Professional Documents
Culture Documents
net/publication/292678002
CITATIONS READS
38 1,216
2 authors:
Some of the authors of this publication are also working on these related projects:
NCHRP 12-114 - Guidance on Seismic Site Response Analysis with Pore Water Pressure Generation View project
Foundation design for 1915 Canakkale bridge (2023-m-long main span suspension bridge to be built in Turkey) View project
All content following this page was uploaded by Hobi Kim on 27 March 2016.
ABSTRACT: This paper presents a numerical study performed to investigate the effect of expanded
polystyrene (EPS) geofoam panels placed over a buried pipe. It is recognized that EPS geofoam panels as
compressible inclusions over a buried pipe are effective in reducing the earth pressure acting on the pipe due
to positive arching action. To date, however, there is no systematic methodology that links the earth pressure
on a buried pipe with the geometry of EPS panels. To investigate the ‘optimal’ geometry of EPS panels, a
two-step numerical modeling approach was employed and calibrated against results of a model-scale
experimental study. First, material properties were estimated for each component used in the model-scale
tests (i.e., soil, EPS geofoam and steel pipe). Second, the model-scale experiments were simulated using the
selected material properties. These simulations resulted in reasonable agreement between model-predicted
and measured vertical and lateral earth pressures. Using the calibrated model, additional cases that were not
covered in the experimental study were investigated to examine different widths and thicknesses of EPS
panels. The numerical analysis provided quantification of the effect of EPS compressible inclusion and a
systematic approach to optimizing the design of buried pipes using EPS geofoam panels.
pipes and culverts (e.g., Vaslestad et al. 1994b; Kim and steel pipe and Jumunjin sand). ‘In-isolation’ (i.e., individ-
Yoo 2005; Sun et al. 2005, 2009). For example, based on ual component) tests were simulated for each of these
numerical analyses for box-type concrete culverts, Sun three materials.
et al. (2005) found that the use of EPS panels reduces
bending moments along both the top and sides of the 2.1. EPS geofoam
culvert. They also generally found that load reduction due The model-scale tests by Kim et al. (2010) used a
to placement of EPS panels is more pronounced for single type of EPS geofoam, which had a density of
gravelly, sandy fill materials than for silty, clayey fill approximately 15 kg/m3. Kim et al. (2010) performed
materials. Sun et al. (2009) performed both instrumented two unconfined compression tests on specimens of this
field tests and numerical analyses to investigate earth geofoam material. Results of these unconfined com-
pressure reduction on concrete culverts with EPS panels in pression tests are reproduced in Figure 1. For reference,
various configurations. Based on these results, Sun et al. test results reported by Duskov (1997) for EPS with
(2009) indicated that EPS geofoam can be used effectively a density of approximately 15 kg/m3 and by Hazarika
to reduce vertical load on rigid culverts resting on rigid (2006) for EPS with a density of approximately 16 kg/m3
foundations. are also reproduced in Figure 1.
However, few studies exist to ‘optimize’ the dimensions The EPS geofoam used in the Kim et al. (2010) model
of an EPS geofoam compressible inclusion placed over a tests was simulated as a linear-elastic/Mohr–Coulomb
buried pipe via a systematic approach. The effect of an material with post-yield strain-hardening using FLAC’s
EPS geofoam compressible inclusion over a buried pipe built-in strain-hardening/softening model (Itasca 2011).
depends on several factors, such as: (1) EPS geofoam Although a bilinear elastic model might have been used to
density and panel thickness, which determine stress–strain produce similar behavior for monotonic loading con-
behavior; (2) EPS geofoam width relative to pipe ditions, a linear-elastic/strain-hardening model was
diameter; and (3) number and spacing of EPS geofoam adopted to allow for the possibility of non-recoverable
panels when multiple EPS panels are placed over the pipe. plastic strain in the event of unloading. In light of the
Although Kim et al. (2010) researched these factors and variation of test results on EPS materials having similar
associated changes in the earth pressure acting on a buried densities illustrated in Figure 1, the assumption of
pipe, this experimental investigation was subject to linear-elastic pre-yield behavior per se was considered
limitations on the measurement of stresses and strains. unlikely to introduce significant error into the model.
