TRANSPORTATION RESEARCH RECORD 1290 113

Modeling Bridge Foundations for Seismic

Design and Retrofitting
IGNATIUS Po LAM,1 GEOFFREY R. MARTIN, 2 AND RoY IMBSEN 3

Past experience and recent research motion (magnitude and phase shift) could
indicates that proper modeling of abutments result in differences in force excitation
and foundations is very important in the levels and differential movements at
evaluation of dynamic response of an support points.
overall bridge structure. This paper
presents simplified procedures with An excellent literature survey (Iwasaki
accompanying design charts for the et al., 1973) chronicles earthquake damage
development of stiffness coefficients for to bridges up until 1971. This report
abutments, piles and spread footing founda- concluded that foundation behavior played a
tions for highway bridges. The presented major role on performance of highway
procedures have been calibrated to design bridges during past earthquakes. Extracts
practice adopted by bridge engineers. from the conclusion sect i on of the report
Several examples are presented in the paper are quoted below to illustrate the
to highlight various sensitivity issues in significance of foundation systems in
abutment and foundation design. seismic performance of bridges:
IllTRODOC'l'IOH o "Seismic damage, particularly to low
bridges, are moat commonly caused by
Highway bridges are known to be highly foundation failures resulting from
susceptible to damage under earthquake excessive ground deformation and/or loss
loading. One major reason for the poor of stability and bearing capacity of the
performance relates to the complexities of foundation soils •••••• "
the bridge structural and substructural
systems as compared to other structures. o "Backfills exert large forces on
Some of these complexities are listed abutments which can at certain times be
below: in-phase with the seismic inertia forces
developed in the superstructures. These
o There are wide- variations in structural forces in combination may cause severe
types and configurations (bridge decks, failures, often of a brittle nature, in
bents and abutments). Furthermore, the the substructures."
need for expansion joints to accommodate
temperature effects increases the o "To minimize damage, bridge structures
potential for collapse and introduces should be designed with proper
significant complexities in design. recognition of the stability and bearing
capacities of foundation soils, force-
o Variations in soil conditions are common deformation and energy absorption
along the length of a highway bridge characteristics of substructure,
which could lead to different types of superstructure and connecting elements,
abutments and foundation systems. The the dynamic nature of structural
wide range of combinations of structures, response, and the dynamic characteristics
abutments, foundations and connections of all forces acting on the complete
introduces many unique design problems soil-structure system."
for bridge engineers.
Poor soil conditions (soft soil and a
o Bridges are supported on multiple support high water table) contributed to moat of
points that are spaced relatively wide the structural and substructural damage
apart. As a result, variations in ground during many of the past earthquakes, often
of a severe catastrophic nature. Sites
1
Principal, Earth Mechanics, Inc., that have sustained heavy damage during
Fountain Valley, CA. past earthquakes due to ground motion
2
Professor, Dept. of Civil Engineering, amplification, slope failure or
Univ. of Southern California, L.A., CA. liquefaction would be likely candidates for
J
President, Imbsen and Associates, Inc., damage in future earthquakes. Therefore,
Sacramento, CA. existing earth-science information
114 TRANSPORTATION RESEARCH RECORD 1290

(seismologic, geologic, geotechnical, Due to the wide range of combinations in

hydrological) should be used in regional connection details (bridge-abutment and
earthquake preparedness and planning abutment-foundation) , many of the above
programs as well as in site specific issues· remain to be resolved and currently,
designs. a wide~ variation in design practice of
abutments prevails.
Geotechnical considerations for highway
bridges can be divided into two categories: Some guidance is currently provided by
the Caltrans Bridge Design Aids Section 14-
o Ground Stability. Site response, lique- "Dynamic Model Assumptions and Adjustments"
faction potential, embankment slope sta- (Caltrans) and the AASHTO · Guide
bility and assessments of the magnitude Specifications for Seismic Design of
of ground deformations (including cyclic Highway Bridges" ( AASHTO, 1983) • Both
and permanent deformations of the free- documents recognize the highly nonlinear
f ield site soil) and the evaluation of behavior in abutments due to failure of the
the nature and the magnitude of founda- backfills and from structural nonlinearity
tion movements and their implications on at expansion joints. The load-displacement
structural loading, are critical elements characteristics for a typical monolithic
in foundation design of highway bridges. and a seat-type abutment are shown in
Figure 1.
o Substructures and Abutment Models.
Evaluation of the fundamental periods and An iterative design procedure as shown in
mode shapes of the bridge structure is Figure 2 and described in the following
needed for the determination of the steps is needed to account for the
seismic force level for design. For some nonlinear behavior of abutment systems in a
bridges, especially the shorter and low linear dynamic response analysis.
bridges, the effects of foundation
stiffness could significantly affect the (1) Assume an Initial Abutment Stiffness.
overall dynamic response characteristics. Thie stiffness should be compatible
Therefore, evaluation of foundation with the configuration of the
stiffness is an important part of the structure and connection deti:lile and
overall bridge structure model. the assumed peak displacement of the
bridge deck.
This paper focuses on modelling
procedures for estimating stiffnesses of For longitudinal loading, it should be
abutments, spread footings and pile recognized that in the course of an
footings (pile groups) for dynamic response earthquake, the stiffness of only one
analysis of an overall bridge. abutment would be mobilized (i.e.,
soil resistance is mobilized when the
ABDTKENTS structure is moving toward the soil,
whereas, no soil resistance is
For many highway bridges, the abutments mobilized when the abutment moves away
attract a large portion of the seismic from the soil). Therefore, the
force, particularly in the longitudinal stiffness would be too high if the
direction. Therefore, the stiffness of the full soil stiffness were used at both
backfill at the abutments must be abutments. For most non-curv1::1tl
considered. However, further research is bridges, as an approximation, one-half
needed to improve our understanding and to of the total stiffness should be
develop improved design procedures related allocated to each abutment at both
to the following aspects: ends. When the "half-half" stiffness
approach is used, the resulting
o Stiffness of abutment systems in both the abutment forces should be doubled for
longitudinal and transverse loading design. For curved bridges, it may be
directions. necessary to assign the full abutment
stiffness at one end of the bridge
o The magnitude and distribution of soil while assigning a zero stiffness on
pressure on end and wing walls. the other end. Two dynamic response
analyses are needed (each run using a
o The relative significance in the induced full stiffness at each abutment) for
soil pressure from inertia forces of the this "full-zero" stiffness approach.
bridge deck versus the inertia forces of
the soil mass acting on the abutment The transverse stiffness should
walls and the appropriate procedures for reflect the potential reduction in
selection of design soil pressure to stiffness arising from the
account for the interplay of the two deformability of wing walls (relative
loading mechanisms. to the bridge) and the partial contact
Lam et al. 115

