You are on page 1of 17

Computers and Geotechnics 69 (2015) 219–235

Contents lists available at ScienceDirect

Computers and Geotechnics


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

Research Paper

Static soil culvert interaction the effect of box culvert geometric


configurations and soil properties
Osama Abuhajar a,b,⇑, Hesham El Naggar a,1, Tim Newson a,2
a
Department of Civil and Environmental Engineering, Geotechnical Research Center, The University of Western Ontario, 1151 Richmond Street, London, Ontario N6A 5B9, Canada
b
Department of Civil Engineering, University of Benghazi, Benghazi, Libya

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

Article history: The response of box culverts to static loads is controlled by soil arching. Soil arching is a result of a com-
Received 16 December 2014 plex soil culvert interaction (SCI) due to the relative stiffness between the culvert and the surrounding
Received in revised form 13 April 2015 soil, and is a critical consideration in culvert design. The factors that affect soil arching on box culverts
Accepted 9 May 2015
include the soil height above the culvert, the geometrical configuration of the box culvert and the prop-
Available online 12 June 2015
erties of the soil around it. Box culverts are typically designed using formulae that assume simplified
behaviors and in some cases rely on considerable empiricism. In the present study, small scale centrifuge
Keywords:
physical model tests were conducted to investigate SCI considering the height and density of soil above
Box culvert
Soil arching
the culvert and the geometry of the culvert. The results of these centrifuge tests were used to calibrate
Centrifuge modeling and verify a numerical model that was used to further investigate the response of box culverts to static
Numerical modeling loads. The results have been evaluated for bending moment and soil culvert interaction factors. The
Static soil pressure results demonstrated that the soil culvert interaction factors are not only a function of the height of soil
Soil–culvert interaction factors column above the culvert, but also a function of the culvert thickness, soil elastic modulus and Poisson’s
ratio. Therefore, the results were used to establish charts and equations that can be employed to assess
the design values of the static soil pressure and static bending moment for box culverts.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction pressures attracted to the culvert. This arching effect can cause
increases or decreases in the soil pressure on the culvert due to
Box culverts are an important transportation infrastructure stress redistribution. Arching is usually ignored in engineering
with significant economic value. The primary usages of box cul- design due to a lack of knowledge, which may lead to damage in
verts include roadways and railways, storm runoff and sewerage the culvert or over design. Typically, the design of culverts is based
conduits, and electrical and telephone lines pathways. Box culverts on formulae that assume simplified behaviors and often rely on
can be classified based on the type of material or based on the considerable empiricism. A box culvert is designed to support
installation type. Based on the material used, the culvert can be loads above and beside it, but it may be required to resist loads that
either rigid or flexible; rigid culverts can be made of reinforced are considerably higher, from adjacent structures and/or construc-
concrete, which can be precast or cast-in-place, while flexible cul- tion activities. SCI should be considered in defining the actual static
verts can be made of steel, cast iron, aluminum, or plastic or other loads that the culvert will be subjected to, which can then be used
materials. Based on the installation method, the culvert can be con- in providing safe and economic culvert designs. Even though the
structed under embankments, trenches or induced/imperfect behavior of flexible circular and arch culverts is well researched,
trenches. Box culverts can be installed as single cell, or multiple the behavior of box culverts, has not received much attention to
cells depending on the use. date.
Depending on the installation condition, soil culvert interaction In this paper, the static response of box culverts constructed
(SCI) can be examined to show the effect of soil arching on the using the embankment installation method was investigated by
performing a series of centrifuge tests and numerical modeling
⇑ Corresponding author at: Department of Civil and Environmental Engineering, using FLAC 2D. A brief description of the centrifuge and develop-
Geotechnical Research Center, The University of Western Ontario, 1151 Richmond ment of the numerical model and its calibration/verification is
Street, London, Ontario N6A 5B9, Canada. Tel.: +1 5194715133. introduced. A detailed numerical parametric study was then per-
E-mail addresses: oabuhaja@alumni.uwo.ca, oabuhajar@gmail.com (O. Abuhajar), formed using the verified numerical model to investigate several
helnaggar@eng.uwo.ca (H. El Naggar), tnewson@eng.uwo.ca (T. Newson).
1
Tel.: +1 5196614219.
factors that affect the static SCI and the results are presented in
2
Tel.: +1 5198502973. the form of static bending moment and soil culvert interaction

http://dx.doi.org/10.1016/j.compgeo.2015.05.005
0266-352X/Ó 2015 Elsevier Ltd. All rights reserved.
220 O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235

factors. The soil fill height to culvert width ratio (H/Bc = 0–6.13) Table 1
was investigated for two different culvert wall thickness values, Formulae proposed for SCI factors for embankment installation method.

referred to as thin (t/Bc = 0.06) or thick (t/Bc = 0.12) culverts. Also, Reference Culvert Formula Condition
the change in culvert thickness to width ratio (t/Bc) and the change Clarke [3] Circular e0:38ðH=BcÞ 1
Fe ¼ CC ¼ 0:38
in soil properties, such as soil relative density, friction angle, dila-
tion angle, elastic modulus, and Poisson’s ratio were examined. Bennett et al. Box 0:38ðH=BcÞ
F e ¼ e0:38ðH=BcÞ
1
; H/Bc 6 2.42
The results of this comprehensive parametric study are pre- [4]
F e ¼ 1:69  0:12 H/Bc > 2.42
H=Bc ;
sented in a series of charts and equations that can be used to aid
design of box culverts. In addition, a design example is presented Tadros et al. Box For top slabs:
to demonstrate the calculation of the soil pressure on the top slab [7] Fe = 0.984 + 0.0019H Silty clay
and side wall of the box culvert required for the structural design Fe = 0.970 + 0.0020H Silty sand
of the culvert. For side walls:
Fe = 0.600 Silty clay
Fe = 0.567 Silty sand
2. Literature review
Kim and Yoo Box Fe = DE = 1.047H0.055 Yielding
[8] foundation
2.1. Soil structure interaction for embankment installation box culverts Fe = DE = 1.200H0.059 Unyielding
foundation
The theoretical vertical soil pressure rv is usually obtained by
multiplying the unit weight of the soil column above the box cul- Kang et al. Box For top slab:
[9] Fe = 0.005(H/Bc) + 1.304 Compacted side
vert times its height. To account for the effect of soil arching, the
fill
actual vertical soil pressure on the culvert top slab may be given Fe = 0.012(H/Bc) + 1.407 Uncompacted
by the theoretical vertical pressure times the soil structure interac- side fill
tion factor Fe. The actual vertical soil pressure can then be given by: For the bottom slab:
Fe = 0.004(H/Bc)2  0.105(H/ Compacted side
rv ¼ F e cS H ð1Þ Bc) + 2.105 fill
Fe = 0.006(H/Bc)2  0.175(H/ Uncompacted
where H is the height of soil fill over the culvert and cS is soil unit Bc) + 2.685 side fill
weight.
Marston and Anderson [1] investigated the soil pressure on the
top of circular culverts and pipes, and proposed a method to eval-
uate it considering the culvert installation type. Spangler [2] com- bottom slabs, with higher values at edges and lower at center;
pared the soil deflection with the culvert deflection and proposed the average pressure on the top slab was 9% higher than the weight
design pressure values as a function of the projection condition, of the soil column (i.e., Fe = 1.09).
various soil properties and the soil height above the culvert H as Tadros et al. [7] also performed a parametric study to evaluate
a ratio of culvert width Bc, i.e., H/Bc. Clarke [3] developed formulae the soil pressure on all four sides of the box culvert. Twelve box
and charts for design of circular sections of buried conduits and culverts ranging in size from 1.20  1.20 m to 3.70  3.70 m were
proposed Fe for embankment installation as shown in Table 1. modeled using both soils silty sand and silty clay. The calculated
Bennett et al. [4] extended the Clarke [3] work to box culverts pressures were found to be higher at the corners and lower in
and suggested Fe for the embankment condition as shown in the middle of the top and bottom slabs. Side pressures distribu-
Table 1. Spangler [2] defined the plane of equal settlement as the tions were approximately trapezoidal. The Fe values for the top slab
horizontal plane in the embankment at which the settlements of increased with fill height from 1 for low fill height to 1.15 at largest
the interior soil prism above the culvert and the exterior soil prism fill height considered. This was attributed to settlement of the side-
outside the culvert are equal. For embankment heights greater fill more than the column of soil directly above the top slab, which
than the plane of equal settlement, the soil structure interaction induces downward drag. The Fe value for the side wall was gener-
factor is essentially constant. In the Bennett et al. [4] formula, ally about 0.6, which was consistent with the at-rest Ko conditions
the H/Bc ratio of 2.42 corresponds to the plane of equal settlement against non-yielding walls. For the bottom slab, Fe values were
being at the ground surface, while Fe varies from 1.64 at H/Bc = 2.42 found to be higher for relatively low fills and approached 1.0 as
to 1.69 at H/Bc = 1. the fill increased. Based on the above parametric study, Tadros
Katona et al. [5,6] analyzed the Kentucky installation of a full et al. [7] presented some equations to obtain the soil pressure on
scale box culvert numerically and the calculated soil pressures all culvert sides as shown in Table 1.
were in good agreement with the measured values at specific loca- Kim and Yoo [8] examined the effect of foundation on the soil
tions on the top and bottom slabs, but achieved less agreement on structure interaction factors. They defined the soil structure inter-
the side walls. They found that the soil pressure distribution on the action factor as ‘effective density’ in their numerical modeling
top slab was parabolic with high values at the edges and decreased study. The culvert H/Bc ratio varied from 6.3 to 25.4 for the
at the center. embankment installations. The soil structure interaction factors
Tadros et al. [7] numerically modeled a box culvert with were found to be most sensitive to the foundation characteristics.
H/Bc = 1.7 considering two soil types: silty sand and silty clay. Based on the regression analysis of their results, they proposed a
The results showed that the pressures for the silty sand model set of equations for effective density DE. For embankment installa-
were smaller at mid span and larger at the corners than for the silty tion, DE is given as a function of the fill height H as shown in
clay model. This was to be expected, since the relative stiffness of Table 1.
silty sand to the culvert was greater than that of the silty clay. They Kang et al. [9] investigated numerically the side friction of a box
also analyzed a box culvert with H/Bc = 5.5 under 8.2 m of silty culvert under different soil fill heights, considering soil properties
sand backfill, considering a linear elastic and Duncan hyperbolic representative of compacted and uncompacted side fills for yield-
soil models to examine the effect of soil arching. The analyses of ing and unyielding foundation soils. The Fe values were established
the two soil models yielded small differences in soil pressures. as a function of the culvert (H/Bc) ratio. The proposed relations for
The calculated soil pressure was not uniform on the top and embankment installation are as shown in Table 1.
O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235 221

