Integral Bridges and the Modelling of Soil-

Structure Interaction
Steve Rhodes and Julian Moses, LUSAS
No standard approach for the analysis of integral bridges appears in AS5100, AASHTO, the Eurocodes or
other international codes. This paper considers the approaches most suitable for modelling common
integral bridge forms, expanding upon recent UK guidance regarding soil-structure interaction approaches.
Issues including material properties, initial stress state and the incorporation of the effects of soil ratcheting
are discussed and both continuum and spring-type (‘subgrade modulus’) finite element models are explored.

portion where horizontal displacements are largest –

1. INTRODUCTION
pressures perhaps an order of magnitude greater than
Owing to durability problems associated with those experienced by the abutments of a non-integral
movement joints, it is widely accepted that short and bridge. For some integral bridge arrangements, it is
medium length bridges are best designed without such sufficient to carry out the design on the basis of some
joints. This has led to a rise in popularity of integral assumed lateral earth pressure distribution (i.e. a
bridges (no movement joints, no bearings) and semi- limiting equilibrium approach, as described in Section 2
integral bridges (no movement joints) for new below). For others, the soil stiffness plays a more
construction internationally and indeed AS5100.4 [1] significant part in the behaviour of the system and an
requires that the number of deck joints in a bridge be analysis which models the behaviour of both soil and
minimised (clause 5). structure – a soil-structure interaction (SSI) analysis – is
required (as described in Section 3)
Both integral and semi-integral bridges accommodate
the thermal expansion and contraction of the Kp K
superstructure by movement of the abutments or end-
8.0

screens, which are retaining structures. ∆ 4.0 ∆

Often retaining structures are analysed representing 2.0

the soil as merely a load – the stiffness of the soil is not H H

modelled. The design proceeds considering only 1.0

limiting active and passive lateral earth pressures .

a
Passive Movement Active Movement
However, if movements/ deflections of the structure
0.5

are insufficient to mobilise the limiting values, K0

0.25
intermediate values of earth pressure occur, as
Ka

illustrated in Figure 1 below. 0.025 0.025 ∆

The lateral earth pressure depends on the strain in the
soil, which in turn depends on movements in the Wall movement ∆ / Height H
structure. Structural movements depend on the Figure 1. Pressure/ deflection curve
b
stiffness of both structure and soil, and on lateral earth
pressures. Unless the assumption of limiting earth The repeated thermal movements of integral bridge
pressures can be deemed conservative and acceptable, abutments cause particle realignment in granular
an analysis which somehow reflects this loop is backfill materials. Year-on-year, the lateral earth
required. pressures each summer increase, as the backfill
becomes stiffer, a phenomenon known as soil
In some retaining structures, use of limiting earth ‘ratcheting’ [6]. The effect of ratcheting is that soil
pressures can be demonstrably either over- stiffness and therefore the maximum lateral pressure
conservative or unconservative. During the summer can be significantly greater than the ‘intermediate’
expansion of the superstructure in an integral bridge, value (K) that would be obtained from a pressure/
lateral earth pressures on the abutments can approach
the theoretical passive state, especially in the upper
b
After NCHRP Report 343 [3]. The lateral earth pressure is illustrated
in Figure 1 by reference to an earth pressure coefficient (K) which
links lateral pressures to vertical stress in the soil without specific
a
Limiting pressures are described fully in Soil Mechanics textbooks – reference to cohesion, for simplicity. The principle stands for various
see, for example, Craig [2] Chapter 6. soil types. Also see LRFD [4] C3.11.1 or EN1997-1 [5] Figure C.3.

1
deflection curve such as Figure 1 for the expected where
movement (Δ) . After 100-200 cycles, the increase in
c
H height of the wall
stiffness tails off, with maximum (summer) pressures
tending to a value which has been empirically linked
d
d'd wall movement range at H/2 below ground level,
to backfill properties, geometry and the movement taken as 0.5 to 0.7 times the design value of the
range. movement range at the top of the wall, based on
an assessment of the rotation and flexure in the
2. LIMITING EQUILIBRIUM APPROACHES FOR system. See Figure 3 below.
INTEGRAL BRIDGES
Ko coefficient of at-rest earth pressure
‘Limiting equilibrium’ approaches for the design of
Kp;t coefficient of passive earth pressure determined
integral bridges generally use an assumed lateral earth
using the design value of the triaxial φ’ .
e
pressure distribution and earth pressure coefficient,
commonly denoted K*. Where the abutment retains C coefficient dependent upon the elastic modulus
granular material, the pressure distribution and value of the subgrade (Es). Where Es is in MPa:
of K* used should be based on a theory which takes
C = 0.51Eୱ + 14.9 subject to 20 ≤ C ≤ 66
ratcheting into account.
No standard approach for earth pressure distribution
d d d
Deck Pinned Fixed Fixed

