Missouri University of Science and Technology

Scholars' Mine
International Conference on Case Histories in (1988) - Second International Conference on Case
Geotechnical Engineering Histories in Geotechnical Engineering

Jun 1st

Axial Load Capacities of Steel Pipe Piles in Sand

Roy E. Olson
University of Texas, Austin, Texas

Khalil Al-Shafei
Aramco, Dahran, Saudi Arabia

Follow this and additional works at: http://scholarsmine.mst.edu/icchge

Part of the Geotechnical Engineering Commons

Recommended Citation
Olson, Roy E. and Al-Shafei, Khalil, "Axial Load Capacities of Steel Pipe Piles in Sand" (1988). International Conference on Case
Histories in Geotechnical Engineering. 52.
http://scholarsmine.mst.edu/icchge/2icchge/2icchge-session6/52

This Article - Conference proceedings is brought to you for free and open access by Scholars' Mine. It has been accepted for inclusion in International
Conference on Case Histories in Geotechnical Engineering by an authorized administrator of Scholars' Mine. This work is protected by U. S. Copyright
Law. Unauthorized use including reproduction for redistribution requires the permission of the copyright holder. For more information, please contact
scholarsmine@mst.edu.
 )roceedings: Second International Conference on Case Histories in Geotechnical Engineering, June 1-5,1988, St. Louis; Mo., Paper No. 6.47

Axial Load Capacities of Steel Pipe Piles in Sand

Roy E. Olson Khalil AI-Shafei
L.. P. Gilvin Professor of Civil Engineering, University of Texas, Civil Engineer, Aramco, Dahran, Saudi Arabia
~ustin, Texas, USA

SYNOPSIS

The 1986 API method was used to predict the capacities of steel pipe piles, in predominently
cohesionless soils, for thirty three axial load tests. The ratio of calculated to measured
capacities (QC/QM) was found to range from 0.15 to 3.0 with a mean QC/QM of 0.74. Reconsideration
of the soil properties in terms of standard penetration resistances, made it possible to reduce the
scatter to the range of 0.65 to ·1.23 with a mean value of 0.93. The large errors previously
existing for short piles were eliminated. Analyses were equally accurate for piles in compression
and tension. The factor of safety required to reduce the probability of overloading to only 1% was
reduced from 4.5 to 1.5.

INTRODUCTION QS = E K a ' tan ( o) •••••••••••••••• (2)

The static design methods recommended by the where K is the ratio of horizontal to verti-
American Petroleum Institute (API, 1986) are cal effective stress, a' is the free field
widely used for estimating pile capacities. vertical overburden effective stress, and o
Previous studies (Olson and Dennis, 1982; is the pile/soil friction angle. The side
Dennis and Olson, 1983a, 1983b; Olson, shearing stress between the pile and soil in
1984; Olson, Dennis, and Winter, 1984) have any layer is limited to a value denoted by
provided estimates of the accuracies of FLIM.
earlier API methods in both cohesive and
cohesionless soils. Also:

This paper contains a study of the probable QP = cr ' NQ AP . . • . . . . . . • . . . • . . . • . . . (3 )

accuracy of the recently revised API method
for estimating the axial capacity of steel
pipe piles in cohesionless soils, and also an
extension to the method that appears to inc-
rease its accuracy significantly for terres-
trial piles.
ANALYTICAL METHOD AND ASSUMPTIONS
API Method
Because the interest here is in piles in
cohesionless soils, load test case histories
were included only when the fraction of the
theoretical side capacity in cohesionless
soils was more than 90% of the theoretical
total side capacity. For analytical purposes,
the general form of the API (1986) method was
used for estimating capacity in cohesionless
soil layers. In cohesive soils, the method
called NCL1 by Dennis and Olson (1983a) was
used. It will not be reviewed here because of
the relative unimportance of capacity derived
from cohesive soils in this study.
The calculated axial pile capacity, QC, is
taken as:
QC = QS + QP ± WP . . . . . . . . . . . . . . . . • .
where QS is the capacity in side shear, QP is
the tip capacity (zero for piles in tension),
and WP is the weight of pile submerged in soil
(+ for tension, - for compression). In turn:
where NQ is the tip bearing capacity factor
and AP is the tip area of the pile. The tip
stress is limited to a value QLIM.
The soil properties assumed when the API
(1986) method is used are summarized, in part,
in Table 1. The earth pressure coefficient, K
is assigned a value of 0.8 for open ended pipe
piles and 1.0 for closed ended pipe piles, and
has the same value in compression and in
tension.
Validity of Soil Properties
It should be recognized that the terms o',
o, and K do not represent real values of over-
burden pressure, friction angle, and earth
pressure coefficient, respectively, but are
instead convenient empirical terms. The
actual vertical stress near a pile du~ing
loading cannot be the free field stress
because of depth dependent load transfer
between the pile and the soil. The actual
value of 8 depends on such factors as stress
level and relative pile/soil movement, both
of which vary with depth. The earth pressure
coefficient must vary with pile displacement
( 1) ratio, installation method, and possibly with
the initial state of stress in the soil.
Regardless of the lack of theoretical rigor
in defining these properties, they prove
convenient to use and, if they result in