2.2. Soil structure interaction for box culverts in codes and Standards Table 2
Arching factors for box sections in standard installations (from CHBDC) [11].

2.2.1. AASHTO Installation type Vertical arching factor, kv Horizontal arching


The American Association of State and Highway Transportation factor, kh
Officials, AASHTO [10] requirements are based on the Marston– Minimum Maximum
Spangler theory to determine the soil structure interaction factors B1 1.20 0.30 0.50
Fe, and depend on the type of installations of the box culvert. For B2 1.35 0.25 0.50
embankment installation, the soil structure interaction factor is
given by:
H 1.8
F e ¼ 1 þ 0:20 ð2Þ 1.7
Bc 1.6
1.5
where Fe should not be greater than 1.15 for installations with com- 1.4
pacted fill at the sides of the box section, and need not be greater

Fe
1.3
than 1.4 for installations with un-compacted fill at the sides of 1.2
1.1
the box section.
1
0.9
2.2.2. CHBDC 0.8
0 1 2 3 4 5 6 7 8 9
The Canadian Highway Bridge Design Code (CHBDC) [11] spec-
H/Bc
ifies the vertical and horizontal arching factors, kv and kh, as shown Bennett et al [4] AASHTO [10] Compacted Sidefill
in Table 2. The maximum and minimum values of kh should be AASHTO [10] Uncompacted Sidefill CHBDC [11] Installation B1
CHBDC [11] Installation B2 Kang et al [9] Compacted Sidefill
used to obtain the maximum positive and negative moments in Kang et al [9] Uncompacted Sidefill Spangler [15]
the culvert walls. The reaction pressure on the bottom of the box Sato et al [16]
Tadros et al [7]
James et al [17]
Dasgupta et al [14]
is assumed to be uniformly distributed. Vasleastad et al [18] Yang [19]
Bennett et al [4] Pimentel et al. [20]
The commentary of CHBDC [12] states that the proposed kv and
kh values are based on previous practice and limited soil–structure Fig. 1. Comparison between different methods for calculating the soil structure
interaction analyses with the finite element method. The soil struc- interaction factors Fe as a function of H/Bc [4,7,9–11,14–20].
ture interaction factors are provided for two standard installation
methods, B1 and B2, for box culverts resting on the soil groups
classified by the code. 1.8

1.7

2.3. Comparison of soil structure interaction factors given in the 1.6


standards and the literature 1.5

1.4
Fe

Both AASHTO [10] and CHBDC [11] deal only with positive pro-
1.3
jecting box culverts. These documents recommend uniform earth
1.2
pressures over the top and bottom slabs of positive projecting
box culverts, with the contact pressure on the bottom equal to 1.1

the sum of the top pressure and the pressure due to the culvert 1
dead load [13]. 0.9
AASHTO makes no classification of the soil as compacted or 0 5 10 15 20 25 30 35
H (m)
uncompacted, and no reference concerning the shape of the pres-
Kim et al [8] Yielding Foundation. Kim et al [8] Unyielding Foundation
sure diagram. However, the results provided by many researchers Tadros et al [7] Silty Clay Tadros et al [7] Silty Sand
(e.g., [4,14,8,7]) show that the pressure is lower at the center of the Girdler [21] Russ [22]
Sato et al [16] James et al [17]
slab and higher at the edges. The validity of the interaction factors Tadros et al [7] Dasgupta et al [14]
proposed in AASHTO [10] is limited to cases of yielding founda- Vasleastad et al [18] Yang [19]
Bennet et al [4] Pimentel et al. [20]
tions. However, the interaction factor for unyielding foundations Spangler [15]
can take higher values [8]. Conversely, CHBDC [11] specifies a
constant arching factor regardless of the H/Bc ratio, and this only Fig. 2. Comparison between different methods for calculating the soil structure
interaction factors Fe as a function of H [4,7,8,14–22].
varies according to installation type, and does not differentiate
between the embankment and trench installation methods.
Figs. 1 and 2 compare the Fe values obtained from instrumented
moment, static soil pressure and soil culvert interaction factors.
culverts with available relations for Fe as a function of H/Bc and H,
The soil fill height to culvert width ratio (H/Bc) was investigated
respectively. All of the relations (except for Bennett et al. [4]) fall
for two different culvert wall thicknesses, and the change in culvert
between the upper and lower limits of the ASSHTO [10] and CHBDC
thickness to width ratio (t/Bc) was explored for several culvert
[11] values. In comparison with some of the presented test data,
thicknesses. The effect of different soil properties was examined.
the ASSHTO and CHBDC Fe values could be considered unconserva-
tive for H/Bc < 5 as can be noted from Fig. 1. Also, the scatter in the
test data indicates that Fe may not be a unique function of H/Bc. For 3. Centrifuge modeling
example, the relative stiffness between the soil and the culvert
could be a significant factor as suggested by Einstein and Schwartz A series of static centrifuge tests was conducted at the
[23] and Karinski et al. [24]. Rensselaer Polytechnic Institute (RPI) geotechnical centrifuge facil-
In this paper, the response of the box culvert investigated in the ity in the USA, to investigate the effect of soil arching on the
centrifuge tests was used to validate a numerical model. A detailed response of box culverts. Four tests were performed involving
parametric study was then performed to investigate the static SCI two box culvert thicknesses buried inside dry Nevada Sand with
and the results are presented in the form of static bending two different relative densities. The static centrifuge model tests
222 O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235

(a)
Free surface

127 H

76.2 Bc Box culvert

127 Bc
Nevada sand

876.3
(b)

Fig. 3. Model configuration: (a) schematic diagram for centrifuge tests (all units are in mm) and (b) photos of instrumentations used.