and sloping configuration of the backfills The above steps outlined the various
at the exterior surfaces of wing walls. design considerations that should be
addressed in abutments. Development of the
(2) Conduct Dynamic Response Analysis. load-displacement characteristics of the
Using the above abutment stiffness, abutment-backfill system forms the basic
conduct a dynamic response analysis of requirement in abutment modeling. It
the overall bridge to determine the depends on three parameters:
forces and displacement•.
(1) The initial abutment-backfill
(3) Check Abutment Force Capacity. Using interaction stiffness prior to soil
peak abutment force and the effective failure.
area of the abutment wall, solve for
the peak soil pressure and check that (2) The ultimate resistance of the load-
the soil capacity has not been displacement characteristics from
exceeded. The soil capacity should be backfill soil failure consideration.
based on the properties of the
backfill. Caltrans have reconunended a (3) The magnitude of the gap at expansion
soil capacity of 7.7 ksf (370 kPa) for joints for seat-type abutments.
typical California abutment-backfill
conditions (sandy soil with shear wave Discussions of the latter two parameters
velocity of about 800 fps or 240 meter (baaed largely on Caltrans recommendations)
/sec). If the peak soil pressure have been presented earlier. Caltrans
exceeds the soil capacity, the Bridge Design Aids Section 14 recommended
analysis should be repeated with a an &Putment-backfill interaction stiffness
reduced abutment stiffness to reflect coefficient of 200 kips per inch of
plastic yielding of the backfill soil. deflection per linear foot (115 MPa) along
Iterations should be conducted until the length of the abutment wall as a
the force levels are below the starting point for iterative analysis.
acceptable capacity of the abutment, This abutment-soil stiffness coefficient
prior to proceeding to the next step. would be appropriate for typical California
abu.t ment backfill conditions (material with
(4) Check Abutment Displacement. Compare shear wave velocity of about 800 ft/sec
the computed displacements against the (240 m/s) and approximately 8 feet (2.4 m)
value assumed in Step 1 in relation to of effective height of abutment walls).
the load-displacement characteristics
(See Figure 1) of the abutment system For abutment configurations and soil
for the configuration of the expansion conditions that differ significantly from
joints and the soil capacity value. the above condition, a more general form of
Thie cross check is needed to ensure abutment wall-backfill stiffness equation,
that the assumed abutment stiffness which considers the passive resistance of
reflects the load-displacement the soil, as recommended by Wilson (1988)
characteristics properly. If the could be used to develop the longitudinal
error in the assumed stiffness is stiffness of the end wall and the
excessive, the analysis should be transverse stiffness of the wing wall. The
repeated with a revised stiffness. equation is given by:
The converged displacement value of
the abutment (with respect to the K8 = E • • • • • • • • • • • ( 1)
nonlinear load-displacement charac- ( 1-u 2 ) I
teristic) should then be evaluated
against the acceptable level of dis- where K8 is soil stiffness per unit
placement. Excessive deformations at deflection per unit wall width; E11 is the
the abutment may cause problems. Young's modulus of the backfill soil; u is
Field inspections after the 1971 San the Poisson ratio of the backfill soil; and
Fernando earthquake suggest that abut- I is the shape factor as shown in Figure 3.
ments which moved up to 0.2 feet (6
cm) in the longitudinal direction into The above equation allows for input of
the backfill soil appeared to survive site specific soil parameters and abutment
with little need for repair. There- wall configurations. As the length to
fore, if possible, this limit should height ratios for wing walls are somewhat
be maintained. Excessive deformations smaller than end walls, the above equation
may create stability and integrity suggests a lower shape factor I, or a
problems both at the abutment and at higher soil stiffness coefficient (K 8 ) for
the bents. Deformations greater than wing walls as compared to end walls.
0.2 feet (6 cm) at abutments should be
evaluated for these effects.
 0
g._........_................_"'T"".._..............................
....
0

-....I.
Ill.

EQUIVALENT LINEAR
STIFFNESS '-v
, ,
,,
~ ~, WAJ.L...son. ~
, • .. S'l'IFf1(ESS
o-1-------,.:.;------+-..;.;;;....;,,;;;..t------1
-1 1 2 l
Defl•CtiOD UllClluJ Deflection Ullda•.• I
llDnolithlc Abut..nt Seat-Type Abutment
Figure 1. Load-Displacement Characteristics of Abutments

a
MODEL THI alUOGI
WIYM INITIAL AaUTMENT
=-• L • Dimurwian, Leng SJ.do of QJl"lt.oct: Anw
a-~. SMXt SW. ot Canr..c:t AE-.

