Geo-Congress 2022 GSP 336 420

Modulus Properties of Granular Materials at Various Strain Levels from Repeated Load
Triaxial Testing with Bender Elements

Mingu Kang1; Han Wang2; Issam I. A. Qamhia3; and Erol Tutumluer4

1
Ph.D. Student, Dept. of Civil and Environmental Engineering, Univ. of Illinois at Urbana-
Champaign, Urbana, IL. Email: [email protected]
2
Ph.D. Student, Dept. of Civil and Environmental Engineering, Univ. of Illinois at Urbana-
Champaign, Urbana, IL
3
Postdoctoral Research Associate, Dept. of Civil and Environmental Engineering, Univ. of
Illinois at Urbana-Champaign, Urbana, IL. Email: [email protected]
4
Abel Bliss Professor, Dept. of Civil and Environmental Engineering, Univ. of Illinois at
Urbana-Champaign, Urbana, IL (corresponding author). Email: [email protected]

ABSTRACT

Modulus or stiffness of granular layers is a fundamental material property needed for

pavement analysis and design and is also used as a quality control indicator of a constructed
pavement layer. This paper presents results from a laboratory study on repeated load triaxial tests
with bender elements (BE) to compare the modulus properties of base, subbase, and subgrade
materials in small and resilient strain levels. Three pavement base course aggregate materials and
one silty-sand subgrade soil were selected to prepare specimens at target maximum densities and
optimum moisture contents. BE pairs were installed at different heights in the specimens to
collect shear-wave velocity data. In addition, a pair of piezoelectric disk elements was installed
in the subgrade sample to collect the compressional wave velocity. Repeated load triaxial testing
was performed following the AASHTO T 307 resilient modulus test procedure. From the 15
loading stages at different stress states, the shear-wave velocities, and compressional-wave
velocity (when applicable), the resilient modulus properties were evaluated at the end of each
stress state. Comparisons were made among modulus values obtained in different strain ranges
such that shear-wave velocities measured by BE were utilized to calculate the modulus in a
small-strain range, while the resilient modulus in a larger strain range was estimated by strain
measurements. The test results regress to a linear relationship between small-strain elastic
modulus and resilient modulus for crushed stone coarse aggregates and silty-sand subgrade soils.
The findings of this research prove the possibility of establishing correlations to obtain the
resilient modulus of granular materials from BE measurements at smaller strains and enable the
utilization of the small-strain modulus measured in situ and instantaneously by BE sensors for
pavement analysis and design purposes.

INTRODUCTION

The stress states existing in pavement layers and their modulus or stiffness characteristics
both as constructed and during trafficking with moving vehicular loads are quite complex. Such
layer modulus profiles of unbound aggregate base/subbase layers and the behavior of subgrade
are of great significance in overall pavement analysis and structural design. For decades, a
considerable amount of research efforts has been devoted to characterizing the resilient behavior
of granular and fine-grained materials under repeated loading (Lekarp et al., 2000). Currently, it