Free surface

20 m
Fixed
Box Culvert
in x

x
Fixed in x and y

52 m

Fig. 4. Typical numerical grid and model components.

focussed on the response of the box culverts to the self-weight of Strain gauges and tactile pressure sensors were used to investi-
sand. The data collected from all of the sensors during the cen- gate the static bending moment and soil pressure around box cul-
trifuge tests were interpreted to understand the main features of vert. The locations of these instruments were kept constant in all
soil–culvert interaction. tests. Strain gauges were used to measure strains on the outside
The 120-Nevada Sand used for the test bed was fine, uniform, and inside faces of the box culvert top slab and side wall. These
and clean. It was poorly graded sand (SP) with particle sizes in strain measurements were then converted to bending moments
the range of 0.075 to 0.550 mm. The maximum and minimum unit using calibration factors. Two tactile pressure sensors were used
weights cmax and cmin were 16.77 kN/m3 and 14.85 kN/m3. The for each culvert model to measure the vertical soil pressure on
critical state and peak friction angles were 32° and 40° respectively the top slab and horizontal soil pressure on side wall of the culvert.
(i.e., peak dilation angle was 8°) [25]. It should be noted that this sensor has the ability to measure the
Aluminum tubes were used to construct the model box culverts. normal stress only (i.e., it does not account for shear stresses).
They had external dimension of 76.2 mm, one was 6.35 mm thick The large centrifuge rigid box used in all tests was 355.6 mm in
(termed the ‘thick’ culvert) and another was 3.18 mm thick (ter- high. The sand was placed in 25.4 mm layers using a raining tech-
med the ‘thin’ culvert). At 60g, these dimensions represented rein- nique (air pluviation) alone to achieve a 50% relative density. To
forced concrete box culverts with 4.572 m external dimension and achieve 90% relative density, each sand layer was additionally
0.533 m and 0.267 m thick. These prototype wall thickness values tamped after air pluviation. The models were built after placing
were evaluated using the similitude relationship: the empty centrifuge box on the centrifuge platform. As the
required level of sand under the culvert model was reached, tactile
t p ¼ Ntm a1=3 ð3Þ
pressure sensors were connected to the culvert model and placed
where t = wall thickness, N = scaling factor. The subscripts ‘m’ and in the middle of the centrifuge box. All strain gauges and tactile
‘p’ refer to model and prototype, respectively and a = Em/Ep. pressure sensors were then checked and connected to the data
O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235 223

Centrifuge FLAC 2D 200


150
180

Vertical Pressure (kPa)


100 160
140
BM (kN.m/m)

50
120
0
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 100
-50 80
Tactile avg
-100 60 Tactile max
40 Tactile min
-150 Strain gauges
20 FLAC 2D
-200
0
Distance (m)
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

Fig. 5. Measured versus computed bending moment on the top slab. Distance (m)

Fig. 6. Measured versus computed pressure on the top slab.

acquisition system. Fig. 3 shows the general configurations of the


test models. Tests 1 and 2 are for the thick culvert with sand rela-
tive density 90% and 50%, respectively. Tests 3 and 4 are for the 600
thin culvert and the sand relative density was 50% and 90%,
respectively. 400 H/Bc = 0
H/Bc = 0.08
The centrifuge was accelerated incrementally and held at the

BM (kN.m/m)
200 H/Bc = 0.38
following acceleration levels, 10g, 20g, 30g, 40g, 50g and 60g to
H/Bc = 0.77
check stability of the sensor readings. Data from all sensors were 0
H/Bc = 1.53
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
recorded continuously during the test. H/Bc = 2.30
-200
H/Bc = 3.07
-400 H/Bc = 6.13
4. Numerical modeling
-600
Numerical modeling has been conducted to investigate the
Distance (m)
effects of several parameters on the soil culvert interaction. The
numerical models consisted of two main materials: the box culvert
(a)
and the soil. The box culvert was modeled using liner elements, 500
which were assumed to behave linearly and therefore the linear 400
elastic model was used to simulate their response. Liner elements H/Bc = 0
300
H/Bc = 0.08
are two-dimensional elements with three degrees of freedom
BM (kN.m/m)

200 H/Bc = 0.38


(x-translation, y-translation and rotation) at each end node, and 100 H/Bc = 0.77
these elements can be joined with the grid [26]. The soil response 0 H/Bc = 1.53
was expected to cover a range of elastic and plastic deformations; -2.5 -2 -1.5 -1 -0.5 0
-100
0.5 1 1.5 2 2.5 H/Bc = 2.30
hence the elastic–perfectly plastic model with Mohr–Coulomb fail- -200
H/Bc = 3.07
H/Bc = 6.13
ure criterion was used to simulate its nonlinear behavior. The over- -300
all boundary value problem was based on plane strain conditions,
-400
and was formulated in terms of effective stresses. The material
Distance (m)
behavior was modeled using elements that obey assigned linear
or nonlinear stress–strain behavior in response to the forces and (b)
boundary conditions.
Fig. 7. Effect of H/Bc ratio on the bending moment on the top slab: (a) t = 0.533 m
The numerical models simulated the centrifuge tests at proto- and (b) t = 0.267 m.
type scale. The soil was modeled using continuum zones and each
zone divided into small grids. The finite difference grids used
around the box culvert were square in shape; and were rectangular
estimate the preliminary values for the interface normal and shear
elsewhere as shown in Fig. 4. The density of the grid was increased
stiffness. These values were adjusted by refining the magnitude of
around the box culvert to improve accuracy. Several trials were
kn and ks to obtain a good match with the centrifuge data. After
performed to refine the grids until there was no noticeable change
several calibrations, the values adopted for both stiffness values
in the results. The boundary conditions for the numerical model
are kn = ks = 5.68  103 MPa/m.
simulated the same conditions in the centrifuge tests. The base
of the model was fixed in x and y directions, while the side bound-
aries were fixed in the x direction only. 4.1. Model parameters
The box culvert–sand interface conditions in the numerical
models were simulated using interface elements. The interface ele- The experimental data were used to calibrate/verify the numer-
ments were modeled with a linear spring system denoted ‘‘glued ical models. The box culvert was modeled using linear elastic ele-
interface’’ in FLAC, which does not allow slipping or gap opening. ments, with mass density of 2548.4 kg/m3, elastic modulus of
It allows only elastic displacement according to the specified stiff- 25.2 GPa and Poisson’s ratio of 0.2. Two box culvert thicknesses
nesses: normal stiffness (kn) and shear stiffness (ks) between the were used: 0.533 m and 0.267 m for the thick and thin culverts.
two planes representing the structures and soil. Itasca [26] recom- The nonlinear elastic–plastic material using Mohr–Coulomb failure
mends a ‘rule-of-thumb’ to estimate the maximum interface stiff- criterion with non-associated flow rule was used to model the dry
ness values kn and ks to be set to ten times the equivalent stiffness Nevada Sand. The initial mass densities obtained during centrifuge
of the stiffest neighboring zone. This procedure was followed to tests for the 50% and 90% relative density were used in the analysis.
224 O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235

BM (kN.m/m) 4.2. Comparison between centrifuge and numerical results


0 100 200 300 400 500
0
To calibrate the numerical model, the sand and culvert proper-
0.5 ties stated above were assigned to the respective elements, and the
H/Bc = 0
1 analysis was repeated numerous times to examine the effect of dif-
H/Bc = 0.08
ferent soil and interface parameters on the results in comparison
1.5 H/Bc = 0.38
with the centrifuge results of Test 1. The model that achieved best
Distance (m)

2 H/Bc = 0.77
H/Bc = 1.53 fit with the experimental results was verified by applying it to all
2.5
H/Bc = 2.30 other centrifuge tests. The calibrated/verified model was then
3 H/Bc = 3.07 employed to conduct an extensive parametric study of static per-
3.5 H/Bc = 6.13 formance of box culverts in sand. Two main parameters were con-
sidered in the calibration/verification process. These parameters
4
are the bending moment and soil pressure. The results from
4.5 repeated tests were found to be approximately the same (e.g., in
(a) terms of measured strains and/or soil pressures), hence only the
results from Test 1 are shown here. A detailed description of the
BM (kN.m/m) centrifuge tests and their results are given elsewhere [32].
-100 0 100 200 300 400 500
0
4.2.1. Static bending moment
0.5 The calculated bending moments from the numerical model
H/Bc = 0
1 H/Bc = 0.08 were compared with the corresponding values evaluated from
1.5 H/Bc = 0.38 the strain gauge measurements on the top slab and side wall dur-
Distance (m)

H/Bc = 0.77 ing centrifuge tests. Fig. 5 compares the calculated bending
2
H/Bc = 1.53 moments obtained using the numerical model that resulted in best
2.5 H/Bc = 2.30 fit with the measured data with those obtained from centrifuge
3 H/Bc = 3.07 results on the top slab of Test 1. The agreement between the calcu-
3.5
H/Bc = 6.13 lated and measured bending moments on the top slab and side
wall were excellent.
4