....
Dl51GN • ITIPINI•
COifflCllWTI

0
,,
~"'
•
ti) i,... ...
cc• ~
0 v
vu NO
u
t-
a:
u.
0
N
• i.. ~
v
~
~
LLJ ~
Q..
a: c I/
- /
RlOUtf A-IUTM(Nf
ITIFFNESS TO
ACCOUNT FOR
NO :c
(/)
• v
jNCfll.A.5-1:0 U:VEL Of.
YIELDING

Ks • E~
0 (t-~2 ) I

~.
. • . 5 ... 1f

lVALUATI
OVlftillLL
Dil!WI 101
LOG CL/Bl
Figure 2. Iterative Procedure for the
Determination of Abutment-Soil Figure 3. Shape Factor for abutment
Interaction (After AASHTO, 1983) Stiffness in Equation 1
Lam er al. 117

Comparison to Caltrane. The soil component Stiffness Coefficient

stiffness (K 8 ) from Wilson's equation can
be compared to Caltrans reco.m mendation Vert. Translation a p 4 GR I (l-u) •• (2a)
through an example calculation for the
longitudinal stiffness of a typical Hor. Translation a P8 GR I (2-u) •• (2b)
California abutment end wall-backfill
system: Tore. Rotation a p 16 G R3 I 3 • • ( 2c)

o Configuration of end wall: 50-ft (15.2 m) Rocking Rotation a P 8 G R3 I 3 (1-u) (2d)

wide by 8-ft ( 2. 4 m) high. For this
configuration, a shape factor, I of 1. 84 where G and u are shear modulus and poieson
is obtained from Figure 3. ratio for an elastic half-space material;
R is the equivalent radius as shown in
o Shear wave velocity of backfill: (800 Figure 4. a and ~ are the embedment and
ft/sec or 240 m/sec). shape correction factors, respectively (See
Figure 5).
o Using a Poisson ratio of O. 3, a Young's
modulus corresponding to a shear wave It should be noted that the design chart
velocity of 800 ft/sec would be about 6.2 was developed for a special case of zero
x 10 6 psf (300 MPa). However, a ground cover thickness (soil thickness
reduction factor is normally needed to above the top of the slab). We recommend
adjust for a soil modulus based on shear that, for conservatism, the thickness of
wave velocity measurement to account for the sla.b be used as the embedment depth (D
nonlinear soil behavior at high~r strain in Figure 4) rather t han the full depth
levels. Some typical soil modulus from ground surface to footing base. Thie
variations with shearing strain curves approximation is needed to avoid the need
have been recommended by Lam and Martin of an extended set of design charts to
(1986). At a typical average shear accommodate the wide combinations of ground
strain value of about 0. 01 percent, a cover and slab thicknesses.
reduction factor of O. 7 would be
reasonable. The correspon-ding reduced
Young's modulus is estimated at about
4.34 x 10 6 psf (210 MPa).
o The soil stiffness coefficient K 8 for the RECTANGULAR
'OOTINQ
above end wall is estimated at 216 kip/in
of deflection/lineal foot of end wall
(124 Mpa) as compared to the 200 kip/in
/ft (115 MPa) as recommended by Caltrans. 2L

It can be concluded that the Wilson's

equation compares favorably with Caltrans
recommendations and provides a rational
basis for extrapolation to non California
design conditions (different soil types). EQJIVALENT
ClllCULAll
'00TING
SPREAD FOOTINGS

The current state-of-practice in analyses

of footings involves the use of stiffness
equations of a rigid footing on a semi-
infinite elastic half apace. Typically,
footings used for highway bridges are IQIJIVALINT llAOIUI:
rectangular in shape and are embedded
beneath a layer of ground cover soil. A TllAHll.ATIONAL; 11 • J~
procedure has been developed for solutions
of stiffnesses of embedded rectangular
footings. The procedure involves first
ROTATIONAL: 11 • [ IU~~LiJ r. . .... . . . . . . . . . ..
C.-AlCI& ROCKING!

solving for the radius of an equivalent R • [ l2l:fQLIJ~............................... ly-IUUS ROCKING!

circular footing as shown in Figure 4. The
cross-coupling ef fecte between moment and ..L l•-2 • •L2t ]"
II • [ tor ....................... 11-AlCIS TORSION!
shear for moat shallow footings are small
and can be neglected. The diagonal
stiffness terms in the stiffness matrix of
an embedded rectangular footing can be Figure 4. Procedures for Equivalent Radius
obtained from equations 2. of a Rectangular Footing
118 TRANSPORTATION RESEARCH RECORD 1290

SEE flGURE 4 fOft DEflNITIOHS Of L, B, R, AHD D PI.LB FOUNDATIONS

Pile footings (pile group systems) are
the most commonly used foundation systems
for support of bridge structures. In a
dynamic response analysis of an overall
bridge structure, a pile foundation can be
incorporated in the bridge model by several
methods, including: (l) equivalent
cantilever models, ( 2) uncoupled base
springs models and ( 3) coupled fouqdation
stiffness matrix models.

