EXPERIMENTAL p-y M O D E L FOR SUBMERGED,

STIFF CLAY
By Todd W. Dunnavant, 1 Associate Member, ASCE,
and Michael W. O'Neill,2 Fellow, ASCE
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

ABSTRACT: A series of full-scale, cyclic, lateral load tests on instrumented piles

of varying diameter at a test site in submerged, overconsolidated clay is described.
A p-y model is derived from the tests that defines the effect of pile diameter,
relative pile-soil stiffness, and number of load cycles on the unit load-deflection
relationship for soil of the type encountered at the test site. The model is partic-
ularly well-suited to large.-diameter piles. It reflects observations that there is a
threshold deflection below which cyclic soil degradation does not occur and the
consequences of the fact that degradation was primarily caused by gap formation,
intensified by hydraulic scour. Rest periods between loading events did not result
in recovery of soil resistance and in fact appeared to produce further degradation
through swelling.

INTRODUCTION

The analysis of laterally loaded piles is frequently carried out by digital

computer with the use of uncoupled, nonlinear, load transfer curves, called
"p-y curves," to represent the soil behavior (Matlock 1970). These "curves"
are mathematical relationships of soil reaction per unit length of pile versus
lateral pile deflection and vary with position along the pile. Criteria for p-
y curves based on commonly measured soil properties have been developed
for different categories of soil: sand (e.g., Reese et al. 1974), soft clay (Ma-
tlock 1970), heavily overconsolidated ("stiff") clay above the water table
(Reese and Welch 1975), and stiff clay below the water table (Reese et al.
1975), and for different modes of loading (viz., static and cyclic). The prin-
cipal criteria in design use today were derived from analysis of tests on full-
sized piles, both driven and bored. Unfortunately, the data base for such
tests is small, so that uncertainty remains in the adequacy of a given criterion
for a particular application. Uncertainties associated with using the available
criteria were analyzed statistically for cohesive soils by Gazioglu and O'Neill
(1984) and for granular soils by Murchison and O'Neill (1984). Analyses
presented by O'Neill and Gazioglu (1984) indicated that the criterion with
the greatest uncertainty is that for submerged, stiff clay (Reese et al. 1975),
which was developed using results of tests performed in expansive soil that
may have imbibed water during the course of testing and which degraded
very severely during cyclic loading.
This paper describes an independent set of full-scale lateral pile load tests
conducted in submerged, stiff clay of different geological character than the
aforementioned tests. The intent is to add to the data base from which p-y
models for design may be derived or tested. A new p-y criterion is proposed,
based on the analysis of these tests, which may be more appropriate than
'Sr. Res. Engr., Exxon Production Res. Co., Houston, TX 77027.
2
Prof., Dept. of Civ. Engrg., Univ. of Houston, Houston, TX 77004.
Note. Discussion open until June 1, 1989. To extend the closing date one month,
a written request must be filed with the ASCE Manager of Journals. The manuscript
for this paper was submitted for review and possible publication on July 27, 1987.
This paper is part of the Journal of Geotechnical Engineering, Vol. 115, No. 1,
January, 1989. ©ASCE, ISSN 0733-9410/89/0001-0095/$1.00 + $.15 per page.
Paper No. 23120.

95

J. Geotech. Engrg., 1989, 115(1): 95-114

 UNDRAINED SHEAR STRENGTH, Su, kPa

0 50 100 150 200 250

Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

INTERPRETED .
SHEAR S T R E N G T H - * ^
PROFILE

t0 . (CONE PENETROMETERS, A\

O
-J
UJ
ffi
I
a.
Q

A UU TRIAXIAL TEST
12 M CTD TRIAXIAL TEST

FIG. 1. Profile of Undrained Shear Strength

the previous criterion for submerged, heavily overconsolidated clay at equi-

librium moisture content before pile installation.
The primary objectives of the testing program were to investigate the ef-
fects of scale, including pile diameter and relative pile-soil stiffness, and
number of loading cycles on the p-y relationships.

SITE CONDITIONS

The test site was the University of Houston Foundation Test Facility. The
soils at the site are natural, overconsolidated, saturated clays of the Beau-
mont formation (Mahar and O'Neill 1983). The soils in the zone of interest
are borderline CL to CH clays and are moderately jointed with small, closely
spaced, discontinuous slickensides and some isolated sand seams and car-
96

J. Geotech. Engrg., 1989, 115(1): 95-114

 UNIT WEIGHT, KG/M 3

1500 2000
~l i ' o Ho ^
o bo
o|o
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