4.5
(b) 4.2.2. Static soil pressure
The numerical model was used to calculate the soil pressure at
Fig. 8. Effect of H/Bc ratio on the bending moment on the side wall: (a) t = 0.533 m the interface between the soil (grid) and the box culvert (liner ele-
and (b) t = 0.267 m. ments) to determine the contact pressure. The calculated pressures
were compared to the soil pressures measured directly through the
tactile pressure sensors and indirectly through the double deriva-
The soil Poisson’s ratio, critical state friction angle and dilation tives of the measured static bending moment. Fig. 6 illustrates
angle were 0.28, 32° and 8°, respectively [25,27]. To avoid numer- the comparison between the measured and calculated soil pres-
ical instability, a small value of cohesion (1 kPa) was used as rec- sures for the top slab of Test 1. The calculated soil pressures were
ommended by Kanungo [28]. slightly less than the values obtained from strain gauges readings,
Two different elastic moduli (10 and 30 MPa) were chosen for but follow the same trend. The calculated soil pressures on the top
the soil to represent two different relative densities and enable slab confirmed that the soil pressure distribution at the top slab is
comparison of the relative stiffness for the different EI values. parabolic.
These moduli were chosen to correspond to a range of strains
of approximately 0.2–2%, which has been found close to the cul- 5. Numerical static parametric study
vert walls/slabs in these tests and previous similar studies (e.g.,
[28]). For dense sand this would give stiffness in the range of The validated numerical model presented above has been used
E50 (Young’s modulus measured as the secant modulus corre- to perform a comprehensive parametric study to examine several
sponding to a stress level of 50% of peak shear strength). Taking factors that affect the soil culvert interaction (SCI). The parameters
Janbu’s pressure related stiffness relationship Eq. (4) [29] with investigated include: the soil fill height to culvert width ratio
appropriate values for a dense sand (m = 0.5 and Eaoed ¼ 35; 000, (H/Bc), the culvert thickness to width ratio (t/Bc), and the change
after Fang [30] and Schanz and Vermeer [31]) and assuming in soil properties, such as soil relative density, friction angle, dila-
E  2/3 Eoed gives a possible range of E across the culvert of 18– tion angle, elastic modulus, and Poisson’s ratio.
38 MPa. A similar estimation for the lower density soil gives A wide range of box culvert thickness values are used in prac-
E = 11–24 MPa. Hence the values chosen are toward the lower tise. Therefore, the thick culvert case (i.e., t/Bc = 0.12) was selected
end of the scale, but still appear to be representative of Nevada as the base case for all models in the current study. In cases where
sand at these strain levels and appear to provide good matches the effect of culvert thickness is important, two different thick-
in the later numerical modeling. nesses similar to centrifuge tests were used and referred to as thin
 m (t/Bc = 0.06) or thick (t/Bc = 0.12) culverts.
r1
Eoed ¼ Eaoed ð4Þ
ra
5.1. Effect of H/Bc ratio
Eaoed ¼ v e ra ð5Þ
To investigate the effect of H/Bc ratio on SCI, eight different val-
where ra = 100 kPa, r1 is the vertical effective stress, Eeod is the ues covering a range of culvert embedment were considered, i.e.,
oedometer modulus, Eaeod the normalized stiffness and ve is a H/Bc ratios of 0, 0.08, 0.38, 0.77, 1.53, 2.30, 3.07, and 6.13. Two dif-
coefficient. ferent culvert thicknesses (0.267 m and 0.533 m) were used to
O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235 225

H/Bc = 0 H/Bc = 0.08

H/Bc = 0.38 H/Bc = 0.77

H/Bc = 1.53 H/Bc = 2.30

H/Bc = 3.07 H/Bc = 6.13

Fig. 9. Effect of H/Bc ratio on the vertical stresses around box culvert (t = 0.533 m) (legend units in Pa).

explore its effect on the SCI factors for the same H/Bc ratio. Several soil arching is more pronounced for the thin culvert for all H/Bc
analyses were conducted to demonstrate the effects of H/Bc ratio ratios. For the thick culvert, the effect of soil arching decreases as
and culvert thickness on the culvert bending moment, soil pressure the H/Bc ratio increases. The horizontal soil pressures on the thick
distribution and the SCI factors, including comparisons with SCI culvert are less than the case for the thin culvert, which reflects the
factors obtained from previous research and design codes. increased soil arching in this case.
The effects of H/Bc ratio on the soil pressure diagrams on the top
5.1.1. Bending moment slab and side wall are presented in Figs. 13 and 14. Generally, as
Figs. 7 and 8 present the effects of H/Bc ratio on the bending the H/Bc ratio increases, the soil pressure values increase for the
moment diagrams on the top slab and side wall of the box culverts, top slab and side wall. The soil pressure values are affected by
respectively. The bending moment values of the top slab increase the culvert thickness; as the culvert thickness decreases and as
as the H/Bc ratio or culvert thickness increases. For the side wall, the H/Bc ratio increases, the horizontal soil pressure distribution
as the thickness of the culvert increases the bending moment val- takes a curved shape reflecting a reduction in the horizontal soil
ues increase in the positive direction. The bending moment of the pressure at the center of the side wall. This behavior was observed
thin culvert is characterized by positive bending moment at the in the thin culvert, while the thick culvert showed a uniform
top and bottom corners and negative bending moment at the cen- increase in the horizontal soil pressure with depth.
ter, except for very low H/Bc ratios, and the bending moment val-
ues increased as the culvert thickness increased. 5.1.3. Soil culvert interaction factors
Fig. 15 presents the effect of H/Bc ratio on the SCI factors
5.1.2. Soil pressure defined at the edge and center of the culvert top. The results show
Soil arching influences the soil pressure distribution on the cul- that all Fe values for the edge are greater than 1.0 (i.e., soil pres-
vert top slab and side wall, which is in turn affected by the H/Bc sures are larger than the theoretical values). At the center, all Fe
ratio. Figs. 9 and 10 present the contour lines of the vertical soil values are less than 1.0 (i.e., soil pressures are less than the theo-
pressure and Figs. 11 and 12 display the horizontal soil pressure retical soil values). The Fe values at the edge from thin culvert
contours. The vertical pressure contours on the top slab show some are larger than those from thick culvert, while at the center, the
stress concentrations at the culvert edges. The results indicate that Fe values for the thin culvert are less than for the thick culvert.
226 O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235

H/Bc = 0 H/Bc = 0.08

H/Bc = 0.38 H/Bc = 0.77

H/Bc = 1.53 H/Bc = 2.30

H/Bc = 3.07 H/Bc = 6.13

Fig. 10. Effect of H/Bc ratio on the vertical stresses around box culvert (t = 0.267 m) (legend units in Pa).

The SCI factors at the edges increase as the H/Bc ratio increases up Generally, the Fe values proposed by AASHTO and CHBDC are
to H/Bc = 0.38, after that the Fe values decrease as H/Bc increases up bounded by the calculated Fe values for the thick and thin culverts.
to H/Bc = 1.53. There is no change in the Fe values for H/Bc > 1.53, All Fe values are presented as a function of H/Bc ratio. It should be
especially for the thick culvert. The Fe values at the center decrease noted, however, that the Fe values from CHBDC are constant (i.e.,
for H/Bc ratios up to 0.77 and remain almost constant for not a function of H/Bc), while the Fe values proposed by AASHTO
H/Bc > 0.77 in both thick and thin culverts. increase linearly for H/Bc 6 0.77 to a maximum of 1.15 and 1.4
Fig. 16 illustrates the effect of H/Bc ratio on the SCI factors on for the compacted and uncompacted side fill cases, and remain
the side wall under the conditions of at rest soil pressures. Gener- constant for H/Bc > 0.77. The calculated Fe values for the thin cul-
ally, the effect of H/Bc ratio on the Fe values is only important for vert are similar to the AASHTO guidelines for H/Bc P 2.4. At
H/Bc < 1.53, with higher Fe values for the thin culvert than the thick H/Bc < 0.38, the calculated Fe values for the thick culvert case are
culvert. considering at rest pressure condition, Fe < 1.0 at the top in good agreement with those proposed by AASHTO.
corner, except at H/Bc = 0.77 where F  1.0. At the bottom corner, Several researchers proposed soil culvert interaction factors for
Fe > 1.0 for H/Bc up to 1.2 and less than 1.0 for H/Bc > 1.2. For the different soil heights above the culvert (e.g., [4,9,7,8]). Fig. 18 com-
top and bottom corners, Fe < 1.0 and increases for H/Bc 6 0.77 pares the current numerical results with the relationships pro-
and decreases slightly for H/Bc > 0.77. posed by Bennett et al. [4] and Kang et al. [9] as a function of
H/Bc, while Fig. 19 compares the results with the relationship pro-
5.1.4. Comparison of soil culvert interaction factors with AASHTO and posed by Tadros et al. [7] and Kim and Yoo [8] as a function of H.
CBDHC and other researchers The relationship proposed by Kang et al. [9] for compacted and
Fig. 17 compares the SCI factors obtained from the verified FLAC uncompacted side fills shows a reduction in the Fe values as the
2D numerical model at the edge of the top slab with those pro- H/Bc ratio increases, while the relationship suggested by Bennett
posed by AASHTO 2002 [10] and CHBDC 2006 [11]. Both codes et al. [4] shows an increase in the Fe values for H/Bc P 2.42, and
assume uniform soil pressure on the culvert top slab, while the very small increase afterward. The Bennett et al. [4] Fe values are
numerical results demonstrate parabolic distribution with Fe > 1.0 overly conservative, especially for H/Bc > 2.0. The Fe values pro-
at the edges and Fe < 1.0 at the center. The comparison is presented vided by the Tadros et al. [7] and Kim and Yoo [8] methods increase
herein in terms of Fe at the edges. as H increases, but the Tadros et al. [7] method (proposed for silty
O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235 227

H/Bc = 0 H/Bc = 0.08

H/Bc = 0.38 H/Bc = 0.77

H/Bc = 1.53 H/Bc = 2.30

H/Bc = 3.07 H/Bc = 6.13

Fig. 11. Effect of H/Bc ratio on the horizontal stresses around box culvert (t = 0.533 m) (legend units in Pa).

