See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/298520519

Seismic performance and retrofit of steel pile to

concrete cap connections. Paper by Ayman A.
Shama, John B. Mander, and Amj....

Article · November 2002

CITATIONS READS

0 107

4 authors, including:

Ayman Shama J.B. Mander

Ammann & Whitney Texas A&M University
23 PUBLICATIONS 141 CITATIONS 226 PUBLICATIONS 6,228 CITATIONS

SEE PROFILE SEE PROFILE

Amjad Aref
University at Buffalo, The State University of N…
125 PUBLICATIONS 991 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Fitness for Service: Probabilistic Fatigue Life Assessment of Steel Railway Bridges View project

Fracture Critical Steel Twin Tub Girder Bridges View project

All content following this page was uploaded by Ayman Shama on 22 March 2016.

The user has requested enhancement of the downloaded file.

 ACI STRUCTURAL JOURNAL
Title no. 99-S6
Seismic Performance and Retrofit of Steel Pile
to Concrete Cap Connections
By Ayman A. Shama, John B. Mander, and Amjad J. Aref
This research is concerned with the seismic performance of steel
pile-to-pile cap connections representative of construction practice
in the eastern U.S. Two perspectives are considered. The first is the
seismic vulnerability of existing pile cap connections, where the
embedment depth of the pile inside the cap beam is small. There-
fore, an initial experimental study was conducted for testing two
specimens that represented existing exterior connections under
cyclic lateral loading. The second perspective is the seismic design
requirements for strong cap beam-to-pile connections. Hence, a
theory that assumes a linear distribution of stresses along the con-
nection embedment depth was developed, and comparisons with a
finite element model were performed. A second experimental pro-
gram was conducted to evaluate the performance of specimens ret-
rofitted in accordance with the theoretical model developed in this
study. The results of the second experimental study validated the
proposed retrofit strategy.
Keywords: concrete; pile; seismic.
RESEARCH SIGNIFICANCE
The behavior and capacity of cyclically loaded steel H-pile-
to-concrete pile cap connections are still not well under-
stood, and the retrofit strategies for these connections have
not been assessed. The present work has two main signifi-
cances. It gives a better understanding of the performance of
the pile-to-cap connection under cyclic loading, and pro-
vides the design engineer with theoretical principles that are
sufficiently accurate, yet simple enough to be used in practice
for predicting the behavior of such connection under lateral
loading conditions.
INTRODUCTION
As a result of the disastrous consequences of earthquakes
during the last decades, research has been focused not only
to enhance the seismic performance of newly designed struc-
tures, but also to develop retrofit strategies for existing ones.
Pile-to-cap connection may exhibit plastic hinging during
earthquakes if the piles are driven in different saturated soil
layers and the first layer is prone to liquefaction. In such a
case, the local performance of the connection will have a major
effect on the overall performance of the structure.
Limited previous work has been conducted on the seismic
evaluation of pile-to-cap connections. Pam and Park 1,2
conducted a series of cyclic tests on prestressed concrete pile-
to-cap connections. The results showed that well-detailed
prestressed concrete piles and pile-to-cap connections are
capable of undergoing large postelastic deformations without
significant loss in strength when subjected to severe seismic
loading. Chai and Hutchinson3 conducted an experimental
program for bridge structures supported on single extended
reinforced concrete pile shafts, considering the pile-soil
interaction in their experiments. They tested four full-scale
400 mm diameter concrete piles with details representative
ACI Structural Journal/January-February 2002
TECHNICAL PAPER
of the current CALTRANS design. Their study showed that
the depth to the plastic hinge within the soil decreases with an
increase in the soil density and an increase in the above-
ground height. Xiao et al.4 conducted a series of tests as a part
of the FHWA-MCEER Highway Project to investigate seismic
behavior of bridge steel pile-to-pile cap connections represen-
tative of construction in California. Five full-scale H-shaped
steel pile-to-cap connection subassemblies were tested during
this experimental study. Two of the full-scale subassembly
specimens were subjected to vertical cyclic load that simulated
axial forces in pile due to footing overturning during a seismic
ground motion. Two others were loaded with cyclic lateral
force and constant vertical load. It was found that the pile-to-
cap connection representative of west-coast construction
could sustain a significant amount of moment.
Steel H-piles are used extensively in bridge pile founda-
tion construction. A common form of practice for bridge
foundations in the eastern U.S. is to embed steel piles approx-
imately 300 mm into the reinforced concrete cap beam. Other
foundations are sometimes designed with 75 to 100 mm
embedment, with the bottom reinforcing bar cage about the
pile. Primarily designed for vertical loading, pile-to-cap connec-
tions may be susceptible to damage during seismic events.
Therefore, it is imperative to appraise the performance of
these connections under cyclic lateral loading through a
comprehensive experimental program and to develop methods
for its retrofit, assessed by experiments to maintain its ductility
during large cyclic drifts. The work presented herein is an
effort to contribute towards both goals.
EXPERIMENTAL STUDY ON EXISITING
PILE-TO-CAP CONNECTIONS
This experimental program consisted of testing two speci-
mens consistent with present as-built practice in the eastern
and central U.S. An HP10X42 steel section was employed in
the experiments, as it is extensively used in pile foundation
practice. Figure 1 illustrates the procedure used to determine
the physical modeling configuration for a prototype steel pile
foundation. By extracting the shaded portion in Fig. 1(d) and
inverting it, a test specimen is formed when anchored to the
laboratory strong floor (Fig. 1(e)). The plastic mechanisms
for these structures in cohesionless and cohesive soils
suggest an average value of 3dp for their effective length L.
Consequently, the cantilever length of the pile foundation
experiments was taken as 0.785 m. The reader may wish to
ACI Structural Journal, V. 99, No. 1, January-February 2002.
MS No. 00-321 received December 20, 2000, and reviewed under Institute publica-
tion policies. Copyright © 2002, American Concrete Institute. All rights reserved,
including the making of copies unless permission is obtained from the copyright pro-
prietors. Pertinent discussion will be published in the November-December 2002 ACI
Structural Journal if received by July 1, 2002.
1