1731
Second International Conference on Case Histories in Geotechnical Engineering
Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu
 reasonable predictions of axial pile capacity, Table 2 - Definitions of Relative Density
the difference between assumed and real values used Here
becomes of reduced interest. Of course,
extrapolation of such empirical values to Corrected
cases not previously covered with load tests, relative density N (blows/foot)
must be performed with caution. very loose . . . . . . . . . . . . 0- 4
loose................. 4 - 10
Use of Standard Penetration Test medium . . . . . . . . . . . . . . . . 10 - 30
The API standard does not specify how dense . . . . . . . . . . • . . . . . . 30- 50
terms like "very loose" are to be defined. very dense . . . . . . . . . . . . 50+
Among the pile load tests used in this study,
the only measure of soil properties that was
available for all cases was a dynamic pene-
tration test, usually the "standard pene- Tip Capacity
tration test", N. Consequently, in spite of For open-ended steel pipe piles, the tip
the well known difficulties in measuring N, capacity was taken as the smaller of the tip
we chose to use N values in defining relative bearing capacity of a closed-ended pile and
density. the side shear capacity of a full plug.
Standard penetration resistances increase The tip capacity is considered to be a net
with depth because of the increasing over- capacity. Consequently, the weight of pile
burden pressure. When pile capacities are submerged in soil was included in the calcu-
correlated directly with N values, the N lations (Eq. 1). It had a minor influence on
values should not be corrected for this stress the results.
increase. However, in the API method, the Tip capacities vary near an interface if
increase in soil strength due to overburden the adjacent soil layers have properties that
pressure is accounted for directly in the differ significantly (Meyerhof, 1976).
equations so it should not be included again However, for the cases studied here, the
in the reported N values. In the absense of a properties of the cohesionless soils tended to
well documented method, we corrected the vary either smoothly, or erratically, with
measured N values using the recommendation of depth. For smooth variations, the interface
Peck, Hanson, and Thornburn (1974, Eq. 5.3, p. corrections seemed inappropriate. In the case
114) . The corrected N value is obtained by of erratic variations, the uncertainties in
multiplying the measured N value by CN where: the local N value directly below a pile tip
are large enough that an interface correction
CN = o. 77 loglo (20/a ' ) . . . . . . . . . . . . . (4) lacks significance. As a result, interface
corrections were not used.
and a' is the effective overburden pressure Closed-ended pipe piles were assumed to be
in units of tons per square foot. empty unless specified to be concrete-filled.
For soils ranging from gravel through Plug heights of 90% of the pile penetration
silt, we used definitions of relative density were assumed for open-ended pipe piles when
proposed by (Peck et al., 1974) for sands and plug heights were undefined. These assump-
(Table 2). tions are required when WP is evaluated, and
The relevant soil properties for an have a relatively minor effect on the calcu-
analysis can then be obtained by combining lated pile capacity.
data in Tables 1 and 2.
Definition·of Failure
The measured axial load capacities of the
piles (QM) were defined as the peak applied
Table 1 - Conditions Used in the API Method load. The ratio of the peak applied load to
the "defined" failure load (Davisson, 1973)
Pile Limit- Limit- ranged from 1.0 to 2.0 so the ratios QC/QM
Soil ing ing from load tests will depend significantly on
Friction Side Tip the choice of methods for defining QM, as well
Relative Stress Stress as on methods used to determine QC.
Density Soil deg. FLIM NQ QLIM
(1) (2) (3) (4) (5) (6) CASE HISTORIES
v l sa 15 1.0 8 40
1 sa-si Thirty three tests were selected for
m si analysis. Brief summary data are included
l sa 20 1.4 12 60 here to provide an indication of the types of
m sa-si data currently available. Selected data are
d si summarized in Table 3. All of the test piles
m sa 25 1.7 20 100 were steel pipe.
d sa-si
d sa 30 2.0 40 200 Arkansas River (Mansur and Hunter, 1970)
v d sa-si Piles were provided with exterior strain
d gv 35 2.4 50 250 gages encased in welded channels. Diameters
v d sa in Table 3 are equivalent diameters taking
into account the channels. The site was
Note: v l = very loose, l = loose, m = medium excavated to a depth of 20 feet prior to pile
d = dense, v d = very dense, sa = sand driving. The subsoils consisted mainly of
si = silt, gv = gravel poorly graded sand but there were strata of
silty sand, sandy silt, and silt. Data used
in this study came from borings made after