1.00L-------!'-------+-----~
Among the three methods, the equivalent
1 1 coupled foundation stiffness matrix model
L/B
is the most general method of
representation of foundation stiffness in a
dynamic response analysis of the overall
bridge. In fact, it can rigorously
represent both the equivalent cantilever
and the uncoupled base spring models.
Because of its generality, it can represent
all types of foundation systems including
abutments, spread footings, pile groups and
drilled shafts. The main draw back relates
to the added effort to develop the
coefficients in the stiffness matrix.
Solutions of stiffness coefficients of a
o. .. o.s full 6 X 6 stiffness matrix as shown in
O.l 0.2 o.s Figure 6a are needed for this method. A
DIR simplified procedure has been developed for
the solution of the stiffness matrix
coefficients of pile groups.
Figure S. Design Charts for Footings

4ii'!lt-H-lx 11'
e.
e '9 G
e .··e··
I
..., • G
11'
1' Diameter

e '9 G
111 81

r '
1K111 0 0 0 0
I I
I0 I K22 0 0 0
, • 120pcf
I I •
Io I o 0 0 0 •• - 30
i ~
I0 I K42 0 0 0 10' 1'Diametar
I I pipe pile
~K11I o 0 0 0 0.25"wall
I I thickness
filled with
\.oj o o o 0
concrete
~FORCE VECTOR FOR a.• 1.0
(a) DEFINITIONS OF COORDINATE SYSTEMS (b) EXAMPLE OF A PILE FOOTING
Figure 6. Form of a Pile Head Stiffness Matrix
Lam et al. 119

,....
The five basic steps involved in this 0
simplified procedure for a pile footing Cul!ff. of Vari&Jtion uf Suil Rl!action -- .
such as that shown in Figure 6b are: Modulus with Depth, f (LB/IN')

( m200.o,
( 1) Solve for the stiffness matrix of a r = 1so.~, ---
I/'"'. ::::::~
single pile under lateral loading. r ~ IOO.«( ~ ::::: ,_
- ( = Hit.I . "" ..
- _, ":.- :: -
,_- --v
r = t\o.q ~ 2 ;--:

(2) Solve for the stiffness matrix of a r = 40.o '

,_ -
r = 20.u" '-' c....-
single pile under axial loading. .....
( ~ 10.1~
"':>.;
~ ,..
P<":: i.-::::-
......... t:::---
~
......... ...... - -
!..-' .....

....
- ----
(3) Superimpose the stiffness of ~ .............
- -
individual piles to obtain the pile
group stiffness. - """
,_
- ~
-

.....,
'~
/

.....-
( 4) Determine the stiffness contribution
of the pile cap (pile footing).
~

..... """'
v v
~
.....~
..... ~ '
..... :
-,1
'\ ~ 5.0 .
(• 1.0
(5) Superimpose the stiffnesses of the v ~

,... "' r. 0.5

/ f ~ O.J
pile cap to the pile group.
vv
Details of the 5-Step Procedures are 1010 1011 1012
described below.
BENDING STIFFNESS, El ( LB-IN1 )
Lateral Load -Def lection Methods.
Currently, in practice, the lateral pile-
soil interaction problem is obtained by
solving the problem of a beam member
(modeled by finite elements, difference
equations, or by discretized mechanical
analog) supported on closely spaced linear
or nonlinear elastic soil springs. Due to
the dominance of the elastic pile stiffness
over the nonlinear soil and the localized
zone of influence (confined to the upper
five to ten diameters), linear solutions
were found to be adequate for pile stiff-
ness evaluations (Lam and Martin, 1986).
Most available linear non-dimensional Figure 7. Pile Translational Stiffness, K6
solutions have been geared
development of the total pile solution
toward
-
....
0
(including distributions of deflection,
slope, moment and shear along the entire E~~Fmlmlmml
of Codf. Variation of Soil Reaction ._
Modulus wi1h o~pth, f (Ul/lN )
1
-... ~
pile length). For the purpose of founda-
tion stiffness evaluation, the total pile
solutions can be simplified to provide
coefficients of pile-head stiffness matrix
as shown in Figures 7 through 9. Pile-head
stiffness coefficients can be obtained for
a combination of bending stiffness of the
pile (EI) and the coefficient of variation
of soil reaction modulus E8 with depth (f).
Recommendations for the coefficient f for
sand are available in the literature
(Terzaghi, 1955; and O'Neill and Murchison,
1983). At normal working load levels
(pile-head deflection between 0.5 to 1.0
inch or 1.3 to 2.5 cm), Terzaghi's
recommendation is considered reasonable for
sand and his recommendation is presented in ,.... v
oJ___J._j...~~1----1--H-H+l#--+-+-H+tttt~t-t'"+lcttttt
Figure 10. Recommendations for the
coefficient f for clays have been developed -109 1010 1011 1012 1013
from correlations of nonlinear computer BENDING STIFFNESS, El ( LB-IN~ )
solutions at a pile-head deflection (fixed
head condition) of about 1-inch (2.5 cm) by Figure 8. Pile Rotational Stiffness, Ke
120 TRA NSPORTATION RESEA RCH RECORD 1290

ar :IO"
•lllCTIClll NIGU ...
.. .,. ...
en
Cl
Coeff. of Variution of Soil Rc11c1ian
Modulu• wi1h Dcp1h. r (LBflN')
-8-r--~oos_EA_~_.___
LOOSi
_____,_____ __.,........__o_EN_w~.....-~_&_"~--f
Mi_D_•Ulll

Recorrmendeo by Terzdgh1, 19SS

-!i
(After O'Ne111 and tllrch1son, 1983)
I• 200.0
f" ISO.O
,,...
r .. 100.0
m~r-.
~
~

~ ~t:>

l~'
60.11
;c f .. 40."
,.. f• 21J.U
10.0
......'
_.. -- , _
,.. f• 5." ,, ,_..
_..... "'
./
.....
~

,,..
0
I.II_
·'-' ' ~-- co
~~t-. ~ ~~
~
~ t:::"' ~ ....... ~
.....0
Cl)
E,, (LB/INJ)

/'. 'V
~

,,, - -
_..
i.,.
IE
Cl) 0
'::#'
/,/:_-:>. "/ ~

v - 8
~ r/L/
/I><
_.. I /[ /

,,,, ~
.....,....

