Computers and Geotechnics 79 (2016) 31–40

Contents lists available at ScienceDirect

Computers and Geotechnics

journal homepage: www.elsevier.com/locate/compgeo

Research Paper

Model test and numerical simulation on rigid load shedding culvert

backfilled with sand
Baoguo Chen a,⇑, Dingbao Song a, Xinying Mao a, Elton J. Chen b, Jun Zhang c
a
Engineering Faculty, China University of Geosciences (Wuhan), Lumo Road 388, Hongshan, Wuhan 430074, China
b
School of Civil Engineering & Mechanics, Huazhong University of Science & Technology, Wuhan 430074, China
c
Key Lab of Highway Construction & Maintenance Technology in Loess Region, Ministry of Transport, PRC, Shanxi Transportation Research Institute, Taiyuan 030006, China

a r t i c l e i n f o a b s t r a c t

Article history: For reducing embankment load on box culvert, a new culvert structure called load shedding culvert (LSC)
Received 20 December 2015 is proposed. First, model tests were conducted to investigate the performance of LSC, which was com-
Received in revised form 20 April 2016 pared with embankment-installation box culvert (EBC) and imperfect-trench-installation box culvert
Accepted 29 May 2016
(IBC). Then, numerical simulations were performed to investigate the stress states of the three culvert
structures and associated important influencing factors. This study reveals that the LSC, compared with
EBC and IBC, can not only reduce the vertical earth pressure on the top slab but also reduce the horizontal
Keywords:
earth pressure on the culvert sidewall.
Rigid culvert
Load shedding
Ó 2016 Elsevier Ltd. All rights reserved.
Earth pressure
Model test
Numerical simulation
Soil arch

1. Introduction
Rigid box culverts under high embankment are widely used in
highway and railway projects. Although culvert structures are rel-
atively simple, the interaction between soil and structure can be
complex. The rigid embankment-installation box culvert (EBC)
has its top projecting the natural ground as shown in Fig. 1(a).
The stiffness of the culvert is higher than that of the adjacent fill
mass, and therefore, the settlement of surrounding soil prism is
larger than that of the central soil prism above the culvert. The
shear stress (or friction stress) between the surrounding soil prism
and the central soil prism results in backfill pressure concentration
on the culvert [1–5]. The extra load on culvert often leads to struc-
tural hazards of the culvert [6,7].
In order to reduce the backfill load on the culvert, several meth-
ods are commonly used in construction process, among which, the
imperfect-trench-installation box culvert (IBC) is usually used in
the embankment engineering. This IBC method was originally pro-
posed by Marston [8,9]. In this method, a compressible layer is set
on the culvert top to generate soil arch in the backfill, which can
transfer the load of the central fill mass over the culvert to the
⇑ Corresponding author.
E-mail addresses: baoguo_chen@126.com (B. Chen), songdingbao0214@163.com
(D. Song), 1150636438@qq.com (X. Mao), eltonjchen@hust.edu.cn (E.J. Chen),
zhangjun_hust@qq.com (J. Zhang).
adjacent fill mass, such that the vertical earth pressure on the cul-
vert is reduced, as shown in Fig. 1(b). Based on of Marston’s work,
further study was carried out by Spangler [10]. The performance of
imperfect-trench-installation culvert were investigated by field
tests, experimental and numerical methods [11–14]. Kim and
Yoo [15] conducted a numerical study to investigate the effect of
several important factors (i.e., the width, thickness, location, and
stiffness of the compressible layer) on the performance of the
box culvert. Kim and Yoo also pointed out that the width of com-
pressible layer should not be greater than 1.5 times the box culvert
width and the greatest load reduction occurred when the com-
pressible layer was placed immediately on the culvert top.
The imperfect-trench-installation method is recognized as one
of the acceptable methods in the prevailing AASHTO LRFD bridge
design specifications [16]. However, there is no guideline to guide
the installation process of the box culvert. Instead, AASHTO sug-
gests using previous experience to determine the load on the cul-
vert. Moreover, the design method of the load reduction is not
given in the prevailing Chinese General Code for Designing High-
way Bridges and Culverts [17].
The current load reduction methods are applied on the normal
culvert structures (i.e. single box culvert, pipe culvert, twin box
culvert). The load of the central fill mass over the culvert is trans-
ferred to adjacent fill mass after the load reduction; however, it
also increases the horizontal earth pressure on the culvert sidewall,