clay) gives low values for Fe that fall well below the range obtained Inspecting Figs. 20 and 21 shows that the bending moments on
in this study for the thick and thin culverts. The Kim and Yoo [8] the culvert top slab and side wall increase with t/Bc but at a
relationships for yielding and unyielding foundations provide Fe decreasing rate. The bending moment diagrams suggest that the
values that fall between those for the thick and thin culvert. ratio of t/Bc  0.1 may be considered to be a limiting ratio beyond
which very little to no further increase in moment is expected.
5.2. Effect of t/Bc ratio Similar observation can be made on the effect of t/Bc on soil pres-
sures on the culvert top slab and side wall as discussed below.
Both experimental and numerical modeling results suggest that Hence, it can be suggested that t/Bc P 0.1 represents the condition
the culvert thickness affects the bending moment, soil pressure of relatively rigid (thick) culvert. On the other hand, at very low
and soil culvert interaction factors significantly. Six different cul- t/Bc ratios of 0.02, the bending moment diagram is slightly differ-
vert thickness values that cover a range of culvert thicknesses used ent than for other cases. Inspecting the deformed shape of culverts
in practise were considered in the parametric study, i.e., t = 0.1, 0.2, with different thicknesses can shed some light on that behavior. It
0.3, 0.4, 0.5 and 0.6 m, normalized by the culvert width, Bc to give appears that for very low t/Bc ratio all sides of the box culvert
t/Bc = 0.02, 0.04, 0.07, 0.09, 0.11, and 0.13. These thicknesses cover deformed inwards (i.e., within the culvert). However, as t/Bc
a wide range of relative stiffnesses. The effects of t/Bc ratio on the increases, the top slab deforms inwards but at a decreasing rate,
bending moment, soil pressure, and soil culvert interaction factors while the side wall deformation changes gradually from inwards
are explored in the following sections. to outwards. These observations suggest that the case of
t/Bc = 0.02 represents a flexible culvert, which is obvious in the
5.2.1. Bending moment amount of horizontal soil pressure attracted to it as will be shown
The effects of t/Bc ratio on the top slab and side wall bending in next section.
moment diagrams are presented in Figs. 20 and 21, respectively.
Fig. 20 shows that the top slab bending moments (both positive 5.2.2. Soil pressure
values at edges and negative values at centers) increase as the Fig. 22 presents the vertical and horizontal soil pressure contour
t/Bc ratio increases. The effect of culvert thickness is even greater lines. The results show that the soil arching is greater for the thin-
on the bending moment of the side wall, as can be noted from ner culvert. As the thickness of the culvert increases, the effect of
Fig. 21. soil arching decreases. This is demonstrated in Fig. 23, which pre-
228 O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235

H/Bc = 0 H/Bc = 0.08

H/Bc = 0.38 H/Bc = 0.77

H/Bc = 1.53 H/Bc = 2.30

H/Bc = 3.07 H/Bc = 6.13

Fig. 12. Effect of H/Bc ratio on the horizontal stresses around box culvert (t = 0.267 m) (legend units in Pa).

sents a horizontal section throughout the model above the top slab with depth and peak values occur at the top and bottom corners.
of the culvert, while Fig. 24 presents the vertical section passing The general trend is that as t/Bc decreases, the horizontal soil pres-
beside the side wall of the culvert. The results display the effect sure increases at the top and bottom corners, while at the center
of soil arching in terms of large increase in soil pressure (i.e., stress the opposite behavior is observed, and the horizontal soil pressure
concentration) at the edges and corners, followed by a reduction in decreases.
the soil pressure toward the center of top slab or side wall. The The effect of t/Bc ratio can be summarized in two distinguishing
results also show that soil arching decreases as the t/Bc ratio behaviors that can be separated at the ratio t/Bc  0.1. For
increases. t/Bc > 0.1, all horizontal soil pressures increase with depth, and
The effect of t/Bc ratio on the soil pressure diagrams on the top their values reduce as t/Bc increases. For t/Bc < 0.1, large increases
slab and side wall are illustrated in Figs. 25 and 26, respectively. in the horizontal soil pressures are observed at the top and bottom
The results show that the vertical soil pressure on the top slab corners, while at the center of the side wall the horizontal soil
decreases at the edges as the ratio of the t/Bc increases and the pressure distribution decreases as t/Bc decreases. For very low
opposite occurs at the center where the vertical soil pressure t/Bc = 0.02, the soil pressures on the top slab and side wall are large
increases as the ratio of t/Bc increases. As the thickness of the cul- at the edges, top and bottom corners, and small at the center.
vert increases especially for the ratios of t/Bc higher than 0.09, the
vertical soil pressure shows very similar values as the t/Bc ratio 5.2.3. Soil culvert interaction factors
increases, while as the thickness of the culvert decreases (i.e., Fig. 27 presents the effect of t/Bc ratio for different H/Bc values
t/Bc < 0.09), the difference between the vertical soil pressure on the soil culvert interaction factors for the culvert top slab. The
increases. Generally, all vertical soil pressure diagrams show that results show that for the edge, Fe > 1.0 and for the center, Fe < 1.0.
the soil pressure distribution on the top slab takes a parabolic Generally, Fe for the edge decreases as t/Bc increases, while Fe for
shape with increases at the edge and decreases at the center. This the center increase as t/Bc increases. Also, Fe for the edge is much
parabolic shape is a function of the thickness of the culvert, i.e., as larger than Fe for the center for t/Bc < 0.1, and this difference
t/Bc increases, the difference between the vertical soil pressure val- diminishes for t/Bc > 0.1.
ues at the edges and at the center decreases and vice versa. On the Fig. 28 shows the effect of t/Bc ratio on the soil culvert interac-
side wall, the horizontal soil pressure for all t/Bc ratios increases tion factors for the top and bottom corners of the culvert side wall
O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235 229

600 H/Bc = 0 1.8


Vertical Pressure (kPa)

H/Bc = 0.08
500 1.6
H/Bc = 0.38
H/Bc = 0.77 1.4
400
H/Bc = 1.53
1.2
H/Bc = 2.30
300
H/Bc = 3.07 1

Fe
200 H/Bc = 6.13
0.8

100 0.6
Edge (t/Bc = 0.06)
0.4
0 Center (t/Bc = 0.06)
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 Edge (t/Bc = 0.12)
0.2
Distance (m) Center (t/Bc = 0.12)
0
(a) 0 1 2 3 4 5 6 7
H/Bc
700
Fig. 15. Effect of the thickness and the ratio H/Bc on the soil culvert interaction
Vertical Pressure (kPa)

600 H/Bc = 0 factors Fe on the top slab.

500 H/Bc = 0.08


H/Bc = 0.38
400 1.4
H/Bc = 0.77
H/Bc = 1.53
300 1.2
H/Bc = 2.30
200 H/Bc = 3.07 1
H/Bc = 6.13
100
0.8

Fe
0
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 0.6
Distance (m)
(b) 0.4 Top (t/Bc = 0.06)
Bottom (t/Bc = 0.06)
0.2 Top (t/Bc = 0.12)
Fig. 13. Effect of H/Bc ratio on the soil pressure on the top slab: (a) t = 0.533 m and
Bottom (t/Bc = 0.12)
(b) t = 0.267 m. 0
0 1 2 3 4 5 6 7
Horizontal Pressure (kPa) H/Bc
0 50 100 150 200 250 300
0 Fig. 16. Effect of the thickness and the ratio H/Bc on the soil culvert interaction
0.5 factors Fe on the side wall at rest pressure Ko.
H/Bc = 0
1 H/Bc = 0.08
1.5 H/Bc = 0.38
Distance (m)

1.7
H/Bc = 0.77 Edge (t/Bc = 0.06)
2
H/Bc = 1.53 Edge (t/Bc = 0.12)
2.5 1.6 AASHTO [10] Compacted Sidefill
H/Bc = 2.30
AASHTO [10] Uncompacted Sidefill
3 H/Bc = 3.07
1.5 CHBDC [11] Installation B1
H/Bc = 6.13 CHBDC [11] Installation B2
3.5
1.4
4
Fe

4.5 1.3
(a)
1.2
Horizontal Pressure (kPa)
0 50 100 150 200 250 300 350 1.1
0

0.5 1
H/Bc = 0 0 1 2 3 4 5 6 7 8 9
1 H/Bc = 0.08 H/Bc
H/Bc = 0.38
Distance (m)

1.5
H/Bc = 0.77 Fig. 17. Comparison between the soil culvert interaction factors Fe determined from
2
H/Bc = 1.53 parametric study with AASHTO and CHBDC.
2.5 H/Bc = 2.30
3 H/Bc = 3.07

3.5
H/Bc = 6.13 corner are lower than the top corner. Generally, Fe decreases as
t/Bc increases and for thick culverts (t/Bc > 0.11), Fe approaches
4
unity.
4.5
(b)

Fig. 14. Effect of H/Bc ratio on the soil pressure on the side wall: (a) t = 0.533 m and 5.3. Effect of shear strength parameters, soil elastic modulus and
(b) t = 0.267 m. poisson’s ratio

for different H/Bc values. The results show that for shallow embed- To investigate the effect of soil strength on the culvert behavior
ment depth (H/Bc = 0.08, 0.38) Fe < 1.0, while for large embedment four friction angle values, i.e., 30°, 35°, 40° and 45°, and 3 different
(H/Bc P 1.67) Fe > 1.0. At t/Bc = 0.02, Fe values for the bottom dilation angles, i.e., 0°, 5° and 10° were considered. The results
230 O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235

2 BM (kN.m/m)
Edge (t/Bc = 0.06)
1.9 -50 0 50 100 150 200
Edge (t/Bc = 0.12)
0
Bennett et al [4]
1.8
Kang et al [9] Compacted Sidefill 0.5
1.7 Kang et al [9] Uncompacted Sidefill t/Bc = 0.02
1 t/Bc = 0.04
1.6

Distance (m)
1.5 t/Bc = 0.07
Fe

1.5
2 t/Bc = 0.09
1.4
t/Bc = 0.11
2.5
1.3 t/Bc = 0.13
3
1.2
3.5
1.1
4
1
0 1 2 3 4 5 6 7 8 9 4.5
H/Bc
Fig. 21. Effect of t/Bc ratio on the bending moment on the side wall.
Fig. 18. Comparison between the soil culvert interaction factors Fe determined from
parametric study with Bennett et al. [4] and Kang et al. [9].
level [33]. To investigate the effect of Es on the culvert behavior,
five values were considered in the analysis, i.e., Es = 10, 20, 30,
40, and 100 MPa.
1.7
Edge (t/Bc = 0.12) Fig. 29 shows the effect of Es on the bending moment diagrams.
1.6
Edge (t/Bc = 0.06)
Tadros et al [7] Silty Clay
Generally, the effect of Es on the bending moment diagrams is
Kim et al [8] Yielding Foundation. noticeable at the center of the top slab and side wall and the bend-
1.5 Kim et al [8] Unyielding Foundation
ing moment decreases as Es increases. At the edges, the bending
1.4 moments are almost the same for all Es values.
Fig. 30 shows the effect of Es on the SCI factors. It is noted that Fe
Fe

1.3
values on the top slab increase at the edge and decrease at the cen-
1.2
ter linearly as Es increases. On the side wall, Fe for both the top and
1.1 bottom corners increase linearly as Es increases.
1
5.3.2. Effect of Poisson’s ratio
0.9
0 5 10 15 20 25 30 To investigate the effect of Poisson’s ratio on culvert behavior,
H (m) four values of Poisson’s ratio were used, covering the range of Pois-
son’s ratio for sand, i.e., 0.2, 0.25, 0.3, and 0.35.
Fig. 19. Comparison between the soil culvert interaction factors Fe determined from Fig. 31 shows the effect of Poisson’s ratio on the bending
the parametric study with Tadros et al. [7] and Kim and Yoo [8]. moment diagrams for top slab and side wall. The results show that
as Poisson’s ratio increases, the positive bending moment at the
edges decreased and the negative bending moment at the center
150 increased. Similar shift of the bending moment on the side wall
was observed.
100 t/Bc = 0.02 Fig. 32 shows the effect of Poisson’s ratio on the soil culvert
50 t/Bc = 0.04
interaction factors for top slab and side wall. Generally, Poisson’s
BM (kN.m/m)

t/Bc = 0.07
0 ratio effect on Fe values for the top slab is moderate, but its effect
t/Bc = 0.09
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 is strong on Fe values for the side wall. On the top slab, Fe at the
-50 t/Bc = 0.11
t/Bc = 0.13 edge decrease linearly as Poisson’s ratio increases, while at the
-100 center Fe increases linearly as Poisson’s ratio increases. The Fe val-
-150 ues on the side wall increase as Poisson’s ratio increases. For Pois-
son’s ratios = 0.2 and 0.25, Fe < 1.0, while for Poisson’s ratios = 0.3
-200
and 0.35, Fe > 1.0.
Distance (m)
It should be noted that the above analyses for different soil elas-
Fig. 20. Effect of t/Bc ratio on the bending moment on the top slab. tic modulus and Poisson’s ratio was repeated for different H/Bc
ratios but only for t/Bc = 0.12 and yielded the same results for dif-
ferent culvert embedment ratios.
indicated that there is no effect on the culvert behavior from the
shear strength parameters indicating that the behavior remained
6. Static design guidelines for box culvert
within the elastic range.
The purpose of this section is to investigate the effect of soil
The results of the static centrifuge and numerical models under-
elastic modulus and Poisson’s ratio on the soil culvert interaction
score several factors that may affect the design of square box cul-
factors. Thus, the analysis are performed using constant culvert
verts. In this section, some potential design guidelines interpreted
thickness of 0.533 m (t/Bc = 0.12) as it is widely used in practice.
from these results are suggested to aid in the design of box culverts
under the effect of static loads. For the static design of a box cul-
5.3.1. Effect of soil elastic modulus vert, the main focus is on the static soil pressure, which can be
The elastic modulus, Es, of fine sand (similar to sand used in the used to determine the internal forces of the structural members
centrifuge tests) ranges from 8 to 30 MPa, and for coarser sands it of the box culvert. These internal forces are required for designers
may range from 10 to 80 MPa depending on the relative density to have a safe and economic design. Several factors affect the shape
O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235 231

t/Bc = 0.02 t/Bc = 0.04

t/Bc = 0.07 t/Bc = 0.09

t/Bc = 0.11 t/Bc = 0.13

(a)

t/Bc = 0.02 t/Bc = 0.04

t/Bc = 0.07 t/Bc = 0.09

t/Bc = 0.11 t/Bc = 0.13

(b)

Fig. 22. Effect of t/Bc ratio on the stresses around box culvert: (a) vertical stresses and (b) horizontal stresses (legend units in Pa).

and value of the soil pressures including: soil properties, H/Bc and soil pressure diagram takes a curved shape at the mid-section sim-
t/Bc ratios, as well as the elastic modulus of the culvert material. ilar to a parabola. This parabolic shape increases with increases in
The proposed static design guidelines are based on these factors the soil height above the culvert. In general for simplicity, by defin-
and summarized in the following sections. ing the top and bottom values of the horizontal soil pressure, the
horizontal soil pressure diagram on the side wall can be considered
6.1. Soil pressure distributions to linearly increase with depth as the curved shape in the
mid-section of the side wall appears more under high soil pressure
The actual shape of the vertical soil pressure on the culvert top columns above the culvert or very small culvert thicknesses that
slab is parabolic. By defining the pressures at the edges and center, are not normally used in design.
the resulting parabola can describe the actual vertical soil pressure
diagram on the top slab. The horizontal soil pressure diagram on 6.2. Soil pressure values
the side wall generally shows an increase in their values with
depth. The increase in the horizontal soil pressure is linear for To determine the soil pressure on the top slab, the theoretical
the thicker culverts, while for the thinner culverts the horizontal vertical soil pressure rv is evaluated based on the soil unit weight
232 O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235

250 Box Culvert Location Horizontal Pressure (kPa)


0 50 100 150 200
0
Vertical Pressure (kPa)

200 t/Bc = 0.02


t/Bc = 0.04 0.5
t/Bc = 0.07 t/Bc = 0.02
150 1
t/Bc = 0.09 t/Bc = 0.04
t/Bc = 0.11 1.5 t/Bc = 0.07

Distance (m)
100 t/Bc = 0.13
Vertical Stress 2 t/Bc = 0.09
t/Bc = 0.11
50 2.5
t/Bc = 0.13
3
0
3.5
0 10 20 30 40 50 60
Distance (m) 4

4.5
Fig. 23. Effect of t/Bc ratio on the vertical stresses at the level of the top slab.
Fig. 26. Effect of t/Bc ratio on the soil pressure on the side wall.