Second International Conference on Case Histories in Geotechnical Engineering 1732

Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu
 he excavation had been completed. Original Florida (Owens and Reese, 1982)
alues of the standard penetration resks- The pile was driven to a depth of 40 feet,
ances, N, generally increased from about 7 where it met refusal, using a .vibratory
pf near the surface to 50 bpf at depths hammer. The soil plug was augered out and
reater than about 50 feet. the casing redriven another 20 feet. The
Piles were tested first in compression and soil was cleaned out to the tip, a cage
hen in tension. Piles for load test numbers equipped with strain gages was lowered into
LTN' s) 8 9 , 9 0, 9 5 , 9 6, 10 2, and 10 3, were place, and the pile was filled with concrete .
.riven using a double acting air/steam hammer The soil consisted of fine to medium sands
Vulcan 140C) whereas the pile for LTN 100-101 with uncorrected standard penetration
·as driven with a Bodine hammer. resistances of 5 to 33 bpf. The water table
was at a depth of 3 feet.

Kansas City (Williams, 1960)

'able 3 - Summary of Geometric Data for The top 4 to 5 feet of soil was excavated
Case Histories prior to pile driving. The soil profile
after preexcavation consisted of silty sand
Pene. Diam. End Tens.or and a small amount of clayey silt overlying a
ite LTN feet inches Cond. Camp. coarse sand containing some gravel. Standard
1) (2) (3) (4) (5) (6) penetration resistances were of the order of
4 bpf in the silty sand and 20 bpf in the
.rkansas River 89 53 14 c c coarse sand and gravel. The water table was
95 53 21 c c at a depth of 8 to 10 feet, after excavation.
100 53 17 c c The piles were filled with concrete after
102 53 17 c c driving.
90 53 13 c T
96 53 20 c T Louisville (Cutter and Warder, 1978)
101 53 16 c T The soil profile varied from a sandy silt
103 53 17 c T near the surface to gravel at a depth of 65
'lorida 553 60 36 c c feet, with "standard penetration resistances"
:ansas City 341 55 13 c c increasing from 7 bpf in the sandy silt to
342 55 14 c c more than 50 bpf in the gravel. Fourteen
346 55 14 c c feet of soil was excavated prior to pile
347 .55 16 c c driving but after the soil borings were made.
.ouisville 805 70 12 0 c We arbitrarily reduced the N values to account
804 70 12 0 T for stress relief. The water table was at a
luskegon 796 58 12 c c depth of 44 feet, after site preparation.
798 58 12 c c
~stang Is. 354 69 24 0 T Muskegon (Cutter and Warder, 1978)
lgeechee Riv. 143 10 18 c c At this site the soil profile consisted of
144 20 18 c c 62 feet of fine to medium sand with an average
145 29 18 c c N value of 8 bpf. The N values were obtained
146 39 18 c c using a non-standard sampler (1.375-in. ID,
147 49 18 c c 1.75-in. OD, and an undefined hammer). The
148 49 18 c T water table was at the ground surface. A pile
lld River 112 65 21 c c (LTN 796) was driven closed ended. A dupli-
114 66 17 c c cate pile was driven open ended to the same
116 65 19 c c depth with a water jet used to clean out the
117 65 18 c c inside during driving, with care not to jet
122 65 19 c T beyond the end of the pipe.
118 65 21 c T
120 66 17 c T Mustang Island (Reese and Cox, 1976)
~s Tanajib 1021 59 24 0 T About five feet of sand was excavated so
~kyo 243 27 8 c c the water table would be at the ground
244 11 8 c c surface. The soil profile consisted of silty
245 34 8 c c sands (N = 28 to 40 bpf) . An open-ended steel
pipe pile was driven to a depth of 38 feet,
rote: Columns are identified as follows: the inside cleaned out, a plate welded across
:1) location of the site the top (leaving 38 feet of empty pipe) ,
:2) load test number extensions added, and the pile driven to a tip
:3) penetration of pile tip into the subsoil depth of 69 feet.
:4) outside diameter of the pile
:5) pile tip condition, O=open, C=closed Ogeechee River (Vesic, 1970)
:6) C=compression, T=tension The soil at this site was mostly clean
fine sand with occasional zones of silty
sand. One boring encountered a layer of
highly plastic clay near the surface.
Standard penetration resistances increased
from about 6 bpf for the top layer to 37 bpf
for the bottom layer. The steel pipe pile was
successively driven, tested, lengthened,
redriven, .... for five tip penetrations
ranging from ten to fifty feet. Finally, a
tension test was performed.