http://dx.doi.org/10.1016/j.compgeo.2016.05.026
0266-352X/Ó 2016 Elsevier Ltd. All rights reserved.
32 B. Chen et al. / Computers and Geotechnics 79 (2016) 31–40
Backfill surface
central
fill mass surrounding
fill mass
BC
Natural ground
(a)
Fig. 1. Sketches of culverts. (a) Embankment-installation Box Culvert (EBC); (b) imperfect-trench-installation Box Culvert (IBC); (c) load shedding Culvert (LSC).
which is even larger than the vertical earth pressure on the top of
the culvert [11,14,18–22].
A new culvert structure, called load shedding culvert (LSC), is
proposed in this paper. In the LSC, two load shedding blocks
(LSB) are set on both sidewalls of box culvert as an integral struc-
ture. The load shedding hole (LSH) between two load shedding
blocks is filled with compressible material, as shown in Fig. 1(c).
In this study, the stress states of the LSC were investigated by phys-
ical model tests and numerical analyses. The results of LSC were
compared with those of EBC and IBC. A parametric study was con-
ducted to analyze the significant influencing factors on culvert per-
formance, including the friction angle of backfill, the heights of
load shedding blocks (for comparison, the thicknesses of compress-
ible material layer in cases IBC and LSC are equal to those of load
shedding blocks), the width of load shedding blocks, and the elastic
modulus of compressible material.
2. Model test
2.1. Model test setup
Model test system consists of a tank and two types of culvert.
The tank is 1500 mm long, 1500 mm wide and 1650 mm high as
shown in Fig. 2. Steel plate (2 mm thick) was used for three side-
walls of the tank and a toughened glass (10 mm thick) was used
for the other sidewall of the tank for observation. Two types of cul-
vert model, i.e., normal Box Culvert and LSC, are shown in Fig. 3.
Aluminum alloy plate (6 mm thick) was used for both of the cul-
vert models.
The tank was placed directly on the concrete floor in laboratory
with a maximum filling height of 1650 mm. In view of the low
compressibility of the ground, it was assumed that the model tests
were conducted on a rigid foundation.
2.2. Model test procedure
Three different cases (EBC, IBC, and LSC) were considered for
comparison and to investigate the performance of the culvert.
The thickness of compressible layer was 30 mm and the widths
were equal to that of the top slab of the culvert in both IBC and LSC.
Backfill surface Backfill surface
central
fill mass
compressible
material
LSB
Compressible
material
BC LSC
Natural ground Natural ground
(b) (c)
A total of 14 miniature pressure cells (LY350 produced by
Changsha Kun Peng Testing Technology Co., Ltd.) were used in
the model test. The miniature pressure cell is 10 mm thick and
28 mm in diameter, with a maximum stress range of 100 kPa and
sensitivity of 0.01 kPa.
The layout of all the earth pressure cells is shown in Fig. 2. The
pressure cells were symmetrically placed along the axis of culvert.
To measure the vertical earth pressure, four pressure cells were set
on the culvert (No. 1 to No. 4) and four pressure cells were placed
in sand within the same plane (No. 5 to No. 8). Another six pressure
cells were set on both sidewalls of culvert to measure the horizon-
tal earth pressure (No. 9 to No. 14).
For comparison, the pressure cells (from No. 1 to No. 8) were
placed in the same plane and 30 mm over the culvert top slab,
which is exactly the thickness of the compressible layer for cases
IBC and LSC. Thus, the maximum height of backfill on pressure cells
(from No. 1 to No. 8) was 1400 mm.
The procedural steps of these model tests are described as
follows:
(1) For all tests, a culvert model was installed in the tank (A nor-
mal box culvert model was used for the EBC and IBC cases).
(2) Backfilling with sand. For the EBC case, the initial filling
height was 30 mm above the top slab, whereas for the IBC
and LSC cases, a 30 mm-thick compressible layer was placed
on the top slab of culvert and the surrounding fill should be
graded to the top of the compressible layer.
(3) The backfill over culvert was installed step by step. In each
step a 200 mm-thick layer was filled in one day, until the
backfill reached a maximum height of 1430 mm.
2.3. Material properties
The sands used in the model tests were from Hankou riverside
of Yangtze River. The properties of the sand are presented in
Table 1. The average unit weight of the fill in the tank is 17.5
kN/m3.
Sponge (polyurethane foam) was used as the compressible
material in the model tests. The density of the sponge was
3.2 kg/m3, the elastic modulus was 300 kPa and the Poisson’s ratio
was 0.01 (all of the mechanical parameters were supplied by the
sponge factory).
 B. Chen et al. / Computers and Geotechnics 79 (2016) 31–40 33

1500 1500
630 240 630 630 240 630

500
Sand
2 1 3

EBC or IBC LSC

1500
390 240 240 390

500

1650
12 9 12 9
13 10 13 10
7 5 4 6 8 14 11 14 11

500
Culvert

Toughed Glass Wall Steel Plate Wall Earth pressure cell

(a) top view (b) side view

Fig. 2. Schematic diagram of the model test (units in millimeter).

the mechanism of load reduction on culverts could be gained and

240 be used to improve the current design methods of load reduction.
240
40 160 40
2.4. Test results
30

2.4.1. Pressure on culvert top slab

6 6 The vertical earth pressures at the center (average pressure
measured by No. 1 and No. 4 cells) and the edge (average pressure
183

183

measured by No. 2 and No. 3 cells) of the top slab are presented in
Fig. 4. It can be found that the vertical earth pressures (VEP) on the
top slab increased with the backfill height.
As shown in Fig. 4(a), the VEP at the center of the top slab of IBC
(a) EBC or IBC (b) LSC or LSC is smaller than that of EBC. The vertical earth pressure ratio
(kvepr, defined as the ratio of the measured vertical earth pressure
Fig. 3. Dimensions of culvert models (units in millimeter). to the overburden pressure) of EBC is larger than 1.0, whereas
those of IBC and LSC are smaller than 1.0. Fig. 4(a) also shows that
kvepr of IBC and LSC decreased with increasing backfill height. It
Table 1
demonstrates that the load reduction effect is enhanced with the
Properties of Wuhan sand. increase of backfill height. The VEP of LSC is 30% less than that of
IBC at the maximum backfill height (1430 mm).
Properties Value
Fig. 4(b) indicates that the VEP on the edge of the top slab
Specific gravity Gs 2.65 increased with the backfill height. The kvepr on the edge of the
Effective diameter D10 = 0.1 mm top slab in case IBC is less than 1.0, whereas that in cases EBC
D60 = 0.8 mm and LSC are larger than 1.0. The results demonstrate that the ver-
Dmax = 2.0 mm
Curvature coefficient Cc = 1.13
tical earth pressure concentrated on the edge of the top slab in
Uniformity coefficient Cu = 8.00 cases EBC and LSC. The distribution of the VEP on the top slab is
Maximum void ratio emax 0.90 not uniform and the VEP on the edge of the top slab is larger than
Minimum void ratio emin 0.55 that at the center for all cases.