Horizontal Pressure (kPa)


1.8
0 20 40 60 80 100 120 140
0 Edge (H/Bc = 0.08)
1.7 Edge (H/Bc = 0.38)
2
t/Bc = 0.02 Edge (H/Bc = 1.67)
1.6
4 t/Bc = 0.04
Edge( H/Bc = 2.30)
6 t/Bc = 0.07 1.5
Edge (H/Bc = 6.13)
Distance (m)

t/Bc = 0.09
Fe
8 1.4
Box t/Bc = 0.11
10 Culvert t/Bc = 0.13 1.3
Location Horizontal Stress
12
1.2
14
1.1
16
18 1
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
20
t / Bc
Fig. 24. Effect of t/Bc ratio on the horizontal stresses at the level of the side wall. (a) Fe values at the edge of the top slab

1.2

300
1
Vertical Pressure (kPa)

250 t/Bc = 0.02


0.8
t/Bc = 0.04
200 t/Bc = 0.07
Fe

0.6
t/Bc = 0.09 Center (H/Bc = 0.08)
150
t/Bc = 0.11 0.4 Center (H/Bc = 0.38)
t/Bc = 0.13 Center (H/Bc = 1.67)
100
0.2 Center (H/Bc = 2.30)
50 Center (H/Bc = 6.13)
0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
0
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 t / Bc
Distance (m) (b) Fe values at the center of the top slab
Fig. 25. Effect of t/Bc ratio on the soil pressure on the top slab. Fig. 27. Effect of the thickness ratio t/Bc on the soil culvert interaction factors Fe on
the top slab.

cS and the height of soil column H above the culvert. The earth 6.3. Soil culvert interaction factors
pressure coefficient K is required to evaluate the theoretical hori-
zontal soil pressure rh on the side wall as shown in Eqs. (6) and Several charts have been provided for the SCI factor, Fe, for the
(7), respectively. edge and center of the top slab, as well as the top and bottom cor-
ners of the side wall. It was found that the Fe value is a function of
ðrv ÞTheoritical ¼ cS  H ð6Þ H/Bc ratio (external pressure) and t/Bc ratio (i.e., culvert relative
stiffness).
ðrh ÞTheoretical ¼ K  cS  H ð7Þ
Once the soil culvert interaction factor Fe is determined, the
design vertical and horizontal soil pressures can be calculated 6.4. Design example
using Eqs. (8) and (9), respectively.
A simple example is presented herein to demonstrate the proce-
ðrv ÞActual ¼ F e  ðrv ÞTheoritical ð8Þ dure for obtaining the static soil pressures and static bending
moments on the top slab of box culverts using the soil culvert
ðrh ÞActual ¼ F e  ðrh ÞTheoritical ð9Þ interaction factors. For the purpose of this example, a box culvert
O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235 233

2.5 width Bc = 4.572 m, its thickness t = 0.533 m and the height of soil
Top (H/Bc = 0.08)
fill H = 7.62 m is assumed.
Top (H/Bc = 0.38)
2
Top (H/Bc = 1.67)
Top (H/Bc = 2.30) Step 1: Calculate the H/Bc and/or t/Bc ratios, and use Fig. 15
1.5 Top (H/Bc = 6.13) and/or Fig. 27 to determine the soil culvert interaction factor
Fe

value Fe. The results obtained are H/Bc = 1.67, t/Bc = 0.12 and
1 the Fe values at the left and right edges of the top slab is 1.08
and at the center is 0.92.
0.5 Step 2: Calculate the theoretical vertical soil pressure using Eq.
(6). Using a soil unit weight of 16.56 kN/m3 the theoretical ver-
0 tical soil pressure is 126.2 kPa.
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
t / Bc
Step 3: Calculate the actual soil pressure on the top slab using
Eq. (8). The actual soil pressure at both edges is 135.87 kPa
(a) Fe values at the top corner of the side wall
and at the center is 116.09 kPa.
1.6
Step 4: Use the three static soil pressures calculated at the edges
Bottom (H/Bc = 0.08) and center to fit a 2nd order polynomial. The fitted equation can
1.5 Bottom (H/Bc = 0.38) be integrated twice to obtain the actual static shear force and
Bottom (H/Bc = 1.67) static bending moment diagrams on the top slab as shown in
1.4 Bottom (H/Bc = 2.30) Fig. 33. The resulting example fitted and integrated equations
Bottom (H/Bc = 6.13)
are:
Fe

1.3

1.2 W ¼ 4:9505x2 þ 3  1013 x þ 116:09 ð10Þ


1.1
SF ¼ 1:65017x3 þ 1:5  1013 x2 þ 116:09x ð11Þ
1
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
BM ¼ 0:4125x4 þ 5  1014 x3 þ 58:045x2  132:06 ð12Þ
t / Bc
(b) Fe values at the bottom corner of the side wall The same procedure can be used to determine the static bend-
ing moments on the side wall. For the side wall the horizontal soil
Fig. 28. Effect of thickness ratio t/Bc on soil culvert interaction factors Fe on side
pressure distribution should be fitted to linear equation in order to
wall at rest pressure Ko.

(a) top slab (b) side wall


BM (kN.m/m)
0 50 100 150 200
150 0

0.5
100
1 Es = 10 MPa
50 Es = 20 MPa
BM (kN.m/m)

1.5
Distance (m)

Es = 30 MPa
0 Es = 40 MPa
2
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Es = 100 MPa
-50 2.5

-100 Es = 10 MPa 3
Es = 20 MPa
3.5
-150 Es = 30 MPa
Es = 40 MPa 4
-200 Es = 100 MPa
4.5
Distance (m)

Fig. 29. Effect of soil elastic modulus on the bending moment on the top slab and side wall.

(a) top slab (b) side wall


1.2 1.16
1.15 1.14 Top
1.1 Bottom
1.12
1.05
1.1
1
Fe

Fe

Edge 1.08
0.95
Center
1.06
0.9
0.85 1.04
0.8 1.02
0.75 1
0 10 20 30 40 50 60 70 80 90 100 110 0 10 20 30 40 50 60 70 80 90 100 110
E s (MPa) Es (MPa)

Fig. 30. Effect of soil elastic modulus on soil culvert interaction factors Fe.
234 O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235

(b) side wall


BM (kN.m/m)
(a) top slab -50 0 50 100 150 200
200
0
150 0.5
100 1
BM (kN.m/m)

50 ν = 0.2

Distance (m)
1.5
ν = 0.25
0 2
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 ν = 0.3
-50 2.5 ν = 0.35
-100 ν = 0.2 3
ν = 0.25
-150 3.5
ν = 0.3
-200 4
ν = 0.35
Distance (m) 4.5

Fig. 31. Effect of Poisson’s ratio on the bending moment on the top slab and side wall.

(a) top slab (b) side wall


1.15 1.6

1.1
1.4
1.05

1 1.2

Fe
Fe

0.95 1
0.9
0.8 Top
0.85 Edge
Bottom
Center
0.8 0.6
0.15 0.2 0.25 0.3 0.35 0.4 0.15 0.2 0.25 0.3 0.35 0.4
ν ν

Fig. 32. Effect of the Poisson’s ratio on the soil culvert interaction factors Fe.
Vertical Soil Pressure (kPa)

180 obtain static bending moment as 3rd order polynomial fit. It is also
160 important to note that the Fe values obtained from the H/Bc and
140
t/Bc charts are for Es = 30 MPa and m = 0.28. The effect of different
120
100 soil parameters has been examined and it was found that there
80 Proposed method is no effect for the H/Bc ratio.
AASHTO [10] Compacted Sidefill
60 AASHTO [10] Uncompacted Sidefill The results of the proposed method which represent the mea-
40 CHBDC [11] Installation B1
CHBDC [11] Installation B2 sured values and shape of soil pressure on the top slab are also
20 Vertical soil pressure compared to the AASHTO, CHBDC and theoretical soil pressure
0
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 methods as shown in Fig. 33. The soil pressures produced by
Distance (m) AASHTO depend on the compaction of the side-fill, while the soil
pressures produced by CHBDC depend on the soil type. As the
400 AASHTO and CHBDC soil culvert interaction factors Fe values are
300 higher than the proposed method, the resulting soil pressures are
200 also higher and have a uniform soil distribution. The higher values
SF (kN/m)

100 of soil pressure therefore produce higher values of shear forces and
0 bending moments. In addition, the theoretical soil pressures only
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 represent an average state between the soil pressure values at
-100 Proposed method
-200 AASHTO [10] Compacted Sidefill
AASHTO [10] Uncompacted Sidefill
the edge and center of the top slab. As explained earlier, AASHTO
-300 CHBDC [11] Installation B1 and CHBDC do not consider the effect of culvert thickness. Hence
CHBDC [11] Installation B2
-400 Vertical soil pressure for a thinner culvert with soil pressure values even higher at the
Distance (m) slab edges and lower at the slab center, the results presented in
the proposed method in Fig. 33 would diverge even further.
200
150
7. Conclusions
100
BM (kN.m/m)

50
Four centrifuge tests were conducted to examine soil culvert
0
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 interaction under static loadings. The results from these tests were
-50
-100
used to calibrate and verify a numerical model created using the
-150
computer program FLAC 2D. The verified model was used to per-
Proposed method
AASHTO [10] Compacted Sidefill -200
form a static parametric study to investigate the effect of several
AASHTO [10] Uncompacted Sidefill
CHBDC [11] Installation B1 -250 factors that may affect the soil culvert interaction factors such as
CHBDC [11] Installation B2
Vertical soil pressure Distance (m) geometric configurations and different soil properties. The investi-
gation includes bending moment, soil pressure, and soil culvert
Fig. 33. Example static soil pressure, shear force and bending moment diagrams. interaction factors.
O. Abuhajar et al. / Computers and Geotechnics 69 (2015) 219–235 235