olio

INTERPRETED
PROFILE

5
o- j
LU
m
x
i- TOTAL
Q.
LU UNIT
• WEIGHT
DRY
UNIT
WEIGHT
10

FIG. 2. Profile of Total and Effective Unit Weights

bonate nodules. Shear strength profiles developed from undrained triaxial

compression tests, interpreted results of cone penetrometer (CPT) soundings
(designated C) and field vane shear tests (designated V) are shown in Fig.
1. The interpreted shear strength profile, upon which the proposed p-y cri-
terion is based, is also shown in Fig. 1. The profile to 5.5-m depth is based
on a visual average line through the triaxial test results. The trend to this
depth is supported by the in situ results, except from 1.5-m to 3.0-m depth.
It is believed that the in situ results in that interval were influenced by the
presence of sand seams and calcareous nodules. The profile below 5.5-m
depth, which is below the zone of principal lateral pile-soil interaction for
the piles tested, was selected to follow the trend of the CPT data because
triaxial test results exhibit significant scatter below this depth due to increas-
ing frequency of slickensides. The apparent lower-bound relation of the triaxial
results to the CPT results above 5.5 m is an artifact of data interpretation.
The factor that divides the cone tip resistance minus the total overburden
pressure to give undrained shear strength (Nk = 13.6) was developed through
correlations with very shallow vane shear tests (1 m or less) throughout the
test pit. That correlation factor also represents approximately a lower 50 per
cent confidence limit to UU triaxial-test-correlated data for soils deeper than
5.5 m from all cone soundings acquired historically at the test site, but it
may not be numerically appropriate for other depth ranges.
The piles were tested in a 0.6-m-deep pit flooded with fresh water. The
natural groundwater depth was 1.53 m below the pit; the flooding procedure
saturated the soil fully from the base of the pit to the water table.
97

J. Geotech. Engrg., 1989, 115(1): 95-114

 0.005 0.010 0.015 0.020

" 0.022

' L '•'
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

INTERPRETED \ o

9 I

• ! "
5
o • i •
• i
i
i
0.022
. •
•
•UU TRIAXIAL TEST
0 ~
« o
" ClU TRIAXIAL TEST

0.059

FIG. 3. Profile of e3(

Fig. 2 gives the unit weights of the test soil, and Fig. 3 presents the profile
of e50, the axial strain at one-half of the peak principal stress difference in
a monotonic, undrained triaxial compression test. The profiles were esti-
mated by applying visual fits to the data. Locations of the breaks in the
profiles were estimated by making use of knowledge regarding stratigraphic
breaks (Mahar and O'Neill 1983). The variability in the stress-strain prop-
erties of the upper soils at the test site is illustrated by a series of selected
UU laboratory triaxial compression test curves in Fig. 4.
The process by which the soil was preconsolidated (desiccation) has left
it with high lateral in situ effective stresses, with the at-rest earth pressure

200

FIG. 4. Typical UU Triaxial Compression Stress-Strain Curves

98

J. Geotech. Engrg., 1989, 115(1): 95-114

 TABLE 1. Characteristics of Test Piles
Wall
Pile Description/O.D. thickness
number penetration (m) (mm) EI (MN-m2) Other
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

(D (2) (3) (4) • (5)

1 Open-ended, 0.273 9.27 138 Plugged at 6.1 m
driven, steel pipe/
11.8 m
2 Open-ended, 1.22 15.9 358,600 Did not plug
driven, steel pipe/
11.4 m
3 Bored, reinforced — 2.53 x 107 24 #11 bars on
concrete/11.4 m (above 0.76 m) 1.65-m-diam.
1.98 x 107 circle and
(below 0.76 m) 12.7-mm-wall
steel casing to
depth of 0.76 m
Note: E = Young's modulus; / = moment of inertia; Piles 1 and 2 were driven into a
pit that had been flooded for five months prior to driving. Pile 3 was bored in a dry area
that was then excavated and flooded for 8 months prior to testing. All flooding water was
fresh water. Heads were unrestrained.

coefficient ranging from 3 at a depth of 0.6 m to 1 at a depth of 10.7 m.

The soil has a sensitivity of approximately 2.0, and laboratory pinhole dis-
persion tests indicate that the clay is nondispersive. Further details concern-
ing the soil behavior, including in situ and laboratory moduli, are given by
Dunnavant (1986).

TEST PILES

Three piles of widely varying diameter and flexural stiffness, described in

Table 1, were tested. Based on measurements taken after pile driving was
complete, Pile 1 plugged at a depth of 6.1 m while driving; Pile 2 did not
plug. These observed values of internal plug depth are supported by cal-
culations of internal plug length required for equilibrium between toe and
internal shaft resistance. In Pile 1 computed plug equilibrium is reached at
a penetration of about one-half of the pile penetration; hence, it is probable
that Pile 1 plugged only after being driven to a penetration of greater than
5 m and that only minor disturbance occurred in the soil around the driven
piles in the zones of highest lateral resistance, which may permit a reason-
able comparison of their behavior with that of the large bored pile (Pile No.
3).

INSTRUMENTATION AND CALIBRATION

Piles 1 and 2 were instrumented for the measurement of bending moment

with full electronic strain gage bridges on a vertical spacing of 305 mm near
the surface, as shown in Fig. 5. Pile 3 was instrumented with an inclinometer
situated near the neutral axis of the pile. The load on all three test piles was
99

J. Geotech. Engrg., 1989, 115(1): 95-114

 3@1.53 M
302 MM 2 1 1 MM

—/MV. m * *—
T
11 BRIDGES
SEISMIC CASING
( 2 . 3 0 M E&W OF
A/AM

O 3 0 5 MM "LOADING LINE)
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

4 BRIDGES
@ 610 MM

2 BRIDGES
@1220 MM
4 BRIDGES
<S>1220 MM

INCLINOMETER
TUBE
• - 1 2 2 0 MM , „ „ „ ...-r 1

• PIEZOMETER
(IN LINE W/LOAD)