2.4.2. Pressure in the adjacent backfill mass

The model testing was conducted on a reduced scale model at
1-g and the principles of the material similarity between the pro-
totype and model were not strictly maintained, and thus there
may be some differences between strains of the prototype and
those of the model. However, this aspect was considered in the
design of the model test. Aluminum plate with 6 mm thickness
was used in the model test, which can ensure that the strain mag-
nitude of model culvert and prototype was at the same level.
Although the results obtained from the model tests may not be
directly extrapolated to real culverts, a better understanding of
Fig. 5 shows the variation of VEP in the adjacent backfill mass
(measured by pressure cells No. 5 to No. 8) as a function of the
backfill height H. As shown in Fig. 5(a), the VEP (average pressure
measured by pressure cells No. 5 and No. 6) right adjacent to the
culvert sidewall is larger in case IBC than that in cases EBC and
LSC. The maximum kvepr right next to the culvert sidewall is
approximately 1.7 in case IBC, while that is slightly smaller than
1.0 (between 1.8 and 1.0) in cases EBC and LSC. Fig. 5(b) shows that
the VEP in the adjacent backfill mass (average pressure measured
by pressure cells No. 7 and No. 8) within one culvert width to
34 B. Chen et al. / Computers and Geotechnics 79 (2016) 31–40

35 1.6 90 4.5
EBC VEP EBC kvepr EBC VEP EBC kvepr
30 1.4
IBC VEP IBC kvepr 75 IBC VEP IBC kvepr 3.6
25 LSC VEP LSC kvepr 1.2 LSC VEP LSC kvepr
60
1.0
VEP (kPa)
2.7

VEP (kPa)
20

kvepr
kvepr
0.8 45
15 1.8
0.6
30
10
0.4
15 0.9
5 0.2
0 0.0 0 0.0
200 400 600 800 1000 1200 1400 200 400 600 800 1000 1200 1400
H ( mm) H (mm)
(a) measured by No. 1 and No. 4 (b) measured by No. 2 and No. 3
Fig. 4. Variations of VEP and kvepr of the top slab with H.

3.0 35 1.3
EBC VEP EBC kvepr EBC VEP EBC kvepr
50
IBC VEP IBC kvepr 2.5 30 IBC VEP IBC kvepr 1.2
40 LSC VEP LSC kvepr 25 LSC VEP LSC kvepr
2.0
VEP (kPa)

VEP (kPa) 20 1.1

30
kvepr

kvepr
1.5
15 1.0
20 1.0
10
10 0.5 0.9
5

0 0.0 0 0.8
200 400 600 800 1000 1200 1400 200 400 600 800 1000 1200 1400
H (mm) H (mm)
(a) measured by No. 5 and No. 6 (b) measured by No. 7 and No. 8
Fig. 5. Variations of VEP and kvepr on the adjacent backfill mass with H.

the culvert sidewall is approximately equal to the overburden
pressure. The kvepr in case IBC is slightly larger than that in cases
EBC and LSC.
2.4.3. Horizontal earth pressure on culvert sidewall
The horizontal earth pressures (HEP) on culvert sidewall (mea-
sured by pressure cells No. 9 to No. 14) are presented in Fig. 6. The
HEP increased with increasing backfill height. When the fill height
varied from 0 to 600 mm, the HEP values of all cases were very
close. However, when the backfill height was over 600 mm, the
HEP in case IBC was markedly larger than that in cases EBC and
LSC. At the maximum backfill height, the HEP in case LSC decreased
by 13% of that in case IBC. However, the horizontal earth pressure
ratios (khepr, defined as the ratio of the measured horizontal earth
pressure to the overburden pressure at the test point) fluctuated in
a wide range before the fill height reached 600 mm, which might
be caused by the disturbance of the backfilling process. When
the fill height is larger than 600 mm, the khepr reached a steady
state in all cases.
3. Numerical simulations
3.1. Numerical model
A series of two-dimensional finite-difference numerical simula-
tion using FLAC(2D) was conducted to investigate the stress state
of the culverts. A model calibration was first implemented to vali-
date the numerical model to the real practice of the culvert-backfill
system, and then a parametric study was carried out to investigate
the significant influencing factors to the stress state of the culverts.
The numerical model was created with the same dimensions as
the physical model shown in Figs. 2 and 3. In the numerical anal-
ysis, layered loading was carried out to simulate the backfilling
process, with the height of each layer = 200 mm.
The vertical boundaries of the model were treated as a fixed
boundary without horizontal displacements. The bottom was con-
strained by fixing both vertical and horizontal displacements. The
sand used in the model tests was assumed to be elasto-plastic
material with Mohr-Coulomb failure criterion. The culvert and
the compressible layer (sponge) were assumed to be isotropic
and linearly elastic material. The continuum zones were used in
the model of backfill and each zone were divided into small grids.
The culverts were modeled using structural element. Interface ele-
ments were used between the culvert and soil or compressible
material. The contact boundary between the culvert and sand
was modeled by a Coulomb shear strength interface. The friction
angle of the aluminum-sand interface was set as 36° [23,24], with
no cohesion or dilation considered. The normal and shear stiffness
of soil-culvert interface were set as approximately 10 times the
stiffness of aluminum [25].
3.2. Material parameters
The material parameters of the sand used in the numerical sim-
ulations were determined by triaxial tests at a confining pressure
of 20 kPa. Following the Duncan-Chang constitutive model [26],
the tangent modulus of sand was expressed as:
 n  2
r3 Rf ð1  sin /Þðr1  r3 Þ
Et ¼ kpa 1 ð1Þ
pa 2c cos / þ 2r3 sin /
 B. Chen et al. / Computers and Geotechnics 79 (2016) 31–40 35

