Professional Documents
Culture Documents
ductile
performance
from precast,
prestressed
concrete
piles
Andrew Budek
and Gianmario Benzoni
64 S um me r 2 0 0 9 | PCI Journal
Axial load
45
Lateral force A Ground level
40
35
Height above fixity/pile diameter
20 B
15
10
Yield Yield
5 moment moment
0
-1500 -1250 -1000 -750 -500 -250 0 250 500 750 1000 1250 1500
Moment, kN−m
Figure 1. This drawing shows the pile moment pattern resulting from axial and lateral loads. Note: 1 m = 3.28 ft; 1 kN = 0.225 kip.
Inadequate detailing of early prestressed concrete piles Ah = area of the spiral steel
may have caused inadequate performance in several earth-
quakes, such as the 1964 Alaska2 and 1972 Miyagi-Ken- d' = core diameter
Oki3 events. The relation of a pile’s detailing deficiency
to its poor performance has been identified by several s = spiral pitch
researchers.4–6
f c' = concrete compressive strength
Unfortunately, the response to inadequate performance has
been either a complete ban on precast, prestressed concrete fyt = yield strength of the transverse reinforcement
piles in seismic applications or the specification of ex-
tremely conservative amounts of transverse reinforcement, Paxial = axial load
which make precast, prestressed concrete piles uncompeti-
tive in the marketplace. Ag = gross cross-sectional area
Piles in seismic regions are governed by the provisions of ρl = ratio of area of distributed longitudinal reinforce-
the American Concrete Institute (ACI) in Building Code ment to gross concrete area perpendicular to that
Requirements for Structural Concrete (ACI 318-05) and reinforcement
Commentary (ACI 318R-05)7 chapter 21. The minimum
transverse reinforcement ratio ρs in section 21.4.4 was However, it is not less than the minimum specified in ACI
modified by Priestly et al.8 as Eq. (1). 318 section 10.9.3, given by Eq. (2).
PCI Journal | S u m m e r 2009 65
0 1 2 3 4 5 6 7 8 9 10 11
40
170 Head displacement, in.
160
35
150 K = 53,400 kN/m3
140
30
130
120
110 25
100 K = 40,840 kN/m3
90 20
80
Prediction
70
Test 1 envelope 15
60
50
º Predicted spalling
40 10
30
20 5
10
0 0
0 50 100 150 200 250
Head displacement, mm
Figure 2. This graph compares the results of the finite-element-model prediction with the in-place pile test. Note: The pile had a 400 mm (16 in.) diameter with 50 mm
(2 in.) of cover, transverse reinforcement ratio = 0.006, longitudinal reinforcement ratio = 0.021, and axial load ratio = 0.1fc' Ag. Ag = gross cross-sectional area; fc' = con-
crete compressive strength; K = subgrade reaction modulus. 1 m = 3.28 ft; 1 kN = 0.225 kip.
Ah = cross-sectional area of a structural member measured If the spacing requirements were relaxed to allow the
out to out of transverse reinforcement use of a lighter-diameter wire at a smaller pitch, another
problem would be created. The use of a small aggregate
In the case of a typical 610-mm-diameter (24 in.) pile, size would be mandated by the need to get a good aggre-
using twenty-four 13 mm (1/2 in.) special tendons and a gate connection between the core and the cover. Failure to
76-mm-thick (3 in.) cover, Eq. (1) would be superseded by provide a solid interface could result in spalled cover at the
Eq. (2), giving a required transverse reinforcement ratio of pile head during driving.
3.5%.
It would, therefore, be advantageous to show that levels of
This amount of spiral steel reinforcement would lead to displacement ductility adequate for competent foundation
congested construction; the spiral pitch in the plastic hinge performance can be achieved using levels of transverse
regions using D9.5 (0.348 in. [8.8 mm]) wire would be reinforcement commensurate with ACI 318 section 21.4.4,
18 mm (0.71 in.) (as opposed to 63.5 mm [25 in.] in Fig. without the restriction that the minimums of section 10.9.3
2). This is not acceptable, as ACI 318 specifies Eq. (3). be met.