© ASCE
Geo-Congress 2022 GSP 336 421

is well recognized that resilient responses, defined by resilient modulus and Poisson’s ratio, are
necessary to illustrate the nonlinear behavior of base, subbase, and subgrade materials under
traffic loading.
The resilient modulus under repeated traffic loading is used to characterize subgrade and
unbound aggregate base/subbase behavior in analyzing multilayered pavement structures in the
currently prevailing design methods including the Mechanistic Empirical Pavement Design
Guide (MEPDG), American Association of State Highway and Transportation Officials
(AASHTO) AASHTOWare Pavement-ME, Federal Aviation Administration’s (FAA’s)
FAARFIELD program and advisory circulars for airport pavement design, and many U.S. state
Departments of Transportation (DOT) procedures (Puppala, 2008). As such, layer modulus is a
key input for the mechanistic analysis which computes critical pavement responses linked to
pavement performance predicted through a mechanistic-empirical (M-E) pavement design.
Nazarian et al. (2015) emphasized the significance of switching from a density-based to a
modulus-based quality assurance approach for an efficient and practical pavement design and
construction alternative.
Repeated load triaxial testing is commonly performed in the laboratory following the
standard AASHTO T 307 test procedure to determine resilient modulus of base/subbase and
subgrade materials. The field determination of resilient modulus is often more challenging and
usually lower in accuracy. The most performed in situ nondestructive testing (NDT) to determine
material modulus include Falling Weight Deflectometer (FWD), Light Weight Deflectometer
(LWD), GeoGauge and seismic techniques, e.g., spectral analysis of surface waves (SASW)
(Byun et al., 2018). The interpretation of results of these tests to determine the constructed
layer/lift modulus often requires multiple assumptions and may lack consistency.
A modulus degradation or reduction curve has been well established in geotechnical
engineering associated with increasing strain levels (Seed and Idriss, 1970; Ishibashi and Zhang,
1993). Establishing proper relationships between small-strain elastic modulus (strains less than
10-3 %) and nonlinear behavior exhibited by geomaterials at larger strains (above 10-3 %) would
facilitate linkages between measured small-strain moduli and resilient moduli (MR) of pavement
base/subbase materials and subgrade soils for pavement design input. Once the normalized
modulus reduction curve (E/E0 vs log 𝜀 or G/G0 vs log 𝛾, where E/G is elastic/shear modulus,
E0/G0 is the maximum elastic/shear modulus, and 𝜀 /𝛾 is axial/shear strain, respectively) for a
specific material is readily available, the elastic modulus of a given material obtained from either
the laboratory or in situ small-strain stiffness measurements can be correlated to the resilient
modulus corresponding to larger strains occurring under traffic loading (Sawangsuriya et al.,
2006). Sawangsuriya et al. (2006) suggested a single reduction factor (RF) to link measured
modulus to the modulus at a desired strain level. Further, Abdallah et al. (2004) developed a
software to determine design modulus from seismic testing including FWD and Seismic
Pavement Analyzer (SPA), considering the nonlinear modulus reduction of asphalt and granular
layers.
This paper aims to investigate the relationship between laboratory-tested resilient modulus
and small-strain elastic modulus (EBE) measured using bender element (BE) piezoelectric
sensors. Repeated load laboratory testing following the AASHTO T 307 procedure using two
triaxial test setups at the University of Illinois (the TX-12 and UI-FastCell) were utilized to
determine the resilient moduli at 15 different stress states. Samples for three unbound aggregate
materials and one silty-sand subgrade material were prepared at different gradations. The BE
method has been proven as a successful method, both in laboratory and field scenarios, to

© ASCE
Geo-Congress 2022 GSP 336 422

determine the small-strain modulus by measuring the shear wave velocity propagating through a
granular medium (Byun et al., 2019; Kang et al., 2020). BE pairs were installed in the samples to
measure the shear wave velocities in this study. At different stages of the repeated load triaxial
test procedure, the measured results of the shear wave velocity, compressional wave velocity,
and resilient modulus were compared. The paper presents the MR and EBE correlations
established, and the ratios (or the correlations) of two moduli at different strain levels are
discussed.

EXPERIMENTAL RESEARCH

Material Properties. Three types of base course aggregate materials utilized in this study
include a crushed dolomite material from an Illinois quarry, a crushed granite from a quarry in
North Carolina (NC), and a subgrade material sampled from a “Smart Runway” pavement test
section study in Hill Air Force Base (AFB) in Utah. A test specimen of crushed dolomite,
referred to herein as Illinois (IL) base, was prepared in accordance with the CA 6 dense-graded
base course gradation specification of Illinois DOT. Two specimens of crushed granites with
different gradations, referred to as NC base 1 and 2, were prepared to satisfy the requirements of
the North Carolina DOT base course specifications. The subgrade soil specimen, referred to as
Utah (UT) subgrade, was a silty sand. The particle size distributions determined by dry sieving
according to ASTM C136 are shown in Figure 1.