1733
Second International Conference on Case Histories in Geotechnical Engineering
Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu
 Old River (Mansur and Kaufman, 1958) RESULTS OF ANALYSES
· The site was excavated to a depth of fifty
fe7t, over.an area of 100 feet by 150 feet, The results of analyses using the API (1986)
pr~or to p~le driving, and the water table
standard, but using values of N, corrected
was drawn down to about excavation depth. for overburden pressure, to classify the soil
S~bsoils were mostly silty sand and sandy
(Tables 1 and 2) are summarized in Table 4
s~lt over clean fine to medium sand, but with
and presented in Fig. 1.
shallow layers of silt and clay. Standard
penetration resistances increasing from 4 bpf Table 4 - Summary of Results of Analyses
near the surface (after excavation) to 80 bpf using the API Standard
near the pile tips. Four of the piles were
left empty, one (LTN 117) was filled with Tens. QC QM
concrete. Site LTN Camp. kips kips QC/QM

Tokyo (BCP Committee, 1971) Arkansas 89 c 217 360 0.60

A 7.87-inch diameter, instrumented steel River 95 c 573 560 1.02
pipe pile was driven through 13 feet of loose 100 c 409 450 0.91
sand fill, and 23 feet of sand, into a gravel 102 c 406 500 0.81
layer. Japanese standard penetration resis- 90 T 148 180 0.82
tances increased from about 4 blows/foot in 96 T 245 230 1. 07
the fill to 55 bpf in parts of the gravel. 101 T 199 230 0.87
Load tests were carried out to large settle- 103 T 207 240 0.86
ments, sometimes exceeding six feet. Old 112 c 950 750 1. 27
River 114 c 684 750 0.91
Ras Tanajib (Helfrich, Wiltsie, Cox, and 116 c 803 700 1.15
Al-Shafei, 1985) 117 c 747 680 1.10
The test was performed at a site adjacent 118 T 321 390 0.82
to the Arabian Gulf in Saudi Arabia. The soil 120 T 271 400 0.68
was sand to silty sand with standard penetra- 122 T 292 370 0.79
tion resistances increasing from 15 bpf near Ogeechee 143 c 30 200 0.15
the surface to 45 bpf at a depth of 18 feet River 144 c 96 480 0.20
and below that 50 blows (the preselected 145 c 156 630 0.25
limit) was achieved in less than 6 inches of 146 c 215 800 0.27
penetration. Cone tip resistances increased 147 c 332 860 0.39
from zero at the ground surface to 1400 ksf 148 T 195 346 0.56
at a depth of 26 feet, with refusal (more than Tokyo 242 c 188 370 0.51
1500 ksf) at greater depths. The pile was 244 c 26 68 0.38
loaded in compression to 3800 kips without 245 c 168 310 0.54
failure, and then loaded in tension until it Kansas City 341 c 269 188 1. 43
failed. 342 c 303 218 1.39
Mustang Is. 353 T 426 455 0.94
Florida 553 c 1007 690 1. 46
Muskegon 796 c 177 240 0.74
798 c 173 110 1. 57
Louisville 805 c 494 280 1. 76
804 T 328 110 2.99
Ras Tanajib 1021 T 893 2850 0.31