Ayman A. Shama is a structural engineer at Parsons Transportation Group Inc., New
York, N.Y. He received his MSc and PhD from the State University of New York at
Buffalo. His research interests include nonlinear finite element modeling and design
of earthquake-resistant reinforced concrete structures.
John B. Mander is a chaired professor of structural engineering in the Department of
Civil Engineering at the University of Canterbury, New Zealand. His research inter-
ests include reinforced and prestressed concrete structures, particularly aspects of
seismic design of bridges.
Amjad J. Aref is an assistant professor in the Department of Civil Structural and
Environmental Engineering at the State University of New York at Buffalo. His
research interests include reinforced concrete and application of fiber reinforced
composite materials in the infrastructure.
refer to Shama5 for a detailed plastic mechanism analysis of
pile-soil interaction.
Design and construction of specimen
Figure 2(a) illustrates a typical piled foundation prototype
for an exterior bridge foundation, taken from Standard Plans
for Highway Bridges.6 As shown in the figure, an exterior
pile in a pile group can be exposed to seismic load along its
strong and weak axes directions. Moreover, the tension uplift
during cyclic loading may be a major factor that can affect
its performance. Based on this criterion, the experimental
program in this study included testing two exterior piles
along the strong and weak axes. One cap beam was constructed
to accommodate both cases. Some slight modifications were
introduced to the model as follows.
Although the bridge foundation design shown in Fig. 2 uti-
lized one layer of mesh reinforcement at the bottom of the
foundation, some other drawings utilized an additional layer
at the top. Consequently, for the present study, two layers
were used in the design of the model specimen. The hatched
Fig. 1—Physical modeling rationale for pile-to-cap connection.
Fig. 2—Bridge foundation prototype and rationale used for devising model.
2 ACI Structural Journal/January-February 2002
area in Fig. 2(c) was employed to design the model speci-
men. Therefore, the transverse steel was welded to a
L2XL2X3/8 angle to represent the continuation adequacy
purposes of the steel in that direction. Figure 3 illustrates the
geometry and reinforcement of the specimen. The pile spec-
imens were provided with the bearings shown in the figure
to provide the rotation adequacy for a lever beam relative to
the pile during the test through its sole plate. The HP10 x 42
piles were obtained from a local pile-driving company. The
piles had experienced driving stresses, but were otherwise
unused. A typical pile-driving procedure involves driving
the pile to the required depth, then cutting the exposed length
to achieve the proper height. The sections left over from this
operation were employed as test specimens for this study.
During the construction of the wooden boxing for the pile
cap beam, a wooden framework was also constructed to keep
up the proper alignment of the pile in the cap beam while
the concrete was poured in. A photograph for the construction
of the specimen is shown in Fig. 4. Following each test of
the as-built specimens, the pile was remounted back in its
location, and any damage to the concrete cap was repaired
prior to the retrofitting.
Experimental setup
Exterior piles are usually exposed to additional vertical
thrust during cyclic loading. The vertical thrust was repre-
sented in the present study by using a hybrid setup that
combines the vertical component of diagonal actuator force
with a variable axial force from vertical actuator acting via a
lever beam system (Fig. 5). This setup is effective when the
space available in the laboratory restricts setting the diagonal
actuator in an angle that furnishes the required vertical
thrust. According to this setup, the vertical prototype applied
load can be expressed as
P v = Pg + υ F = Pv ′ + P da sin θ (1)
where Pg = the gravity load on pile; υ = ratio of the required
vertical thrust to the horizontal applied load; F = the horizon-
tal applied load evaluated as F = Pda cos θ; θ = available
inclination angle to fit the space in the laboratory; Pda = the
force applied by diagonal actuator; and Pv′ = the force trans-
ferred from the vertical actuator to the specimen, can be
quantified as
Pv ′ = Pg + ω Pda (2)
where ω = a proportionality factor that adjusts the fraction of
load transferred from the diagonal to the vertical actuator. By
substituting Eq. (2) into (1), ω can be quantified as
ω = ν cos θ – sin θ (3)
tical actuator increased the axial force in the column when
the lateral actuator was pushing and decreased it during pull.
A W10X88 steel beam was used to anchor the specimen to
the strong floor to provide the sufficient restraint against
translation and uplift during the tests. The beam was an-
chored at one end to the strong floor using two 25 mm high-
alloy prestessing threadbars. It was anchored from the other
end to the strong floor, using one 32 mm high-alloy prestess-
ing threadbar. The two 25-mm bars were prestressed to 90 kN
each. This prestressing force resulted in a 480 kN axial an-
choring force on the specimen. Figure 6 shows a photograph
of the test rig.
The instrumentation used for this experiment consisted of
sonic transducers and load cells. The load cells were 500 and
650 kN devices supplied with the actuators, respectively.
Two sonic transducers (ST) were used for the measurement
of lateral displacement. The upper one was ± 203 mm
stroke and was mounted at the same height as the centerline of
the inclined actuator. The second transducer was ± 150 mm
stroke and was mounted at the same height of the point of ac-
tion of the actuator that is at the lateral loading inflection