Figure 1. Particle size distributions of four geomaterials studied

The moisture-density characteristics for all geomaterials were investigated in accordance

with the modified Proctor test method specified in ASTM D1557 standard. The optimum
moisture content (OMC), maximum dry density (MDD), median particle size (D50) and
AASHTO classifications of all materials are listed in Table 1. The OMC and MDD values were
targeted when test specimens were compacted and prepared in the laboratory.
Test Setup. Bender element (BE) sensors were used as shear wave transducers to determine
the small strain stiffness characteristics of the tested geomaterials. The convertibility between
applied electrical voltage across BE piezoelectric sheet and its deformation amount allows a BE
sensor to work either as a transmitter or a receiver. BE pairs with one BE transmitter at one end

© ASCE
Geo-Congress 2022 GSP 336 423

and one BE receiver on the other end are therefore designed and utilized. The shear waves
propagating through geomaterials between the two BE sensors are collected to calculate shear
wave velocity. The dimensions of a BE sensor used in the study were 20 mm × 10 mm × 0.6 mm
(0.8 in. × 0.4 in. × 0.02 in.) for length, width, and thickness, respectively. Meanwhile, a pair of
piezo disk elements (PDE) were used to collect the compressional waves propagating through
the subgrade soil. The metal plate of the PDE was 27 mm (1.6 in.) in diameter and 0.3 mm (0.01
in.) in thickness. The shear/compressional wave measuring system includes a signal generator,
an oscilloscope, and a filter-amplifier. The signal generator provides input signal to BE
transmitter and an oscilloscope. Sinusoidal signals with the resonant frequencies of the specimen,
which were determined by the square wave test, are applied as an input signal for stable and clear
shear wave measurement (Lee and Santamarina, 2005). The recorded signal through BE receiver
over the oscilloscope is then used to determine the first arrival time regarding the excitation time
illustrated by original input signal. The recorded signal through BE receiver is filtered and
amplified to help eliminate signal noise and achieve more accurate estimates of the first arrival
time of the shear wave.

Table 1. Moisture-density and properties for the four geomaterials studied

IL base NC base 1 NC base 2 UT subgrade

OMC (%) 4.0 5.8 5.6 (5*) 9.8
MDD (kN/m3) 21.3 22.7 24.3 (23.5*) 19.3
D50 (mm) 9.27 8.40 4.00 0.14
AASHTO Soil Classification A-1-a A-1-a A-1-a A-2-4
*Achieved value

Figure 2. Schematic drawings of the repeated load triaxial test setups for the shear and
compressional wave measurement systems

© ASCE
Geo-Congress 2022 GSP 336 424

Repeated load triaxial testing was performed next to determine the resilient moduli of these
geomaterials. Cylindrical specimens with a diameter of 152 mm (6 in.) and a height of 305 mm
(12 in.) were prepared for the IL base and NC base 1, and the repeated load tests were conducted
using the TX-12 (stands for 12-in. tall specimen) triaxial test apparatus. Cylindrical specimens
with a diameter of 150 mm (5.9 in.) and a height of 150 mm (5.9 in.) were prepared using the NC
base 2 and UT subgrade, and the repeated load triaxial testing were performed using the UI-
FastCell, which is an innovative advanced triaxial testing device allowing dynamic stresses to be
applied independently in the vertical and horizontal directions (Tutumluer et al., 1999). UI-
FastCell was utilized to provide a static confining pressure and pulsed vertical deviator stress on
the specimens in this study.
For the TX-12 apparatus, three BE pairs were installed diametrically on the cylindrical specimens
in the horizontal direction at different heights: at 10 mm (0.4 in.), 60 mm (2.4 in.), and 110 mm (4.3
in.) from the mid-specimen height, respectively (see Figure 2). For the UI-FastCell setup, only one
BE pair was installed at the mid-specimen height. For UT subgrade specimen, a PDE pair was
installed at mid-specimen height at a perpendicular direction to the BE pair to measure the
compressional wave signal of the specimen. The complete test setups are shown in Figure 2.

d

3

d
3 3

 time
0.1sec 0.9sec 0.1sec 0.9sec

Figure 3. Schematic drawing of the loading sequence during repeated load triaxial tests

Table 2. AASHTO T 307 resilient modulus test sequence and target stress states

Confining Deviator stress Number of load

Stage Bulk stress (kPa)
pressure (kPa) (kPa) applications
Conditioning 103.4 103.4 413.6 1,000
1 20.7 20.7 82.7 100
2 20.7 41.4 103.4 100
3 20.7 62.1 124.1 100
4 34.5 34.5 137.9 100
5 34.5 68.9 172.4 100
6 34.5 103.4 206.8 100
7 68.9 68.9 275.8 100
8 68.9 137.9 344.7 100
9 68.9 206.8 413.7 100
10 103.4 68.9 379.2 100
11 103.4 103.4 413.7 100
12 103.4 206.8 517.1 100
13 137.9 103.4 517.1 100
14 137.9 137.9 551.6 100
15 137.9 275.8 689.5 100

