Professional Documents
Culture Documents
Abstract: Although rammed aggregate piers (RAPs) have attracted much attention as a cost-beneficial ground improvement technique, their
construction process and the other affecting parameters on the behavior of these geotechnical structures during the construction and loading
phases have not been properly investigated and only some predictions for downward and radial expansion resulting from the construction pro-
cess have been proposed to be used in numerical simulations. The goal of this paper is to numerically simulate the RAP construction process to
investigate the effect of construction sequences on RAP behavior. Keeping this aim in mind, a hybrid distinct element–finite difference
algorithm is developed to simulate the placing and ramming of the RAP’s clean crushed stone layers in detail. Introducing several numerical
models, a parametric study was then conducted to compare the behavior of RAPs during the construction process for different pier geometries
and soil properties. The results of this research not only help to understand the RAP behavior during the ramming process but can also be used as
a guide for predicting the amount of downward and lateral expansion of RAPs after the ramming process. DOI: 10.1061/(ASCE)GM.1943-
5622.0000429. © 2014 American Society of Civil Engineers.
Author keywords: Numerical methods; Distinct element; Rammed aggregate piers (RAPs); Construction process.
Introduction is then introduced into the drilled cavity in 300-mm lifts. In the first
lift, the bottom bulb is formed by ramming the placed stone layer
Over the past century, significant progress has been achieved to using a downhole impact hammer equipped with a specially designed,
make the construction of buildings possible over soft and compress- high-energy beveled tamper foot. The pier is completed by in-
ible soils. Some of the advanced techniques, such as deep foundations, troducing more thick lifts of aggregate over the bottom bulb and
stone columns, ground improvement, and soil replacement, are rela- densifying the aggregate with the beveled tamper. During ramming,
tively expensive and not economical in many cases. An alternative although the beveled tamper compacts the aggregates, it forces them
to these methods is the rammed aggregate pier (RAP) technique. laterally into the sidewall of the excavated cavity, resulting in an
This method was developed in the early 1980s and has been used for increase in the lateral stresses in the surrounding soil and the soil
different geotechnical applications, such as settlement control, matrix is densified and prestressed. In granular soils, casing is often
bearing capacity improvement, slope stabilization, uplift enhance- required to provide the stability of the cavity sidewall during con-
ment, and liquefaction mitigation (Hoevelkamp 2002; Fox 2004; struction. A steel casing slightly larger than the drill tool is either
Farrell and Taylor 2004; Farrell et al. 2008; Girsang et al. 2004; vibrated or lowered into the cavity during drilling and is withdrawn
White et al. 2007; Majchrzak et al. 2009). This technique is different incrementally during aggregate tamping.
from similar techniques, such as stone columns. The difference is The behavior of RAPs has been previously investigated using
mainly in the construction process, material, and their behaviors, as both experimental and numerical approaches (White et al. 2007;
discussed in past studies (White et al. 2002). The construction Pham and White 2007; Handy and White 2006a, b; Suleiman and
procedure of RAP systems is well-described in the literature White 2006; Chen et al. 2009; Halabian et al. 2012a, b). However,
(Lawton et al. 1994; Lawton and Fox 1994). Aggregate piers are despite this expectation that the installation process has a significant
constructed using a variety of methods, such as vibroreplacement, effect on RAP behavior, most of the numerical studies have focused
predrilling, vibration, and ramming, the vibrocomposer, the cased on the behavior of these structures after installation, neglecting the
borehole, and finally the Geopier reinforcing elements. In general, construction process effect.
RAP construction is initiated by drilling 600- to 900-mm-diameter Limited studies have considered the pier’s construction effect, in
holes into the ground to depths ranging from 2 to 8 m below the which the ramming effect on the soil matrix has been induced by cavity
planned footing bottom elevation. Well-graded clean crushed stone
expansions along the shaft and at the bottom of the cavity (Pham and
White 2007; Chen et al. 2009; Halabian et al. 2012a; Pham 2005).