For these data, the mean value of QC/QM was

0.74 (antilog of the mean of the logs of
>:
.j.J QC/.QM) with a range in QC/QM from 0.15 to
..-I
0 0 2. 99. s·ome individual values of QC/QM differ
a!
p. significantly from the values obtained using
a!
u a subjective classification of the soils, but
G) the mean value of QC/QM and the range in
r-1
or-f values, are almost identical (0.72 mean,
ll.< range from 0.13 to 2.86 when the subjective
"d 100
G) approach was used) .
1-4
::I
Cll
a! Based on the assumption of a normal distri-
~ bution for log(QC/QM), and that a correction
is applied for bias (Olson and Dennis, 1983)
the factor of safety required to reduce the
probability of failure (overloading) to 1% is
4.7.
1,000
Calculated Pile Capacity, kips Arkansas and Old Rivers
For the Arkansas River tests, the average
Fig. 1 Comparison of the Measured Capacities QC/QM was 0.84 for piles in compression and
of Steel Pipe Piles with Capacities 0.90 for piles in tension. For the Old River
Calculated using the API (1986) tests, the average QC/QM was 1.12 in compres-
Standard Method RP-2A. Soil relative sion and 0.76 in tension. The relatively
densities were defined using standard accurate predictions for these two sites are
penetration resistances. anticipated because the API method was prob-
ably based, in part, on these two sets of
tests.

1734
Second International Conference on Case Histories in Geotechnical Engineering
Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu
 ~lorida (Olson and Dennis, 1982), the mean value of
The capacities of the Florida pile was QC/QM was still only 0.48. The data are
::>ver-predicted by 46%. The tip capacity of similar to those at Ogeechee River in that
this pile was· measured to be 28 kips whereas QC/QM was quite low for the shortest pile.
the calculated value was 563 kips. It seems
likely that the cleaning out of the pile TRIAL ANALYSES
~nloaded the soil at the tip. Thus, when the
pile was tested, the side shear developed at For each site, a set of trial analyses were
a deformation much less than required for the performed in which values of K, o , and NQ,
tip and the fraction of the ultimate tip were varied successively, and repeatedly,
bearing capacity that was developed was until the average value of the ratio of
probably small. The tip settlement at peak calculated to measured pile capacities
applied pile load was 0.62 inch. Estimated (QC/QM) was essentially 1.0.
settlements at failure for a pile (prestressed
tip) and pier are 0.4 inch and 3.5 inches, For sites involving only a single load test,
respectively. and in other sites when there were multiple
If the theoretical tip capacity is set soil layers, there are numerous combinations
equal to 28 kips, then QC/QM=0.67. of assumptions that all yield values of QC/QM
near one. The "best" combination was selected
Kansas City as the one that led to reasonable variations
For the Kansas City sites, capacities were in o and NQ as functions of sand density, with
rather uniformly overpredicted with QC/QM an effort made to find consistency between
averaging 1.41. sites as well as at single sites.

Louisville Efforts were made to correlate properties

For the pile at the Louisville site, the with such factors as pile length, mean
large values of QC/QM may indicate a tendency effective stress, and mean value of N, but the
to overpredict the capacity of long piles. combination of multiple variables (K, o , FLIM,
NQ, QLIM), scatter in the data, and a small
Muskegon data base, made such correlations uncertain.
For piles at Muskegon, QC/QM was 0.74 for For open ended pipe piles (Louisville, Mustang
a pile driven closed ended but was 1.57 for a Island, Ras Tanajib), there was an excellent
pile of the same diameter and length but which relationship between the fitted value of K and
had mild internal jetting, although we assumed mean N (Fig. 2), of the form:
a tip capacity of zero in the latter case.
Apparently, even mild jetting reduces sand K =A+ B N •••••••••••••••••• (5)
compaction during pile driving and thus
reduces side capacity. where A=0.2 and B=0.012. No such correlation
was clear for closed ended pipe piles.
Mustang Island
For the Mustang Island test, the capacity
was underpredicted by 6%.