4db ≤ s ≤ 6db (3) The research presented in this paper gives the results of
a parametric study using a pile modeled with transverse
where reinforcement slightly less than that required by ACI 318
section 21.4.4. The parameters that varied are axial load,
db = main bar diameter lateral soil stiffness, and the type of pile-cap connection.
66 S um me r 2 0 0 9 | PCI Journal
Analytical modeling
Finite-element analysis (FEA) for inelastic soil-pile inter- Node n-2 K n-2
action has been verified through field studies.9 FEA was
chosen due to its flexibility in representing both pile and Node n-3 K n-3
soil properties.
Node 5 K5
Budek et al. used nonlinear, inelastic constitutive models
for both the pile and soil and described them in detail.1 Node 4 K4
Briefly, the change in flexural stiffness of the pile as
inelastic action took place was extracted from the moment- Node 3 K3
curvature data as the slope of the moment-curvature curve.
Lateral load was applied in a series of steps. The elements’ Node 2
K2
flexural stiffnesses were modified as necessary after each Node 1
load step, according to the elements’ respective average
moment. Pile yield was defined by tendon stress reaching
85% of its ultimate value.
A bilinear soil model was used in which the lateral stiff- Figure 3. The prestressed pile analytical model is shown with indicated forces.
The notch in the cap of the prestressed model is for illustrative purposes, to allow
ness of the soil (that is, individual spring stiffness) was re- the top spring to be shown. Note: K1 = subgrade reaction modulus associated with
duced to one-fourth of its original value when the displace- node 1; K2 = subgrade reaction modulus associated with node 2; K3 = subgrade
ment at a node associated with a given spring exceeded reaction modulus associated with node 3; K4 = subgrade reaction modulus as-
25.4 mm (1 in.). sociated with node 4; K5 = subgrade reaction modulus associated with node 5; Kn
= subgrade reaction modulus associated with node n; Kn-1 = subgrade reaction
modulus associated with node n – 1.
Soil stiffness as expressed by the subgrade reaction modu-
lus K ranged from 3200 kN/m3 to 48,000 kN/m3
(20 kip/ft3 to 300 kip/ft3). The nondimensional system stiff- son.13 In these tests, 0.406-m-diameter (1.33 ft) piles were
ness KD6/D*EIeff is used to describe the properties of the embedded in soil placed under controlled conditions into a
soil-pile system. It includes cracked-section flexural stiff- purpose-built soil box and tested under combined axial and
ness EIeff of the pile shaft, and normalizes the pile diameter lateral load. Figure 2 shows the force-displacement enve-
D against a reference pile diameter D* of 1.83 m (6.0 ft). lope and predicted response from the first of these tests.
Budek et al. describes its derivation.10
The general trends of the test results—such as elastic stiff-
The FEA program used was written specifically for this ness, beginning of the softening branch, and incipient fail-
research program and was validated through comparison ure—are reproduced by the prediction. Some discrepancies
with laboratory testing10–12 and against results from in-place may be attributed to conservative modeling assumptions.
testing.9,13 First, the initial stiffness of the test pile was higher than
that predicted. This comes from the action of the small-
The laboratory testing program consisted of a series of 16 strain modulus of elasticity of the soil, which is estimated
tests,10 which included a number of precast, prestressed at four times the large-strain value.14 This factor-of-four
concrete pile shafts of a size and configuration similar to difference is quite close to that shown in Fig. 2. Second,
those examined in this study. The purpose of the investiga- the model underpredicted the maximum lateral force by
tion was to characterize the effect of external confinement about 12%. This can be directly attributed to the use of the
(as may be provided by competent soil) on the flexural 28-day concrete strength to generate the prediction, rather
ductility available in the pile shaft at the subgrade hinge than day-of-test strength. This was the prediction placed
(Fig. 1). The finite-element model in this study successful- on the plotter at the test site.
ly predicted pile-shaft response under the imposed loading.