FIG. 5. Test Pile Elevations

measured by using electronic load cells, of 1.3-kN accuracy for Pile 1 and
6.7-kN accuracy for Piles 2 and 3. The cyclic load controller was driven by
signals from LVDT's at the loading level. Linear potentiometers of 0.1-mm
accuracy were placed above the level of load application to permit mea-
surement of pile-head rotation. Data were acquired by a digital computer at
the test site, except for the inclinometer data, which were acquired manually.
The strain gage circuits for Piles 1 and 2 were calibrated by loading the
piles in the shop in simple beam bending and relating the applied moment
at each gage station to the strain gage voltage response. Calibration constants
for the gages on Pile 1 were developed by making eleven separate calibration
loadings; six loadings were used for Pile 2. Each pile was exercised 10 times
prior to applying the calibration loadings, which included variations in the
inclination of the load to the plane of the gages. Calibration constants for
both piles were consistent between calibration runs. The constants for Pile
1 were all within 5.3% of constants that would be computed from beam
theory. The constants for Pile 2 were within 6.0% of those computed from
beam theory, except within 3 m of the pile head, where the proximity to
the pile-head reaction and egging of the cross-section influenced the results
significantly. Pile 2 was therefore calibrated in the field as a cantilever after
the load tests to define the calibration constants with the field loading fixture
in place. The field calibration constants, which were used in the data re-
duction, ranged from 27% (at the surface) to 98% (at 3.3-m depth) of the
shop values.
Soil instrumentation consisted of an array of pore pressure cells and seis-
mic instrument casings near Pile 2 (Fig. 5). The latter instruments were used
to infer relative magnitudes of softening in the soil mass in front of the pile
during cyclic loading.

SITE LAYOUT

Site layout details are provided by Dunnavant (1986). Piles 1 and 2 were
located within 6 m of each other, and Pile 3 was located about 15 m from
100

J. Geotech. Engrg., 1989, 115(1): 95-114

 them. Soil investigations were concentrated within 5 m of Piles 1 and 2, but
one soil boring was made adjacent to Pile 3 to confirm the continuity of the
profile. For Piles 1 and 2 the nearest 0.6-m-high wall of the test pit was 6
and 4 diameters away, respectively. Boundary effects are considered to have
been minimal for these piles, which were the ones that were used primarily
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

in the criterion development. The nearest boundary to Pile 3 was 1.3 m, in

the direction of loading, but, based on the observed localized nature of soil
failure around each of the test piles, it is considered that this relatively close
boundary influenced Pile 3 very little.

TESTING PROCEDURES

All tests were conducted with a water depth in the test pit of about 150
mm to simulate offshore or river-bottom conditions. Most pile loads were
applied using two-way displacement-controlled cycling under free-head con-
ditions. Cycle periods of from 1 to 100 s were used. Fixed- or restrained-
head tests were not performed, but, based upon conclusions reported by Ma-
tlock (1970), the p-y relations derived from unrestrained-head conditions should
be essentially equivalent to those for fixed-head conditions.
General features of the loading schedules for the three piles are presented
in Table 2. Specific items for each pile are summarized below.

Pile 1
As is indicated in Table 2, three loading series—"primary," "healing"
and "sand"—were performed, p-y criteria were developed using the results
of the primary series, while the healing series was performed after a gap
had developed around the pile to investigate the effects of the lapse of time
between major loading events. In the sand series, pile-soil gaps were filled
with fine mortar sand to investigate changes in pile behavior caused by fill-
ing the gaps. Localized plastic yielding of the pile, assessed using strain
gage readings, occurred during the last increment of the loading prior to the
healing series (head deflection = 1 0 4 mm). Data from this loading were not
used in synthesizing the p-y criterion.

Pile 2
Some one-way cycling (Table 2) was applied manually because the pri-
mary closed loop hydraulic system did not have adequate capacity to displace
the pile far enough to develop the complete p-y relations to a significant
depth. Primary, healing, and sand tests were also conducted, and a second
set of primary tests was conducted after the healing tests.

Pile 3
This pile had an irregular loading history over a period of more than one
year. Healing and sand tests were not conducted. The large-amplitude de-
flections were applied using a one-way, manually controlled jack. Based
upon large changes in the pile slope measured with the inclinometer, the pile
was observed to be cracked at a depth of about 6 m for pile loads greater
than 1,330 kN.
In all of the tests the behavior measured on the first cycle was interpreted
as the "static" behavior, although at some deflection levels cycling at lower
deflections undoubtedly influenced the pile response. This assumption per-

101

J. Geotech. Engrg., 1989, 115(1): 95-114

 TABLE 2. Pile Loading Schedules
Pile Head deflection Number
number levels (mm) of cycles Dates Remarks
(1) (2) (3) (4) (5)
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