behind integral abutments, or for determination of K*

appears in AS5100.3 [8], AASHTO [4] or the Eurocodes
[5] and practices vary [9, 10], with some methods
H/2

making no allowance for ratcheting. The NZ Bridge

d'=0.5d d'=0.5d d'=0.7d
H

Manual [11] refers (in section 4.8.3) to the UK guidance

document BA42/96 [7], however that guidance has
been thoroughly reviewed [12], resulting in the Footing Pinned Fixed Pinned

publication of PD6694-1 [13], which be referred to

throughout this paper. Pure rotation Pure pressure Combined flexure

Figure 2 below shows the assumed pressure

and rotation

distribution given by PD6694-1 [13] for a full height Figure 3. Comparison of different types of rotational
abutment on flexible foundations. and flexural abutment movements after [12] Stage 1
Report, Fig 9
For the shorter height bank pad abutments that
accommodate thermal movements by translation
without rotation, a simple triangular pressure
distribution may be assumed and PD6694-1 [13] clause
9.4.4 gives the following expression for K*:
଴.ସ
40d′ୢ
K = K୭ + ቆ
∗
ቇ K ୮;୲
H
These expressions and the recommendations of
PD6694-1 [13] in general are much improved by
Figure 2. Assumed earth pressure distribution for full comparison to the previous UK guidance in BA42/96
height abutment on flexible foundations after [7] , as described by Denton et al [15].
f

PD6694-1 [13] Fig 5

The study [12] which preceded the publication of
PD6694-1 [13] clause 9.4.3 indicates that for this case, PD6694-1 [13], found limiting equilibrium approaches
K* can be conservatively calculated using: based on K* to be appropriate for some common
Cd′ୢ
଴.଺ integral bridge abutment types such as those
K ∗ = K୭ + ቆ ቇ K ୮;୲ illustrated in Figure 4 below. It is usual for these
H
abutment types to be constructed with a free-draining

c
By the same token, the relative movements required to reach active
e
Reference to PD6694-1 [13] clause 9.4.1 is recommended
or passive conditions suggested by LRFD [4] Table C3.11.1-1 are not f
BA42/96 including Amendment 1 [7] was itself a significant
applicable. improvement upon the first edition [14]. Removal of the limit K*≥
d
Similar peak values are reached even if the backfill was not very Kp/3 of [14] clause 3.5.4, which had caused some engineers to regard
well compacted at placement (BA42/96 [7] clause 3.2) the method as crude and over-conservative, is of particular note.

2
granular material used for backfill – it should be noted
that the effects of soil ratcheting may be ignored when
the material behind the abutment is a cohesive soil.