Use of the nondimensional system stiffness term KD6/
The analytical model was further validated by comparison D*EIeff was also validated through comparison with this
with a series of four in-place tests of free-head reinforced in-place testing program. One observation from previous
concrete pile-columns performed by Chai and Hutchin- analytical work was that the center of the plastic hinge
0.50
0.00
0 5 10 15 20 25 30 35 40
Equivalent system stiffness
Figure 4. This graph compares the predicted and observed plastic-hinge depths from in-place pile tests. Note: D = pile diameter; H = depth of plastic hinge.
in the pile shaft (that is, the subgrade hinge) would move center of the plastic region after testing, fall almost directly
toward the surface as plasticity progressed from the origi- on the curves.
nal location of the point of maximum subgrade moment.
Its terminal location when the ultimate inelastic capacity In this study, the structure was modeled as a 0.61-m-
of a free-head pile is known (in which the subgrade hinge diameter (2 ft) continuous pile. Depth to the pile base was
controls response) can therefore be predicted. It is consis- set at 24.4 m (80 ft), which is equal to 40D and exceeds
tently at a depth of 70% of the point of maximum subgrade the depth corresponding to long-pile response. The pile
elastic moment, a value that is insensitive to either soil head was assumed to connect to a cap at ground level.
stiffness or above-grade height of the pile head. The actual Axial load varied from zero to 0.4 f c' Ag to represent forces
depth is, of course, a function of both of these variables. from global overturning moments transferred from the
superstructure.
Figure 4 shows curves predicting subgrade hinge depths
for the above-grade heights tested in the relevant range of The pile section used in the analysis was a typical 0.61-m-
nondimensional system stiffness for the four tests con- diameter (2 ft) round pile as used in California. Figure
ducted by Chai and Hutchinson. The depths of the centers 2 shows the details. A transverse reinforcement ratio of
of the plastic hinges, determined by digging down to the 1% was chosen for this study, slightly less than the 1.2%
that would result from the use of Eq. (1). Nominal effec-
tive section prestress after losses was 9.3 MPa (1.35 ksi).
24 with 13.2-mm-diameter tendons Figure 5 shows the section geometry.
1860 MPa ultimate
1302 MPa yield W11 A82 D9.5, 565 MPa nominal
Prestressed at 1060 MPa at 63.5 mm pitch Three different types of pile–pile cap connections were
considered: pile head embedded in cap, tendons embed-
76.2 mm ded in cap, and tendons stressed through cap. Each was
examined with and without the presence of mild-steel
f 'c = 41.3 MPa
longitudinal bars providing dowel reinforcement through
the pile-cap connection and down through the area of the
0.61 m subgrade plastic hinge (Fig. 4). The piles’ flexural respons-
es were analyzed using the Mander15 model for confined
Figure 5. This drawing illustrates the prestressed concrete pile section.
concrete modified for prestressed sections. Modifications
Note: fc' = concrete compressive strength; D9.5 = 8.8 mm. 1 mm = 0.394 in.; 1 m allowed for appropriate positioning of tendons with the
= 3.28 ft; 1 MPa = 0.145 ksi.
68 S um me r 2 0 0 9 | PCI Journal
correct level of prestressing force applied through the sec- was handled similarly. Active prestress went from zero
tion analysis, which produced the moment-curvature data to its full value over the transfer length, and the effect of
used as input for the FEA. stressing the tendons was modeled by applying an extra
increment of axial load to each section (in addition to the
Previous experimental work6,16–20 has shown that similar design load), such that the combination of this additional
pile-cap connections can support ductile response up to a axial and prestress load would remain constant (that is,
displacement ductility level of six. 40% of prestress at 40% of db was supplemented by an
axial load equivalent to 60% of the prestressing force).
Modeling of the different connections was addressed
through variation of either section prestress or axial load Strain penetration of the longitudinal reinforcement into
in calculating moment-curvature data. For the pile head the cap is an important part of modeling. In the response of
embedded in the cap, full prestress was assumed at the actual structures it permits a larger rotation at the pile-cap
bottom of the cap (that is, full transfer was assumed at the connection. It was modeled in this study by decreasing the
interface) (Fig. 6). To model embedment of the tendons, stiffness of the pile’s top element as yielding of the tensile
effective prestress was assumed to go from zero at the top reinforcement occurred.
of the pile to its full value at the end of the transfer length
of 115db.21,22 This part of the pile was, therefore, modeled For plastic hinges forming against supporting members,
in 10 sections, adding 10% of the effective prestress each such as footings or cap beams, theoretical and experi-
time. Therefore, the input data for this case consisted of 11 mental studies have led to the development of Eq. (4) for
sets of moment-curvature data (output from the Mander plastic hinge length lp.8
model analysis) applied to the relevant section of the pile.
lp = 0.08L + 0.022fydbl (4)
The practice of stressing the tendons through the pile cap
0% prestress
115db
100% prestress
Full
prestress
Figure 6. Three prestressed concrete pile–pile cap connections were considered in this study and are shown without reinforcing steel dowels. Note: The nominal embed-
ment length of a pile head embedded in cap is two pile diameters. db = main bar diameter.
0.7
0.6
0.5
Paxial = 0.2f 'c Ag
0.4 Pile head embedded in cap
Tendons embedded in cap
0.3 Tendons stressed through cap
Pile head (with dowels) embedded in cap
0.2 Tendons and dowels embedded in cap
Tendons stressed through, dowels embedded
0.1
0.0
0 1 2 3 4 5 6 7 8 9 10 11 12
Nondimensional system stiffness KD /D*EIeff × 10
6 3
Figure 7. This graph compares Eq. (1) results with nominal design values for a prestressed pile with head embedded in cap with no nonprestressed, longitudinal reinforce-
ment. Note: Ag = gross section area; D = pile diameter; D* = reference pile diameter; EIeff = cracked-section bending stiffness of the pile; fc' = concrete compressive
strength; K = subgrade reaction modulus; Paxial = axial load.
45
K = 32,000 kN/m3 Ground level
40
Height above fixity/pile diameter
35
30
Paxial = 0
25
20
K = 32,000 kN/m3
15
10
5
yield
0
-1500 -1250 -1000 -750 -500 -250 0 250 500 750 1000 1250 1500
Moment, kN-m
Figure 8. This graph plots the moment versus height for a pile head embedded in cap with no reinforcing dowels. Note: fc' = concrete compressive strength; K = subgrade
reaction modulus; Paxial = axial load. 1 m = 3.28 ft; 1 kN = 0.225 kip.
70 S um me r 2 0 0 9 | PCI Journal
45
K = 32,000 kN/m3 Ground level
40
30
Paxial = 0.2f 'c Ag
25
20 K = 32,000 kN/m3
15
10
5
yield
0
-1500 -1250 -1000 -750 -500 -250 0 250 500 750 1000 1250 1500
Moment, kN-m
Figure 9. This graph plots the moment versus height for a pile head embedded in cap with no reinforcing dowels. Note: Ag = gross section area; fc' = concrete compres-
sive strength; K = subgrade reaction modulus; Paxial = axial load. 1 m = 3.28 ft; 1 kN = 0.225 kip.
Used in this context, Eq. (4) demonstrates one of the The response of piles to lateral loading is best introduced
inaccuracies associated with elastic analyses of piles: through examination of moment patterns. In the case of
when yielding is reached in the pile’s critical section, the a fixed-head pile, the maximum moment, which controls
structure softens and a lesser depth of soil needs to be mo- overall response, is generated at the pile-cap connection,
bilized, thus moving the point of contraflexure toward the and a secondary moment maximum forms below grade.
surface. If an appropriate correction is made, Eq. (4) gives Modeling inelastic pile response resulted in the formation
a reasonable prediction of lp. of a plastic hinge at the pile-cap connection, after which
30
Paxial = 0.4f 'c Ag
25 K = 32,000 kN/m3
20
15
10
5
yield
0
-1500 -1250 -1000 -750 -500 -250 0 250 500 750 1000 1250 1500
Moment, kN-m
Figure 10. This graph plots the moment versus height for a pile head embedded in cap with no reinforcing dowels. Note: Ag = gross section area; fc' = concrete compres-
sive strength; K = subgrade reaction modulus; Paxial = axial load. 1 m = 3.28 ft; 1 kN = 0.225 kip.