1
Associate Professor, Faculty of Civil Engineering, Isfahan Univ. of Halabian et al. (2012a) showed that the stress path for an adjacent soil
Technology, 8415683111 Isfahan, Iran (corresponding author). E-mail: element located at shallow depths of the pier in a pier group is affected
mahdi@cc.iut.ac.ir by pier installation (Fig. 1). During pier installation, the radial normal
2
Ph.D. Student, Dept. of Geotechnical Engineering, Univ. of New stress (sH ) increased beyond the vertical normal stress (sv ), whereas
Hampshire, Durham, NH 03824; formerly, Associate Professor, Faculty the shear stress also increased. By the end of the pier installation
of Civil Engineering, Isfahan Univ. of Technology, 8415683111 Isfahan, process, the stress state of the soil element approached the Rankine
Iran. E-mail: pj1@wildcats.unh.edu.
passive condition. The stress states of the soil element’s subsequent
Note. This manuscript was submitted on August 10, 2013; approved on
June 2, 2014; published online on July 11, 2014. Discussion period open application of foundation loading, captured by the upwards stress
until December 11, 2014; separate discussions must be submitted for path continuing from the passive condition, which can be charac-
individual papers. This paper is part of the International Journal of terized by a rapid increase in the vertical normal stress and reduction
Geomechanics, © ASCE, ISSN 1532-3641/04014085(14)/$25.00. in the radial normal stress following another reverse to the shear
of the neighboring soil matrix. A comprehensive parametric study As it was pointed out, the installation process of RAPs is a phe-
was conducted to compare the ramming effects for different pier nomenon of considerable significance and was studied experimen-
tally by a number of researchers (Handy and White 2006a, b). To
further investigate the installation effects, the soil-stress path during
construction has also been studied for isolated piers by some
researchers (Pham and White 2007; Chen et al. 2009). As part of an
experimental study at a test site in Neola, Iowa, a single RAP that was
3 m long and had a 0.76-m diameter was constructed and various in
situ and laboratory tests were performed to evaluate the load-
deformation behavior of the pier. Verification of the model de-
veloped in the current study was performed based on the field and
laboratory test results at this test site (Pham 2005). The site consisted
of an upper 1-m-thick desiccated fill and a lower 13-m soft alluvial
clay overlying glacial till and weathered shale bedrock. Table 1
shows the geotechnical properties of the soil and compacted ag-
gregate that had been obtained from the conducted in situ and
laboratory tests (Pham 2005).
Fig. 2 shows the sequence of RAP construction in Neola
(Majchrzak et al. 2004). The RAP construction in this site had been
initiated by boring a 760-mm-diameter hole into the ground to
Fig. 1. Stress path for selected matrix soil element near pier (data from a depth of 3 m below the ground surface. Well-graded clean crushed
Halabian et al. 2012a) stone was then introduced into the drilled cavity in 300-mm lifts.
Each layer was rammed with a beveled tamper that was attached to a
Table 1. Summary of Geotechnical Properties of the Compacted Aggregate, Alluvial Clay, and Desiccated Soil
Unified Soil Classification
System (USCS) classification g (kN=m3 ) wn (%) e0 c9 (kPa) f9 c (degrees) Cc ref
E50 (MPa) ref
Eur (MPa) K0 n9
(GP) aggregate 21 6 0.33 4 47 12 — 90 180 — 0.2
(CL) alluvial clay 18.9 36 1.0 2 24 0 0.1 3 9 0.6 0.2
(ML) desiccated soil 18.9 36 1.0 2 35 0 0.05 9 27 1.0 0.2
Note: Cc 5 compression index; c9 5 effective cohesion; Eur ref
5 unloading/reloading modulus; E50
ref
5 deviatoric reference modulus; e0 5 initial void ratio;
K0 5 lateral earth pressure coefficient at rest; n9 5 effective Poisson’s ratio; wn 5 natural water content; g 5 unit weight above the groundwater table;
f9 5 effective friction angle; c 5 dilation angle.
was also installed and so displacements at the top of the pier and at features of granular flow, using these sophisticated techniques is
the telltale plate were recorded during the load test. usually a challenging task. A general PF model simulates the me-
chanical behavior of crushed aggregates as a collection of arbitrarily
shaped particles. The model adopted in this study is composed of
Numerical Modeling distinct particles with a finite amount of space that interacts at the
interfaces between the particles. The method uses the soft contact
As it was mentioned, the current study seeks to have a better un- approach, which assumes that elements have a finite normal stiff-
derstanding of the construction effects of a single RAP during the ness, and represents elastic flattening at contacts by allowing the
ramming process using a hybrid PF model capable of simulating bodies to overlap. The mechanical behavior of such a system is de-
aggregate movements and the FD method for modeling the soil scribed in terms of the movement of each particle and the interparticle
matrix (Fig. 3). A methodology similar to what was presented by forces acting at each contact point (Fig. 4). A stiffness model, bonding
Halabian et al. (2012b) is considered here, with some exceptions, to model, and slip model are included in the constitutive representation of
take the ramming effect into account in a systematic manner. A series contact points between the elementary circles (in a 2D nature), which
of two-dimensional (2D) distinct element–FD models of isolated are the basic building blocks. Newton’s second law is used to de-
piers embedded in different soil conditions were developed to termine the motion of each particle arising from the contact and body
investigate the pier construction. Quadratic brick elements were forces acting upon it, whereas the force-displacement law is used to
used to model the soil matrix in a FD algorithm, whereas all nu- update the contact forces arising from the relative motion at each
merical computations of elements were performed using the reduced contact. In 2D cases, only two force components and one moment
numerical integration. The zone and subzone dimensions and the component exist.
number of nodes and elements in the descritized soil matrix domain
were decided on after performing a number of sensitivity analyses.
The elements’ size at the vicinity of the pier should be small to
capture the intense stress and deformation gradients. The discretized
Fig. 3. (a) Hybrid PF model–FD algorithm of RAP examined at test Fig. 4. Interparticle forces acting at each contact point during ramming
site of Neola, Iowa; (b) contributed models in PF and FD models process
force and shear elastic force increment over a time step at the contact by the FD approach. The computed solution produced by both
of two entities, respectively. The normal stiffness, K n , and shear methods will reach a unified response only if the time step is kept the
tangent stiffness, k s , at a contact are computed, assuming that the same in both solutions. This is possible if the differential density
stiffnesses of the two contacting objects act in series scaling approach is used to solve the equations in the PF model. By
employing this technique, the time step in both approaches is taken
½A ½B ½A ½B
kn kn ks ks as unity and then the calculated displacements in each time step in
Kn ¼ ½A ½B
ks ¼ ½A ½B
(2) both methods will be the same.
kn þ kn ks þ ks
The solution calculated in the PF model will remain stable if
the time step is limited to a critical time step that is related to the
where kn½A and kn½B 5 normal stiffness of Objects A and B, which are maximum eigenvalue of the total system. The actual time step
in contact; and kn½A and kn½B 5 shear stiffnesses. The total normal used in any cycle is taken as a fraction of this estimated critical
and shear forces that the contact can carry are limited by enforcing value
bond strengths. The maximum tensile force that the bond can sustain
in tension and the maximum shear force that it can withstand before (pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
breaking are specified when the bond is created and may be modi- m=k tran ðtranslational motionÞ
fied at any time during the simulation. The slip model also allows tcrit ¼ pffiffiffiffiffiffiffiffiffiffiffi (6)
I=k rot ðrotational motionÞ
the particles to slip over each other. Slip occurs when the shear
component of the force exceeds the allowable shear strength, which
is where tcrit 5 critical time step; m and I 5 mass and moment of inertia
of the particle; and k tran and k rot 5 translational and rotational
Fsmax ¼ m j Fin j (3)
stiffnesses, respectively.