Figure 5. Typical 3D (shell) model of a full-height

abutment integral bridge
From such a structure-only model, load effects
(i) Full height integral abutment on pad footing (bending moments, shear forces etc) may be obtained
for design purposes. Care must be taken to ensure
that all components – superstructure, abutment walls
or end screens, foundation members/ piles – are
designed considering suitable maximum or minimum
earth pressures in combination with all other
applicable loads, including the corresponding bridge
temperature.
The pressures which should be considered in
conjunction with bridge expansion and contraction are
illustrated in PD6694-1 [13] Figure 6 and the Finnish
Transport Agency Guideline [18] Figure 12. Notably
the latter identifies the minimum pressure as being
close to zero (rather than Ka) with not only contraction
of the bridge but also freezing of the soil at the lowest
(ii) Full height integral abutment on piles
temperatures.
PD6694-1 [13] suggests that the lateral earth pressure
on wingwalls of abutments which support K* pressures
should themselves be subject to a pressure distribution
similar to that illustrated in Figure 2, but calculated
using the greater of Ko or Ka×K* (clause 9.9).
In skew integral abutment bridges, the earth pressures
(iii) Bank pad
(normal to the walls) create a couple which, unresisted,
Figure 4. Integral abutment types which can typically would cause the bridge to rotate on plan.
be designed with K* approaches Furthermore, for bridges such as types (i) and (ii) of
both Figures 4 and 6, thermal expansion results in
In the limiting equilibrium approach, the assumed
twisting of the top of the abutment walls relative to
earth pressures are applied to an analysis model of the
their bases, along with any intermediate piers.
structure in question. A 2D frame (beam element)
PD6694-1 [13] discusses the rotation and twisting
model or 3D model may be used. The grillage (grid)
(clause 9.8) but a simple calculation approach is not
analogy is routinely used for bridge superstructure
given. Where such effects are deemed to be
analysis and some texts advocate extending such
significant, a 3D analysis and perhaps SSI analysis may
models into three dimensions to include abutment
be appropriate.
walls [16]. However, in-plane effects would be
expected to be significant and these can create
3. SOIL STRUCTURE INTERACTION APPROACHES
misleading local in-plane distortions of grid members in
a 3D analysis [17]. Therefore 3D finite element models The limiting equilibrium approach is not appropriate
would probably be more appropriate (as in Figure 5 where soil stiffness plays a more significant part in the
below). But, whatever type of model is used, the soil is behaviour of the system. This is the situation for a
represented only as a load – this is not a soil-structure number of common integral bridge types such as those
interaction analysis; it is a way of avoiding a soil- illustrated in Figure 6 below.
structure analysis.

3
 – type (iii) in Figure 6 – are perhaps the most popular
integral bridge type where space allows. For these, a
‘subgrade modulus’ model where the soil stiffness is
represented using springs is probably most
appropriate. Some authorities require the upper
portion of piles to be sleeved in an effort to reduce the
effects of SSI. As a result, the movement of the bank
Wall pads in and out of the backfill will be relatively greater,
due to the lower resistance from the foundations. Soil
ratcheting will occur, resulting in increased earth
pressures on the bank pads – but the total lateral load
arising will be quite limited due to the low height of the
pads. Structural models with springs and K* pressures
(i) Embedded wall integral abutment may be adequate, as discussed in section 3.2 below.
The ratcheting effect may be further mitigated by the
use of pea gravel or other specialised backfill.
Embedded walls (contiguous piled, secant or
diaphragm) – type (i) in Figure 6 – are perhaps most
popular for short-span underpasses in congested urban
areas [9]. For such structures, a full continuum model
may be more suitable.

3.1 FE continuum models

3.1.1 Element Selection and Considerations

For the representation of soil masses, beam or plate
(ii) Full height integral abutment on single row of piles elements (most familiar to bridge engineers) are
inappropriate and continuum elements must be used.
In some bridges, 3D effects are of concern, however,
often 2D models are sufficient and are certainly
recommended for preliminary studies of SSI issues –
for example, to assess the sensitivity of results to
possible variation in certain parameters. Figure 7
shows the analysis model for an embedded wall
underpass (type (i) in Figure 6), where 2D plane strain
elements are used to represent the soil and 2D beam
Discrete g

elements used to represent the structural members.

piles

(iii) Bank pad on single row of piles

Figure 6. Integral abutment types which typically
require SSI analysis
SSI can be handled by a number of calculation methods
including closed form solutions, although these
typically consider only simple cases, not extending to
integral bridge arrangements. Most real project cases
require numerical integration for which the finite Figure 7. Typical 2D model of a single-span underpass
difference, boundary element or finite element (FE) (part model shown)
methods may be employed. FE approaches to SSI tend
to fall into those which represent the soil using g
Plane strain elements are suitable for thick body problems as it is
continuum elements and those which represent the assumed that the out-of-plane strain is zero and so the out-of-plane
soil using springs. stress is non-zero. By contrast, plane stress elements are for thin
body problems as they are based on the assumption that the out-of-
Bank pad abutments, often supported on steel H-piles plane stress is zero and so the out-of-plane strain is non-zero.