© ASCE
Geo-Congress 2022 GSP 336 425

Test Procedure. The repeated load triaxial testing was performed according to the standard
AASHTO T 307 resilient modulus test procedure. As part of the specimen conditioning, 1,000
load cycles were applied first at a confining pressure of 103.4 kPa (15 psi) and a deviator stress
of 103.4 kPa (15 psi). Following the conditioning stage, 100 load cycles were applied
sequentially at each of the 15 target stress states, as listed in Table 2. One load cycle consists of
0.1-second load pulse followed by a 0.9-second rest period (see Figure 3). In total, 2,500 load
cycles were applied on the compacted specimens during each test, and both the applied loads and
axial strains were recorded at each pulsed stress state. The P-waves measured by PDE (only for
UT subgrade specimen), and the shear waves measured by BE pairs were collected after the 100
load applications were completed at each stress state.

TEST RESULTS AND DISCUSSION

Resilient strains were recorded from all the 2,500 load cycles during the repeated load
triaxial tests. Resilient moduli were estimated using the average resilient strain and deviator
stress values obtained from the last five cycles of the 100 load repetitions applied at each stress
state. The resilient modulus (MR) values were calculated using Equation 1:

𝑀 (1)

where 𝜎 is deviator stress and 𝜀 is the recoverable resilient strain.

(a) (b)

Figure 4. Recorded elastic waves for UT subgrade; (a) shear waves from stages 1 to 15 (top
to bottom, respectively), and (b) compressional waves from stages 8 to 15 (top to bottom)

Figure 4 shows shear and compressional wave signals collected after each test stage from the
Hill AFB subgrade test. Note that the compressional wave signals were only recorded from states
8 to 15 (see Table 2). The first arrival time of the elastic wave signals gradually decreased as the
magnitudes of bulk stress applied to the specimen increased up to stage 15. From the distance
between the source shear and compressional wave transducer and the receiver wave transducer,
elastic wave velocity can be calculated using Equation 2:

© ASCE
Geo-Congress 2022 GSP 336 426

V &V (2)

where 𝑡 is the first arrival time of the elastic wave, and L is the distance between the two
transducers. Confining stress influences directly the shear wave velocity of a granular material.
Shear wave velocity can be expressed in terms of the confining stress applied to the specimen or
the shear modulus of the specimen as indicated in Equation 3 (Santamarina et al., 2001):

. .
V (3)

where 𝜌 is density of the specimen, 𝐺 is a small strain shear modulus, a and b are model
parameters determined from regression analysis, and 𝜎 is the confining stress applied on the
specimen. Thus, a shear modulus can be calculated using shear wave velocity using Equation 4:

𝐺 𝜌𝑉 (4)

Based on theory of elasticity, elastic modulus (in a small-strain range) of an isotropic

material is a function of shear modulus and Poisson’s ratio, and can be calculated as follows:

E 2𝐺 1 𝜐 (5)

where 𝐸 is the elastic modulus from BE sensor in the small strain range, referred to herein by
BE modulus, and 𝜐 is the Poisson’s ratio of the aggregate material. Therefore, a small-strain
elastic modulus of the aggregate layer can be calculated from shear wave velocity measurements.
The Poisson’s ratio of the isotropic material can be calculated from shear wave velocity and
compressional wave velocity as follows:

.
𝜐 (6)