12 0.7 15 0.8
EBC HEP EBC khepr EBC HEP EBC khepr
10 IBC HEP IBC khepr 0.6 IBC HEP IBC khepr 0.7
12
LSC HEP LSC khepr LSC HEP LSC khepr
8 0.5 0.6

HEP (kPa)
HEP (kPa) 9

khepr
khepr
6 0.4 0.5
6
4 0.3 0.4
3 0.3
2 0.2

0 0.1 0 0.2
0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400
H (mm) H (mm)
(a) measured by No. 9 and No. 12 (b) measured by No. 10 and No. 13

15 0.8
EBC HEP EBC khepr

12 IBC HEP IBC khepr 0.7

LSC HEP LSC khepr
0.6
HEP (KPa)

khepr
0.5
6
0.4
3 0.3

0 0.2
0 200 400 600 800 1000 1200 1400
H (mm)
(c) measured by No. 11 and No. 14
Fig. 6. Variations of HEP and khepr of the culvert sidewall with H.

where k = modulus number; pa = atmospheric pressure; n = modu- Table 3

lus exponent; Rf = failure ratio; c = cohesion; / = friction angle; Properties of the materials used in numerical simulations.

r1 ; r3 = maximum and minimum principal stress. Material Elastic Poisson’s Internal Unit weight
In Eq. (1), the values of k, n, and Rf are determined from stress- modulus ratio friction angle (kN/m3)
strain relationship of the sand. The stress-strain parameters (MPa) (°)

derived from the results of the triaxial tests are summarized in
Table 2.The material properties used in the numerical simulations
are presented in Table 3.
3.3. Model calibration
The comparison between the model tests and the numerical
results is presented in Table 4. The earth pressures measured in
the model tests are compared with those obtained from the
numerical simulations. The comparisons show reasonable
agreement.
3.4. Culvert performance
The average vertical earth pressure on the top slab and the aver-
age horizontal earth pressure on the sidewall are investigated by
numerical simulations in this study. In case LSC, two different aver-
age vertical earth pressures are considered: one is the average ver-
tical earth pressure on the load shedding hole (LSC-LSH), and the
other is the average vertical earth pressure on the entire top slab
(LSC-WTS), including load shedding hole and load shedding blocks.
Table 2
Summary of stress-strain parameters for Wuhan Sand.
Parameters n Rf k c
Wuhan sand 0.6 0.63 365
Aluminum 70,000 0.32 – 27.50
alloy
culverta
Sand 7.27 0.30 39 17.50
Spongea 0.30 0.01 – 3.20
a
The properties of the sponge and aluminum are applied by the respective
factories.
As shown in Fig. 7(a), the average vertical earth pressures (in
cases EBC, IBC, and LSC) increased with increasing backfill height.
In case EBC, the average vertical earth pressure was larger than
those in cases IBC and LSC-LSH. Fig. 7(b) shows that the average
vertical earth pressure coefficients (kvepc, defined as the ratio of
the average vertical earth pressure on the culvert to the overbur-
den pressure) in cases EBC and LSC-WTS increased slightly with
increasing backfill height and approached asymptotic value at
higher backfill height. The kvepc in case LSC-WTS is approximately
equal to 1.0, which means that the average earth pressure in case
LSC is around the overburden pressure. Fig. 7(b) also shows that
the kvepc in cases IBC and LSC-LSH decreased with increasing back-
fill height and reached a steady value at higher backfill height.
Fig. 7(c) shows that the average horizontal earth pressures
increased with increasing backfill height. The horizontal earth
pressure in case LSC was reduced by 17% of that in case IBC at
the maximum backfill height. There was no significant difference
in the horizontal earth pressures between cases EBC and LSC.
/
Fig. 7(d) shows that the horizontal earth pressure coefficient (khepc,
0 39
defined as the ratio of the average horizontal earth pressure on the
36 B. Chen et al. / Computers and Geotechnics 79 (2016) 31–40

Table 4
Comparison of numerical results with model test measurements.

Cases EBC IBC LSC

Avg. pressure Model test/kPa FLAC/ The difference (%) Model test/kPa FLAC/ The difference (%) Model test/kPa FLAC/ The difference (%)
kPa kPa kPa
No. 1 & No. 4 26.28 24.35 7.3 12.38 11.5 7.1 8.47 8.52 0.6
No. 2 & No. 3 46.17 45.13 2.3 15.07 12.95 14.1 67.23 68.71 2.2
No. 5 & No. 6 21.05 22.14 5.2 40.73 41.65 2.3 23.10 20.56 11.0
No. 7 & No. 8 22.31 23.02 3.2 26.24 25.91 1.3 23.04 23.35 1.3
No. 9 & No. 12 6.95 6.75 2.9 8.58 8.75 2.0 6.25 5.94 5.0
No. 10 & No. 13 10.57 10.41 1.5 11.98 12.04 0.5 10.27 10.63 3.5
No. 11 & No. 14 10.72 9.69 9.6 12.50 13.05 4.4 10.87 10.32 5.1
Average vertical earth pressure (kPa)

40 1.4
EBC
IBC 1.2
32
LSC-LSH
LSC-WTS 1.0
24 EBC

kvepc
0.8 IBC
16 LSC-LSH
0.6 LSC-WTS

8 0.4

0 0.2
200 400 600 800 1000 1200 1400 200 400 600 800 1000 1200 1400
H (mm) H (mm)