By invoking differential density scaling, the inertial mass of each
One of two vector equations of motion relates the resultant force to particle is modified at the start of each cycle such that the stability
the translational motion, and the other one expresses the resultant criterion of Eq. (6) is satisfied with a time step of unity. The laws of
moment in terms of the rotational motion for each particle motion given by Eqs. (4) and (5) are modified as
where
2=5 ðspherical particleÞ
b¼
1=5 ðdisk-shaped particleÞ
(9)
ðtÞ validation process in the PF model is different and more compli-
ðtþDt=2Þ ðt2Dt=2Þ M3
v3 ¼ v3 þ Dt cated than the mesh-based continuum models because the micro-
bðmi ÞR2
parameters of the materials, such as the cohesion or friction angle,
could not be evaluated from laboratory tests without engineering
where x_ i and v3 5 translational and angular velocity, respectively; judgments and, consequently, the model parameters such as normal
and t and Dt 5 time and time step, respectively. and shear contact stiffness, as well as the particle friction coef-
As mentioned earlier, in this study the Mohr-Coulomb model ficients, are not certainly obtained. In other words, selecting con-
was used to characterize the plastic behavior of the surrounding soil. stitutive microparameters of granular materials in a PF model to
Following this adoption, a stage analysis consisting of six steps was simulate the macroscopic behavior of laboratory samples under
used to model the construction stages of the piers different stress states needs a trial-and-error process and, because of
1. An axisymmetric soil continuum in the FD space is developed. current knowledge at this stage, is a complicated process.
Having the surrounding soil affected by the ramming process, Keeping the same aspect ratio with the experimental samples,
the edge length and height of the model are assumed to be 14 some numerical models were made, considering the RAP con-
and 8 m, respectively (Fig. 3). The model dimensions were struction procedure. Triaxial tests were simulated on numerical
decided on after performing a number of sensitivity analyses, models at the corresponding experimental confining pressures.
taking into account the wave attenuation. Full fixity is applied Fig. 7 plots the result of the simulation in the q, ɛ a -plane in the triaxial
along the base of the model, whereas the vertical boundaries test. In this simulation, the microparameters given in Table 2 are
were taken to be fixed in the horizontal direction, as shown in
Fig. 3. After assigning the soil parameters, the model is
allowed to get balanced under its weight.
2. In this step, in the FD space, the excavation is modeled.
Because of weakness of the soil and the possibility of it falling
into the cavity, casing has been modeled by fixing the edges of
the cavity. The model is then allowed to reach equilibrium
under this condition.
3. In this step, having both solving algorithms linked, the coor-
dinates of the grid points constructing the cavity’s sidewalls
are transmitted from the FD space to the PF model. In fact, each
of the two neighboring grid points is modeled as the boundary
walls in the PF model.
4. The backfilling and ramming of the aggregate is modeled in the
PF model, and the resultant forces and moments at each wall
are transmitted to the soil matrix space.
5. The applied forces and momentum are converted to equivalent
forces at the grid points and then applied to the corresponding
points in the FD model (Fig. 5). The surrounding soil matrix is
deformed due to the new amount of forces and the resultant
deformations transmitted to the PF model.
6. Applying grid point velocities, the geometry of the pier in the
PF model is modified and then Steps 4–6 are repeated.
Fig. 6 presents the details of these steps.
Model Verification
Fig. 7. Comparison of simulated and experimental results for samples in triaxial tests on RAP’s aggregates at different confining pressures
Table 2. Microparameters of RAP’s Aggregate in PF induced to the soil matrix after RAP construction as well as the load-
Parameter Value settlement curves of the top and tip of the pier were compared with
the corresponding values and curves from the experimental test
Kn
6 3 108 N=m (Pham 2005). Fig. 10 shows the radial expansions along the height of
Ks 5 3 108 N=m the pier after the construction is done. As can be seen, the radial
m 0.7 expansion at the tip (near the ground) is approximately 4 cm, which
is in good agreement with the experimental measurements in the
used. As can be seen, the numerical simulations could predict the field (5% of the pier’s diameter). Fig. 11 also shows the downward
ultimate strength and postpeak softening behavior of samples during expansion after the pier’s construction, which is pretty much con-
the loading phases. sistent with 10% of the pier diameter suggested by the experimental
results. When it comes to workability of the model developed here to
Verification on Soil Matrix simulate the load-settlement behavior of the pier during vertical
loading, as is shown in Fig. 12, a good agreement between the
In the second phase, to use an appropriate constitutive model for the experimental and numerical analyses results were achieved for both
soil matrix in the hybrid soil-pier model, two constitutive models the tip and top of the pier, showing the accuracy of numerical
(Mohr-Coulomb and Cam clay) (Potts and Zdravkovic 1999) were modeling.
selected to be employed in the FD domain. The result of the
consolidated-drained (CD) triaxial tests conducted at confining pres-
sures of 25.5, 41, and 60 kPa on undisturbed alluvial clay samples
extracted at a depth of 4.2 m from the ground surface (Pham 2005) were Discussion of the Numerical Results
used to verify the workability of both constitutive models and validate
their model parameters for the soil matrix. Figs. 8 and 9 show the results
Radial Expansion Distribution
of this validation. Tables 3 and 4 also present the Mohr-Coulomb and
Cam-clay model parameters, respectively. As can be seen, both models The radial displacement distribution along the pier, which is given in
can predict the ultimate strength of the samples; however, the Mohr- Fig. 10, has demonstrated that radial displacement distribution is not
Coulomb model is capable of capturing the load-displacement curve uniform along the RAP and clearly decreases with depth, which can
for all samples when compared with the experimental results (Pham be due to an increase in the amount of overburden with depth,
2005). Accordingly, the Mohr-Coulomb model is chosen in the hybrid resulting in higher confining stresses. In fact, Fig. 10 shows a critical
soil-pier model to simulate the soil matrix behavior. length for radial expansions of the RAP, whereas beyond this length,
the radial expansion is not remarkable and can be neglected. Because
Verification on Hybrid Soil-Pier Model of these results, the uniform radial expansion along the RAP’s height
suggested in previous studies (Pham 2005) is far from reality. The
In the last phase, based on the presented constitutive model jagged behavior of the radial displacement distribution seen in
parameters for aggregates in the PF domain (pier) and soil matrix in Fig. 10 could be related to the points between two layers of placing.
the FD domain, the results of the load-settlement test conducted in In other words, by ramming each layer, the upper points of the layer
Neola, Iowa, on an isolated pier (Pham 2005) were compared with have more displacements due to more received energy compared
the numerical predictions of the hybrid soil-pier model. The size of with the lower points, which makes the sequential bracket shape of
the examined pier in the test was 0.80 m in diameter and 3.20 m in the pier after construction (Fig. 2).
length and was capped with a concrete footing having a 0.46-m
thickness. According to field measurements in Neola, Iowa, the
values of expansion along the shaft and at the bottom of the cavity
Radial Stress Distribution
were determined to be equal to 5 and 10% of the nominal diameter
of the cavity, respectively (Pham 2005; White et al. 2003). The To better understand the stresses induced to the soil matrix, the
construction procedure sequence, which was similar to what was radial stress distribution along the RAP is studied during the se-
conducted in Neola, was implemented in the hybrid soil-pier model quential ramming process of construction and at the end of RAP
developed in this study, and radial and downward expansions construction.
Fig. 8. Comparison of simulated and experimental results for samples in triaxial tests on soil matrix using the Mohr-Coulomb model at different
confining pressures
Fig. 9. Comparison of simulated and experimental results for samples in triaxial tests on soil matrix using the Cam-clay model at different
confining pressures
Table 3. Mohr-Coulomb Model Parameters for Soil Matrix Table 4. Cam-Clay Model Parameters for Soil Matrix
Parameter Value Parameter Value
C 2 kPa rwet 1,924 kg=m3
Tension 0 kPa Elastic bulk modulus, K 1,000 MPa
f 33° Elastic shear modulus, G 3.75 MPa
c 0° Poisson’s ratio, y 0.2
rwet 1,924 kg=m3 Slope of ncl, l 0.043
E 3.5 MPa Slope of url, k 0.0108
y 0.2 Slope of csl, M 0.94
Reference pressure, Pref 25,500 Pa
Initial specific volume, v0 2
Note: csl 5 critical state line; ncl 5 normal critical line; url 5 unloading-
Radial Stress Distribution in the Soil Matrix during Ramming reloading line.
of Each Layer
Fig. 13 shows the radial stress distribution during the ramming
that stresses in all elements will not exceed this line. This line cor-
process of each layer. Lines K00 and KP0 are referred as at-rest and
responds to a passive pressure line of soil with an equivalent friction
passive pressures based on
angle of 44°. As can be noted, the stresses induced into the soil matrix
K0 ¼ 1 2 sin f (10) during the ramming process increases uniformly for each layer at
different depths, except for the first layer, in which the stress increase is
more pronounced due to a greater ramming effect, as described pre-
Kp ¼ tan2 ð45 þ f=2Þ (11) viously. This means that the friction of the angle of soil in shallower
layers upgrades to 44°, showing an increase of approximately 40%.
Based on the fact that stresses in the soil matrix cannot be more than the However, this is not the case in deeper layers because the radial
passive limit if no failure happens, Fig. 13 also presents the KP1 line so stresses do not exceed the passive pressure line (KP0 ).
Parametric Analyses
Fig. 11. Downward expansion at base of RAP Fig. 17 shows the variations of the RAPs’ downward expansion
as a percentage of their diameters in terms of different values of pier
Fig. 12. Verification results of isolated pier (data from Pham 2005)
Fig. 15. Radial stress distribution in adjacent soil element before and after pier construction in different radial distances
increases, especially up to a friction angle equal to 18°; however, more uniformly in the surrounding soil and thus the RAP ramming
the downward expansion percentage is not that much sensitive to process varies little with changes to the soil cohesion.
the cohesion. The radial expansion percentage also decreases
when the friction angle increased. This mostly happened in the Summary and Conclusions
shallow layers, where it is not sensitive to the cohesion values
taken in this study. Developing the ramming process in RAP construction, this paper
Fig. 22 also demonstrates that the nonuniform behavior of the aimed to present the results of a comprehensive numerical study that
radial expansion distribution decreases and the critical depth in- was conducted to examine the effects of pier geometry and the
creases with increasing cohesion. In other words, by increasing the mechanical properties of the soil matrix on RAPs’ radial and
cohesion, the induced soil matrix displacements after pier con- downward expansion induced to the soil matrix. The major results
struction are uniform and then the critical depth increases. from this study can be summarized as follows:
Tables 9 and 10 present the variations of the soil matrix’s im- 1. The PF model as one of the distinct elements method could be
proved properties in terms of different soil friction angles and one of the suitable methods to simulate the granular material
cohesions. As it is seen, the soil’s modified friction angle varies behavior, such as crushed sands under different static and
between 38 and 42° for the initial values considered in this study, dynamic stress states.
showing the effectiveness of using RAPs in general soft soils. 2. The RAP construction process has a significant effect on the
Table 10 also highlights that the higher soil cohesion does not result stress path and behavior of the soil matrix around the piers, and
in a bigger mechanically improved soil matrix zone. This could refer neglecting this stress phase could cause unreliable results in
to the fact that with increasing soil cohesion, stresses are distributed the design of RAPs.
Fig. 16. Soil matrix stress path at different depths during pier’s construction
Fig. 18. Radial expansion distribution along RAP for different lengths and diameters
Table 6. Improved Friction Angle for Different RAPs Table 8. Depth of Improved Area for Different RAPs
RAP diameter RAP diameter
RAP length, L (m) d 5 60 cm d 5 80 cm d 5 100 cm RAP length, L (m) d 5 60 cm (cm) d 5 80 cm (cm) d 5 100 cm (cm)
1:6 40 40 41 1:6 0:24H 5 40 0:28H 5 45 0:38H 5 60
3:2 43 44 43 3:2 0:22H 5 70 0:22H 5 70 0:26H 5 83
4:8 45 43 41 4:8 0:18H 5 86 0:18H 5 86 0:2H 5 96
6:4 44 43 43 6:4 0:18H 5 86 0:18H 5 86 0:18H 5 86
Cohesions
Fig. 20. Downward expansion percentage for different cohesions Parameter, Improved friction Radius of improved Depth of improved
c (Pa) angle (degrees) area (cm) area (cm)
2,000 44 30 0:22H 5 70
4,500 44 40 0:24H 5 76
7,000 43 40 0:24H 5 76
10,000 44 45 0:26H 5 83
References
Fig. 21. Radial expansion distribution along RAP for different friction Chen, J.-F., Han, J., Oztoprak, S., and Yang, X.-M. (2009). “Behavior of
angles single rammed aggregate piers considering installation effects.” Comput.
Geotech., 36(7), 1191–1199.
Farrell, T., Fitz Patrick, B., and Kenney, W. (2008). “Uplift testing of
rammed aggregate pierÒ systems.” Proc., Geotechnical Earthquake
Engineering and Soil Dynamics IV Congress, D. Zeng, M. T. Manzari,
and D. R. Hiltunen, eds., ASCE, Reston, VA, 1–14.
Farrell, T., and Taylor, A. (2004). “Rammed aggregate pier design and
construction in California—Performance, constructability, and eco-
nomics.” Proc., SEAOC 2004 Convention, Structural Engineers Asso-
ciation of California, Sacramento, CA, 147–154.
Fox, N. S. (2004). “Geopier soil reinforcement system—Case histories of
high bearing capacity footing support and floor slab support.” Proc., 5th
Int. Conf. on Case Histories in Geotechnical Engineering, S. Prakash,
ed., Vol. 4, Univ. of Missouri-Rolla, Rolla, MO, 13–17.
Girsang, C. H., Gutierrez, M. S., and Wissmann, K. J. (2004). “Modeling of
the seismic response of the aggregate pier foundation system.” Proc.,
Geo-Support 2004, ASCE, Reston, VA, 485–496.
Halabian, A. M., Naeemifar, I., and Hashemolhosseini, S. H. (2012a).
“Numerical analysis of vertically loaded rammed aggregate piers and
pier groups.” Proc. Inst. Civ. Eng., 165(3), 167–181.
Halabian, A. M., Naeemifar, I., and Hashemolhosseini, S. H. (2012b). “Nu-
merical analysis of vertically loaded rammed aggregate piers and pier
groups under dynamic loading.” Soil Dyn. Earthquake Eng., 38, 58–71.
Handy, R. L., and White, D. J. (2006a). “Stress zones near displacement
piers: I. Plastic and liquefied behavior.” J. Geotech. Geoenviron. Eng.,
10.1061/(ASCE)1090-0241(2006)132:1(54), 54–62.
Fig. 22. Radial expansion distribution along RAP for different Handy, R. L., and White, D. J. (2006b). “Stress zones near displacement
cohesions piers: II. Radial cracking and wedging.” J. Geotech. Geoenviron. Eng.,
10.1061/(ASCE)1090-0241(2006)132:1(63), 63–71.