- I/"'
I/)
0
o-1-----~------+------+------1------1
109 1012
0 20 60 80 JOO
BENDING STIFFNESS, El ( LB-IN1 )
Relative Density, Dr (Percent)

Figure 9. Pile Cross-Coupling Stiffness, K69 Figure 10. Recommendations for

coefficient f for Sands
(Note: 1 LB/IN 3 = 0.27 N/cm 3 )
the authors. This recommendation and
results of the correlation for clay are
shown in Figure 11. Only the upper five
diameters of soils (eoil type and ground
water) need to be considered in
8---------.-------.---------------.....------.
usage of the presented design charts. ..- Based on correlation of non11nedr p1le
':2:; solutions using Matlock's soft clay
Limitations of Approach. There are ~ ~ +-...,c~r1•tmer.-i_a_(_1_9_7o_>_ _ _ _-t---.----;Jll'G)
several simplifying assumptions in the
presented approach. The coefficient f ie
not an intrinsic soil parameter. The ~ ~
·~::f 0 ....q,...........</;
recommendations for f presented in Figures ...
Cl) co { ~' <>§;'!) ~-----ii
~:\,<::>
10 and 11 are appropriate for piles in 0

typical highway bridge foundations (i.e.

·o
Uj
., (LBnN'>
~
....
4v;''ti:'
~ ....~
smaller piles). Furthermore, the embedment
effect has not been taken into account in
the procedure. Therefore the recommenda-
8 g-1------...----,--_..,.."--~----1----
tions are conservative and appropriate for
shallow e.m bedment conditions (say less than Purameten in Correlulion Solulio111:
5 feet or 1.5 m).
(I) 12-inch Concrete Pile
(2) Fixed Pile-Hcud Condilion
Although correlations for the coeff'icient (3) Mutloclr.'s Soft Clay Criterion
f can be conducted for other conditions (4) I-inch Pile Heud Dcnection
(e.g. larger piles and bigger embedment
depths), the additional complexity negates
the merits of the use of simplified linear
o. 1. 2. 3. 5.
elastic solutions. For such cases, com- Cohesion (ksf)
puter solutions, which can readily accomo-
date nonlinear effects and more general
boundary conditions, are reconunended.
Comparison to Caltra.ns Practice. The Figure 11. Recommendations of
above procedure can be compared to the Coefficient f for Clays
practice adopted by Caltrans. In Caltrans (Note: 1 LB/IN 3 = 0.27 N/cm 3 )
Lam et al. 121

Bridge Design Aids Section 14, a pile head Role of Axial Stiffness. The role of
stiffness of 40 kips per inch deflection lateral loading on piles is usually
( 70 kN/cm) is recommended for a standard emphasized for earthquake consideration.
16-inch ( 40-cm) CIDH (Cast-in-Drilled-Hole) However, in a pile group system, the base
pile. The bending stiffness (EI) of the moment from structural loading is reacted
above pile ie estimated at 9.7 x 10 9 in 2 -lb largely by the moment couple from variation
(2.8 x 10 11 N-cm 2 ). For a typical sandy in axial load among individual piles as
soil condition (friction angle c 30°), the compared to reaction from pile-head
coefficient f, obtained from Figure 10, is bending. Experience from recent earth-
about 10 pci (2.7 N/cm 3 ). For the above quakes and research have provided ample
combination of bending stiffness and evidence (Roeenblueth, 1986; and Douglas et
subgrade stiffness, a lateral pile-head al. 1984) that the rotational stiffness of
stiffness coefficient (Ka) of 42 kips per a pile group, which is related to the axial
inch (73.5 kN/cm) of deflection is obtained pile stiffness, will have a dominating
from Figure 7. The 42 kips per inch effect on the overall structure as compared
compares favorably with the Caltrana to the lateral stiffness. Therefore,
recommendation of 40 kips per inch ( 70 evaluation of axial pile etif fnees is
kN/cm). critical for realistic modeling of
foundation stiffness.
Sensitivity of Boundary Condition. The
above stiffness coefficient, K6 , represents For a generalized soil-pile condition,
the lateral stiffness of a fixed-head pile the following factors need to be accounted
(zero rotation). The rotational and the for in evaluation of the overall axial pile
cross-coupling stiffness terms for the behavior and pile head load-displacement
above (16-inch or 40.6-cm CIDH) pile can be characteristics:
obtained from Figures 8 and 9, respectively
and sUJ1UJ1arized in the following paragraph: o The stiffness characteristics of the pile
(AE) and pile length (L).
o Rotational stiffness (Ke) (From Figure 8)
= 2.3 x 10 8 in-lb/rad (26 MN-m/rad) o The shear transfer-displacement charac-
teristic of the soil along the length of
o Croes-coupling stiffness (K68 ) (From the pile shaft (related to the cumulative
Figure 9)= 2.3 x 10 6 lb (10 MN) ultimate akin- friction capacity).
The above coefficients can be used to o The end-bearing load-displacement
solve for the load/moment-deflection/rota- characteristic of the soil at the pile
tion relationship for any combination of tip (related to the ultimate end-bearing
shear and moment or pile-head constraint capacity).
condition. For example, if the pile head
condition is a pinned head connection In a normal design condition, a pile foun-
detail, the zero pile-head moment leads to dation derives a significant portion of the
the following relationship between pile soil reaction throughout the pile shaft as
head deflection (a) and rotation (6): well as the pile tip. Unless the pile is
very lightly loaded, plastic slippage at
6 = - ( Kae I Ke ) a (3) the pile-soil interface will occur along a
significant upper portion of the pile.
The above relationship can be substituted Furthermore, the nonhomogeneous, layering
into the pile head force equation in nature of soil deposits must be accounted
Figures 7 through 9 for the lateral for in axial stiffness evaluation. Due to
stiffness (ratio of force to pile-head these complexities, linear analytical
deflection) of a free-head pile as procedures would be of limited practical
presented below: applications and nonlinear analyses are
preferred for axial pile response.
Lateral stiffness of free head pile
= Ka - Ka 8 2 I K8 = 19 kips/in (33.3 kN/cm) Solutions for Axial Stiffness.
Uncertainty in axial soil-pile interaction
analysis relates largely to uncertainties
From the example, it can be observed that in soil parameters including the ultimate
the lateral stiffness could vary from 42 to pile capacity (akin-friction and end-
19 kips per inch (70 to 33.3 kN/cm) for a bearing) and load-displacement relation-
fixed versus a free pile-head condition. ships. Computer solutions can be used for
It can be concluded that a realistic a rigorous nonlinear solution. An
representation of the pile-head connection approximate nonlinear graphical solution
is very important, and often of more method has been developed and presented by
significance than the selection of soil Lam and Martin (1984, 1986). It will be
parameters. described below. The procedure is
122 TRANSPORTATION RESEARCH RECORD 1290

schematically shown in Figure 12 and

involves the following steps:
0
(1) Soil Load-Displacement Relationships. 0
(f) 'ltJTAL PIU: CNl&:ri"t, Q u • 278 k
Side-friction and end-bearing load-
displacement curves are constructed
for a given pile capacity scenario
(accumulated skin-friction
ultimate tip resistance).
and
Various
re
forms of curve shape recommended by _... a 202 k
researchers can be used to develop the (I) 0
above load-displacement curves. o.o
or1 N
Vij ayvergiya • s recommendation ( 1977) ~
was adopted (for simplicity) in the
example shown in Figure 12. F • Fmax<2 ~zlzc - zlzc>

(2) Rigid Pile Solution. Using the above

0
a: ~
D
...J
... zc• 0.2 1n.

load-displacement curves, the rigid _J

pile solution can be developed by
summation of the side-friction and the
end-bearing resistance values at each
displacement along the load-
- ...8
cc
><
a:
TIP RESISTANCE CURVE:
displacement curves. 0
II> Q• °'"4x<zlzc>1/3
(3) Flexible Pile Solution. From the CYCLIC LOAD (70kl z,• 0.6 1n.
rigid pile solution, the flexible pile
SECANT l«JOULUS
solution can be developed by adding an ~..........:
· ~1~·~20~0~k~/~1~n.:..,.~~~+-~~--t~~~~
additional component of displacement C-1
at each load level (Q) to reflect the 0.00 0.25 0.50 o. 75 1.00 J.25
pile compliance. For the most DISPLACEMENT Cin.l
flexible pile scenario, corresponding
to a uniform thrust distribution along
the pile shaft, the pile compliance is
given by Equation-4. Figure 12. Graphical Solution of Axial
Pile Stiffness
Pile Compliance (6 0 ) = 2...b • • • • (4)
A E level (without the static bias load) would
be appropriate for dynamic response
where L is the pile length; A is the analyses.
cross-sectional area, and E is the
Young's modulus of pile. Stiffness Mat rix of a Singl e Pile. Using
the above described procedures, results of
the pile-head stiffness coefficients of a
(4) Intermediate Pile Stiffness Solution. single typical pile for the pile footing
The "correct" solution, as indicated shown in Figure 6b are tabulated in Table
by the computer solution, is bounded 1. It should be noted that the torsional
by the above rigid pile and flexible stiffness of a single individual pile can
pile solutions. In most cases, a good usually be ignored and assumed to be zero.
approximation can be developed by The torsional moment on a pile group is
averaging the load-displacement curves usually reacted by the torsional moment
for the rigid and the flexible pile couple from variation in lateral shears
solutions. The above graph i cal method among individual piles in a pile group.
can be used to solve for the load-
d is pl ac eme n t curve for any Stiffness for a.Pile Group. The single
combinations of pile/soil situations pile stiffness matrix in Table l can be
(end-bearing and friction piles as used in the next step to develop the pile
well as any pile type: long or short group stiffness matrix. For a vertical
and any pile material). pile group such as that shown in Figure 6b,
the form of the stiffness matrix will be
( 5) Selection of Secant Stiffness. The identical to the individual pile (as shown
secant axial pile stiffness appro- in Figure 6a). Also the stiffness
priate for use in dynamic response summation procedure is relatively straight
analyses should reflect the unloading forward. For battered-pile systems,
and reloading behavior. The secant computer solutions are recommended. A
stiffness from the origin of the load- PILECAP computer program that can be used
displacement curve to the cyclic load to conduct the summation of individual
Lam et al. 123