The vertical actuator force Pva can be evaluated as

Pva = λ Pv ′ (4)

where λ = coefficient to account for the effect of the lever

beam. To operate the vertical actuator according to the rela-
tionship of Eq. (4), the actuator analog control system required
input from the load cell of the diagonal actuator. This load
cell value is then multiplied by the value ω. A fixed offset is
employed by the vertical actuator controller to account for
the gravity load.
Figure 5 illustrates the test rig employed for this experimental
program. Lateral load was applied directly at the theoretical
inflection point of the column and provided by a 1100 kN MTS
hydraulic actuator anchored to a reaction frame at an angle
of 42 degrees to the horizontal and connected directly to the
specimen. The vertical load was provided by a 50 kN capacity
± 50 mm stroke servo-controlled hydraulic actuator, operated
in load control and connected to a W10X77 lever beam. The
W10X77 gravity load beam was anchored to the strong floor
at one end using a 32 mm diameter high-alloy prestressing
threadbar. The axial load transferred from the vertical actua-
tor to the pile (in kN) was given by 135 + 0.276Pda . The ver-

Fig. 4—Photograph of construction of specimen.

Fig. 3—Geometry and reinforcement of pile foundation as-

built specimen. Fig. 5—Test rig employed for experiments.

ACI Structural Journal/January-February 2002 3

point and was used to provide the input signal for the inclined ac- cycle as the pile walked out of the socket. Failure of the con-
tuator, which was operated in displacement control. nection was evident by the end of the second cycle at 5% drift.
The maximum tension uplift experienced during this experi-
Materials ment was 132 kN at a maximum pull lateral force of 205 kN.
The test specimen was constructed with ready-mix concrete The maximum vertical compression force reached was 430
and Grade 60 (fy = 414 MPa) steel reinforcement. The con- kN at a maximum push lateral force of 227 kN. A photograph
crete’s ultimate compressive strength was determined from of the specimen after failure is shown in Fig. 7(b).
the results of three 150 x 300 mm cylinder tests as 30 MPa. Specimen PW—Specimen PW was tested along its weak
Four steel coupons were cut from the pile flanges for the axis orientation under the same loading conditions as
evaluation of its material properties. The specimens were Specimen PS. The overall connection behavior was governed
constructed according to the ASTM standard dimensions for by the response of the steel section up to the end of the 2%
tension testing of metallic materials. A tension test was drift test. The steel section flanges exhibited an inelastic
conducted for each specimen in an axial MTS 445 kN behavior at the push of the first cycle during the 2% drift.
closed-loop servocontrolled hydraulic test system. The average
This was characterized by visible yield lines on the white-
yield and ultimate stresses were determined as 315 and
washed compression flanges along a distance of 159 mm
475 MPa, respectively.
above the concrete cap surface. At 3% drift, during the pull
cycle, a brittle failure occurred, accompanied by a loss of
Experimental program and results
The experimental program consisted of testing the speci- strength of approximately 50 kN, as shown in Fig. 8(a), and
mens with two reversed cycles at drift amplitudes of ± 0.25, the strength continued to drop during the 4 and 5% drift tests.
0.5, 1, 2, 3, 4, and 5% drift. Testing was conducted under dis- Five more drifts were performed at the 5% level. An obvious
placement control, where the specimens were first pushed, slipping out of the steel section from the concrete occurred
then pulled. The command signal was provided by an analog during the pull of the first cycle of that test. This slipping was
function generator in the form of a positive sine wave with a accompanied by centralization of cracks along the region of
cyclic period of 1 min. contact between concrete and steel. It was apparent, by the
Specimen PS—Specimen PS represents an exterior pile end of the fifth cycle at this drift, that the connection had no
foundation with a typical steel pile cap connection. The
specimen was tested along its strong axis orientation with a
variable axial load considered to be representative for this
class of piles. The specimen had a clear height of 785 mm
above the concrete base surface. The specimen behaved
linearly prior to the 1% drift amplitude. A brittle failure of
concrete was pronounced at the push of the first cycle at the
2% drift. This failure was characterized by two wide cracks
formed in the concrete starting at the flange tips and propa-
gated away from the pile towards the transverse edge of the
cap beam to a 450-mm distance above the concrete’s lower
surface. This failure was accompanied with an approximate
30% loss of strength in the hysteretic load-displacement
loops (Fig. 7(a)). The pinching shown in the force displace-
ment loops evidenced the significant cracking of the pile
cap. During the course of the 4 and 5% drifts, slipping of the
steel section within the embedment zone was apparent, and
tension uplift made it more pronounced during the pull of each
(a) Lateral-load displacement relationship