Oqeechee River
For the Ogeechee River tests, capacities Note
were underpredicted by a substantial amount
with an average QC/QM of 0.25 for compression ~ G open ended piles
tests, but the degree of underprediction
clearly varies with pile penetration as shown "c:l ~ closed ended piles
below: ~
4-1
2.0
QJ
0
Load Pile u
Test Penetration QJ A
Number feet QC/QM ~(/)
(/)
QJ
143 10 0.15 J:: 1.0
144 20 0.20
145 29 0.25 -5
146 39 0.27 ~
[il
147 49 0.39
0.0~_______.___________________________
Ras Tanajib
For Ras Tanajib, the sand was apparently 0 50 100 150 200
too dense to have its properties measured Standard Penetration Resistance, blows/foot
reliably using either the standard penetration
test or the quasi-static cone test. Further,
the required soil properties would be "off
scale" in the API design chart.
Fig. 2 Backcalculated Earth Pressure
Tokyo Coefficients in Terms of the Weighted
Part of the reason for the low values of Mean Standard Penetration Resistance.
QC/QM for the Tokyo tests was the use of a The weighting was performed in terms
different definition for the standard of the total calculated load transfer
penetration test. However, even after in each layer.
multiplying the Japanese N values by four

1735
Second International Conference on Case Histories in Geotechnical Engineering
Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu
 The best-fit site values were then tabulated SASI 4 ( 5) (0.5) 4) ( 20)
and used as a starting point for general 10 10 (1. 0) 8) ( 40)
analyses. The general analyses involved 30 (15) (1. 4) 12) ( 60)
repeated trials where values of o, FLIM, NQ, 50 ( 25) (1. 7) 20) (100)
and QLIM were varied for each of six grain 100 ( 35) (2. 0) 40) (200)
size classifications (gravel through silt) 1000 ( 38) (2. 4) 60) (250)
and six ranges in N values. However, it
became clear that some of the soil categories SILT 4 ( 5) (0.3) 2) ( 10)
were not encountered, and that pile capaci- 10 10 (0.5) 4) ( 20)
ties were mainly governed by deep soil layers 30 15 (1. 0) 8) ( 40)
where the effective stresses were high, and 50 20 (1. 4) 12) ( 60)
these layers were predominently classified as 100 (25) (1. 7) 20) (100)
sand. Pile tips were not left in· silty soils 1000 (30) (2.0) 40) (200)
in any case so no empirical data could be
generated for values of NQ and QLIM in such
materials.
Table 6 - Summary of Results of Analyses
After the overall best-fit properties were using the Revised Method
determined, these values were tabulated and
it was found that there were inconsistencies Tens. QC QM
resulting from normal scatter and the small Site LTN Comp. kips kips QC/QM
size of the data base. The inconsistencies
were slightly smoothed out and a final table Arkansas 89 c 366 360 1. 02
of soil properties prepared (Table 5). River 95 c 686 560 1. 23
Numbers shown in parentheses are based on 100 c 463 450 1. 03
interpolation and extraplation, not on direct 102 c 460 500 0.92
data. Values of K were found to be the same 90 T 191 180 1. 06
in compression and tension and were given by 96 T 257 230 1.12
the following equations: 101 T 184 230 0.80
103 T 192 240 0.80
open ended K 0.16 + 0.012 N (6a) Old 112 c 861 750 1.15
River 114 c 639 750 0.85
closed ended K 0.70 + 0.012 N (6b) 116 c 734 700 1. OS
117 c 688 680 1. 01
The properties shown in Table 5, and values 120 T 319 400 0.80
of K in Eq. 6, were then used with the data 122 T 340 370 0.92
base to produce the results shown in Table 6 118 T 375 390 0.96
and Fig. 3. The mean value of QC/QM (antilog Ogeechee 143 c 198 200 0.99
of the mean of the logs of QC/QM) is now 0.93 River 144 c 386 480 0.80
and the scatter in QC/QM is reduced to the 145 c 440 630 0.70
range of 0.65 to 1.23. After correcting for 146 c 524 800 0.65
bias, a factor of safety of 1.5 would reduce 147 c 624 860 0.73
the probability of overloading to one per- 148 T 276 346 0.80
cent. Tokyo 242 c 356 370 0.96
244 c 59 68 0.87
245 c 334 310 1. 08
Table 5 Soil Properties Extracted from the Kansas City 341 c 210 188 1.12
Empirical Data Base. Properties shown in 342 c 241 218 1.11
parentheses are interpolated or extrapolated. Mustang Is. 353 T 460 455 1. 01
Florida 553 c 538 690 0.78
Soil N FLIM QLIM Muskegon 796 c 245 240 1. 02
Type bpf deg. ksf NQ ksf Louisville 805 c 239 280 0.85
804 T 110 110 1. 00
SAGV 4 ( 20) (1. 4) ( 12) ( 60) Ras Tanajib 1021 T 2565 2850 0.90
10 (25) {1. 7) ( 20) (100)
30 25 (2. 0) 150 230
50 (35) (2. 4) (200) ( 250)
100 40 3.5 250 300 It would not be difficult to change a few
1000 45 5.0 (250) (300) properties slightly and raise QC/QM precisely
to 1.00 but the small size of the data set
SAND 4 (15) (1. 0) ( 80) ( 40) and its inherent uncertainties make such
10 30 1.1 110 110 refinement inappropriate. In any case, use
30 35 2.3 120 190 of a "defined" failure point, as opposed to
50 35 2.6 140 190 the peak applied load, would raise values of
100 35 3.6 150 190 QC/QM substantially.
1000 40 3.6 220 650
The analyses used the same properties for
SISA 4 15 (1. 0) 8) ( 40) tension and compression tests. Average
10 10 (1. 4) 12) ( 60) values of QC/QM for sites with both comp-
30 15 (1. 7) 20) (100) ression and tension tests, are shown below.
50 25 (2. 0) 40) (200)
100 ( 35) (2. 4) ( 60) (250)
1000 38 (20. 0) (300) (600)