4
As in all FE analyses it is important to ensure that the To describe a granular material using a Mohr-Coulomb
number of elements used is sufficiently large that any material model, elastic properties must be given (E', ν')
inaccuracy arising from the division strategy may be along with the initial and final internal angles of
deemed negligible by comparison to other friction (φ'1, φ'2), cohesion (c’) and dilation angle (ψ)
j k

assumptions inherent in the analysis. For FE [21]. Effective stress parameters are referenced and
continuum analyses we must also consider that the concept of effective stress is important because
elements give best accuracy at an aspect ratio of 1:1 the stiffness of a saturated soil is dependent on
and equal internal angles, although pragmatically ratios whether an increase in load may be carried by fluids,
up to 1:3 are usually acceptable in areas of interest, fluids and soil skeleton or skeleton only and the shear
and ratios up to 1:10 may be acceptable in remote strength of a soil is dependent on the effective normal
regions of the model. stress.
When modelling a structure interacting with a soil Simple test models can be used to show that the
mass, the extent of the model is not straightforward to internal stresses in a block of Mohr-Coulomb material
define: vertical and horizontal boundaries must be correspond to active and passive pressures as
imposed on the soil mass at some distance from the predicted by Rankine-Bell equations. Figure 8
structure . Where such boundaries cannot be
h
illustrates this, showing a test case (plane strain soil
reasonably defined to match physical boundaries (e.g. block of 10m×10m) with active pressures achieved by
free soil face, bedrock) they need to be determined by movement of the right boundary; movement in the
comparing key results from several models which are opposite direction will similarly achieve the expected
identical except for the assumed width or depth. passive pressure.
Where the stiffness of the soil (E’) has been assumed
constant with depth, the predicted deflection under
vertical load at the surface will increase as the depth of
soil below the structure is increased so other key
results should be used for comparison. If E’ increases
with depth, this effect is less pronounced.

3.1.2 Soil Material Properties and Initial Stresses

The simplest SSI models can assume isotropic linear
elastic material properties for the soil. These would
require only two parameters in their definition: elastic
modulus, (E’, perhaps varying with depth) and
Poisson’s ratio (ν), or shear modulus and bulk modulus.
However, clearly most situations which demand SSI
analysis also demand a more realistic mathematical
representation of the soil to be employed.
There are many nonlinear material models designed to
represent soil behaviour. Amongst these, the Mohr-
Coulomb model is probably the most widely used ([19]
section 3.4.1). Yield is based on a critical shear stress
which is dependent on the normal pressure, making it
Figure 8. Results from Mohr-Coulomb test model
applicable for soils where strength increases with
where ࣘ’=30
balanced confining stresses. When yield is exceeded,
volumetric plastic straining (dilatancy ) occurs, and The results of Figure 8 may be corroborated simply:
i

isotropic hardening may be assumed.

Vertical stress at 10m depth
SY = ρgH = 1.8 × -9.81 × 10 = -176.58
h
Where dynamic effects are required to be considered, non-
reflective boundaries are typically required. Such considerations are
outside the scope of this paper.
i
Soils comprise a high proportion of void (filled with single or multi-
j
For detailed information on friction hardening in Mohr-Coulomb
phase fluid): a typical medium dense sand might be 1/3 void by models, see [20].
volume; a normally consolidated clay might edge towards 1/2. When
k
The dilation angle ψ describes the amount of volumetric change
the soil is subject to straining, some rearrangement of particles that occurs during plastic straining or shearing. For plastic
occurs, accompanied by change in volume: this is dilatancy. Either a deformation at constant volume, ψ = 0; for soils that contract when
reduction in volume or an increase in volume can occur, depending they are sheared plastically, ψ < 0; for soils that expand when they
on the particle shape and packing arrangements. See [19] section are sheared plastically, ψ > 0; for most real soils, ψ < φ'. See [19]
2.6 section 3.3.4