(b) Damage occurred to the pile cap after test

Fig. 6—Photograph of test rig. Fig. 7—Performance of Specimen PS under cyclic loading.

4 ACI Structural Journal/January-February 2002

more resistance. The maximum vertical tension and com- F M
pression forces attained were 86 and 356 kN, respectively. – --- – ------j ≥ – f c (6)
A Sx
The maximum lateral forces associated with these forces were
172 and 170 kN in the pull and push directions, respectively.
A photograph of this specimen after failure is illustrated whereupon rearranging
in Fig. 8(b).
A general observation can be inferred from this experi- 6Mj
mental program that a pile-to-pile cap connection is very F - + ---------------
f c ≥ ------------- - (7)
susceptible to damage from cyclic lateral loading. Therefore, b f l emb b l 2
f emb
in zones of moderate to high seismicity, it is prudent to per-
form any properly designed seismic strengthening for this
in which fc = the allowable concrete compressive stress at
class of connection. The goal is to provide a more ductile
connection that can possess a large deformation capability the extreme fiber in the front face of the connection; F= the
that permits dissipation of seismic energy during large earth- applied horizontal load; bf = the flange width of the steel
quakes. Hence, a theory is developed with the aim that pile section; lemb = embedment depth of the steel pile sec-
plastification takes place only in the steel pile. The theory is tion inside the concrete cap beam; and Mj = the nominal
explained in the following section. moment capacity of concrete. The applied lateral load can
be expressed as
PILE-TO-CAP CONNECTION EFFICIENCY
The plastic crushing moment capacity of the concrete cap
beam as well as the plastic moment capacity of the steel
pile control the performance of the connection under high
drifts. Because steel is more ductile than concrete, failure
of the connection under severe seismic drifts is expected to
occur when the concrete stress reaches its crushing capacity,
while the steel pile section is still capable of absorbing
more energy. Consequently, connection efficiency is defined as

M
ρ = ------j- (5)
Mp

where Mj = moment capacity of the concrete-pile connection

(joint); Mp = fy Zp = nominal moment capacity of the pile;
fy = the yield stress of the steel pile section; and Zp = the plastic
modulus of the steel pile section. If ρ > 1, then plastification (a) Lateral-load displacement relationship
in the pile is expected, whereas if ρ < 1, damage to the
connection is expected.

THEORETICAL MODEL
The theoretical model was developed for the strong axis
bending direction, and the same approach can be used for the
weak axis direction. The following assumptions are made in
the analysis:
1. The lateral applied load is relatively high when com-
pared with the axial gravity load. Hence, axial load effect
is ignored;
2. Both concrete bending and compressive stresses along
the embedment depth of the steel section will counteract the
lateral applied load;
3. The stresses and strains, at any load stage, continue to
be closely proportional, that is, the stress distribution along
the embedment depth is linear;
4. The stresses developed at the back face of the con-
nection are small enough to be considered negligible dur-
ing the analysis; and
5. The overall behavior of the connection is governed by
the concrete compressive stress at the extreme fiber in the
front face of the connection.
The mechanism based on these assumptions is shown in
Fig. 9. According to the last assumption, an expression for
the compressive stress of the concrete can be written at the (b) Damage occurred to the pile cap after test
extreme fiber in the front face of the connection in terms of
the stress blocks as follows Fig. 8—Performance of Specimen PW under cyclic loading.