the moment was redistributed into the shaft. This resulted Analyses of other end conditions and considering the
in the formation of a secondary hinge in the shaft. Figures presence of dowels show similar trends. Ultimate moment
8 through 10 demonstrate this, and the effect of varying versus height for three levels of axial load Paxial (0,
soil stiffness and axial load on pile response is represented 0.2 f c' Ag, and 0.4 f c' Ag) and the yield moment for the pile
for the case of the pile head embedded into the cap. shafts is represented. Increasing soil stiffness increased the
45
Ground level
40
Height above fixity/pile diameter
35
30
25
Paxial = 0.2f 'c Ag
K = 25,600 kN/m3
20 Plastic analysis
Elastic analysis
15 (design strength)
Yield
10
0
-1500 -1250 -1000 -750 -500 -250 0 250 500 750 1000 1250 1500
Moment, kN-m
Figure 11. This graph plots the moment versus height in comparing plastic and elastic analyses for a pile head embedded in cap with reinforcing dowels. Note: The plastic
analysis implies the development of a subgrade hinge. Ag = gross section area; fc' = concrete compressive strength; K = subgrade reaction modulus; Paxial = axial load.
1 m = 3.28 ft; 1 kN = 0.225 kip.
72 S um me r 2 0 0 9 | PCI Journal
45
Ground level
40
30
Paxial = 0.2f 'c Ag
25 K = 25,600 kN/m3
20
Plastic analysis
15 Elastic analysis
(design strength)
10
0
-1500 -1250 -1000 -750 -500 -250 0 250 500 750 1000 1250 1500
Moment, kN-m
Figure 12. This graph plots the shear versus height in comparison of elastic and inelastic analyses for pile head embedded in cap with reinforcing dowels. Note: Ag = gross
section area; fc' = concrete compressive strength; K = subgrade reaction modulus; Paxial = axial load. 1 m = 3.28 ft; 1 kN = 0.225 kip.
10
8
Subgrade hinge depth/pile diameter
0
0 1 2 3 4 5 6 7 8 9 10 11 12
Nondimensional system stiffness KD 6/D*EIeff × 103
Figure 13. This graph shows the depth of maximum subgrade moment of the three pile–pile cap connections tested with no dowel reinforcement. Note: Ag = gross sec-
tion area; D = pile diameter; D* = reference pile diameter; EIeff = cracked-section bending stiffness of the pile; fc' = concrete compressive strength; K = subgrade reaction
modulus; Paxial = axial load.
10
8
Subgrade hinge depth/pile diameter
0
0 1 2 3 4 5 6 7 8 9 10 11 12
Nondimensional system stiffness KD /D*EIeff × 10 6 3
Figure 14. This graph shows the depth of maximum subgrade moment of the three pile–pile cap connections tested with dowel reinforcement. Note: Ag = gross section
area; D = pile diameter; D* = reference pile diameter; EIeff = cracked-section bending stiffness of the pile; fc' = concrete compressive strength; K = subgrade reaction
modulus; Paxial = axial load.
74 S um me r 2 0 0 9 | PCI Journal
45
K = 48,000 kN/m3
40
35
Height above fixity/pile diameter
30
25
20 K = 32,000 kN/m3
15
Pile head embedded in cap
Tendons embedded in cap
10
Tendons stressed through cap
0
-1500 -1250 -1000 -750 -500 -250 0 250 500 750 1000 1250 1500
Moment, kN-m
Figure 15. This graph compares ultimate inelastic moment patterns for different pile–pile cap connections with no reinforcing dowels. Note: Ag = gross section area; fc' =
concrete compressive strength; K = subgrade reaction modulus; Paxial = axial load. 1 m = 3.28 ft; 1 kN = 0.225 kip.