The effect of the geometric parameters is presented in the form [2] Spangler MG. Underground conduits – an appraisal of modern research. Proc
Am Soc Civ Eng 1947;73:855–84.
of dimensionless numbers H/Bc and t/Bc. The ratio H/Bc = 1.53 can
[3] Clarke NWB. The loads imposed on conduits laid under embankments or valley
be considered to be a limiting ratio for the top slab and side wall fills. Proc Inst Civ Eng Struct Build 1967;36:63–98.
because there is no change in the Fe values after that, while before [4] Bennett RM, Wood SM, Drumm EC, Rainwater NR. Vertical loads on concrete
that it is variable. The Fe values obtained from the H/Bc ratio box culverts under high embankments. J Bridge Eng – ASCE 2005;10(6):643–9.
[5] Katona MG, Vittes PD, Lee CH, Ho HT. CANDE-1980: box culverts and soil
increases at the edges and decreases at the center of the top slab models. Report no. FHWA/RD-80/172. Washington, DC: Federal Highway
as the culvert thickness decreases. Similar observations were made Administration; 1981. p. 214.
for H/Bc ratio on the side wall. Similar observations to the H/Bc [6] Katona MG, Vittes PD. Soil-structure analysis and evaluation of buried box-
culvert designs. Transport Res Rec Transport Res Board 1982;878:1–7
ratio were also made in terms of the thickness effect. Even though [Washington, DC].
the Fe values at the center of the top slab are less than 1.0, it was [7] Tadros MK, Benak JV, Gilliland MK. Soil pressure on box culverts. ACI Struct J
observed that the Fe values increase as the t/Bc ratio increase. At 1989;86(4):439–50.
[8] Kim K, Yoo CH. Design loading on deeply buried box culverts. J Geotech
the edge of the top slab and the top and bottom corners of the side Geoenviron Eng – ASCE 2005;131(1):20–7.
wall, it was observed that the Fe values decrease as the t/Bc ratio [9] Kang J, Parker F, Kang YJ, Yoo CH. Effects of frictional forces acting on sidewalls
increase. of buried box culverts. Int J Numer Anal Methods 2008;32(3):289–306.
[10] American Association of State Highway and Transportation Officials Inc.
The soil density, friction angle, and dilation angle have no AASHTO standard specifications for highway bridges, 17th ed. Washington,
noticeable effect on the Fe values and have a constant Fe value for DC; 2002.
the H/Bc and t/Bc conditions that were used in the investigation. [11] Canadian Standards Association. Canadian Highway Bridge Design Code
(CHBDC). CAN/CSA-S6-06, 10th ed. Mississauga, ON; 2006.
The Fe values increase as the elastic modulus increases at the edges
[12] Canadian Standards Association. Commentary on the Canadian Highway
of the top slab and top and bottom corners of the side wall, while at Bridge Design Code (CHBDC), CAN/CSA-S6-06, 10th ed. Mississauga, ON; 2006.
the center of the top slab, the Fe values decrease as the elastic mod- [13] McGuigan BL, Valsangkar AJ. Centrifuge testing and numerical analysis of box
ulus values increase. The Fe values decrease at the edge of the top culverts installed in induced trenches. Can Geotech J 2010;47(2):147–63.
[14] Dasgupta A, Sengupta B. Large-scale model test on square box culvert
slab as Poisson’s ratio increase, while at the center of the top slab backfilled with sand. J Geotech Eng – ASCE 1991;117(1):156–61.
as well as the top and bottom corners of the side wall, the Fe values [15] Spangler MG. Field measurements of the settlement ratios of various highway
increase as Poisson’s ratio values increase. Several charts showing culverts. Ames, IA: Iowa Engineering Experiment Station; 1950 [Bulletin 170].
[16] Sato Y, Iwasaki Y. A measure for reducing the vertical earth pressure on the
the effect of these factors were produced to show their effect on reinforced concrete culvert. Isuchi to Kiso (Earth and Foundation); 1981.
the Fe values. [17] James RW, Brown DE, Bartoskewitz RE, Coyle HM. Earth pressures on
The Fe values calculated from the numerical modeling were reinforced concrete box culverts. TAMU System, College Station, TX: Texas
Transportation Institute; 1986.
compared to the AASHTO and CHBDC values. The Fe values from [18] Vaslestad J, Johansen TH, Holm W. Load reduction on rigid culverts beneath
AASHTO and CHBDC are greater than 1.0 and therefore were com- high fills: long-term behavior. Transport Res Rec Transport Res Board 1993
pared to the values obtained at the edges of the top slab as it is also [Washington, DC].
[19] Yang MZ. Evaluation of factors affecting earth pressures on buried box
greater than 1.0. The results show that the Fe values from AASHTO culverts. PhD dissertation. Knoxville, TN: University of Tennessee; 2000.
and CHBDC are bound by the results calculated for the thick and [20] Pimentel MPCCF. Behavior of reinforced concrete box culverts under high
thin culverts analyzed. embankments. J Struct Eng – ASCE 2009;135(4):366–75.
[21] Girdler HF. Loads on culverts under high embankments. Research Rep No 386,
Based on the static parametric study, several design charts are
Interim Rep No KYHPR-72-68; HPR-1(9), Part II. Lexington, KY: Department of
proposed. These design charts can be used to define the Fe values Transportation; 1974.
at specific extreme points on the box culvert for different cases. [22] Russ RL. Loads on culverts under high embankments, positive projection
These results can be used to evaluate the actual soil pressure dis- without imperfect trench. Research Rep No 431, Interim Rep No KYHPR-72-68;
HPR-1(11), Part II. Lexington, KY: Department of Transportation; 1975.
tribution around box culverts under static loads. Using these soil [23] Einstein HH, Schwartz CW. Simplified analysis for tunnel supports. J Geotech
pressure values, the static bending moment can be easily Eng Div – ASCE 1979;105(4):499–518.
calculated. [24] Karinski YS, Dancygier AN, Leviathan I. An analytical model to evaluate the
static soil pressure on a buried structure. Eng Struct 2003;25:91–101.
It should be noted that the results presented in this paper are [25] Arulmoli K, Muraleetharan KK, Hossain MM, Fruth LS. VELACS: Verification of
valid for the embankment method of installation of box culverts, liquefaction analyses by centrifuge studies, laboratory testing program, soil
and the specific soil properties and model configurations that were data report. Rep prepared for the Earth Technology Corporation, Project No 90-
0562, Irvine, CA; 1992.
used. For any other installation methods, culvert configuration and [26] Itasca Consulting Group. FLAC, Fast Lagrangian Analysis of Continua. Version
different soil conditions similar analysis should be performed. 5.0. Minneapolis, Minnesota: Itasca Consulting Group; 2005.
[27] Lai T, Elgamal A, Yang Z, Wilson DW, Kutter BL. Numerical modeling of
dynamic centrifuge experiments on a saturated dense sand stratum. In: Doolin
Acknowledgements D, Kammerer A, Nogami T, Seed RB, Towhata I, editors. Proc 3rd int conf on
earthquake geotechnical engineering. Proc 11th int conf on soil dynamics &
The authors would like to show their great appreciation and earthquake engineering and 3rd int conf on earthquake geotechnical
engineering, January 7–9, vol. 1. Berkeley: University of California; 2004. p.
thanks for all the staff members of the Geotechnical Centrifuge
558–65.
Facility at Rensselaer Polytechnic Institute (RPI), Troy, NY, USA, [28] Kanungo M. Soil-structure interaction for buried box culverts. MSc
for providing all the help and support during the experimental part dissertation. London, ON: University of Western Ontario; 2008.
of this research. The authors would also like to thank the Natural [29] Janbu N. Soil compressibility as determined by oedometer and triaxial tests.
In: Proc of the 3rd European conf on soil mechanics and foundation
Sciences and Engineering Research Council of Canada (NSERC) for engineering, Wiesbaden, Germany, vol. 1; 1963. p. 19–25.
their financial support. [30] Fang HY. Foundation engineering handbook. Springer; 1990. pp. 953.
[31] Schanz T, Vermeer PA. On the stiffness of sands. Geotechnique 1998;48:383–7.
[32] Abuhajar OS. Static and seismic soil culvert interaction. PhD dissertation.
References London, ON: University of Western Ontario; 2013.
[33] Hunt RE. Geotechnical engineering investigation handbook. Taylor & Francis
[1] Marston A, Anderson AO. The theory of loads on pipes in ditches and tests of Group; 2005.
cement and clay drain tile and sewer pipes. Ames, IA: Iowa Engineering
Experiment Station; 1913 [Bulletin 31].

You might also like