ACI Structural Journal/January-February 2002 5

 M in which tf = the pile steel section flange thickness; and dp =
F = ------j (8) the pile steel section depth. Therefore, the connection effi-
*
L ciency can be quantified as

where L* = the distance from the point of application of the fc  d p  l emb 2
lateral load to the neutral axis of the joint. Substituting Eq. (8)  --- ----- ---------
 fy  t f   d p 
in (7) gives ρ = -------------------------------------- (13)
 6 + --------
l emb
-
 * 
f c ≥ M j  -------------------- + ----------------
1 6 L
(9)
 * 2
L b f l emb b f l emb
Based on the study by Shama5 on the locations of plastic
Solving for the joint moment gives hinges within the pile, an average value of 0.3 can be taken
for lemb/L*. Substitute in Eq. (13) an expression for the
2
elastic (precrushing) efficiency is obtained
f c b f l emb
M j ≤ -----------------------
- (10)
 6 + --------
l emb
-
f d l emb 2
ρ = 0.16  ---c   ----p-  --------
- (14)
 *   fy   t f   d p 
L

Introducing the connection efficiency ρ as defined in By inverting Eq. (14), one can obtain a design expression
Eq. (5), therefore for determining the required normalized embedment length

2
f c bf l emb l emb ρ   fy   t f 
 ---------
ρ = ---------------------------------- (11) --------- = - --- -----
 0.16  fc  d p
(15)
l emb dp
f y Z p  6 + -------- -
 * 
L
FINITE ELEMENT MODELING OF
PILE-TO-CAP CONNECTION
The plastic modulus Zp of the steel pile section can be ap- The three-dimensional finite element model described
proximately expressed as herein represents a short exterior pile with a 300 mm embed-
ment depth in a concrete pile cap. The portion of Specimen
Zp ≈ tf b f dp (12) PS up to the location where it was anchored to the laboratory
strong floor was represented in this model. Therefore the
length, width, and height of the pile cap were taken as 1450,
1000, and 900 mm, respectively. Fixed boundary conditions
were assigned to all the nodes at the interior edge surface to
be consistent with the experimental setup. The lever arm for
the pile was taken as 785 mm from the concrete surface to
conform to the experiment. The direction of the lateral load
was set so that the bending was along the pile’s strong axis.
The geometrical properties of the steel section was taken as
HP10X42, which was the same as the experiment.
Finite element mesh for the model was generated using a
finite element (FE) analysis package, and the finite element
analysis was performed using the ABAQUS Code Version
5.7. An eight-node linear brick element of the type C3D8
was used to limit the computational time to a reasonable
extent. 82.5 x 82.5 x 80 mm brick elements were used for the
concrete cap beam. The same dimensions, except for the
thickness, were assigned to the steel section. 18,270 nodes
and 13,260 elements were used. Contact gap elements of the
type GAPUNI were used to model the steel-concrete interac-
tion during loading. These elements were necessary to
transfer the lateral forces properly from the steel pile to the
concrete beam during lateral loading, hence leading to the
appropriate distribution of stresses along the embedment
depth. Each gap element allows for contact between two
nodes. One node is located on the steel section, and the other
one is located in the same location on the concrete beam. Each
gap element was defined by specifying the two nodes forming
the gap and providing geometric data defining the initial
state of the gap. In the present study, the initial state of the gap
was set to zero; that is, the surfaces are initially bonded. A
Fig. 9—Assumed theoretical linear stress distribution. friction property is associated with each gap element.