Some of these limitations can be addressed through a Post-yield strain-hardening behavior was assumed to be
numerical modeling approach. For example, model-scale linear (i.e., linearly increasing strength with increasing
tests provide stress and strain measurements at discrete strain level). This assumption of linear strain-hardening is
points; numerical analyses can provide continuous stress considered to be reasonably accurate up to strain levels
and strain distributions over a cross-section. Moreover, a in the vicinity of 30% based on the test results shown
limited set of EPS panel configurations was evaluated in in Figure 1 and those reported by other researchers
the model-scale tests; numerical analysis provides an (e.g., Horvath 1995).
efficient means of evaluating a significantly larger set of For comparison with the unconfined compression test
configurations. Consequently, additional study is war- results presented by Kim et al. (2010), a numerical model
ranted to develop design guidelines for ‘optimal’ com- was developed to simulate unconfined compression of an
pressible inclusion geometry. EPS specimen. As shown in Figure 2, the simulation
The purpose of this study was to evaluate earth pressure results are in reasonable agreement with the test results
reduction on a buried pipe for various EPS geofoam reported by Kim et al. (2010).
compressible inclusion configurations and to illustrate
‘optimization’ of EPS panel geometry using a numerical
approach. A numerical model was developed using
FLAC. Calibration was performed for each material 150
used in the Kim et al. (2010) model-scale tests (i.e., EPS
geofoam, Jumunjin sand and the pipe) using laboratory
test data reported by Kim et al. (2004, 2010) and Park
Axial stress (kPa)
0
2. EVALUATION OF MATERIAL 0 10 20 30
PROPERTIES Axial strain (%)
A material model was calibrated for each of the three Figure 1. Unconfined compression test results on EPS geofoam
materials used in the model-scale test (i.e., EPS geofoam, specimens from the literature
Geosynthetics International
Parameter Value
Axial stress (kPa)
Figure 2. Comparison of unconfined compression test results for The Jumunjin sand was modeled as a nonlinear-elastic/
EPS reported by Kim et al. (2010) with FLAC model results Mohr–Coulomb material. Material properties for the
Jumunjin sand were estimated based on test results
reported by Kim et al. (2004, 2010) and Park et al.
The Young’s modulus of the EPS was estimated based (2008) and on recommendations by Duncan et al. (1980).
on the stress–strain response of the Kim et al. (2010) EPS The elastic behavior of the sand was controlled by a
specimens up to an axial strain level of approximately stress-dependent tangent Young’s modulus, Et, and a fixed
2.4%. Following this approach, the Young’s modulus for value of Poisson’s ratio, ν. The tangent Young’s modulus
EPS was estimated to be approximately (Ef )avg = 2800 kPa value was calculated as the product of the initial Young’s
(two test results showed 3190 and 2500 kPa, respectively). modulus value, Ei, and a hyperbolic degradation function.
This value is in reasonable agreement with the density– The initial Young’s modulus value was calculated as a
stiffness relationships proposed by other researchers. For function of mean effective stress, void ratio and Poisson’s
example, Hazarika (2006) and Horvath (1995) proposed ratio. Following the general form adopted by
the following equations: Papadimitriou et al. (2001), which is based on the
Ef ¼ 041ρf 28 ð1Þ equation proposed by Hardin (1978), the initial shear
modulus (Gi) is calculated as:
Ef ¼ 045ρf 3 ð2Þ 0 0:5
Bpatm p
Gi ¼ ð4Þ
where Ef is the EPS Young’s modulus in MPa and ρf is 03 þ 07e2 patm
the EPS density in kg/m3. Based on these equations,
where B is the shear modulus number, patm is atmospheric
suggested Young’s modulus values are Ef = 3350 kPa and
pressure (assumed to be 101.3 kPa), e is the void ratio and
Ef = 3750 kPa, respectively. These values are approxi-
p′ is the mean effective stress. The initial Young’s modulus,
mately 20–34% higher than the selected value.
Ei, is related to the initial shear modulus according to
According to Horvath (1995), the Poisson’s ratio of the
Equation (5):
EPS geofoam was expressed in terms of EPS density as:
Ei ¼ 2Gi ð1 þ νÞ ð5Þ
νf ¼ 00056ρf þ 00024 ð3Þ
where ν is Poisson’s ratio.
where νf is the EPS Poisson’s ratio and ρf is the EPS density
Young’s modulus degradation with increasing shear
in kg/m3. A Poisson’s ratio value of approximately
stress was incorporated following the hyperbolic model
νf = 0.09 was selected using this equation.
proposed by Duncan and Chang (1970):
Strain-hardening parameters were estimated based on
the unconfined compression test data reported by Kim Rf ð1 sin ϕÞðσ 01 σ 03 Þ 2
et al. (2010). Strain-hardening was incorporated in the Et ¼ 1 Ei ð6Þ
2c cos ϕ þ 2σ 03 sin ϕ
model by setting EPS cohesion to cf = 41 kPa at a plastic
strain level of zero (i.e., at yield) and linearly increasing where Et is the tangent Young’s modulus, Rf is the failure
EPS cohesion to cf = 64 kPa at a strain level equivalent to ratio, ϕ is the soil friction angle, σ1′ and σ3′ are the major
30% axial strain. The material properties assumed for the and minor principal effective stresses, respectively, and c is
EPS geofoam model are summarized in Table 1. the soil cohesion.
The Poisson’s ratio of the Jumunjin sand was estimated
based on the ratio of horizontal to vertical stresses for
2.2. Soil (air-pluviated Jumunjin sand)
Test 1 (i.e., test box filled with sand only) reported by Kim
Kim et al. (2010) backfilled the test box using Jumunjin et al. (2010) using the equation:
sand air-pluviated from a drop height of approximately σh ν
50 cm. This backfilling technique resulted in a sand dry ¼ ð7Þ
σv 1 ν
unit weight of approximately γd = 14.8 kN/m3, corre-
sponding to a void ratio of approximately e = 0.74 based where σh and σv are the measured horizontal and vertical
on a specific gravity of 2.63. stresses, respectively, and ν is Poisson’s ratio. A value of
Geosynthetics International
3. SIMULATION OF MODEL-SCALE pipe axis) by 140 cm wide by 90 cm high. The base and
TESTS three of the four walls of the box were made of steel, while
one face of the four walls was made of transparent rigid
The model-scale tests performed by Kim et al. (2010) were acrylic plate.
simulated using the calibrated material models for EPS The model-scale test procedure consisted of three steps:
geofoam, Jumunjin sand and the steel pipe. Photographs (1) pipe placement and side fill; (2) EPS panel placement
of the test box illustrating the testing procedure are shown and backfilling; and (3) surcharging. During the first step,
in Figure 5. The details of the test chamber and test the pipe was placed along the base of the test box, and
procedure were presented by Kim et al. (2010). The box Jumunjin sand was pluviated into the box until the sand
had interior dimensions of 100 cm long (i.e., along the fill level reached the pipe crown. During the second step,
an EPS geofoam compressible inclusion was placed at the
Table 3. Material properties assumed for model-scale pipe level of the pipe crown (as applicable), and Jumunjin sand
was pluviated into the box from a drop height of
Parameter Value approximately 50 cm until the sand level reached the top
of the box. During the final step, a surcharge load was
Young’s modulus, E 200 GPaa applied in three stages (i.e., 49, 98 and 147 kPa) to the top
Moment of inertia, I 47 × 10−12 m4/ma
of the Jumunjin sand fill using a steel plate advanced by a
Cross-sectional area, A 1.5 × 10−3 m2/mb
Density, ρ 8 Mg/m3a
hydraulic system.
Two pressure transducers approximately 5 cm in diam-
a
Assumed as typical steel material properties. eter were mounted on the pipe. One transducer was fixed
b
Estimated based on measurements reported by Kim et al. (2010). at the pipe crown and was oriented to measure vertical
(a) (b)
(c)
Figure 5. Model test procedure by Kim et al. (2010): (a) cross-section of soil box, (b) air-pluviation Jumunjin sand, and (c) the placement
of EPS geofoam
Geosynthetics International
stress. The other transducer was attached to the pipe shear strength was assumed to be one-half of the estimated
springline and was oriented to measure horizontal stress. soil shear strength. The interface normal and shear
Both vertical and horizontal stress measurements were stiffnesses were estimated iteratively by fixing stiffness
recorded at the end of backfilling and after each surcharge values and comparing model-predicted stresses against
loading stage. stresses in Test 1 (i.e., test box filled with sand only)
reported by Kim et al. (2010). Assumed properties for the
3.1. Geometry and boundary conditions pipe–soil interface are listed in Table 4.
A plane-strain model was developed using FLAC (Itasca
2011) to simulate the Kim et al. (2010) model-scale test 3.3. EPS–soil interaction behavior
sequence. Due to symmetry, only half of the test box
EPS–soil interaction was modeled assuming a no-slip
was included in the simulation. The geometry and
condition. However, preliminary simulations were per-
boundary conditions of the numerical model are shown
formed to evaluate the effect of this assumption on
in Figure 6.
model-predicted stresses. The effect was evaluated as
Each simulation was separated into two phases:
change from the ‘baseline’ (i.e., no-slip assumption)
a pluviation phase and a surcharging phase. The pluvia-
model-predicted stresses at the pipe crown and springline.
tion phase consisted of a sequence in which one row of
For these preliminary simulations, an EPS–soil inter-
soil numerical zones at a time (for a lift thickness ranging
face was included in the model, and interface shear and
from 1.0 cm to approximately 3.3 cm) was activated
normal stiffness values as well as interface friction angle
and the model was allowed to reach static equilibrium.
values were varied. Interface stiffness values were varied
This sequence was performed for the first (i.e., lowest) row
from 600 × 103 kN/m/m to 9600 × 103 kN/m/m, and
of numerical zones and was repeated until all soil
interface friction angle values were varied from 1° (i.e.,
numerical zones in the model were active (i.e., rows of
similar to a full-slip condition) to 34° (i.e., similar to a
numerical zones were activated sequentially from the
no-slip condition). For the parameter ranges evaluated,
bottom to the top of the model). During the surcharging
model-predicted stresses generally showed modest sensi-
phase, a vertical stress boundary condition was applied
tivity to interface stiffness (i.e., approximately 5–15%
along the top of the model. The magnitude of this vertical
variation from ‘baseline’ values) and interface friction
stress was increased gradually (i.e., in increments of
angle (i.e., approximately 5–10% variation from ‘baseline’
approximately 2.5 kPa, allowing the model to equilibrate
values). In light of this modest sensitivity the no-slip
between increments) until reaching the maximum sur-
assumption was considered to be adequate for the
charge load applied in the model tests by Kim et al.
purposes of this study.
(2010).
Displacement constraints were applied as shown in
Figure 6. A horizontal displacement constraint was used 3.4. Effects of soil and EPS constitutive behavior
along the plane of symmetry. The lateral and rotational Although the model results presented here were developed
degrees of freedom were fixed for the structural nodes using nonlinear-elastic/Mohr–Coulomb soil constitu-
lying on this boundary at all stages of the analyses. During tive behavior, preliminary simulations were performed to
the pluviation phase, the lower boundary was fixed in the evaluate the effects of using linear-elastic/Mohr–Coulomb
vertical direction, and the right boundary was fixed in the soil constitutive behavior on model-predicted stresses.
horizontal direction. However, these boundaries were For these simulations, constant values of Young’s
fixed in both horizontal and vertical directions during modulus and Poisson’s ratio were assigned to all soil
the surcharging phase. It was assumed that friction zones in the numerical grid. Poisson’s ratio was fixed at a
developed along these boundaries could be significant value of ν = 0.39, while Young’s modulus was varied
(e.g., USACE 1994, recommendations suggest a sand– between a ‘lower-bound’ value of approximately 48 MPa
steel interface friction angle of approximately one-half of and an ‘upper-bound’ value of approximately 183 MPa.
the sand internal friction angle, or approximately 17° to These Young’s modulus values were taken as representa-
18° for the sand considered here) and could be idealized tive of model-calculated (using the nonlinear-elastic/
using a no-slip condition. It is noted that boundary Mohr–Coulomb soil model) values at the end of back-
friction effects are evident in the test results by Kim filling and the end of surcharging, respectively.