1 ±1.0, 2.2, 3.2, 101, 260, 200, Feb. 2-Feb. 28, Primary loading
3.8, 5.1, 6.1, 102, 221, 114, 1983 series
7.6, 10.2, 110, 106, 25,
14.2, 17.8, 103, 107, 3,
25.0, 30.5, 37, 104, 7, 3, 100,
48, 76, 84, 10
104
1 ±1.0, 25, 37 104, 102, 100 May 3, 1983 Healing loading
series
1 ±1.0, 37 103, 100 May 5, 1983 Sand loading
series
2 ±0.6, 1.4, 3.8, 1, 107, 203, April 7-15, 1983 Primary loading
6.6, 10.7, 13, 105, 205, 1, series
17 106
2 ±8.9, 11.2, 20.3 104, 104, 100 May 9 - 1 1 , 1983 Healing loading
series
2 36, 46, 56 15, 15, 16 May 16, 1983 Second primary
loading series
(one-way)
2 36, 51 15, 15 May 19, 1983 Sand loading
tests
3 ±0.3, 0.6, 0.8, 100, 200, 200, May 16, 1984 One year after
1.8, 1.7 200, 200 pile cast
3 ±0.3, 1, 0.8 100, 100, 1 Oct. 18, 1984 Load frame
buckled
3 Loads from 140 800 total Dec. 13, 1984 See note
to 805 kN
3 2.1, 27, 38, 47 20, 1, 20, 1 July 24, 1985 One-way loading
Note: Loads for Dec. 13 test produced deformations comparable to those for May 16
test. These data were not used in development of the p-y criteria.

mitted the development of criteria for both "static" p-y curves and envelopes
for "cyclic" p-y curves (Matlock, 1970) from the data at the test site.
DATA REDUCTION AND ANALYSIS PROCEDURES
Raw p-y curves were developed from the measured moment versus depth
relations, pile-head deflection and pile-head slope through double differen-
tiation (for soil reaction per unit length, p) and double integration (for cor-
responding deflection, v) of local cubic polynomials that fit 5 to 9 contiguous
data points. The resulting raw p-y curves were then observed for character-
istic shape and variations with depth and pile diameter of ultimate soil re-
sistance and cyclic p-y curve degradation characteristics. Several trial math-
ematical formulations were developed and fitted to these raw p-y curves and
examined with respect to their ability to back-predict the load-deflection,
load-rotation, moment-depth (pipe piles), and rotation-depth (bored pile) re-
lationships. Further details are given by Dunnavant (1986).

102

J. Geotech. Engrg., 1989, 115(1): 95-114

 RESULTS

Pile-head, load-deflection results are summarized in Fig. 6, and typical

raw, Cycle-1 p-y curves (for Pile 1) are shown in Fig. 7. Some scatter is
evident in both the load-deflection data and the p-y curves for Cycle 1, prob-
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

ably due to the effects of prior cycling at lower loads. Upper bounds for
both of these relationships were inferred as the best possible representations
of "static" behavior.
It is evident from the data for Piles 1 and 2 that significant degradation
due to cyclic loading did not occur in the load-deflection curves until the
head deflection reached about one percent of the pile diameter. The rapid
rate of degradation at larger deflections appeared to be associated with the
formation of a permanent gap around the pile due to (1) Plastic deformation
of the soil; and (2) hydraulic scour during cyclic loading. Shear wave ve-
locities 1.53 m in front of Pile 2 decreased from 192 to 177 m/s to a depth
of 2.5 pile diameters during the course of the testing, and particularly after
the pile had deflected in excess of 1% of its diameter, suggesting a reduction
in shear modulus in the soil mass of about 15% at that location. Excess pore
water pressures developed during loading at the piezometer locations (Fig.
5) were less than 7 kPa.

2000

/ PILE 3 (1.83 M 0)
1500 • CYCLE 1
A CYCLE 15
CREEP A CYCLE 20
(240 S) O CYCLE 100
O CYCLE 200
A VELOCITY - 1.3 MM/S
B VELOCITY = 6.4 MM/S
a.
<
in
x 1000
CO
Q
<
HI
I

500

0 25 50 75 100

PILE - HEAD DEFLECTION, MM

FIG. 6. Pile-Head Load-Deflection Curves

103

J. Geotech. Engrg., 1989, 115(1): 95-114

 150
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

A - - X - 1 . 8 2 M (DEPTH BELOW TEST PIT)

b i

f
I \U-—X-2.12 M

2
100 h
£0
S
X-1.51 M

Z
o
O
<
CC

o 50
CO

10 20 30 40
DEFLECTION, Y, MM
FIG. 7. Raw Static p-y Curves for Pile 1

A map of the gap zone around Pile 2 at the end of the second primary
test sequence (maximum pile deflection of 56 mm) is shown in Fig. 8. The
gap width was slightly less than the maximum pile deflection in the direction
of loading, suggesting either partial elastic recovery or lateral swelling, or
both. Larger zones were evident elsewhere around the perimeter, as shown,
and complete separation was visible on the sides of the pile parallel to the
direction of loading, apparently produced by the hydraulic scour. The prin-
cipal range of particle size of eroded soil was 0.01—0.07 mm.
The peak resistance values from smoothed, static, p-y curves, normalized
by the product of local undrained shear strength (su) and pile diameter (B)
(defined as Np, which can be viewed as a lateral bearing capacity factor) are
plotted versus depth in Fig. 9 for the tests on Piles 1 and 2. A common
104

J. Geotech. Engrg., 1989, 115(1): 95-114

 38 MM; 5 MM WIDE AT 1.5 M DEPTH
65 MM] 5 M M WIDE AT
1.1 M DEPTH
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

100 MM; 5 MM
WIDE AT
1.4 M DEPTH

25 MM; 5 MM WIDE AT 1.2 M DEPTH

TO ACTUATOR

FIG. 8. Map of Gap Zone; Pile 2 Following First Primary Loading Series

linear function specifying that relationship is also described, in which s,la =

averaged undrained shear strength between the surface and the depth in ques-
tion,CT'V= vertical effective stress and x = depth. The relation was estab-
lished using Np values for Pile 2, which was subjected to fewer significant
loading cycles than Pile 1 (Table 2), such that the "static" (first cycle) re-
sistances for Pile 2 were influenced less by previous cycling than were cor-
responding resistances for Pile 1 and were therefore more appropriate for
estimating static p-y curve parameters.
A typical raw cyclic p-y curve (Cycle 100, Pile 1, depth = 1.21 m) is
shown in comparison to the corresponding curve for Cycle 1 in Fig. 10.

Nc

FIG. 9. Np versus Depth

105

J. Geotech. Engrg., 1989, 115(1): 95-114

 60

50
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

40

S
5
z
3 0
o: CYCLE 100

20

10
©CYCLE 1 ACYCLE 20
* CYCLE 2 aCYCLE 50
•CYCLE 5 -CYCLE 100
oCYCLE 10 OCYCLE 200
0 i , , , ,
0 5 10 15 20

PILE DEFLECTION, Y, MM

FIG. 10. Raw p-y Curves for Cycle 1 and Cycle 100; Pile 1, Depth = 1.27 m

This result is typical of all depths for both piles in which moments were
measured. In particular, the initial slopes of the static and cyclic curves were
essentially identical, but the peak value was greatly reduced for the cyclic
relation. The effect of pre-cycling on soil resistance during subsequent load-
ings is evident in Fig. 10. Cycling at a local pile deflection of 6-7 mm, in
particular, significantly influenced the "static" resistance out to a deflection
of about 15 mm, as evidenced by the dip in both curves. It is believed that
the significance of that dip is that cycling at the deflection of 6-7 mm rep-
resented the onset of hydraulic scour.
The ratio of Ncm to N„ for Piles 1 and 2 is shown in Fig. 11, where Ncm
is the ratio for cyclic loading corresponding to Np for static loading at 100
cycles of applied load. The ratio is lower than unity to the maximum depth
to which Pmax could be measured. The relation appears to follow a common
function of depth for both piles. It is suspected that this relation is a function
of site-specific soil characteristics, especially susceptibility to hydraulic scour.
Each cyclic p-y curve appeared to reach a residual value of p, denoted pr.
The ratio of pr to pcm (= Ncm/s„B) for Pile 1, used in developing the p-y
criteria, is shown in Fig. 12.
An increase in number of loading cycles generally produced a decrease in
pcm according to the logarithm of the number of load cycles applied. That
observation is incorporated in the p-y criteria.
The healing series of tests did not produce strengthening of the p-y curves,
implying that the soil gap that formed was permanent. Furthermore, abrupt
softening is observed in the load-deflection data for Pile 2 after the prolonged
lapse period following initial formation of the gap after the first primary
loading series (Table 2), suggesting that exposure of the subsurface soil to
free water had caused swelling, reducing the soil's strength, or had increased
the soil's scour susceptibility. Likewise, the tests conducted with sand placed
106

J. Geotech. Engrg., 1989, 115(1): 95-114

 0.4 0.6 O.f
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

A PILE 1
9 PILE 2

FIG. 11. NCJNP versus Depth

into the gap did not produce significant strengthening, perhaps because the
upward velocity of water in the gap zone reduced the effective stress in the
sand to the degree that it liquefied. The use of a coarser sand may have been
more effective in this respect.
p-y CRITERIA
The optimum p-y criteria deduced from analysis of the tests are described
in Eqs. 1-5 (static criterion) and Eqs. 6-11 (modifications for cyclic load-

5
o

FIG. 12. Pr/Pcm versus Depth

107

J. Geotech. Engrg., 1989, 115(1): 95-114

Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

•CYCLE 1 ACYCLE 20
A CYCLE 2 QCYCLE 50
• CYCLE 5 *CYCLE 100
OCYCLE 10 OCYCLE 200

5 10 16 20

PILE DEFLECTION, Y, MM

FIG. 10. Raw p-y Curves for Cycle 1 and Cycle 100; Pile 1, Depth = 1.27 m

This result is typical of all depths for both piles in which moments were
measured. In particular, the initial slopes of the static and cyclic curves were
essentially identical, but the peak value was greatly reduced for the cyclic
relation. The effect of pre-cycling on soil resistance during subsequent load-
ings is evident in Fig. 10. Cycling at a local pile deflection of 6-7 mm, in
particular, significantly influenced the "static" resistance out to a deflection
of about 15 mm, as evidenced by the dip in both curves. It is believed that
the significance of that dip is that cycling at the deflection of 6-7 mm rep-
resented the onset of hydraulic scour.
The ratio of Ncm to Np for Piles 1 and 2 is shown in Fig. 11, where Ncm
is the ratio for cyclic loading corresponding to Np for static loading at 100
cycles of applied load. The ratio is lower than unity to the maximum depth
to which Pmax could be measured. The relation appears to follow a common
function of depth for both piles. It is suspected that this relation is a function
of site-specific soil characteristics, especially susceptibility to hydraulic scour.
Each cyclic p-y curve appeared to reach a residual value of p, denoted pr.
The ratio of pr to pcm (= Ncm/s„B) for Pile 1, used in developing the p-y
criteria, is shown in Fig. 12.
An increase in number of loading cycles generally produced a decrease in
pcm according to the logarithm of the number of load cycles applied. That
observation is incorporated in the p-y criteria.
The healing series of tests did not produce strengthening of the p-y curves,
implying that the soil gap that formed was permanent. Furthermore, abrupt
softening is observed in the load-deflection data for Pile 2 after the prolonged
lapse period following initial formation of the gap after the first primary
loading series (Table 2), suggesting that exposure of the subsurface soil to
free water had caused swelling, reducing the soil's strength, or had increased
the soil's scour susceptibility. Likewise, the tests conducted with sand placed

106

J. Geotech. Engrg., 1989, 115(1): 95-114

 Ncn'Np

0.4 o.e
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

A PILE 1
9 PILE 2

FIG. 11. NCJNP versus Depth

into the gap did not produce significant strengthening, perhaps because the
upward velocity of water in the gap zone reduced the effective stress in the
sand to the degree that it liquefied. The use of a coarser sand may have been
more effective in this respect.
p-y CRITERIA
The optimum p-y criteria deduced from analysis of the tests are described
in Eqs. 1-5 (static criterion) and Eqs. 6-11 (modifications for cyclic load-

5
o

FIG. 12. Pr/Pcm versus Depth

107

J. Geotech. Engrg., 1989, 115(1): 95-114

 a.
ui
o P = P„ FOR Y = 8Y S
z
<
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

co
CO
HI
cc
_J

o
CO
o
UJ
N
_l
<
DC
o
z
"0 1 8

NORMALIZED DEFLECTION, Y/Y 5 0

FIG. 13. Static p-y Curve (Site Criterion)

a.
Q-

LU P/Pu = 1.02 T A N H [ o . 5 3 7 ( Y / Y 5 0 ) '']

o
z
£ 1.0
CO
CO
HI
DC " " c m ' °u
o
CO
Q
UJ
N
_l
<
5
d
O
z
°Yom/Y50 12
NORMALIZED DISPLACEMENT, Y/Y 5 0

FIG. 14. Cyclic (Envelope) p-y Curve (Site Criterion)

ing). Figs. 13 and 14 describe the criteria graphically. Relative pile-soil stiff-
ness (KR, Eq. 5) is included as a parameter to account for elastic coupling
of the p-y curves. The inclusion of relative stiffness is supported by ana-
lytical results reported by Baguelin et al. (1977), which show that the mod-
ulus of subgrade reaction (initial p-y curve stiffness) increases for piles in
an elastic continuum as KR increases. The term y50, the deflection corre-
sponding to one-half of the /Jmax (static), is therefore not linearly dependent

108

J. Geotech. Engrg., 1989, 115(1): 95-114

Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

D
<
O

Q
<
ID
I

MEASURED
DATA TREND
SITE CRITERION
SOFT CRITERION
STIFFA CRITERION
S T I F F B CRITERION

40
PILE-HEAD DEFLECTION, MM

FIG. 15. Comparison of Pile-Head Load-Deformation Predictions; Cycle 100,

Pile 1

on the pile diameter, as has been suggested by others (e.g., Matlock 1970).
, . 0.70-

p= 1.02p„tanh 0.537(— J , y < 8 y50 (1)

where
Pu = NpsuB (2)
crv x
Np = 2 + — + 0.4 - < 9 (3)
Sun B
y50 = 0.0063e 50 S^« 0.875 (4)
and
EI
KR = (5)
E~I\
where Es = a strength-correlated soil modulus; and L = the pile length,
limited numerically to 3B(EI/ESBY2*6 (Gazioglu and O'Neill 1984) which
is the "critical pile length" beyond which the presence of additional pile
length has negligible effect on pile-head behavior. Np in Eq. 3 is limited
numerically to 9 based upon a consensus of theoretical solutions and earlier
p-y criteria. The maximum value that can be justified from the data is about
6, because the piles could not be deflected far enough to mobilize ultimate
resistances at depths for which Np greater than 6 is predicted (Fig. 9). The
cyclic load envelope (Fig. 13) is then defined by

109

J. Geotech. Engrg., 1989, 115(1): 95-114

Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

1000

z
o"
<
o
_J
O
< "HEALING" TEST, 24 DAYS
111 AFTER PRECEDING SERIES
X
I
500

MEASURED
- - SITE CRITERION
— SOFT CRITERION
— STIFF A CRITERION
— STIFF B CRITERION

10 20 30 40 50 60

PILE - HEAD DEFLECTION, MM

FIG. 16. Comparison of Pile-Head Load-Deformation Predictions; Cycle 1, Pile 2

"" 0.1 + 0 . 3 6 - < 0.76 (100 load cycles) (6)

N,

= 1- 0.45 - 0 . 1 8 - logN< 1 (N cycles) (7)

p I N

(8)
Pcm
1.02 tanh 0.537 (9)
Pu \yso
pr 0.14*
— = 0.5 + < 1 (100 load cycles) (10)
Pcm -*o

and

—) = 1 - [ 0 . 2 5 - 0 . 0 7 - ) log/V< 1 (N cycles) (11)

where x is in m and x„ = 1 m. Dunnavant (1986) gives details regarding

this development. It should be noted from the cycle-dependent form of the
cyclic criterion (Eqs. 7 and 11) that degradation continues with increasing
numbers of cycles. This is supported by the test data, which showed thatp-
y curve degradation continued for at least the 200 cycles of load applied in
some of the loading increments.

110

J. Geotech. Engrg., 1989, 115(1): 95-114

 1000
z
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

d
<
o
_i
Q
<
HI
» MEASURED ON CYCLE 1
"f 500 • MEASURED AT 100 CYCLES
UJ o MEASURED AT 15 CYCLES
SITE CRITERION
0.
SOFT CRITERION
STIFF A CRITERION
— x— STIFF B CRITERION

0 10 20 30 40 50 60
PILE-HEAD DEFLECTION, MM

FIG. 17. Comparison of Pile-Head Load-Deformation Predictions; Cycle 100,

Pile 2

Typical relationships computed for Piles 1 and 2 using a numerical model

for laterally loaded pile simulation (Reese and Sullivan 1980) are given in
Figs. 15-18. Computed behavior is shown using the new (Site) static and
cyclic criteria and using existing clay criteria: Soft (Matlock 1970), Stiff A
(Reese et al. 1975), and Stiff B (Reese and Welch 1975). The cyclic Site
p-y curves were formulated for the case of 100 loading cycles. The Site
criterion fits the Cycle-100 data extremely well for Pile 1 (Fig. 15). For Pile
2 (Figs. 16-18) the primary data are matched well except for the static and
Cycle 100 data that correspond to a 17-mm pile-head deflection and the
deflections corresponding to deflections greater than 25 mm. The discrep-
ancy at 17 mm is believed caused by pre-cycling effects on the test pile at
the 11-mm deflection, an interpretation supported by the proximity of the
17-mm point for Cycle 1 to the Cycle-100 curve computed from the Site
criteria (Fig. 17) and examination of detailed raw p-y data that are not pre-
sented here. It is believed that the large-displacement data (>25 mm) are
softer than predicted due to soil softening following an extended rest period.
These data were acquired 31 days after the first sequence of primary data,
following the healing tests (Table 2). Raw p-y data, not shown, reveal sig-
nificant softening in this zone, and it should be noted that the large-dis-
placement p-y data (second primary series) from Pile 2 were not used in the
formulation of the new p-y criterion. The development of the Site criterion
emphasized fitting the earlier data points, which were not significantly in-
fluenced by the rest period and consequent soil modification that apparently
occurred after the application of the 587 kN load.
The prediction of the behavior of Pile 3 with the Site criterion at Cycle
15 is described in Fig. 19. One set of predictions is based on an analysis of
the pile with an uncracked section, while another set was obtained using
section properties reduced for cracking effects. The second analyses were
111

J. Geotech. Engrg., 1989, 115(1): 95-114

 2 - WAY, „ I . 1 -WAV,
AUTOMATIC I MANUAL

/
1000 "
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

500
" ///
L V
.// • MEASURED AT 100 CYCLES
o MEASURED AT 15 CYCLES
SITE CRITERION
y SOFT CRITERION
— - STIFF A
— x— STIFF B CRITERION

MAXIMUM MOMENT, M - MN

FIG. 18. Comparison of Maximum Bending Moment Predictions; Pile 2

performed because inclinometer data indicated marked increases in pile cur-

vature at 5.2-7.0 m below the ground surface at loads greater than 619 kN.
This suggested concrete cracking in that zone. The Site criterion provides
an excellent prediction of pile-head behavior up to the point of crack initi-
ation, as does cyclic criterion Stiff C (Stevens and Audibert 1979), a large-
diameter-pile criterion that specifies y50 to be a function of B05. The Site
criterion also provides close predictions of pile-head behavior at deflections
exceeding 19 mm (load exceeding 1,200 kN), at which point significant
cracking had likely occurred. Based upon these and other comparisons (Dun-
navant 1986), particularly with measured bending moment diagrams, the Site
criterion appears to be the most appropriate p-y curve criterion for the test
soil for the wide range of test conditions considered.

UNCRACKED (U)

9 1000
CRACKED (C)

SITE
SOFT
STIFF A
STIFF C
o MEASURED

10 20 30 40

PILE-HEAD DEFLECTION, MM

FIG. 19. Comparison of Pile-Head Load-Deformation Predictions; Cycle 15,

Pile 3

112

J. Geotech. Engrg., 1989, 115(1): 95-114

 APPLICATION

The proposed criteria appear appropriate for soils similar to those at the
test site and for nondisplacement piles. It may be possible to extend the
criteria to other overconsolidated clay soils and to displacement piles by
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

employing site-specific, full-scale, uninstrumented lateral load tests. The shape