5
Maximum (initial) lateral stress at 10m depth b) Contact slidelines. Contact algorithms in software
such as LUSAS enable the proximity of elements to
SX = SZ = Ko×SY where Ko was taken to be 0.5
l
each other to be detected, allowing transfer of
Thus SX (initial) = 0.5×-176.58=-88.29 load between one ‘component’ and another
without adjacent elements actually sharing nodes.
Minimum (final) lateral stress at 10m depth
In the context of SSI, the components would be
SX = SZ = Ka×SY where the soil and the structure, and frictional slidelines
would typically be of interest.
1 − sinϕ′
Kୟ = = 0.333
1 + sinϕ′ c) Elasto-plastic interface materials. A layer of
elasto-plastic material (assigned to plane strain or
Thus SX (final) = 0.333×-176.58=-58.86
3D continuum elements) can represent the
The displacement of the wall (or in this case, boundary) friction-contact relationship between the soil and
which is required to mobilise the limiting active or the structure. The material reproduces the
passive pressure must be understood to be dependent nonlinear response of a system containing planes
upon not only the elastic modulus of the material (as of weakness governed by Mohr-Coulomb type
might be immediately anticipated) but also upon the laws.
initial stress in the soil. By varying the assumed value
Whichever of these options is utilised, for retaining
of Ko, the displacements required for full active or
structures and integral bridges, the crucial
passive pressure to be reached is altered, as shown in
consideration is usually back of wall friction, δ. The
Table 1.
value of δ cannot be less than zero (a notionally
Table 1: Effect of assumed value for Ko on movement smooth wall) nor exceed that of φ' for the material
required to mobilise limiting earth pressures being retained. For many retaining structures, lower
bound φ' and δ are deemed critical for design and δ=0
Limiting pressure at 10m depth
is used, as suggested in PD6694-1 [13] clause 7.2.2 and
Ko mobilised at (mm)
CIRIA C580 [22] section 4.1.4, whereas for integral
Active Passive bridges upper bound values may also be critical. For
the design of integral bridge abutments, BA42/96 [7]
0.1 0 90
clause 3.3 states that wall friction should be taken as
0.5 6 90 δ=φ'/2.
1.0 22 70
3.1.4 Representing the Structure
2.0 85 35
It is important to remind ourselves that relative
stiffness is crucial to the distribution of loads in an FE
model. In the case of a wholly concrete structure, for
The values of Table 1 illustrate the importance of initial
example, the accuracy of the value used for elastic
stresses in SSI analyses. The applied loads for the
modulus affects deflections but generally has little
initial (equilibrium) state in an FE analysis must include
effect on load distribution, since the relative stiffness is
an initial stress which varies with depth, usually based
accurate. However, in the case of a concrete structure
on an assumed Ko.
in contact with the ground, a reasonably accurate
relative stiffness may demand more consideration of
3.1.3 Representing Interfaces in Continuum-Based
issues such as concrete cracking and creep
Models
deformation. While these are considerable topics in
The interface between soil and structure typically their own right, it should be underlined that reinforced
needs some special consideration in any SSI analysis. concrete (RC) is generally cracked and therefore has a
Comprehensive FE systems such as LUSAS offer a range stiffness significantly less than that which would be
m

of options in this area such as [21]: assessed using the gross section and the elastic
modulus from a code of practice (see [24] section
a) Joint elements & materials. Known as ‘link’, ‘hook’
2.4.2).
or ‘fuse’ elements in some software, joint
elements notionally have no length but instead
provide a means of connecting two adjacent
elements without full fixity, introducing options
such as frictional or yielding behaviour.
m
The cracked stiffness depends upon reinforcement provided but as
an indication, ACI 318-08 [23] clause 10.10.4.1 suggests that cracked
l
Various expressions for Ko exist, see Craig [2] Chapter 6, EN1997-1 RC columns and walls may be treated as having a stiffness ~70% of
[5] clause 9.5.2 etc the calculated gross stiffness; cracked slabs having only ~25%.

6
3.1.5 When and How to Incorporate the Effects of Soil Nonlinear springs or joints may be used within an FE
Ratcheting model to generate lateral earth pressures for a
retaining wall design based on a pressure/ deflection
Often, the popular embedded wall integral abutment
relationship such as that in Figure 9 below:
(Figure 7 and type (i) in Figure 6), is constructed as part
of a top-down scheme in cohesive soil. As noted Lateral pressure σ
earlier, for such soils, the effects of strain ratcheting
may be ignored (see PD6694-1 [13] clauses 9.4.5.2 and
A3.2) and so a suitable SSI analysis (perhaps utilising a
Passive Movement
σ'p
Cam Clay material model) may be used with no further
special considerations. The software used must be
capable of modelling the staged construction process
in conjunction with the use of the preferred nonlinear kh
soil material.
σ'0
For integral bridges with embedded walls (type (i) in Active Movement