(a) average vertical earth pressures (b) kvepc

Average horizontal earth pressure (kPa)

0.48
14 EBC
0.45
12 IBC
LSC
10 0.42 EBC
IBC
khepc

8 0.39 LSC
6
0.36
4
0.33
2

0 0.30
200 400 600 800 1000 1200 1400 200 400 600 800 1000 1200 1400
H (mm) H (mm)
(c) average horizontal earth pressures (d) khepc
Fig. 7. Variations of earth pressure with H.

culvert sidewall to the overburden pressure on the mid-height of the pressures on the load shedding hole and load shedding blocks.
the culvert) in case IBC is larger than that in cases EBC and LSC. Fig. 8(b) shows the distribution of horizontal earth pressures on
The khepc of EBC and LSC decreased slightly with increasing backfill culvert sidewalls at the maximum backfill height of 1430 mm.
height. However, the khepc in case IBC increased slightly with The maximum horizontal earth pressures in cases LSC and EBC
increasing backfill height and larger than that in cases EBC and occurred at the mid-height of the sidewall, but the difference of
LSC. The results also demonstrate that the khepc in case IBC is more the horizontal earth pressures between cases LSC and EBC was
easily affected by backfill height than those in cases EBC and LSC. small. In case IBC, however, the maximum horizontal earth pres-
Fig. 8(a) shows the distribution of vertical earth pressure on the sure appeared at the bottom of the culvert sidewall. Fig. 8(b) also
top slab at the maximum backfill height of 1430 mm. The load shows that the horizontal earth pressure in case IBC was markedly
shedding blocks showed in Fig. 8(a) are only for case LSC. In case larger than that in cases EBC and LSC. The maximum difference of
EBC, the vertical earth pressure showed a parabolic distribution the horizontal earth pressure between cases LSC and IBC is approx-
on the top slab. In case IBC, the vertical earth pressure was approx- imately 30%.
imately uniformly distributed on the top slab. In case LSC, the ver- The variations of average contact pressure on the bottom slab
tical earth pressure on the load shedding block was larger than that and contact pressure coefficients (kecp, defined as the ratio of the
on the load shedding hole and there was a sharp change between average contact pressure on the bottom slab to the sum of average
 B. Chen et al. / Computers and Geotechnics 79 (2016) 31–40 37

70 180
EBC
60
IBC 150

Culvert height (mm)

50 LSC
120
VEP (kPa) 40
90
30 EBC
60 IBC
20 LSC
10 30

0 0
0 40 80 120 160 200 240 4 5 6 7 8 9 10 11 12 13 14
LSB
Culvert top slab width (mm) LSB HEP (kPa)
(a) VEP on the top slab (b) HEP on the culvert sidewall
Fig. 8. Distribution of the earth pressure on the culvert.

36 1.8
Average contact pressure (kPa)

32
EBC 1.6
28 IBC
LSC
24 1.4

20
kepc

1.2
EBC
16
IBC
12 1.0 LSC

8
0.8
4
0 0.6
200 400 600 800 1000 1200 1400 200 400 600 800 1000 1200 1400
H (mm) H (mm)
(a) (b)
Fig. 9. Variation of contact pressure on the bottom slab with H: (a) average contact pressure (b) kepc.

vertical earth pressure on the top slab and culvert dead load) are In the parametric study, the dimensions of numerical models,
shown in Fig. 9. As shown in Fig. 9(a), the average contact pres- boundary conditions, and constitutive models for materials are
sures in all of the three cases increased with increasing backfill consistent with the aforementioned numerical simulation. For
height. Fig. 9(b) shows that kecp in cases EBC and LSC increased each series of simulations, only the variable of interest will be
slightly with increasing backfill height but the reverse is true in changed, while other variables are held constant and equal to those
case IBC. In case EBC, kecp is approximately 1.62 which is larger for the baseline case.
than the vertical earth pressure coefficient on the top slab (kvepc
is 1.35 as shown in Fig. 7(b)) at the maximum backfill height.
Values of kecp and kvepc are 1.01 and 1.36 in case IBC, 0.45 and 3.5.1. Friction angle of the backfill
0.78 in case LSC, respectively. In all of the three cases, the contact Fig. 10 presents the effect of the friction angle of backfill on the
pressure coefficients on the bottom slab are larger than the vertical culvert performance. As shown in Fig. 10(a), kvepc in cases LSC-LSH
earth pressure coefficients on the top slab. The numerical results and IBC is smaller than 1.0 and decreases nonlinearly with increas-
indicate that the drag forces on the sidewalls lead to increase of ing backfill friction angle. These results indicate that the load
the contact pressure on the bottom slab. reduction effect is amplified with increasing backfill friction angle.
Moreover, kvepc in case LSC-WTS increases with increasing backfill
3.5. Analysis of influencing factors friction angle and is much higher than that in cases LSC-LSH and
IBC. Fig. 10(b) shows that khepc decreases nonlinearly with increas-
A parametric study was carried out to investigate the effect of ing backfill friction angle. It can be seen that khepc in case LSC is
influencing factors on the performance of the LSC, and the results smaller than that in case IBC. khepc in cases LSC and IBC decreases
are compared with that of the IBC since the case IBC can also sharply even when the backfill friction angle increases in a small
reduce the vertical earth pressure on culvert. The influencing fac- range (approximately from 20 to 25). When the backfill friction
tors include the friction angle of backfill, the elastic modulus’s of angle is high, however, khepc in case IBC decrease with increasing
compressible material and backfill, the width of load-shedding friction angle, while that in case LSC increases slowly with increas-
blocks, as well as the height of load shedding blocks (or the thick- ing friction angle. The results demonstrate that increasing the
ness of compressible material layer). The investigated variables of backfill friction angle can help reduce the horizontal earth pressure
the culvert performances include the vertical earth pressure coef- on the culvert sidewall and enhance the load reduction effect of the
ficient (kvepc) and the horizontal earth pressure coefficient (khepc). vertical earth pressure on the top slab.
38 B. Chen et al. / Computers and Geotechnics 79 (2016) 31–40

1.2 0.60
LSC
1.0 0.55
IBC

0.50
0.8 LSC-LSH
kvepc

khepc
LSC-WTS 0.45
0.6 IBC
0.40

0.4 0.35

0.2 0.30
20 25 30 35 40 45 20 25 30 35 40 45
φ φ
(a) kvepc (b) khepc
Fig. 10. Influence of the backfill friction angle u on the culvert performance.