pile-head stiffness for an overall pile support characterist ics. There exists a
group stiffness matrix has been documented wide range of opin ions among geotechnical
in the FHWA report by Lam and Martin engineers on the significance of the "group
(1986). The program can also be used to effects". The importance of "group
distribute the overall foundation load to effect~" would depend on many factors
individual piles. including the configuration of the pile
group (number of piles, spacing, direction
For a vertical pile group, the stiffness of loading in relation to the group
for the translational displacement terms configuration), soil types and pile
(the two horizontal and the vertical installation methods. In view of the lack
displacement terms) and the cross-coupling of evidence that "group effects"
terms can be obtained by merely multiplying contributed to failure of bridges and the
the corresponding stiffness components of lack of well proven approaches for
an individual pile by the number of piles. treatment of "group effects"; and above all
However, the rotational stiffness terms for the sake of simplicity, we have
(the two rocking and the torsional neglected "group effects" in our presented
rotations) require consideration of an procedures.
additional stiffness component. In
addition to individual pile-head bending Stiffness of Pile Cap. So far, the
momenta at each pile head, a unit rotation presented procedure deals with stiffness of
at the pile cap will introduce transla- the pile-soil system. In a typical highway
tional displacements and corresponding bridge situation, the:ce will be additional
forces at each pile head (e.g. vertical stiffnesses arising from the pile cap. Our
forces for rocking rotation and lateral experience indicates that the lateral
pile forces for torsional rotation). These stiffness from ( 1) passive res istance on
pile-head forces will work together among the vertical surface/a and (2) tractional
the piles and will result in an additional shear forces at the base of a pile cap
moment reaction on the overall pile group. could be very significant as compared to
The following equation can be used to the lateral pile stiffness. Wilson's
develop the rotational stiffness terms of a equation for abutment stiffness presented
pile group: earlier can be used for evaluation of the
passive resistance component of cap
stiffnesses. The spread footing procedure
N for an unembedded (surface) footing can be
R9 = N RP + I: K an Sn2 • • • • • • • • • ( 5) used for evaluation of the cap stiffness
n=l for soil reactions at the cap base. The
pile footing example problem shown in
Figure 6b is used to illustrate the
where R9 and RP are the Rotational Stiffness relative stiffness of the pile group versus
of the pile group and an individual pile, the pile cap. The stiffness contribution
respectively. N is the No. of piles in the from piles have been developed earlier (See
pile group. K an is the appropriate Table 1) . The pile-cap stiffnesses
translational stiffness coefficient of an including: (1) the lateral passive
individual pile, and Sn is the spacing resistance component developed from
between the nth pile and the point of Wilson's method (Equation l) and (2) the 6
loading (center of the pile group). degrees-of-freedom soil reaction at the
base of the cap (from spread footing
The subscript n denotes the pile no. Equations 2) are presented in the following
Summation is conducted for all the piles in table for comparison. A shear modulus of
the pile group in the above equation. 7.2 x 10 5 psf or 34.5 Mpa (a conservative,
or low value for compacted backfills in
Using the described procedure, the pile- most construction practice) and Poisson
group stiffnesses of the overall pile group ratio of 0. 3 5 were used in the pile-cap
system shown in Figure 6 is developed and stiffness evaluation presented in Table 2.
presented in Table l. It can be observed
that the rocking rotational stiffness From Table 2, it can be concluded that,
coefficients of the pile group are the vertical pile stiffness dominates the
dominated by translational stiffnesses of response of the vertical translational and
piles. The rotational stiffness from pile the rocking rotational stiffnesses of the
bending can virtually be ignored for most overall foundation. The influence of the
bent foundations. pile cap is relatively minor for these two
modes. However, the lateral footing
The above presented procedure for a pile stiffness is quite high relative to the
group does not account for the "group lateral pile stiffness and dominates the
effects" which relate to the influence of lateral and the torsional rotation
the adjacent piles in affecting the soil stiffness terms.
 Table 1. Pile Stiffness Solution
Single
Stiffness Coefficient Pile Pile Group
Lateral Translation 42 9 x 42 - 378
k 11 .. k 22 , (kip/in)
Vertical Translation l,2aa 9 x l,2aa = la,0aa
kJJ I (kip/in)
N
N ~+
2
Rocking Rotation 193,aoa l: Kan Sn (Eq. 5)

k 44 =k55 (in-kip/rad) = 1.74 x la + 1.66 x la 7

6

= 1.83 x 107

Torsional Rotation a 4x42x48 2 + 4x42x(4B 2+4B 2)

k 6 5, (in-kip/rad) = 1.16 x la 6
Crose-Coupling -2,25a 9 x -2,25a = -2a,250
k15=k51 =-k24=-k42 I (kip)

See Figure 6 for definition of stiffness coefficients and example problem. Note that
the pile size is different from that used in earlier discussion on Caltrans standard
16-inch CIDH pile.

Table 2. Comparison of Pile vs. Cap Stiffness

From Pile Cap

Passive
Base of Pressure
Stiffness Component From Piles Footing Vert. Face
Lateral Translation (k/in) 378 1,833 1,167
Vertical Translation (k/in) la,0aa 2,333 N/A
Rocking Rotation (in-k/rad) 1.8 x la 7 5.2 x la 6 N/A
Torsional Rotation (in-k/rad) 1.2 x 106 1.2 x la 7 N/A
Cross Coupling Between (k) 2.a x la 4 a.a N/A
Lateral Trans. and
Rocking Rotation
See Figure 6b for configuration and Table 1 for pile stiffnesses.
Notes: 1 k/in = 1.75 kN/cm; 1 in-k/rad = 11.3 cm-kN/rad; l k = 4.45 kN
Lam et al. 125