Figure 6) or full height integral abutments on a single

σ'a
row of piles (type (ii) in Figure 6), retaining granular
materials, guidance for a suitable SSI analysis Horizontal deflection
incorporating ratcheting is given in PD6694-1 [13]
Annex A with further background given in [12] Stage 2 Figure 9. Nonlinear ‘soil joint’ pressure/ deflection
report, Section 5. The recommended approach graph [21]
essentially entails the soil being modelled as a
Critically Figure 9 incorporates not only the modulus of
continuum, with an elastic modulus (E’) which varies
subgrade reaction, kh, but also at-rest earth pressures
with depth according to an assumed lateral earth
(σo, based on Ko). Neither quantities are considered
pressure profile which can be regarded as a ‘quasi-
when retaining walls are designed using limiting earth
passive’ limit, similar in nature to the K* profile
pressure methods, however they are essential
described for the limiting equilibrium method above.
components of SSI analyses (see CIRIA C580 [22],
Along with this, lateral earth pressures are restricted to
section 5.1). Typically all the quantities represented in
lie between the active limit and the quasi-passive limit.
Figure 9 – active and passive ‘yield points’, the spring
Therefore a continuum model with Mohr-Coulomb
stiffness, kh, and the at-rest pressure – vary with depth.
material can be used, together with joint elements –
option (a) in section 3.1.3 above – which yield at the The yielding spring approach illustrated is advocated by
quasi-passive limit. The software used must be Frank et al [25]. It is also applied to integral bridge
capable of handling the variation of the material analysis [26] and, with suitable values of Ka and K*
properties (E’, K*) with depth in the Mohr-Coulomb (varying with depth) as the yield points, suitable
and joint materials. stiffness, kh (varying with depth) and initial stresses,
might provide an alternative to continuum analyses for
3.2 Use of Springs to represent Soil embedded wall integral abutments (type (i) in Figure
6). However it is in the analysis of abutments on piles
For a bank pad abutment on piles (type (iii) in Figure 6),
(types (ii) and (iii) of Figure 6) that the use of springs
lateral movement at the pile head does not infer plane
seems most appropriate.
lateral movement of the soil in the way that lateral
movement of a retaining wall does. There may be Where piles are installed in level ground, the at-rest
some arching of soil between piles but judgments are pressures are in equilibrium and so are not of interest
needed. This makes 2D plane strain models less for most analysis/design purposes. When lateral
appropriate for such bridges and the use of spring strains are expected to be small, it may be reasonable
models more attractive. to model piles using beam elements, supported by
linear elastic lateral springs. For cohesive soils, it is
SSI analyses where soil is represented using springs are
generally considered that spring stiffness may be
widely referred to as ‘Winkler spring’ or ‘subgrade
assumed constant with depth (Finnish Guidelines [18]
reaction’ models. The springs may be used to
section 4.3.5.1, Rombach [24]). For granular soils, a
represent the vertical or horizontal resistance of the
linear variation with depth may be used (the Finnish
soil; in the context of retaining structures and integral
guideline [18] suggests that a linear variation up to a
bridges it is the horizontal stiffness, characterised by a
depth of 10d and thereafter a constant value for kh). It
spring stiffness, kh (force/length³), which is of interest.
may be helpful to consider the variation to be a
Spring models are suggested in EN1997-1 [5] clause
polynomial of the form:
9.5.4.
k ୦ = A + Bz ச

7
In this, A, B and κ are empirical constants. Typical loads, and gives p-y curves for both static and cyclical
variations of kh with depth for different soil types are loading.
suggested by Rombach [24] and illustrated in Figure 10.
For bank pad abutments on piles, then, suitable lateral
κ κ = 0 for cohesive soil under earth pressures for the end screen may be determined
moderate loads by adopting a K* approach (PD6694-1 [13] clause
κ = 0.5 for medium cohesive soil and 9.4.6), while the pile-soil interaction can be handled
κ non-cohesive soil above water table using linear or nonlinear springs (or ‘joints’) to
represent the soil as illustrated in Figure 12 below.
κ = 1.0 for non-cohesive soil below
the groundwater level or under
Such approaches to pile-soil modelling have been used
κ greater loads for integral bridges – and validated using field
measurements and sub-models in the past – see
κ = 1.5-2.0 for loose non-cohesive
κ Jayaraman [30], Krizek and Studnička [31], and Albhaisi
soil under very high loads
[32].
Figure 10. Typical variations of modulus of subgrade
reaction with depth for piles
(after Rombach [24] Figure 2.39)
Values for kh are notoriously difficult to obtain, since
the spring stiffness is not a fundamental soil property.
However some guidance may be found in the Finnish
Guidelines [18] section 4.3.5.1 (including a correlation
between φ' and kh), while RP2A [27] section 6.8
describes methods for defining pressure-deflection (p-
y) curves for laterally loaded piles appropriate to
various soil materials. Comprehensive FE software is
capable of handling such curves within a nonlinear
joint material as in Figure 11 below and the matter is
covered in more detail by Reese & Van Impe [28]. Figure 12. 3D FE model of bridge deck integral with
rkn bank pads on piles
f N