Second International Conference on Case Histories in Geotechnical Engineering

1736
Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu
 QC/QM (No. of Tests) 5. A simple static method, such as the one
Site Compression Tension recommended by API, can be used with
Arkansas River ... 1.05 (4) 0.95 (4) tolerable accuracy if there is a reason-
Louisville....... 0.85 (1) 1. 00 (1) ably accurate method for correlating
Ogeechee River ... 0.73 (1) 0.80 (1) properties of cohesionless soils. In
Old River........ 1.02 (4) 0.90 (3) the absense of a substantial data base
for other tests, the standard penetra-
tion test may be used.

6. The assumed soil properties are the same

for piles in compression and tension.

7. The apparent value of K increases with N

and is higher for pipe piles with closed
ends than for open ended pipe piles.

(/)

]' 1,000 ACKNOWLEDGEMENTS

>: The junior author was supported by Aramco

+.1
•r-1
through a major portion of this study (Al-
cJ
cO
Shafei, 1987). The study was initiated by a
~ grant from the American Petroleum Institute
0 (Olson, 1988). The authors express
Q)
.-1
appreciation to both Aramco and API. The
•r-1
p., views expressed here are those of the authors
"0
and do not necessarily represent views within
Q)
1-1
either Aramco or API.
ffi
~ REFERENCES