et al. (2010) test: for Test 1 (i.e., soil only), the measured The effect of the soil constitutive model was evaluated
vertical earth pressure at the end of surcharging as change from the ‘baseline’ (i.e., nonlinear-elastic/
(i.e., 139.3 kPa) was approximately 13% less than the Mohr–Coulomb) model-predicted stresses at the pipe
sum of the self-weight and surcharge loads (i.e., crown and springline. While agreement of model-
12.1 kPa + 147.1 kPa = 159.2 kPa). Thus, the model predicted stresses with the ‘baseline’ was reasonably
allowed the development of friction between the sand good for the ‘lower-bound’ Young’s modulus value (i.e.,
fill and the right boundary. approximately 20–25% variation from ‘baseline’ values),
the variation was significant for the ‘upper-bound’
3.2. Pipe–soil interaction behavior Young’s modulus value (i.e., approximately 60–65%
Pipe–soil interaction was modeled as a frictional interface variation from ‘baseline’ values). In light of this
using the FLAC software’s built-in interface logic. Based sensitivity to the constant stiffness value selected, it is
on recommendations by USACE (1994), the interface recommended to assume nonlinear-elastic soil behavior,
Geosynthetics International
Surcharge load
(surcharging phase)
Jumunjin
sand
EPS
Model boundary fixed
Structural nodes geofoam
-
in y-direction
fixed x- and r- (pluviation phase) and
directions Steel pipe with x- and y-directions
frictional pipe - (surcharging phase)
soil interface
70 cm
(a)
Surcharge load
(surcharging phase)
Jumunjin
sand
70 cm
(b)
Figure 6. Geometry and boundary conditions for the test box model, (a) single layer of EPS geofoam, and (b) double layer of EPS
geofoam
Geosynthetics International
Table 5. Model-scale test variables (D denotes diameter of pipe, Wgeofoam denotes the width of EPS, and SEPS panel denotes spacing
between EPS panels
The presence of EPS geofoam inclusion Test factor Variables Test no.
No EPS Stress distribution of fill soil Sand deposit without pipe Test 1
Sand deposit with pipe Test 2
Single layer of EPS The width of EPS Wgeofoam = 1.0D Test 3
(single layer of EPS geofoam inclusion) Wgeofoam = 1.5D Test 4
Wgeofoam = 2.1D Test 5
Double layer of EPS Spacing between EPS geofoam inclusions SEPS panel = 0.5D Test 6
(for all cases, Wgeofoam = 1.0D) SEPS panel = 1.0D Test 7
SEPS panel = 1.2D Test 8
SEPS panel = 1.5D Test 9
or compressible inclusion).
• Test 2 – the test box contained sand and the pipe (i.e., 150
Horizontal stress at pipe springline (kPa)
σv, crown
(a) σh,springline
σv,crown
σv, crown
Figure 8. Schematic illustration showing possible pipe deformation patterns for (a) uniform stress around the pipe; (b) vertical stresses
larger than horizontal stresses; and (c) horizontal stresses larger than vertical stresses. Solid and dashed lines indicate original and
deformed pipe cross-section
schematically in Figure 8. Therefore, ‘optimization’ of Considering both ‘optimization’ criteria, it appears that
EPS panel dimensions should consider at least the an ‘optimal’ wgeofoam/Dpipe ratio should be as narrow
following criteria: (1) the magnitude of vertical stress at as possible but at least wgeofoam/Dpipe = 1.5. Therefore,
the pipe crown, σv,crown, should be reduced to the extent for the conditions considered here, the ‘optimal’ value of
practicable; and (2) an acceptable ratio of horizontal stress the wgeofoam/Dpipe = 1.5 ratio is approximately wgeofoam/
at the pipe springline to vertical stress at the pipe crown, Dpipe = 1.5 (i.e., wgeofoam = 15 cm).
σh,springline/σv,crown, should be maintained. Regression equations may be used to facilitate the
In general, a lower overburden stress is preferred to ‘optimization’ process. For the configurations examined
a higher overburden stress. Additionally, based on the here, the effect of the normalized EPS panel width
σh,springline/σv,crown ratio reported by Kim et al. (2010) for (wgeofoam/Dpipe) on the normalized vertical and horizontal
Test 2 (i.e., sand and pipe only), it appears that σh,springline/ stresses at the pipe crown and springline, respectively, may
σv,crown = 1.0 ± 0.1 is a reasonable target. Therefore, the be approximated by:
goal of this example ‘optimization’ is to minimize σv,crown
σ v;crown wgeofoam
while maintaining 0.9 ≤ σh,springline/σv,crown ≤ 1.1. ¼ 008 ln þ 027;
σ surcharge Dpipe
ð9Þ
wgeofoam
05 25
4.1. Effects of EPS panel width Dpipe
The first variable investigated during the ‘optimization’ and
procedure was the ratio of EPS panel width to pipe
diameter, wgeofoam/Dpipe, with a fixed EPS panel thickness σ h;springline wgeofoam 0:39
¼ 133 ;
of tgeofoam = 5 cm. Numerical results for the variation of σ v;crown Dpipe
wgeofoam/Dpipe are shown in Figure 9. For comparison, ð10Þ
wgeofoam
model test results by Kim et al. (2010) are also plotted in 05 25
Figure 9. Dpipe
As shown in Figure 9a, the magnitude of σv,crown Following this approach, the ‘optimization’ objective is
initially decreases with the addition of an EPS panel but to minimize σv,crown/Δσsurcharge, by changing wgeofoam/
increases gradually as the EPS panel is widened. Applying Dpipe, subject to the constraint 0.9 ≤ σh,springline/
the first ‘optimization’ criterion (i.e., significant reduction σv,crown ≤ 1.1. The constraint is satisfied on the interval
of σv,crown), a narrow panel is preferable over a wide panel of approximately 1.5 ≤ wgeofoam/Dpipe ≤ 2.5. For this inter-
(i.e., the panel should be as narrow as possible). val, σv,crown/Δσsurcharge is minimized at a value of wgeofoam/
Meanwhile, as shown in Figure 9b, the σh,springline/ Dpipe = 1.5.