i N+1 4. CONCLUSION
i+1
Typically the abutment walls or endscreens of integral
bridges are backfilled with granular material, where
the effects of soil ratcheting should be taken into
account. Where cohesive soil lies behind an embedded
σ wall, ratcheting can be neglected. Limiting earth
2 pressure and SSI analysis options have been explored
and suggested approaches are shown in Table 2 below.
1

1 2

rk1

Figure 11. Piecewise joint definition for nonlinear

joint material, from [21]
PD6694-1 [13] clause 6.4.9 indicates that SSI analysis is
required for bank pad abutments on piles (type (iii) in
Figure 6), and draws attention to a particular problem:
forward of the piles, the ground slopes away and so
the ground stiffness and limiting pressures would be
different on each side of the line of piles (see CIRIA
C580 [22], clause 7.2).
The remolding of soil over many thermal cycles should
also be taken into account as described by Wasserman
[29]. This is reflected in RP2A [27], which describes
cyclic loads as causing a deterioration of lateral
resistance as compared to that observed for static

8
 Table 2. Suggested Analysis Approach by Integral Abutment Type
Limiting
Integral Abutment Type SSI Notes and reference within PD6694-1 [13]
equilibrium
Full height wall on pad
Yes Granular backfill. Assumed earth pressure
footing
distribution from Fig 5 and K* from clause
Full height wall on piled 9.4.3 incorporates ratcheting.
Yes
footing
Granular backfill. Triangular earth pressure
Bank pad Yes distribution and K* from clause 9.4.4
incorporates ratcheting.
Soil modelled using continuum. For granular
soils or backfill, modify E’ with depth to suit
Embedded wall Yes Annex A and restrict pressures to K* from
clause 9.4.3. Alternatively nonlinear spring
model with similar considerations.
Soil modelled using nonlinear springs. For
Full height wall on single
Yes granular backfill, modify stiffnesses and
row of piles
limiting pressures to suit Annex A.
Soil modelled using nonlinear springs.
Reduced stiffness and limiting pressures for
Bank pad on single row front face of piles.
Yes
of piles Granular backfill to end screen. Triangular
earth pressure distribution and K* from clause
9.4.4 incorporates ratcheting.

7. Highways Agency. 2003. BA42/96 Amendment No

5. REFERENCES
1, The Design of Integral Bridges, The Stationery
1. AS5100.4 Bridge Design Part 4: Bearings & Deck Office, London, UK
Joints, Standards Australia International, Sydney,
8. AS5100.3 Bridge Design Part 3: Foundations and
NSW., 2004
soil-supporting structures, Standards Australia
2. Craig R.F. 1992. Soil Mechanics 5 Edition, International, Sydney, NSW., 2004
th

Chapman & Hall, London, UK

9. Connal, J. 2004. Integral Abutment Bridges -
3. Barker, R.M., Duncan, J.M., Rojiani, K.B., Ooi, Australian and US Practice. Austroads 5th Bridge
P.S.K., Tan, C.K. and Kim, S.G. 1991. NCHRP Conference, Austroads, Sydney, Australia.
Report 343: Manuals for the Design of Bridge
10. Kunin, J. and Alampalli. 1999. Special Report 132:
Foundations, Transport Research Board/ National
Integral Abutment Bridges: Current Practice in the
Research Council, Washington D.C., USA
United States and Canada, Transportation
4. AASHTO LRFD Bridge Design Specifications, Research and Development Bureau, New York
American Association of State Highway and State Department of Transportation, New York,
Transportation Officials, Washington D.C., 2012 NY, USA
5. CEN (Comité Européen de Normalisation). 2004. 11. Bridge Manual SP/M/022 3 Edition Amendment
rd

BS EN1997-1:2004 Eurocode 7: Geotechnical 1. Transport Agency, Wellington, NZ, 2014.