1.6 0.55
LSC-LSH LSC
1.4 LSC-WTS IBC
IBC 0.50
1.2

1.0
0.45
kvepc

khepc

0.8

0.6 0.40

0.4
0.35
0.2

0.0 0.30
0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140
Ef / E S Ef / ES

(a) kvepc (b) khepc

Fig. 11. Influence of the compressible material and backfill moduli on the culvert performance.

3.5.2. Elastic moduli of compressible material and backfill
The influence of moduli of the compressible material and back-
fill on the culvert performance is shown in Fig. 11. Fig. 11(a) shows
that values of kvepc decrease nonlinearly with increasing Ef/Es ratio
(the elastic modulus of backfill to that of the compressible mate-
rial) and approach asymptotic values at large Ef/Es ratio. Moreover,
kvepc in case LSC-WTS is larger than that in cases LSC-LSH and IBC.
The results demonstrate that the elastic moduli of compressible
material and backfill have significant influence on kvepc. The varia-
tions of khepc with Ef/Es ratio are shown in Fig. 11(b). Value of khepc
in case IBC increases nonlinearly with increasing Ef/Es ratio and is
larger than that in case LSC. However, khepc in case LSC increases
slightly with Ef/Es ratio. In addition, khepc in case IBC is more easily
affected by Ef/Es ratio than that in case LSC. The results in case IBC
demonstrate that decreasing the elastic modulus of compressible
material layer will decrease the vertical earth pressure on the top
slab, but increase the horizontal earth pressure on the culvert
sidewall.
3.5.3. Width of load shedding blocks
In the analysis of the influence of the load shedding block width
(b) on the culvert performance (Fig. 12), the width of the culvert (B)
remains constant. Since there is no load-shedding block in case IBC,
it is assumed that the non-dimensional ratio of the load shedding
block width to the culvert width (b/B) has no influence in case
IBC and the earth pressure coefficients in case IBC are constant.
Fig. 12(a) shows that kvepc in case LSC-LSH, which is smaller
than that in case IBC, decreases linearly with increasing b/B ratio.
However, kvepc in case LSC-WTS increases significantly with
increasing b/B ratio. Values of kvepc in cases LSC-LSH and LSC-
WTS gradually approach that in case IBC when the b/B ratio
approaches zero. Moreover, the case LSC-WTS is more susceptible
to B/b ratio than the case LSC-LSH. Fig. 12(b) shows that khepc in
case LSC slightly decreases with increasing b/B ratio. It also can
be seen that khepc in case LSC is smaller than that in case IBC. The
results demonstrate that the b/B ratio has little influence on the
horizontal earth pressure in case LSC.
3.5.4. Height of load shedding block (or thickness of compressible
material layer)
In this section, the height of culvert sidewall (h) remains con-
stant. The height of load shedding blocks (d) and the thickness of
compressible material layer (d1) are changed simultaneously, i.e.,
d is equal to d1.
The influence of the d/h (or d1/h) ratio on the culvert perfor-
mance is shown in Fig. 13 and indicates that kvepc in cases LSC-
LSH and IBC decreases with increasing d/h (or d1/h) ratio. kvepc in
case IBC is markedly higher than that in case LSC-LSH. kvepc in case
LSC-WTS is between 0.9 and 1.1, which slightly declines with
increasing d/h (or d1/h) ratio. Therefore, the influence of d/h (or
d1/h) ratio on kvepc in case LSC-WTS is insignificant. Fig. 13(b)
shows that khepc in cases LSC and IBC increases with increasing
d/h (or d1/h) ratio. khepc in case LSC is significantly smaller than that
 B. Chen et al. / Computers and Geotechnics 79 (2016) 31–40 39

1.4
LSC-LSH LSC
1.2 0.48
LSC-WTS IBC
1.0 IBC
0.45
kvepc 0.8
0.42

khepc
0.6
0.39
0.4

0.2 0.36

0.0 0.33
0.08 0.16 0.24 0.32 0.40 0.48 0.56 0.64 0.08 0.16 0.24 0.32 0.40 0.48 0.56 0.64
b/B b/B
(a) kvepc (b) khepc
Fig. 12. Influence of the load shedding block width on the culvert performance.

1.2 0.55

1.0 0.50

0.8 LSC-LSH
0.45
kvepc

LSC-WTS
khepc LSC
0.6 IBC IBC
0.40

0.4
0.35
0.2
0.30
0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.05 0.10 0.15 0.20 0.25 0.30 0.35

d/h or d1/h d/h or d1 /h

(a) kvepc (b) khepc

Fig. 13. Influence of the d/h (or d1/h) ratio on culvert performance.

in case IBC. It is also found that khepc in case IBC is more susceptible jV max  V min j
N¼  100% ð2Þ
to d/h ratio than that in case LSC. The results in case IBC demon- ðV max þ V min Þ=2
strate that increasing the thickness of the compressible material
layer will increase the horizontal earth pressure on the culvert where V max = the maximum value of induced variable, and
sidewall, although the vertical earth pressure on top slab will be V min = the minimum value of induced variable.
reduced. The details of calculating method for the influencing degree can
be found in Refs. [27,28]. The calculated results for each influenc-
ing factor are tabulated in Table 4. The influence degree on a per-
3.5.5. Discussion
formance parameter is divided into three grades: low (less than
An influencing factor on the vertical earth pressure and hori-
30%), medium (between 30% and 60%), and high (greater than
zontal earth pressure is considered as a positive or negative effect
60%). The influencing degree is presented in Table 5 and the posi-
according to its influence on the performance in cases LSC or IBC.
tive sign (+) indicates a positive influence while the negative sign
The positive effect is defined as reducing earth pressures on cul-
() indicates a negative influence.
vert, whereas negative effect is defined as increasing earth pres-
Tables 5 and 6 show that u, b, Es and d (or d1) have medium to
sure on culvert.
high influence on the average vertical earth pressure on the load-
The influence degree of each factor on the investigated perfor-
shielding hole in case LSC. Es and b have medium to high influence
mance parameters (vertical earth pressure and horizontal earth
on the average vertical earth pressure on the whole top slab in case
pressure) is defined as the ratio, N, and is expressed as:
LSC, while u and d (or d1) have low influence. Moreover, u, b, Es and
Table 5 d (or d1) have low influence on the average horizontal earth pres-
The influence of factors. sure in case LSC.
Factors u Es b d or d1
VL (LSC-LSH) 50.62 158.08 40.40 87.46 Table 6
VL (LSC-WTS) 25.65 35.89 88.16 9.81 The degree of influence.
VI 29.95 145.55 – 69.90
HL 16.88 3.12 7.12 2.43 Factors u Es b d or d1
HI 15.45 20.45 – 30.07 VL (LSH) Medium (+) High () Medium () High (+)
VL(WTS) Low () Medium () High (+) Low (+)
Note: VL (LSC-LSH)-average vertical earth pressure over the load-shedding hole in
VI Low (+) High () – High (+)
case LSC; VL (LSC-WTS)-average vertical pressure over the whole top slab in case
HL Low (+) Low(+) Low (+) Low ()
LSC; VI-average vertical earth pressure in case IBC; HL-average horizontal earth
HI Low (+) Low (+) – Medium ()
pressure in case LSC; HI-average horizontal earth pressure in case IBC.
40 B. Chen et al. / Computers and Geotechnics 79 (2016) 31–40
4. Conclusions
In this study, a new culvert structure named load shedding cul-
vert (LSC) is proposed. Model tests and numerical simulations were
carried out to investigate the performance of LSC, and were com-
pared with those of embankment-installation box culvert (EBC)
and imperfect-trench-installation box culvert (IBC). The effect of
influencing factors on performances of LSC were also investigated.
The following conclusions can be drawn:
(1) The average vertical earth pressure on the top slab in case
LSC was smaller than those in cases EBC and IBC. The vertical
earth pressure on the surrounding soil prism at the culvert
top slab plane in case LSC was smaller than the overburden
pressure and was similar to that in case EBC, whereas, in the
case IBC, the vertical earth pressure on the surrounding soil
prism was larger than the overburden pressure. However, in
case LSC, it is noted that the vertical earth pressure on the
load shedding blocks (LSB) was larger than the overburden
pressure because the vertical load over the load shedding
hole (LSH) was transferred to the LSB. Moreover, the vertical
earth pressure ratio in case EBC was larger than 1.0, whereas
it was less than 1.0 in cases IBC and LSC. The vertical earth
pressure ratio in case LSC decreased with increasing backfill
height and reached a steady value at a large backfill height,
which demonstrates that the load reduction effect was
enhanced as the backfill height increased.
(2) The distribution of horizontal earth pressure on the culvert
sidewall in case LSC is similar to that in case EBC, the max-
imum horizontal earth pressure occurred at the mid-height
of culvert sidewall. In case LSC, the horizontal earth pressure
was remarkably smaller than that in case IBC. The horizontal
earth pressure ratio in case LSC decreased with increasing
backfill height.
(3) Average vertical earth pressure coefficients (kvepc) and hori-
zontal earth pressure coefficients (khepc) in cases LSC-LSH
and IBC decreased nonlinearly with increasing backfill fric-
tion angle, whereas kvepc in case LSC-WTS increased nonlin-
early with increasing backfill friction angle. It demonstrates
that the backfill load was concentrated on the load shedding
blocks in case LSC. The Ef/Es ratio and the d/h (or d1/h) ratio
have great influence on kvepc in cases LSC and IBC. Large val-
ues of Ef/Es or d/h (or d1/h) ratios can help reduce the vertical
earth pressure over culvert while resulting in a larger hori-
zontal earth pressure in case IBC. However, the influence
of the Ef/Es and d/h (or d1/h) ratios on the horizontal earth
pressure was negligible in case LSC. kvepc in case LSC-LSH