The average Poisson’s ratio of the UT subgrade calculated from the shear and compressional
wave velocity was 0.43, and the Poisson’s ratio of the base aggregate materials were assumed to
be 0.35 for the purpose of the bender element modulus calculation because compressional wave
data were not measured in these laboratory tests.
Figure 5 (a) presents the resilient moduli determined for the four geomaterials as a function
of bulk stresses. The resilient moduli increased with the increase in bulk stress, and the trendline
fitting was determined in accordance with the K-θ model suggested by Hicks and Monismith
(1971). Resilient modulus (MR) and BE modulus (EBE) of the four geomaterials at five different
confining pressures, including stages 3, 6, 9, 12 and 15 in the AASHTO T 307 test procedure,
were compared in Figure 5 (b). The geomaterials were divided into two groups: aggregate base
materials and the silty-sand subgrade material. For all four geomaterials, BE modulus was
clearly higher than the resilient modulus. Note that geomaterial modulus varies with the strain
level (Seed and Idriss, 1970; Ishibashi and Zhang, 1993). BE modulus was measured in a very

© ASCE
Geo-Congress 2022 GSP 336 427

small strain range, which is the range of the maximum modulus, whereas the resilient modulus
was measured in a higher strain range (Sawangsuriya et al., 2006; Clayton, 2011).
Interestingly, the aggregate base materials indicated a clear linear relationship between BE
modulus and resilient modulus with a coefficient of determination (R2) of 0.76 for the grouped
data, even though these materials had different particle size distributions and were tested at
different densities and moisture contents. The equation of the best fit linear trendline is shown in
Equation 7.

M base 0.21𝐸 65.42 (7)

While further research would be required to verify this preliminary correlation, this empirical
correlation could be useful to convert the small-strain moduli of the aggregate base materials to
the resilient moduli. In addition, the BE moduli and resilient moduli of UT subgrade soil also
show a clear linear relationship with a coefficient of determination R2= 97% given in Equation 8.

M subgrade 0.17𝐸 17.40 (8)

Note that the MR-EBE correlations suggested for the base and subbase materials are only
applicable over the range of moduli measured in this study. Since the modulus degradation curve
of a geo-material is typically nonlinear, these correlations are not necessarily valid for other
strain ranges (see Figure 6).

(a) (b)

Figure 5. Relationships between MR and EBE for the four tested materials

Figure 6 shows the elastic modulus reduction curve according to the strain level, and Figure 7
presents the BE modulus to resilient modulus ratio for the four geomaterials tested in this study.
The modulus ratio varies from 2 to 5 among the four geomaterials, but the values are relatively
stable with respect to confining pressure levels. It has been reported that the modulus
degradation curve is affected by confining pressure, void ratio, and plasticity index, among other
factors. The normalized modulus at the higher strain range increases with an increased confining

© ASCE
Geo-Congress 2022 GSP 336 428

stress (Ishibashi and Zhang, 1993; Abdallah et al., 2005). For example, Figure 6 (a) shows an
increased modulus reduction curve of the NC base 1 material with the higher confining pressure
measured at high stress states. On the other hand, the resilient strain of the specimen increases
with the higher stresses applied on the specimen, which reduces the normalized modulus [see
Figure 6 (b)]. In other words, one can observe the effect of increased confining pressure and that
of increased strain offsetting each other. As a result, the BE modulus to resilient modulus ratio is
relatively steady with changing confining pressures (see Figure 7). Note that the stress-hardening
subgrade material shows the highest BE modulus to resilient modulus ratio.

Figure 6: Elastic modulus reduction curve for tested materials; (a) the normalized modulus
at the high strain range increases with an increase in confining stress, and (b) decreases
with axial strains getting larger.

Figure 7. Ratios of BE modulus to resilient modulus varying with confining pressure

SUMMARY AND CONCLUSIONS

This paper presented findings from an experimental study aimed to investigate the
relationship between laboratory-tested resilient modulus (MR) and small-strain elastic modulus
(EBE) measured by bender element pairs. Three unbound aggregate material and one subgrade
soil specimens were prepared at their observed gradations. The repeated load triaxial tests were

© ASCE
Geo-Congress 2022 GSP 336 429

performed following the AASHTO T 307 procedure to determine the resilient modulus. Bender
element (BE) pairs were installed in test specimens to measure the shear wave velocity in this
study. In addition, a Piezo Disk Element (PDE) pair, which is a P-wave transducer, was utilized
for the subgrade material testing. At 15 different stages or pulsed stress states of the AASHTO T
307 test procedure, shear wave velocity, compressional wave velocity, and resilient modulus test
results were compared.
 Small strain BE modulus (EBE) and Poisson’s ratio were measured at each AASHTO T
307 stress state using both shear wave and compressional wave velocities.
 The comparison between resilient modulus (MR) and BE modulus indicated a clear linear
relationship for base aggregates and the stress-hardening silty-sand subgrade soil,
respectively. A linear regression model for base and subgrade geomaterials were
developed from the test results.
 BE modulus to resilient modulus ratio is relatively stable and does not vary with
confining pressure owing to the offset of the effect of confining pressure and the effect of
strain level.
 A regression model is useful to transform the small-strain bender element modulus to
resilient modulus, which is a design input value for most pavement design methods.

REFERENCES

Abdallah, I., Meshkani, A., Yuan, D., and Nazarian, S. (2005). Design Modulus Values Using
Seismic Moduli (SMART User’s Manual). No. 1780-1784. Center for Transportation
Infrastructure Systems, University of Texas at El Paso.
Abdallah, I., Yuan, D., and Nazarian, S. (2004). Validation of Software Developed for
Determining Design Modulus from Seismic Testing. Center for Transportation Infrastructure
Systems, University of Texas at El Paso.
Byun, Y. H., Tutumluer, E., Feng, B., Kim, J. H., and Wayne, M. H. (2019). “Horizontal
Stiffness Evaluation of Geogrid-Stabilized Aggregate Using Shear Wave Transducers.”
Geotext. Geomembr., 47(2): 177-186.
Byun, Y. H., and Tutumluer, E. (2018). Characterization of As-Constructed Aggregate Base
Layer Stiffness Using Embedded Shear Wave Transducers. In IFCEE 2018: 567-573.
Clayton, C. R. I. (2011)."Stiffness at small strain: research and practice.” Géotechnique 61(1): 5-
37.
Hicks, R. G., and Monismith, C. L. (1971). “Factors influencing the resilient properties of
granular materials.” Transport. Res. Rec. 15–31.
Ishibashi, I., and Zhang, X. (1993). “Unified Dynamic Shear Moduli and Damping Ratios of
Sand and Clay,” Soils and Foundations, Vol. 33, No. 1, pp. 182-191.
Kang, M., Kim, J. H., Qamhia, I. I. A., Tutumluer, E., and Wayne, M. H. (2020). “Geogrid
Stabilization of Unbound Aggregates Evaluated Through Bender Element Shear Wave
Measurement in Repeated Load Triaxial Testing.” Transp. Res. Rec. 2674(3): 113-125.
Lee, J. S., and Santamarina, J. C. (2005). “Bender Elements: Performance and Signal
Interpretation.” J. Geotech. Geoenviron. Eng. 131(9), 1063-1070.
Lekarp, F., Isacsson, U., and Dawson, A. (2000). “State of the art. I: Resilient response of
unbound aggregates.” Journal of transportation engineering 126(1), 66-75.
Lekarp, F., Isacsson, U., and Dawson, A. (2000). “State of the art. II: Permanent strain response
of unbound aggregates.” Journal of transportation engineering 126(1), 76-83.

© ASCE
Geo-Congress 2022 GSP 336 430

Nazarian, S., Mazari, M., Abdallah, I. N., Puppala, A. J., Mohammad, L. N., and Abu-Farsakh,
M. Y. (2015). Modulus-based construction specification for compaction of earthwork and
unbound aggregate. Transportation Research Board.
Puppala, A. J. (2008). Estimating Stiffness of Subgrade and Unbound Materials for Pavement
Design. NCHRP Synthesis 382, Transportation Research Board, Washington, D. C.
Sawangsuriya, A., Bosscher, P. J., and Edil, T. B. (2006). “Alternative testing techniques for
modulus of pavement bases and subgrades.” In Geotechnical Applications for Transportation
Infrastructure: Featuring the Marquette Interchange Project in Milwaukee, WI: 108-121.
Seed, H. B., and Idriss, I. M. (1970). “Soil Moduli and Damping Factors for Dynamic Response
Analyses,” Report No. EERC-70-10, EERC, University of California, Berkeley, CA.
Tutumluer, E., and Seyhan, U. (1999). “Laboratory Determination of Anisotropic Aggregate
Resilient Moduli Using an Innovative Test Device.” Transp. Res. Rec., 1687(1): 13–21.

© ASCE