6 ACI Structural Journal/January-February 2002

This friction property controls the slipping of the surfaces in
contact relative to each other. In the present study, the basic
Coulomb friction model was used to represent the friction
between the interacting surfaces. A value of 0.6 was used
for the coefficient of friction. A total number of 180 gap
elements were used to represent the contact between the steel
section and the concrete beam.
The material properties used for the steel are as follows:
modulus of elasticity = 200,000 MPa; Poisson’s ratio = 0.28;
and yield stress = 315 MPa. Those values are equal to those
obtained from the coupon test of the HP10X42 steel section.
The plastic behavior of the steel material was modeled using
the classical metal plasticity model in ABAQUS. This model
uses Mises yield surface with associated plastic flow and
isotropic hardening.
The material properties used for the concrete are as follows:
compressive strength after 28 days = 30 MPa; plastic strain
at failure = 0.002; and Poisson’s ratio = 0.18. The modulus
of elasticity of concrete was determined according to the
ACI-318 equation (Ec = 4700 √f c′ MPa), in which Ec = mod-
ulus of elasticity of concrete, and f c′ = concrete compressive
Fig. 10—Compressive stresses in MPa for connection from
three-dimensional FEM.
ACI Structural Journal/January-February 2002
strength. The concrete cap beam is characterized by little
reinforcement. Therefore, the tension stiffening was defined
in terms of a maximum displacement that a certain point in
the cap beam might reach, at which a linear loss of strength
after cracking gives zero stress. The determination of this
maximum displacement was based on an energy approach in
which the concrete’s brittle behavior is characterized by a
stress-displacement response rather than a stress-strain re-
sponse. One can refer to the ABAQUS 5.7 theory manual for
more details about this formulation. For the present study,
this maximum displacement was taken as 0.5 mm. This value
is well above the default value used in ABAQUS (0.1 mm).
It was adequate, however, to avoid early divergence problems
of the nonlinear algorithm that might prevent the solution’s
progress at early stages.
Comparisons between the FE models with the cracked
elastic theory are in terms of the compressive stress at the
concrete surface and the distribution of stresses along the
steel-concrete interface. Equation (10) can be rearranged and
written in the following form
l emb
M 0  6 + -------- -
 * 
L
f c ≥ -------------------------------
2
- (16)
b f l emb
in which M0 = the applied moment by the lateral load.
Equation (16) is compared with the FE model in terms of the
compressive stresses at the concrete surface and the distribu-
tion of stresses along the embedment depth. To agree with
the theoretical assumptions, the FE model was only analyzed
under a 120 kN lateral load; that is, no axial load was
considered. This value was sufficient to maintain the
stress distribution within the linear embedment depth and
was enough to induce cracking within the damaged elasticity
concrete model in ABAQUS. One needs to remember that
the lever arm of this specimen was taken as 785 mm. There-
fore, the applied moment can be quantified as 94.2 kNm. By
substituting this value in Eq. (16), the compressive stress at
the concrete beam surface can be obtained as 26 MPa. Contour
plots of the compressive stresses from the FE analysis are
shown in Fig. 10. One can observe that the distribution of
stresses along the flange width is not uniform. Consequently,
the comparison is made in terms of the average stresses
along the flange width. The distribution of the compressive
stresses along the flange width ranged from 6 MPa at the
flange edge to 36 MPa at the center. An average value of 20
MPa was taken for the compressive stress at the concrete
surface. This value agrees reasonably with the value obtained
using the proposed theoretical formula.
On the basis of the stress contours along the steel-concrete
interface, the distribution of the compressive stresses is
plotted in Fig. 11. The distribution can conform to linear
distribution. The figure also indicates that the stresses gener-
ated at the back face of the connection are very diminutive
compared with the linear stresses along the front face, and
hence, they assess the assumptions implemented in the theory.
DESIGN REQUIREMENTS AND RETROFIT
Figure 12 presents relationships between the normalized
embedment depth and the connection efficiency for an allow-
able stress of fc = 0.85 f c′ . Using this stress ensures that
premature concrete crushing will not occur. If an efficiency of
7
ρ = 1.6 is adopted for design (this value allows for connection to determine the required total embedment length of the pile
overstrength as shown in the figure), then the required embed- into the concrete foundation and hence, the additional
ment depth can be obtained by substituting in Eq. (15) as embedment length for retrofit.
Additional reinforcement is required to prevent a gap
from beginning to open and subsequent spalling of concrete
fsu   t f 
- = 3.4  -----
l emb at edge piles under cyclic loading. Assuming that the tension
-------- ----- (17)
dp  fc′   dp  force in the steel reinforcing bars will resist the moment
capacity of the connection, that is, the plastic moment capacity
of the steel H-pile section in Fig. 13, then
in which fsu = the ultimate stress of the steel pile material.
The figure also indicated a possible vulnerability of the pile-
to-cap connections designed for a 300-mm embedment that 2
M p = Tjd = --- Tl emb (18)
agrees with the outcomes of the tests on the as-built speci- 3
mens. To avoid the expected damage to cap beams under
seismic lateral load, seismic strengthening of the pile-to-cap and hence
connection is necessary. Therefore, the objective is to develop
practical and economical seismic retrofit measures to main- M
tain a serviceable connection during earthquakes. The retro- T = As fyh = 1.5 --------p- (19)
fit strategy adopted in the present study is to increase the cap l emb
beam strength to the level required to shift the plastic hinge
location to the steel pile rather than to the concrete cap. This Therefore, the area of steel required can be quantified as
strategy ensures a better ductile connection that can possess
a large deformation capability and permit much more dissi- 1.5f su Z p
pation of seismic energy. Therefore, Eq. (17) was employed A s = ------------------- (20)
f yh l emb

in which fsu = the ultimate stress of the H-pile section that

gives the overstrength demand (Zp fsu); and fyh = the yield
stress of reinforcing bars.

Fig. 11—FEM distribution of compressive stresses along

embedment length.

Fig. 12—Theoretical efficiency of pile-to-cap connections. Fig. 13—Steel required for edge piles.