Significant engineering judgement is specification of minimum seat widths by

required on the use of the above estimated Caltrans and bridge designers should be
stiffness coefficients for a pile footing. cognizant of this issue. Provision of a
Considerations regarding interaction sufficient seat width represents excellent
between piles and the footing and other earthq\Jake design practice and can be
factors such as the contact condition accommodated economically for most new
between the cap bottom and soil in the designs. The role of foundation stiffness
presence of the piles introduce becomes more important in retrofit
uncertainties on the use of the footing situations especially when a more realistic
base stiffnesses (the second column Table analysis approach is warranted to reduce
2). However, it would be prudent to add the level of conservatism associat~d with
the pile-cap lateral passive soil uncertainties in analytical procedures.
resistance (the third column) to the pile
stiffness in design practice. Caltrans The above procedures on foundation
have adopted similar view point in their stiffness can also be applied for
design practice. Ignoring the footing base temperature loading evaluations.
is considered conservative. It can be Significant reduction (relaxation) in the
observed that even if the bottom tractional structural stresses can usually be realized
stiffness is ignored, the lateral passive if the foundations are allowed to deform.
resistance of the pile cap would dominate Therefore introduction of foundation
the lateral stiffness of the pile group. stiffnesses in bridge analysis would
usually lead to a more economical design.
CONCLUSIONS AND REOOHMENDATIONS
There are other geotechnical considera-
Procedures and accompanying design charts tions, especially those related to ground
to facilitate practical solutions of stability that have not been addressed in
abutment and foundation stiffnesses for this paper. Cooperation between structural
dynamic response analyses of typical and geotechnical engineers is strongly
highway bridges have been developed and recommended to address such issues,
presented. The presented procedures are especially for poor soil conditions. In
relatively simple and emphasis has been addition, a number of sensitivity issues
placed on hand and graphical solution have been discussed in this paper.
methods for easy application. An in-house Although an attempt has been made to ensure
Earth Mechanics project to computerize the our example problems reflect real typical
presented procedures is presently being situations, one must be careful in
undertaken. extrapolating the presented discussions and
results to other design conditions.
In view of uncertainties in ground motion
which could lead to differential foundation ACKNOWLEDGEMENT
movements and other geotechnical concerns
(e.g. ground stability and liquefaction The first author, Ignatius Po Lam, is
problems), ductility design has significant indebted to Mr. Hudson Matlock, retired,
technical merit for earthquake resistance. and formerly Professor and Chairman of
Allowance for ductility tends to lead to a Civil Engineering Department from the
design more tolerant to foundation University of Texas at Austin for the
movements. However, ductility design opportunity to learn and exchange
requires realistic evaluation of the background knowledge on pile foundations
magnitude of displacements and deformations over the past ten years. Mr. James Gates
and design provisions (e.g. allowance of of Caltrana has been helpful in discussions
minimum seating widths as recommended by of sensitivity issues in abutment and
Caltrans) to accommodate the displacements. foundation design. Dr. Wen David Liu of
Imbsen and Associates, Inc. developed the
Incorporation of foundation and abutment design charts for spread footings. The
stiffness in design and retrofit analyses presented design procedures and the
of highway bridges leads to an improved accompanying design charts were developed
solution of the overall seismic load level in the course of performance of two
and the distribution of the overall load projects funded by The Federal Highway
among various bents and abutments. More Adminstration ( FHWA) : ( 1) a contract to
importantly, it leads to better estimates develop aseiamic design procedures for
of displacements. However, uncertainties bridge foundations (Lam and Martin, 1986)
in ground motion and other geotechnical and ( 2) a contract to develop teaching
concerns warrant an even more prudent materials and implementation of a training
approach to provide for potential bridge course currently administered by Imbsen and
displacements. This has led to the Associates, Inc.
126 TRANSPORTATION RESEARCH RECORD 1290

RBFBREN'CES 7. Matlock, Hudson, "Correlations for

Design of Laterally Loaded Piles in
1. AASHTO, "Guide Specifications for Soft Clay," Proceedings, Offshore
Seismic Design of Highway Bridges," Technology Conference, Houston Texas,
Highway Subcommittee on Bridges and 1970, Paper No. 1204.
Structures, 1983.
8. 0' Neill, M. W., and J. M. Murchison,
2. Douglas, Bruce, Mehdi Saiidi, Jamee "An Evaluation of p-y Relationships in
Richardson, Jamee Hart, "Results from Sands," A Report to the American
High Amplitude Dynamic Teets and Petroleum Institute (PRAC 82-41-1),
Implications for Seismic Design," May 1983.
extracted from Seismic Research for
Highway Bridges (US-Japan Program), 9. Roeenblueth, E. c., "The Mexican
supported by National Science Founda- Earthquake: A Firsthand Report," Civil
tion Grant Number CEE-8303659, Com- Engineering Magazine, ASCE, January,
plied by John F. Flemming, June 1984. 1986.
3. Caltrans, "Bridge Design Aide Manual". 10. Terzaghi, Karl, "Evaluation of
Coefficients of Subgrade Reaction,"
4. Iwasaki, T., Penzien, J., and Clough, Geotechnigue, Vol. 5, No. 4, pp. 297-
R.W., "Literature Survey Seismic 326, 1955.
Effects on Highway Bridges," EERC
Report No. 72-11. University of 11. Vij ayvergiya, v. N. , "Load-Movement
California, Berkeley. 1972, and FHWA- Characteristics of Piles," Paper
RD-73-13. November 1973. presented in the Port 77 Conference,
Long Beach, California, March 1977.
5. Lam, Ignatius and Geoffrey R. Martin,
"Seismic Design for Highway Bridge 12. Wilson, John c., "Stiffness of Non-
Foundations," Proceedings, Lifeline Skewed Monolithic Bridge Abutments for
Earthquake Engineering: Performance Seismic Analysis," Earthquake
Design and Construction, ASCE Engineering and Structural Dynamics,
Convention, 1984, San Francisco. Vol. 16, 1988, pp.867-883.
6. Lam, Ignatius (Po) and Geoffrey R.
Martin, "Seismic Design of Highway
Bridge Foundations," FHWA Report Noe.
FHWA/RD-86/101, FHWA/RD-86/102,
FHWA/RD-86/103, 1986,