Al-Shafei, Khalil A.-R. (1987), "Axial Load

100 1000
Calculated Pile Capacity, kips
Fig. 3 Comparison of the Measured Capacities
of Steel Pipe Piles with Capacities
Calculated using the Method Recommended
in this paper.
CONCLUSIONS
Conclusions derived from this study include:
1. Static methods of estimating pile
capacity are needed in cases where test
piles are not used, for estimating
lengths for test piles, and for
estimating lengths when wave equation
analyses are used for driving control.
2. The data base of useful pile load tests
for steel pipe piles in cohesionless
soils is small and tests often involve
significant uncertainties in such
aspects as stratigraphy and the details
of testing methods.
3. Internal jetting of an open ended pipe
pile in sand appears to reduce its
capacity even when the jetting is
performed carefully (Muskegon)
4. Cleaning out the inside of a pipe pile
that was driven open ended, and then
filling it with concrete, is likely to
result in a substantial reduction in
tip capacity for acceptable values of
butt settlement (Florida) . The size of
error should increase as pile diameter
increases.
10,000 Capacities of Steel Pipe Piles in Sand,"
M.S. Thesis, Univ. of Texas at Austin.
American Petroleum Institute (1986), "API
Recommended Practice for Planning,
Designing, and Constructing Fixed Offshore
Platforms, RP2A," API, ProductionDiv.,
211 North Ervay, Suite 1700, Dallas, TX
75201.
BCP Committee (1971) , "Field Tests on Piles
in Sand," Soils and Foundations, Vol. 11,
No. 1, pp. 29-49.
Cutter, W. A., and D. L. Warder (1978),
"Friction Piles in Sand - a Review of
Static Design Procedures," Proc., Ninth
Ohio River Valley Soils Seminar,
Louisville, KY.
Davisson, M. T. (1973), "High Capacity
Piles," Soil Mech. Div., Ill. Sect., ASCE,
Chicago, pp. 81-112.
Dennis, N.D., and R. E. Olson (1983a),
"Axial Capacity of Steel Pipe Piles in
Clay," Geotechnical Practice in Offshore
Engineering, Amer. Soc. of Civil Engrs.,
Geot. Engr. Div. Specialty Conference,
Austin, Texas, pp. 370-388.
Dennis, N.D., and R. E. Olson (1983b),
"Axial Capacity of Steel Pipe Piles in
Sand," Geotechnical Practice in Offshore
Engineering, Amer. Soc. of Civil Engrs.,
Geot. Engr. Div. Specialty Conference,
Austin, Texas, pp. 389-402.
Helfrich, S. C. 1 E. A. Wiltsie, W. R. Cox,
and K. A. Al-Shafei (1985), "Pile Load
Tests on Dense Sand: Planning,
Instrumentation, and Results," Proc.,
Seventh Annual Offshore Technology
Conference, Vol. 1, pp. 55.
Mansur, C. I., and A. H. Hunter (1970), "Pile
Tests-Arkansas River Project," Jour.,
Soil Mech. and Found. Div., ASCE, Vol. 96,
No. SMS, pp. 1545-1582.

Second International Conference on Case Histories in Geotechnical Engineering 1737

Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu
 Mansur, C. I., and R. I. Kaufman (1958),
"Pile Tests-Low Sill Structure, Old River,
Louisiana,", Trans., ASCE, Vol. 123, pp.
715-748.
Meyerhof, G. G. (1976), "Bearing Capacity and
Settlement of Pile Foundations," Jour. of
the Geotechnical Engineering Division,
ASCE, Vol. 102, pp. 195-228.
Olson, R. E. (1984), "Analysis of Pile
Response under Axial Loads," Final Report
to the American Petroleum Institute on
Project 83-42B.
Olson, R. E. (1988), "Comparison of Measured
Axial Load Capacities of Steel Pipe Piles
in Sand with Capacities Calculated using
the 1986 API Recorcunended Practice (RP2A)",
Final Report on Project 86-29A.
Olson, R. E., and N.D. Dennis (1982),
"Review and Compilation of Pile Test
Results, Axial Pile Capacity," report to
the American Petroleum Institute, Geot.
Engr. Report No. GR81-14, Department of
Civil Engineering, Univ. of Texas, Austin,
TX.
Olson, R. E., N. D. Dennis, and D. G. Winter
(1984), "Prediction of Axial Pile
Capacity Based on Case Histories,"
Intern. Conf. on Case Histories in
Geotechnical Engr., University of
Missouri-Rolla, s. Prakash-ed., Vol. 1,
pp. 191-200.
Olson, R. E., and N.D. Dennis (1983),
"Reliability of Pile Foundations",
Preprint SC-3, Structures Congress,
Amer. Soc. of Civil Engrs., Houston, Tx.
Owens, M. J., and L. c. Reese (1982), "The
Influence of a Steel Casing on the Axial
Capacity of a Drilled Shaft," Res. Report
No. 255-1F, Proj. 3-5-80-255, Ctr. for
Transp. Res., Univ. of Texas, Austin, TX.
78712.
Peck, R. B., T. H. Thornburn, and w. E.
Hansen (1974), Foundation Engineering,
John Wiley and Sons, NYC.
Reese, L. C., and w. R. Cox (1976), "Pullout
Tests of Piles in Sand," Proc., Offshore
Technology Conf., pp. 527-538.
Vesic, A. s. (1970), "Tests on Instrumented
Piles, Ogeechee River Site," Jour., Soil
Mech. and Found. Div., ASCE, Vol. 96, No.
SM2, pp. 561-584.
Williams, J. A. (1960), "Report on Test Pile
Program Conducted by Kansas and Missouri
State Highway Departments," HRB Bulletin
279, Highway Research Board, pp. 63-85.

1738
Second International Conference on Case Histories in Geotechnical Engineering
Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu