easily and economically replaced. Additionally, the PreWEC system can be designed to
obtain a moment capacity comparable to that of a monolithic reinforced concrete wall ensuring a
fully resilient system [7][8].

Previous Testing of Post-Tensioned Anchorages

Prompted by the premature fracture of unbonded PT strands within a coupled wall subassembly
and the general lack of published research in the area of PT strand/anchorage performance
evaluations, the University of Notre Dame (Walsh and Kurama) undertook a research program
titled Behavior and Design of Unbonded Post-Tensioned Strand/Anchorage Systems for Seismic
Applications [7]. Following the International Code Council Evaluation Service (ICC-ES)
Acceptance Criteria for Post-Tensioning Anchorages and Couplers of Prestressed Concrete
[13], Walsh and Kurama conducted a multitude of monotonic tensile load tests on monostrand
PT strand/anchorage assemblies. The strand/anchorage parameters examined were strand
diameter, anchor type (cast and barrel type anchors) and physical properties, number of wedge
pieces, and the presence of a binding ring around wedges. Several loading parameters were also
investigated including load rate, eccentricity between strand ends, post-yield cyclic loading, and
initial strand stress. Some of the most relevant conclusions and recommendations of the final
report were:
The ultimate stress and strain capacity of the strands are limited by the fracture of
individual strand wire(s) due to the notching effect inside the anchor/ wedge assembly.
Due to the brittle nature of the individual wire fracture, a considerable amount of scatter
exists within the test data presented. Fracture strains ranged from 0.01 in./in. (or slightly
less for a small number of samples) to 0.04 in./in.
Cast anchors of the same type and provided by the same manufacturer but with different
casting date codes can perform significantly differently.
Most notably, the report suggested that the monostrand anchorage systems tested are not
capable of achieving the ultimate strength and strain limits required for certification.
Specifically, strands would often fracture prior to achieving the 0.95fpu strength requirement and
the 2.0% strain requirement of ACI 318-08 Section 18.21.1.

Experimental Program

The experimental program consisted of two primary components: testing of monostrand PT

anchorages, and testing of multi-strand PT anchorages. Two different manufactures products
were tested and all systems were subjected to monotonic and short duration, high intensity
cyclical loads. Other variables were also considered, but they were unique to either the
monostrand or multistrand systems and are described below.

Monostrand Testing

The monostrand testing program evaluated the capacity of 0.5 inch cast anchorages supplied by
two different manufacturers. The systems were tested under monotonic and a 3 different cyclical
loads. The load ranges from the cyclical loads were based off anticipated stress ranges for PT
strands being used in the PreWEC wall system describe previously. In addition, both coupled
and uncoupled systems were evaluated. Table 1 shows the test configurations and number of
samples conducted on the systems provided be each manufacturer.

Table 1. Testing Matrix for Monostrand PT Anchorage Testing

Repetitions per
Loading Scheme Type
Manufacturer
Anchor only 6
Monotonic
Coupler 3
Anchor only 3
40-90
Coupler 3
Cyclic Anchor only 3
40-95
(percent of fpu) Coupler 3
Anchor only 3
20-95
Coupler 3
MONOSTRAND TESTS CONDUCTED
27
PER MANUFACTURER

The strands were tested in a 50 kilo pound servo hydraulic testing machine. Anchorages
were placed at both ends of the strand and restrained using the fixtures shown in Figure 1 to
produce a free length of approximately 36 inches. The wedges were seated within the cast
anchorages to a preload of 800 pounds using a hydraulic jack to ensure that the wedges were
even spaced and seated consistently for each specimen. Further seating of the wedge occurred as
the test progressed. This behavior is representative of the dead or non-jacked end of a post-
tensioned system. If a coupler was required, two lengths of strand were used, each with a free
length of approximately 16 inches as shown in Figure 1(c). All couplers, wedges and anchorages
were carefully stored to ensure that the wedges supplied with each product were used with that
product. Both manufacturers supplied 2 part wedges with their cast anchorages and couplers.

(a) (b) (c)

Figure 1. Test setup for monostrand PT testing. (a) shows the overall setup, (b) shows cast
anchorage with wedges in test setup and (c) shows the test setup with a coupler

The monotonic tests were conducted at a displacement rate of 0.362 inches/minute. This
corresponds to a strain rate that meets the ICC-ES requirement of 0.0047 to 0.021 in/in/minute
[13]. The cyclical tests consisted of 50 cycles within the ranges specified in Table 1 as a
percentage of fpu or the guaranteed ultimate tensile strength (GUTS) of the strand at a frequency
of 2 hertz. Prior to the cycles, the strand was loaded using the monotonic loading rate until the
load reached the lower cyclical threshold. Upon completion of the cyclical loading, the samples
were loaded until failure using the monotonic loading rate. Load was recorded for both tests
using a load cell integrated into the testing frame.

Multistrand Testing

The multistrand testing program evaluated the capacity of multistrand anchorages configured for
7-0.6 inch post-tensioning strands. As with the monostrand system, two different manufacturers
products were tested under both monotonic and cyclical loads. As shown in Table 2, some of the
samples were tested with the anchorages aligned, while others were tested with the anchorages
rotated relative to each other by 135 degrees. When the anchorages are aligned, the strands are
vertical when tested; however, when the anchorages are rotated the strands exit the wedges at an
angle approximately 6 degrees from vertical. The final variable evaluated was the wedges that
were used with the anchor. All anchorages were tested with the wedges typically provided by
the manufacturer. These were two part (2P in Table 2) for Manufacture A, and 3 part (3P) for
Manufacturer B. In addition, a modified geometry wedge was tested with each anchorage
system. This modified wedge geometry was developed by Hayes Industries to increase the
failure strength and strain of the post tensioned system.
 Table 2. Testing Matrix for Multistrand PT Anchorage Testing
Loading Scheme Manufacturer Alignment Wedge Type Repetitions
Aligned 3
A 2P-A Standard
Rotated 3
Aligned 3
Monotonic A 2P-A Balanced
Rotated 3
3P-B Standard 3
B Aligned
3P-B Balanced 3
Aligned 3
A 2P-A Standard
Rotated 3
Aligned 3
Cyclic A 2P-A Balanced
Rotated 3
3P-B Standard 3
B Aligned
3P-B Balanced 3
TOTAL MULTISTRAND TESTS CONDUCTED 36

Standard wedges have a taper angle that matches the angle of the wedge receiving bore of
the anchorage head. Using these traditional wedges, failure of the strand typically occurs at the
smallest portion of the wedge, referred to as the nose. The modified wedge geometry uses a
small angle differential in the taper of the wedge and the wedge receiving bore. This causes the
wedge to begin gripping the strand at the back of the wedge first. As the load is increased the
wedge progressively grips more of the strand until a more uniform load distribution along the
length of the wedge is reached as the strand approaches failure. Another variation in the wedge
geometry is the gap between the wedge pieces at failure. Standard wedges are designed to free
float or have a gap between the wedges even as the system approaches failure. The modified
wedge geometry uses a reduced gap so that the wedge pieces begin to come in contact with one
another as the system approaches failure. This is designed to prevent over penetration of the
wedges into the strand, reducing the damage the wedge causes to the strand resulting in an
increased capacity of the system.
Testing of the multistrand system occurred in a 600 kip servo hydraulic testing machine.
The strands were placed in the anchorages with a free length of 42-45 inches and the wedges
were preseated using an 800 pound load as described for the monostrand system. The monotonic
and cyclic loading schemes were based on the ICC-ES guidelines[13]. The testing procedures
and requirements presented in the ICC-ES document are specifically for monostrand
assemblages. No such document exists for the testing of multistrand assemblages. Thus, the
monostrand guidelines in the ICC-ES document were applied as closely as possible to the
multistrand tests discussed in this report. The monotonic loading protocol consisted of a
constant, displacement-controlled load rate of 0.361 in./min., which corresponds to a strain rate
of 0.0086 in./in./min. The cyclical loading protocol subjected the specimen was to 50 cycles
from 0.20fpu to 0.85fpu at 0.1 Hertz. The frequency of loading during these tests were much
slower than for the monostrand tests do to the limited speed of the machine being used to
conduct the testing. Upon completion of 50 cycles, the specimen was pulled to failure at a
constant, displacement-controlled load rate of 0.361 in./min, which is consistent with the static
tests. The test setup is shown in Figure 2, including the rotation restrain that was provided to
prevent the rotation that is induced as the strands are loaded as well as from the misalignment of
the anchorages in some samples.

Figure 2. Test setup for multistrand PT testing

Experimental Results

Monostrand Results

The results of the monostrand tests are shown in Figure 3. The red dash line corresponds to the
ICC-ES limit of 95% of the free length fracture of the strand, and the black line corresponds to
the ACI 318-08 limit of 95% of fpu. The free length fracture strength was determined
experimentally to be 274 ksi. Each bar in Figure 3 represents the average of at least 3 data
points, with the error bars showing the range of data recorded. All averages, and all but two
individual sample meet the ACI requirements, while 3 average values and 8 individual samples
failed to meet the ICC-ES limit.
Two-tailed Students t-tests were used to evaluate the null hypothesis of equivalent
means for various variables. For all tests, it was assumed that the two populations are unpaired
and have about the same spread (i.e., they are homoscedastic). The results of this analysis show
that there is a significant difference in behavior between the two manufactures products. Also,
the higher performing Manufacturer 1 anchorage appears to suffer a reduction in capacity under
cyclical load that brings the performance more in line with Manufacturer 2 anchorages. While
there does appear to be a statically significant reduction for Manufacturer 1, the magnitude of the
reduction is minimal, and the anchorages still meet the performance requirements. Manufacturer
2 does not suffer a reduction in capacity from the cyclical loads. The performance does not
appear to be affected by the presence of the coupler, or the magnitude of the cyclical loads for
the range of loads examined.
 Figure 3. Monostrand anchorage test results.

Multistrand Results

When testing a single strand, failure occurs when an individual wire fractures within the
anchorage at the nose of the wedge. The peak load at the time of failure is then the ultimate load.
However, in multistrand systems where several strands are loaded simultaneously, it is possible
for the system to achieve a higher load (ultimate load) than was achieved at the time of initial
fracture of an individual wire (fracture load). For this reason, stress capacity results are
presented in terms of both fracture and ultimate values. Figure 4 shows the fracture and ultimate
stress for multistrand tendons under static loads, while Figure 5 shows the results for cyclical
loads. Each bar in these figures corresponds to the results for 1 sample. Each sample consists of
7-0.6 inch strands, and the stresses shown were calculated assuming an even distribution of load
among the seven strands. The black line on the figures corresponds the ACI requirement of 95%
of fpu and the red line corresponds to the ICC-ES requirement of 95% of the free length fracture
stress which was experimentally determined to be 283 ksi for the 0.6 inch strand being used.
It is apparent from Figures 4 and 5 that the performance of the standard wedges in
Manufacturer As anchorage does not meet either the ACI or the ICC-ES requirements for any of
the samples tested. All of the other samples meet the ACI requirements and many meet the ICC-
ES requirements. Student T-tests were again run to further evaluate the data, and they showed a
statically significant improvement based on wedge geometry for Manufacturer A. For
Manufacturer B, the average stress at failure was higher when using the modified wedges, but
the results of the T-test were inconclusive. There appeared to be no significant effect from the
type of loading applied or the misalignment of the anchorages. The result that the anchorages
supplied by Manufacturer A performed very well with a modified wedge geometry indicates that
the low stresses with the traditional wedges were not the result of a problem with the anchorages.
The likely cause of the reduced performance was an incompatibility between the wedges
supplied and the anchorage.
Figure 4. Multistrand anchorage monotonic results

Figure 5. Multistrand anchorage cyclic results

 Conclusions

The paper presented the result of testing completed on both monostrand and multistrand post-
tensioned anchorage to determine their behavior under monotonic and cyclical loads. A variety
of factors were examined to determine if they influenced the capacity of the system. Based upon
the results of the research presented, the following conclusions can be drawn.
The high intensity, short duration loads such as those applied in this study to represent the
behavior of the post-tensioning in a rocking wall system do not significantly affect the
capacity of the post-tensioned system
The performance of the anchorage hardware can vary significantly between manufactures
with the compatibility of the anchorage and wedge being the primary factor predicting
performance
Using a modified wedge geometry appears to increase the capacity of the systems
allowing a stress at failure that is closer to the free length fracture strength of the strand
The use of couplers and the misalignment of multistrand systems have no significant
impact on the capacity of the PT system

References

1. PTI. Acceptance Standards for Post-Tensioned Systems. Phoenix: Post-Tensioning Institute, 1998.
2. Housner, George W. "The Behavior of Inverted Pendulum Structures During Earthquakes." Bulletin of the
Seismological Society of America 1963; 53 (2); 403-417.
3. Priestley, M. J. Nigel, and Jian Ren Tao. "Seismic Response of Precast Prestressed Concrete Frames With
Partially Debonded Tendons." PCI Journal January-February 1993; 38; 58-69.
4. Priestley, M.J. Nigel, Sri Sritharan, James R. Conley, and Stefano Pampanin. "Preliminary Results and
Conclusions From the PRESSS Five-Story Precast Concrete Test Building." PCI Journal November-
December 1999; 44; 42-67.
5. Holden, Tony, Jose Restrepo, and John B. Mander. "Seismic Performance of Precast Reinforced and
Prestressed Concrete Walls." Journal of Structural Engineering March 203; 129(3); 286-296.
6. Kurama, Yahya. "Seismic Design of Partially Post-Tensioned Precast Concrete Walls." PCI Journal 2005;
50(4); 100-125.
7. Walsh, Kevin Q., and Yahya C. Kurama. Behavior and Design of Unbonded Post-Tensioning
Strand/Anchorage Systems for Seismic Applications. Master's Thesis, Notre Dame: University of Notre
Dame, 2009.
8. Sritharan, Sri, Sriram Aaleti, Rick Henry, Keh-Chyuan Tsai, and Kuang-Yen Liu. "Introduction to PreWEC
and Key Results of a Proof of Concept Test." M.J. Nigel Priestley Symposium August 2008; 95-106.
9. Kurama, Yahya, Brad Weldon, and Qiang Shen. "Experimental Evaluation of Post-Tensioned Hybrid
Coupled Wall Subassemblages." Journal of Structural Engineering 2006; 132(7); 1017-1029.
10. Restrepo, Jose I., and Amar Rahman. "Seismic Performance of Self-Centering Structural Walls
Incorporating Energy Dissipators." Journal of Structural Engineering 2007; 133; 1560-1570.
11. Aaleti, S., and S. Sritharan. "A Precast Wall with End Columns (PreWEC) for Seismic Applications." 8th
Pacific Conference on Earthquake Engineering (8PCEE). Singapore, 2007.
12. Henry, Richard S., Sriram Aaleti, Sri Sritharan, and Jason M. Ingham. "Concept and Finite-Element
Modeling of New Steel Shear Connectors for Self-Centering Wall Systems." Journal of Engineering
Mechanics February 2010; 136(2); 220-229.
13. ICC-ES. Acceptance Criteria for Post-Tensioned Anchorages and Couplers of Prestressed Concrete.
Evaluation Report AC303, Whittier: International Code Council Evaluation Service, 2007.