σv,crown ratio is relatively high for a narrow EPS panel
but approaches a value of σh,springline/σv,crown that is
practically close to 1.0 as the panel is widened. Applying 4.2. Effects of EPS panel thickness
the second ‘optimization’ criterion (i.e., 0.9 ≤ σh,springline/ The second variable investigated during the ‘optimization’
σv,crown ≤ 1.1), the ratio of EPS panel width to pipe procedure was the thickness of the EPS panel, tgeofoam,
diameter should be at least wgeofoam/Dpipe = 1.5. with a fixed ratio of EPS panel width to pipe diameter of
Geosynthetics International
1.2 1.2
0.8 0.8
σv,crown /σsurcharge = 0.08 ln(Wgeofoam /Dpipe ) + 0.27 σv,crown /σsurcharge = –0.09 ln(t geofoam ) + 0.45
0.6 (R2 = 0.99) 0.6 (R2 = 0.98)
Test 2 Test 2
0.4 0.4
0.2 0.2
Test 3 Test 4 Test 5 Test 4
0 0
0 0.5 1 1.5 2 2.5 3 0 4 8 12 16 20
Wgeofoam /Dpipe (unitless) t geofoam (cm)
(a) (a)
2.1 2.1
1.5 1.5
1.2 1.2
0.6 Test 1
0.6 Test 1
σh,springline /σv,crown = 1.33 (Wgeofoam /Dpipe )–0.39 σh,springline /σv,crown = 0.94 (t geofoam )0.12
2
(R = 0.95) 0.3 (R2 = 0.95)
0.3
0 0
0 0.5 1 1.5 2 2.5 3 0 4 8 12 16 20
Wgeofoam /Dpipe (unitless) tgeofoam (cm)
(b)
(b)
Figure 9. Effects of varying ratio of EPS panel width to pipe Figure 10. Effects of varying EPS panel thickness while main-
diameter while maintaining constant EPS panel thickness taining constant ratio of EPS panel width to pipe diameter
(tgeofoam = 5 cm) on (a) vertical earth pressure measured at the pipe (wgeofoam = 1.5Dpipe) on (a) vertical earth pressure measured at the
crown; and (b) ratio of horizontal stress at the pipe springline to pipe crown; and (b) ratio of horizontal stress at the pipe springline
vertical stress at the pipe crown to vertical stress at the pipe crown
Test 7 Test 8
were performed. Spacing between panels (spanels, i.e.,
σh,springline /σ
distance from the top of the lower panel to the bottom 0.9
Geosynthetics International
Vaslestad, J., Johansen, T. H. & Holm, W. (1994a). Load reduction on Behaviour of Flexible Large-Span Culvert, Director of
rigid culverts beneath high fills: long-term behavior. Transportation Public Roads, Norwegian Road Research Laboratory, Oslo,
Research Record, No. 1415, 58–68. Norway.
Vaslestad, J., Johansen, T. H. & Holm, W. (1994b). Load Zarnani, S. & Bathurst, R. J. (2007). Experimental investigation of EPS
Reduction on Buried Rigid Pipes; Load Reduction on Rigid geofoam seismic buffers using shaking table tests. Geosynthetics
Culverts Beneath High Fills: Long Term Behaviour; Long-Term International, 14, No. 3, 165–177.
The Editor welcomes discussion on all papers published in Geosynthetics International. Please email your contribution to
discussion@geosynthetics-international.com.
Geosynthetics International