8 ACI Structural Journal/January-February 2002

 CONSTRUCTION OF RETROFIT
The required embedment length was determined using
Eq. (17)

l emb fsu   tf 
--------- = 3.4  ----- 475 × 10.7 = 2.9
----- = 3.4 -------------------------
dp  fc′   dp  28 × 246.4

Therefore, the additional embedment length was taken as

450 mm. The longitudinal reinforcement required to prevent
concrete spalling at the pile cap edges under cyclic loading
was calculated using Eq. (20)

1.5f su Z p 1.5 × 475 × 791000 2

A s = ------------------- = ----------------------------------------------- = 1787mm ,
f yh l emb 414 × 762 Fig. 15—Pile cap retrofit construction.
assume four bars per pile, dbar = 25 mm.

One should keep in mind that pile cap foundations are usu-
ally constructed below the ground level and that during the
retrofit of these foundations, soil will be excavated to the level
of the overlay depth. Under these circumstances, and to
facilitate the concrete placing, 300 mm was added to the pile
cap width at the edges. Figure 14 illustrates the geometry and
reinforcement of the retrofit. The construction of the retrofit
in the lab was carried out in three steps: 1) the main longitu-
dinal L-shaped 25-mm reinforcement was set in place and
fastened together; 2) additional longitudinal reinforcement
consisting of 13-mm reinforcing bars at 150 mm was added
to complete the reinforcing cage; and 3) the construction of
the retrofit was concluded by setting the formwork in place
and placing the concrete. A photograph of the retrofit con-
struction is shown in Fig. 15.
Since the retrofit process will be carried out in the field in
an inverted position, soil excavation must be wide enough to
allow for the retrofit construction. Also, good compaction of
(a) Lateral-load displacement relationship
the new concrete is necessary during placing to ensure
bonding and avoid any potential gaps between the new
and existing concrete.

EXPERIMENTAL PROGRAM AND RESULTS FOR

RETROFITTED PILES
The experimental program for testing the retrofitted speci-
mens was similar to the one used for as-built connections. For
consistency, the same loading procedures were employed.

(b) Local buckling occurred to steel pile in the hinge zone

Fig. 16—Performance of Specimen RePS under cyclic

Fig. 14—Geometry and reinforcement of retrofit. loading.