design — Part 1: General rules, British Standards
12. Highways Agency/ Arup, 2009. Integral Bridges –
Institute, London, UK.
Best Practice Review and New Research, Phase 2b
6. England, G.L. and Tsang, N.C.M. 2001. Technical - Review of Existing Data, Back-Analysis of
Paper No 2: Towards the Design of Soil Loading for Measured Performance and Recommendations
Integral Bridges, Concrete Bridge Development (Stages 1, 2 and 3)
Group, Camberley, Surrey UK
13. BSI. 2011. PD 6694-1:2011 Recommendations for

9
 the Design of Structures Subject to Traffic Loading American Petroleum Institute, Washington D.C.
to BS EN 1997-1:2004, British Standards Institute,
28. Reese, L.C. and Van Impe, W. 2000. Single Piles
London, UK
and Pile Groups Under Lateral Loading, Taylor &
14. Highways Agency. 1996. BA42/96, The Design of Francis, London, UK.
Integral Bridges, HMSO, London, UK
29. Wasserman, E.P. 2007. Integral Abutment Design
15. Denton, S., Riches, O., Christie, T. and Kidd, A. (Practices in the United States), First U.S.-Italy
2010. Developments in Integral Bridge Design, Seismic Bridge Workshop, Pavia, Italy.
Bridge Design to Eurocodes: UK Implementation,
30. Jayaraman, R., Merz, P.B. and McLellan Pte Ltd.
ICE Publishing, London, UK
2001. Integral Bridge Concept Applied to
16. Nicholson, B.A. 1998. Integral Abutments for Rehabilitate an Existing Bridge and Construct a
Prestressed Beam Bridges, Prestressed Concrete Dual-use Bridge, 26 Conference on Our World in
th

Association, Leicester, UK Concrete & Structures, Singapore.

17. O’Brien, E.J. and Keogh, D.L. 1999. Bridge Deck 31. Krizek, J. and Studnička, J. 2007. Integral Bridges
Analysis, E&FN Spon, London, UK and Soil–Structure Interaction, Modern Building
Materials, Structures and Techniques 9th
18. Liikenneviraston ohjeita (Transport Agency
International Conference, Vilnius, Lithuania.
Guidelines) 11/2012. Sillan geotekninen
suunnittelu - Sillat ja muut taitorakenteet, 32. Albhaisi, S., Nassif, H., and Hwang, E. 2012. Effect
Liikennevirasto, Helsinki 2012 of Substructure Stiffness on Performance of Steel
Integral Abutment Bridges Under Thermal Loads,
19. Muir Wood, D. 2004. Geotechnical Modelling,
Journal of the Transportation Research Board,
Spon Press, New York, NY, USA
Transportation Research Board of the National
20. Vermeer P.A. and De Borst R. 1984. Non- Academies, 2313: 22-32
Associated Plasticity for Soils, Concrete and Rock,
Heron, 29 (3): 3-64
21. LUSAS. 2013. Theory Manual Volume 1, LUSAS,
Kingston Upon Thames, Surrey, UK
22. Gaba, A.R., Simpson, B., Powrie, W. and Beadman,
D. R. 2003. CIRIA C580 Embedded Retaining
Walls – Guidance for Economic Design,
Construction Industry Research and Information
Association, London, UK
23. ACI. 2008. ACI 318-08, Building Code
Requirements for Structural Concrete, American
Concrete Institute, Farmington Hills, MI, USA
24. Rombach, G.A. 2011. Finite-element Design of
Concrete Structures: Practical Problems and their
Solutions (Second Edition), ICE Publishing, London,
2011
25. Frank, R., Bauduin, C., Driscoll, R., Kavvadas, M.,
Krebs Ovesen, N., Orr, T. and Schuppener, B.
2004. Designers' Guide to EN 1997-1, Eurocode 7:
Geotechnical design - General rules, Thomas
Telford Ltd, London, UK
26. Faraji, S., Ting, J.M., Crovo, D.S., and Ernst., H.
2001. Nonlinear Analysis of Integral Bridges: Finite
Element Model, ASCE Journal of Geotechnical and
Geoenvironmental Engineering, 127 (5): 454-461
27. API Recommended Practice RP2A-WSD.
Recommended Practice for Planning, Designing
and Constructing Fixed Offshore Platforms –
Working Stress Design, 21 Edition, Dec 2000.
st

10