of the static p-y curve is most strongly influenced by the factor Np (Eq. 3),
the coefficient of the v50-expression in Eq. 4 and the exponent of (y/y50) in
Eq. 1. If a monotonic lateral load test is conducted to large displacement
(10% or more of the pile diameter), it may be possible to match the measured
head load-deflection relation computationally by varying these terms. For
example, increasing the coefficient of y50 softens the curve and reduces the
slope of the computed load-deflection curve for small, finite values of y,
which may be appropriate for a displacement pile in a clay more sensitive
than that at the test site. Shifting the function for Np by changing the constant
factor (2) in Eq. 3 may produce an improved prediction of displacement at
large load, and varying the exponent in Eq. 1 may further improve the entire
relation, since it controls the rate at which the p-y function approaches pu.
It should be understood that these factors all influence the entire load-de-
flection curve, so the proper combination must be obtained.
For cyclic loading Ncm is the most important factor. Its value can be ad-
justed on a site-specific basis to match measured cyclic load test results once
the aforementioned static factors have been established. It is not recom-
mended that other factors in either the static or cyclic criteria be varied with-
out extensive research.

CONCLUSIONS

An improved practical p-y criterion for submerged, stiff clay of the type
at the test site (overconsolidated, saturated and moisture stable clay with
closely spaced, discontinuous slickensides) was developed (Eqs. 1—11 and
Figs. 13 and 14). This criterion indicates slightly less stiff initial behavior
but lower post-peak degradation than the "Stiff A" criterion for submerged
stiff clays and, consequently, predicts considerably different pile-head be-
havior. The criterion appears particularly well-suited to the prediction of the
behavior of piles of very large diameter.
Test data indicated that appreciable cyclic degradation did not begin until
the pile-head displacements had reached about 0.01 B, but, once started, did
not appear to stabilize within 200 cycles. The principal source of degradation
was the development of a permanent gap around the piles, intensified by
hydraulic erosion. Rest periods between loadings after the opening of a gap
between the pile and the soil resulted in weakened soil response, although
this effect is not included explicitly in the proposed criterion.

ACKNOWLEDGMENTS

Financial support for the study was provided by Amoco Production Com-
pany, Chevron Oil Field Research Company, Exxon Production Research
Company, Gulf Oil Exploration and Production Company, Houston Lighting
and Power Company, Marathon Oil Company, McClelland Engineers, Inc.,
Mobil Field Research Laboratory, Pennzoil Exploration and Production

113

J. Geotech. Engrg., 1989, 115(1): 95-114

 Company, Raymond Technical Facilities, Inc., Shell Oil Company, and
Texaco, USA.

APPENDIX. REFERENCES
Downloaded from ascelibrary.org by Queens University Libraries on 05/02/16. Copyright ASCE. For personal use only; all rights reserved.

Baguelin, F., Frank, R., and Said, Y. H. (1977). "Theoretical study of lateral re-
action mechanism of piles." Geotechnique, 27(3), 405-434.
Dunnavant, T. W. (1986). "Experimental and analytical investigation of the behavior
of single piles in overconsolidated clay subjected to cyclic lateral loads." Thesis
presented to the University of Houston, at Houston, Texas, in partial fulfillment
of the requirements for the degree of Doctor of Philosophy.
Gazioglu, S. M., and O'Neill, M. W. (1984), "Evaluation of p-y relationships in
cohesive soils." Analysis and Design of Pile Foundations, J. R. Meyer, ed., ASCE,
San Francisco, Calif. 192-213.
Mahar, L. J., and O'Neill, M. W. (1983). "Geotechnical characterization of desic-
cated clay." ASCE 109(1), 56-71.
Matlock, H. (1970). "Correlations for design of laterally loaded piles in soft clay."
Preprints, Second Annual Offshore Tech. Conf, I, Houston, Texas, 577-594.
Murchison, J. M., and O'Neill, M. W. (1984). "Evaluation of p-y relationships in
cohesionless soils." Analysis and Design of Pile Foundations, J. R. Meyer, Ed.,
ASCE, San Francisco, Calif. 174-191.
O'Neill, M. W., and Gazioglu, S. M. (1984). "An evaluation of p-y relationships
in clays." Rept., PRAC 82-41-2, Amer. Petroleum Inst., Dallas, Tex.
Reese, L. C , Cox, W. R., and Koop, F. D. (1974). "Analysis of laterally loaded
piles in sand." Preprints, Sixth Annual Offshore Tech. Conf., II, Houston, Tex.,
473-484.
Reese, L. C , Cox, W. R., and Koop, F. D. (1975). "Field testing and analysis of
laterally loaded piles in stiff clay." Proc. Seventh Annual Offshore Tech. Conf.,
II, Houston, Tex., 671-690.
Reese, L. C , and Sullivan, W. R. (1980). "Documentation of computer program
COM624." Geotech. Engrg. Software GS80-1, Geotech. Engrg., Ctr, The Univ.
of Texas at Austin, Tex.
Reese, L. C , and Welch, R. C. (1975). "Lateral loading of deep foundations in
stiff clay." J. Geotech. Engrg. Div., ASCE, 101(GT7), 633-649.
Stevens, J. B., and Audibert, J. M. E. (1979). "Re-examination of p-y curve for-
mulations." Proc. Eleventh Annual Offshore Tech. Conf., Houston, Tex., 397-
404.

114

J. Geotech. Engrg., 1989, 115(1): 95-114