decreased with increasing width of the load shedding block
while the variation of kvepc in case LSC-WTS was opposite. It
also shows that the backfill load was concentrated on load
shedding blocks. Furthermore, khepc in case LSC decreased
with increasing width of the load shedding block.
In summary, the load shedding culvert (LSC), can not only
reduce the vertical earth pressure on the top slab but also reduce
the horizontal earth pressure on the culvert sidewall. The load
reduction effect and the stress states of load shedding culvert
(LSC) are better than those of the embankment-installation box
culvert (EBC) and the imperfect-trench-installation box culvert
(IBC).
Acknowledgments
This work was supported by the National Natural Science Foun-
dation of China (Grant No. 51108434), the Fundamental Research
Funds for the Central Universities (Grant No. CUG130408), the
Key Lab of Highway Construction & Maintenance Technology in
Loess Region, Ministry of Transport, PRC, Shanxi Transportation
Research Institute (Grant No. KLTLR-Y13-5), the Scientific Research
Foundation for the Returned Overseas Chinese Scholars, State Edu-
cation Ministry (Grant No. 2013-47). The authors express their
appreciation to the financial support.
References
[1] Spangler MG. Underground conduits-an appraisal of modern research. Trans
Am Soc Civil Eng 1948;113(1):316–45.
[2] Trollope DH, Speedie MG, Lee IK. Pressure measurements on Tullaroop dam
culvert. In: Proceedings of 4th Australia–New Zealand conference on soil
mechanics and foundation engineering [S.I.]: [sn]. p. 81–92.
[3] Penman ADM, Charles JA, Nash J, Humphreys JD. Performance of culvert under
Winscar dam. Geotechnique 1975;25(4):713–30.
[4] Dasgupta A, Sengupta B. Large-scale model test on square box culvert
backfilled with sand. J Geotech Eng 1991;117(1):156–61.
[5] Bennett RM, Wood SM, Drumm EC, Randy Rainwater N. Vertical loads on
concrete box culverts under high embankments. J Bridge Eng 2005;10
(6):643–9.
[6] Chen BG, Sun L. The impact of soil properties on the structural integrity of
high-fill reinforced concrete culverts. Comput Geotech 2013;52:46–53.
[7] Chen BG, Sun L. Performance of a reinforced concrete box culvert installed in
trapezoidal trench. J Bridge Eng (ASCE) 2014;19(1):120–30.
[8] Marston A, Anderson AO. The theory of loads on pipes in ditch and tests of
cement and clay drain tile and sever pipe Bulletin No. 31. Ames: Iowa
Engineering Experiment Station; 1913.
[9] Marston A. The theory of external loads on closed conduits in the light of the
latest experiments. In: Highway research board proceedings. p. 9.
[10] Spangler MG. Field measurements of the settlement ratios of various highway
culverts Bulletin 171. Ames, Iowa: Iowa Engineering Experiment Station;
1950.
[11] Vaslestad J, Johansen TH, Holm W. Load reduction on rigid culverts beneath
high fills: long-term behavior. Transport Res Rec 1992:1415.
[12] Okabayashi K, Ohtani W, Akiyama K, Kawamura M. Centrifugal model test for
reducing the earth pressure on the culvert by using the flexible material. In:
The fourth international offshore and polar engineering
conference. International Society of Offshore and Polar Engineers; 1994.
[13] Hansen P, Miller L, Valsangkar AJ, Bourque S, Mac Leod T. Performance of
induced trench culverts in New Brunswick. 2007 Annual conference and
exhibition of the transportation association of Canada: transportation-an
economic enabler (Les Transports: Un Levier Economique) 2007.
[14] McGuigan BL, Valsangkar AJ. Centrifuge testing and numerical analysis of box
culverts installed in trenches. Can Geotech J 2010;47(2):147–63.
[15] Kim K, Yoo CH. Design loading on deeply buried box culverts. J Geotech
Geoenviron Eng 2005;131(1):20–7.
[16] AASHTO L. Standard specifications for highway bridges. DC: American
Association of State Highway and Transportation Officials Washington; 2010.
[17] China Highway Planning and design Institute (CHPDI). General code for design
of highway bridges and culverts, Beijing; 2004 [in China].
[18] McAffee RP, Valsangkar AJ. Field performance, centrifuge testing, and
numerical modelling of an induced trench installation. Can Geotech J
2008;45(1):85–101.
[19] Bourque SW. Centrifuge and numerical modeling of induced trench twin
conduits. University of New Brunswick, Department of Civil Engineering;
2002.
[20] McGuigan BL, Valsangkar AJ. Earth pressures on twin positive projecting and
induced trench box culverts under high embankments. Can Geotech J 2011;48
(2):173–85.
[21] Turan A, EI Naggar MH, Dundas D. Investigation of induced trench method
using a full scale test embankment. Geotech Geol Eng 2013;31(2):557–68.
[22] Chen BG, Song DB, Jiao JJ, Zhang J. Vertical earth pressure on high fill culverts
under load reduction condition. J Huazhong Univ Sci Technol (Nat Sci Ed)
2015;43(10):112–6 [in Chinese].
[23] Barchard J. Centrifuge modelling of piled embankments on soft soils M.Sc.E.
thesis. Fredericton, N.B., Canada: University of New Brunswick; 2002.
[24] Bourque SW. Centrifuge and numerical modeling of induced trench twin
conduits M.Sc.E. thesis. Fredericton, N.B., Canada: University of New
Brunswick; 2002.
[25] Itasca Consulting Group, Inc. FLAC: theory and background. Minneapolis,
Minn: Itasca Consulting Group Inc.; 2001.
[26] Duncan JM, Chang CY. Nonlinear analysis of stress and strain in soils. J Soil
Mech Found Divis 1970;96(5):1629–53.
[27] Huang J, Han J. Two-dimensional parametric study of geosynthetic-reinforced
column-supported embankment by coupled hydraulic and mechanical
modeling. Comput Geotech 2010;37(5):638–48.
[28] Zhang J, Zheng JJ, Chen BG, Yin JH. Coupled mechanical and hydraulic modeling
of a geosynthetic-reinforced and pile-supported embankment. Comput
Geotech 2013;52:28–37.