ACI Structural Journal/January-February 2002 9

Specimen RePS
Specimen RePS (previously PS) was tested under variable
axial load with two reversed cycles at each drift amplitude of
± 0.25, 0.5, 1, 2, 3, 4, and 5%, concluding with one cycle at
± 6% and another half cycle at +7%. This specimen showed
a ductile behavior in the steel pile with a few insignificant
cracks in the concrete beam cover in a small region sur-
rounding the steel pile. Yielding of the steel section of this
specimen occurred prior to the 1% drift level. The inelastic
behavior of the pile was pronounced during the 2% drift.
This was characterized by diagonal striations in both the
whitewashed flanges and the web along a 125 mm distance
above the concrete cap beam surface. These striations extended
to a distance of 200 mm above the concrete surface during
the first cycle at 3% drift, with strain-hardening of the steel
material noticeable in the force-displacement loops. Local
buckling started to occur at the first cycle of the 3% drift, but
(a) Lateral-load displacement relationship
(b) Local buckling occurred to steel pile in the hinge zone
Fig. 17—Performance of Specimen RePW under cyclic
loading.
10
it was very pronounced along a distance of 200 mm during
subsequent cycles. The force displacement of the retrofitted
pile is shown in Fig. 16(a). It should be noted that the concrete
base did not experience any major damage during these
cycles, except for some surface cracks within the cover
zone. The maximum tension uplift was 475 kN and occurred
at a 480 kN lateral force. A photograph of the specimen after
the test is shown in Fig. 16(b). The effectiveness of the retrofit
strategy developed in the present study is demonstrated by
comparing Fig. 16 to Fig. 7. The as-built connection showed a
very indigent performance under cyclic loading. On the other
hand, the retrofitted connection exhibited ductile behavior,
which is preferable for seismic considerations.
Specimen RePW
Specimen RePW (formerly PW) was tested under variable
axial load with two reversed cycles at drift amplitude of
± 0.25, 0.5, 1, 2, 3, and 4%, concluding with 12 cycles at
± 5% (Fig. 17(a)). The specimen was tested up to 5% drift
without any significant damage in the concrete cap beam.
This behavior also validated the retrofit strategy adopted.
The overall performance of the connection was governed by
the ductile behavior of the steel pile (Fig. 17(b)). Inelastic
action on the section was very pronounced during the push
of the first cycle at 2% drift and continued through the 3%
cycle. The diagonal striations on the whitewashed flanges
extended to a 200 mm distance above the concrete beam
surface during the 3% drift and was accompanied by a
noticeable strain-hardening of the steel material in the
force-displacement loops at a maximum lateral force of
205 and 395 kN vertical compression force. The maximum
tension uplift (164 kN) occurred during the first cycle at
4% drift and was associated with a maximum lateral force
of 236 kN. The test stopped after the specimen completed
12 cycles at 5% drift.
CONCLUSIONS
Based on the results presented herein, the following con-
clusions are drawn:
1. Steel pile-to-concrete cap connections representative of
construction in the eastern U.S. may be vulnerable to damage
during future earthquakes, if the soil is liquefiable. Hence,
subsequent damage of the superstructure may also be expected;
2. A simplified linear elastic theory developed in the
present study was effective in predicting the performance of
such connections under lateral load;
3. Experimental results from the limited number of full-
scale test specimens presented in this paper indicated that
the retrofitted specimens possessed a superior perfor-
mance in terms of ductility with respect to the as-built
specimens. Therefore, the conceptual elastic cap/elasto-
plastic steel pile retrofit strategy proposed in this study is
considered to be validated;
4. It should be emphasized that the retrofitted pile-to-cap
connections investigated in the present study were tested to
high drift amplitudes of ± 6%. In an actual earthquake, the
structure may not exhibit such drifts. Local buckling failure
was attained in the experimental study as a result of low
cycle fatigue under lateral loading, which may not occur
during an actual seismic event; and
5. Further research is necessary for other kinds of piles
such as steel pipe piles and timber piles.
ACI Structural Journal/January-February 2002
 ACKNOWLEDGMENTS Pg = gravity load on pile
This research has been conducted by the Multidisciplinary Center for Pv = vertical prototype applied load
Earthquake Engineering Research (MCEER) and the State University of P v′ = force transferred from vertical actuator to specimen
New York at Buffalo (SUNY). Financial support by the Federal Highway Pva = vertical actuator force
Administration (FHWA) is gratefully acknowledged. The authors would Sx = section modulus of steel pile-concrete interface
like to thank Doug Shaffer, a former graduate student at the Department of T = tension force in steel reinforcing bars for connection retrofit
Civil Structural and Environmental Engineering at the University at Buffalo, tf = pile steel section flange thickness
for his help in the construction of the pile foundation specimen and also for Zp = plastic modulus of steel pile section
his assistance in the construction of its retrofit. λ = coefficient to account for effect of lever beam
υ = ratio of required vertical thrust to horizontal applied load
NOTATION θ = available inclination angle to fit space in laboratory
A = interface area between pile flange and concrete within embed- ρ = connection efficiency
ment depth ω = proportionality factor that adjusts fraction of load transferred
bf = flange width of steel pile section from diagonal to vertical actuator
dp = pile steel section depth
Ec = modulus of elasticity of concrete REFERENCES
F = horizontal applied load on pile 1. Pam, J. H., and Park, R., “Flexural Strength and Ductility Analysis of
Fy = yielding capacity of steel pile section Spirally Reinforced Prestressed Concrete Piles,” PCI Journal, V. 35, No. 4,
Fp = plastic capacity of steel pile section July-Aug. 1990, pp. 64-83.
fc = allowable concrete compressive stress at extreme fiber in front 2. Pam, J. H., and Park, R., “Simulated Seismic Load Tests on Pre-
face of connection stressed Concrete Piles and Pile-Pile Cap Connections,” PCI Journal,
f′c = concrete compressive strength V. 35, No. 6, Nov.-Dec. 1990, pp. 42-61.
fsu = ultimate stress of the steel pile material 3. Chai, Y. H., and Hutchinson, T. C., “Flexural Strength and Ductility of
fyh = yield stress of reinforcing bars Reinforced Concrete Bridge Piles,” Report No. UCD-STR-99-2, Depart-
fy = yield stress of steel pile section ment of Civil and Environmental Engineering, University of California--
jd = lever arm of acting forces on joint Davis, Calif., 1999.
lemb = embedment depth of steel pile section inside concrete cap beam 4. Xiao, Y.; Mander, J. B.; Wu, H.; and Martin, G., “Experimental
L = distance from point of application of lateral load to concrete Study on Seismic Behavior of Bridge Pile-to-Pile Cap Connections,”
surface 15th U.S.-Japan Bridge Engineering Workshop, Tsukuba City, Japan,
L* = distance from point of application of lateral load to neutral axis Nov. 9-10, 1999.
of joint 5. Shama, A. A., “On the Seismic Analysis and Design of Pile-to-Cap
M0 = applied moment by lateral load Connections,” PhD dissertation, State University of New York at Buffalo,
Mj = moment capacity of concrete-pile connection Buffalo, N.Y., 2000.
Mp = nominal moment capacity of steel pile section 6. ________ Standard Plans, “Standard Plans for Highway Bridges,”
Pda = force applied by diagonal actuator Timber Bridges, V. III, USDOT, FHWA, Washington, D.C., 1990.

ACI Structural Journal/January-February 2002 11

View publication stats