45
K = 48,000 kN/m3
40
35
Height above fixity/pile diameter
30
25
20
K = 32,000 kN/m3
15
Pile head embedded in cap
Tendons & dowels embedded
10 Tendons stressed through cap
dowels embedded
5
Paxial = 0.2f 'c Ag
0
-1500 -1250 -1000 -750 -500 -250 0 250 500 750 1000 1250 1500
Moment, kN-m
Figure 16. This graph compares ultimate inelastic moment patterns for different pile–pile cap connections with reinforcing dowels. Note: Ag = gross section area; fc' =
concrete compressive strength; K = subgrade reaction modulus; Paxial = axial load. 1 m = 3.28 ft; 1 kN = 0.225 kip.
4
Displacement ductility capacity
0
0 1 2 3 4 5 6 7 8 9 10 11 12
Nondimensional system stiffness KD6/D*EIeff × 103
Figure 17. This graph shows the displacement ductility capacity versus the nondimensional system stiffness for different pile–pile cap connections with no reinforcing
dowels. Note: 3200 kN/m3 < K < 48,000 kN/m3 (20 kip/ft3 < K < 300 kip/ft3). Ag = gross section area; D = pile diameter; D* = reference pile diameter; EIeff = cracked-
section bending stiffness of the pile; fc' = concrete compressive strength; K = subgrade reaction modulus; Paxial = axial load.
4
Displacement ductility capacity
0
0 1 2 3 4 5 6 7 8 9 10 11 12
Nondimensional system stiffness KD /D*EIeff × 10
6 3
Figure 18. This graph shows the displacement ductility capacity versus the nondimensional system stiffness for different pile–pile cap connections with reinforcing dowels.
Note: Ag = gross section area; D = pile diameter; D* = reference pile diameter; EIeff = cracked-section bending stiffness of the pile; fc' = concrete compressive strength; K =
subgrade reaction modulus; Paxial = axial load.
76 S um me r 2 0 0 9 | PCI Journal
on moment patterns for these configurations. Curves are Prestressed concrete piles can be considered to be ductile
shown for both soft and stiff soil at a moderate axial load. structural elements. This has been shown experimentally
The greatest demand on the pile shaft was from embedding through previous research and has been verified through
the tendons into the pile cap (Fig. 15) due to the reduced the analytical work described in this study. With appropri-
flexural strength of the embedded tendon condition. In ate detailing, prestressed concrete piles may be used as
this case, prestress was assumed to be developed over the energy-dissipating elements in structures in which ductile
transfer length of 115db, beginning at the bottom of the cap behavior may be demanded. Detailing that would allow
and extending upward into the cap. The effective prestress sufficient ductility capacity can be specified in both rein-
force at the connection is, therefore, zero, which gives a forcement specification (through relevant design codes)
lower strength and flexural stiffness over the plastic hinge and location (through knowledge of the location of the
length of 0.5D at the pile-cap connection, requiring greater point of maximum subgrade moment).
mobilization of the pile shaft. The transfer length was
about 2.2 pile diameters, so the active prestress over the The ductility capacity of prestressed concrete piles is
hinge length of 0.5D did not exceed 30% of total prestress. limited by the performance of the soil-pile system. Con-
sideration of ductile response of individual parts (pile-cap
The presence of reinforcing dowels in the three categories connection or pile shaft) may be misleading. Ductility is
of pile–pile cap connection reduced the differences between affected by both the connection type chosen and axial load.
subgrade moment maxima (Fig. 16). In both cases, the dif- The connection type is intrinsic to the structure of which
ferences between subgrade moment maximums among the the piles form a part, but axial load may be subject to
different end conditions were larger for softer soils. variation in a given pile.
78 S um me r 2 0 0 9 | PCI Journal
22. Ghosh, S. K., and M. Fintel. 1986. Development
Length of Prestressing Strands, Including Debonded
Strands and Allowable Concrete Stresses in Preten-
sioned Members. PCI Journal, V. 31, No. 5 (Septem-
ber–October): pp. 38–57.
Notation
D = pile diameter
fy = yield strength
s = spiral pitch
80 S um me r 2 0 0 9 | PCI Journal