Determination of Bearing Capacity of Open-Ended Piles
Qag|nLoT in Sand
O_r^oH Kyuho Paik1 and Rodrigo Salgado, M.ASCE2
$:%*gY4~76 Abstract: The bearing capacity of open-ended piles is affected by the degree of soil plugging, which is quantified by the incremental
Qq`3S> filling ratio ~IFR!. There is not at present a design criterion for open-ended piles that explicitly considers the effect of IFR on pile load
kdK*MUB capacity. In order to investigate this effect, model pile load tests were conducted on instrumented open-ended piles using a calibration
?dp-}3/G chamber. The results of these tests show that the IFR increases with increasing relative density and increasing horizontal stress. It can also
w$DG=! be seen that the IFR increases linearly with the plug length ratio ~PLR! and can be estimated from the PLR. The unit base and shaft
L0X&03e=e: resistances increase with decreasing IFR. Based on the results of the model pile tests, new empirical relations for plug load capacity,
t Y:G54d=_ annulus load capacity, and shaft load capacity of open-ended piles are proposed. The proposed relations are applied to a full-scale pile load
T4V[RN
test performed by the authors. In this load test, the pile was fully instrumented, and the IFR was continuously measured during pile
bajC-5R1k driving. A comparison between predicted and measured load capacities shows that the recommended relations produce satisfactory
kN'|,eKH4 predictions.
bh= \ DOI: 10.1061/~ASCE!1090-0241~2003!129:1~46!
vqrBRlZ CE Database keywords: Bearing capacity; Pile load tests; Sand.
a6;gBoV Introduction
]}nu9z< When an open-ended pile is driven into the ground, a soil plug
L/qZ ; { may develop within the pile during driving, which may prevent or
RtW4n:c partially restrict additional soil from entering the pile. It is known
QT`fix{ that the driving resistance and the bearing capacity of open-ended
Bv;I0i:_
piles are governed to a large extent by this plugging effect.
Q;XXgX#l Many design criteria for open-ended piles, based on field tests,
2"T8^r|U chamber tests or analytical methods, have been suggested @e.g.,
'4af
], Klos and Tejchman 1977; Nishida et al. 1985; American Petroleum
)v${&H Institute ~API! 1991; Randolph et al. 1991; Jardine et al.
2B6^]pSk 1998#. For example, in the case of API RP2A ~1991!, which is
/'-:=0a generally used for offshore foundation design, the bearing capacity
Eem 2qKj of an open-ended pile can only be estimated for either the fully
1k!D0f3qb coring mode or the fully plugged mode of penetration. In practice,
bcq@N most open-ended piles are driven into sands in a partially plugged
Zr\2BOcc.l mode. Stefanoff and Boshinov ~1977! suggested the use of onedimensional
1t0bUf;(M plug analysis, in which the soil plug is treated as a
re7!p(W?, series of horizontal thin discs and the force equilibrium condition
R!sNg is applied to each disc, to calculate plug capacity of an openended
<2n'}&F pile.
Rb{+Ki There have been modifications of one-dimensional plug analysis
qsI{ b<n to improve predictive accuracy, such as the introduction of the
*
zd. concept of the wedged soil plug ~Murff et al. 1990; O’Neill and
s ;48v Raines 1991; Randolph et al. 1991!. Many test results show that
M#=Y~PU the soil plug can be divided into a wedged plug zone and an
I|$'Q$m~ unwedged plug zone. While the wedged plug zone transfers load
3wV86tH% to the soil plug, the unwedged plug zone transfers no load but
"EJ\]S]$X provides a surcharge pressure on top of the wedged plug zone.
dz8-): However, it is not easy to apply the one-dimensional analysis to
0Wa#lkn$I practical cases, because of the sensitivity of the method to the
{\P?/U6~f lateral earth pressure coefficient, which is not easily estimated
!?yxh/>lM ~Brucy et al. 1991; Leong and Randolph 1991!. De Nicola and
5fU!'ajaN7 Randolph ~1997! addressed this by proposing a profile of the
Jm?l59bv
v lateral earth pressure coefficient K along the soil plug length.
eT;AAGql An alternative design method can be based on the incremental
;_x2Ymw filling ratio ~IFR!. The degree of soil plugging is adequately quantified
@8|~+y8, using the IFR ~Paikowsky et al. 1989; Paik and Lee 1993!
72`/d` defined as
)8:n}w IFR5
!$xzAX,
DL
*^n^nnCwp DD
!>\9t9 3100~%! (1)
N0]z/}hd@ where DL5increment of soil plug length ~L! corresponding to a
pMOD\J:l, small increment DD of pile penetration depth D ~see Fig. 1!. The
IoQr+:_R fully plugged and fully coring modes correspond to IFR50 and
Z&TD+fT< 100%, respectively. A value of IFR between 0 and 100% means
8a7YHUL<3i that the pile is partially plugged. A series of model pile tests,
] OUD5T using a calibration chamber, were conducted on model openended
TV<Aj"xw piles instrumented with strain gauges in order to investigate
Xz8$Xz,O the effect of IFR on the two components of bearing capacity: base
4 uShM0qa load capacity and shaft load capacity. Based on the calibration
aWdUuid chamber test results, empirical relationships between the IFR and
k9cK bf@ the components of pile load capacity are proposed. In order to
6s'[{Ov verify the accuracy of predictions made using the two empirical
[S%J*sz~ relationships, a full-scale static pile load test was conducted on a
&.hoCPo$ fully instrumented open-ended pile driven into dense sand. The
&/HoSj>HS predicted pile load capacities are compared with the capacities
E`~i-kf measured in the pile load test.
*`%4loW 1Associate Professor, Dept. of Civil Engineering, Kwandong Univ.,
OthG7+eF Kangwon-do 215-800, South Korea ~corresponding author!. E-mail:
dZF8R pkh@kwandong.ac.kr 9-B@GFB;8 2Associate Professor, School of Civil Engineering, Purdue Univ., West
k@7kNMl Lafayette, IN 47907-1284. E-mail:
rodrigo@ecn.purdue.edu gEE9/\>%- Note. Discussion open until June 1, 2003. Separate discussions must
eVTO#R*'| be submitted for individual papers. To extend the closing date by one
[;<<4k(nL month, a written request must be filed with the ASCE Managing Editor.
cY{I:MA+h@ The manuscript for this paper was submitted for review and possible
]Uu
aN8 publication on July 23, 2001; approved on May 23, 2002. This paper is
: sFo
part of the Journal of Geotechnical and Geoenvironmental Engineering,
f7.m=lbe Vol. 129, No. 1, January 1, 2003. ©ASCE, ISSN 1090-0241/2003/1-
ZH% we 46–57/$18.00.
Sq<3Rw 46 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2003
_'&k#Q Soil Sample Preparation
k/U>N|5 Soil Properties
:|=- (z Han river sand, a subangular quartz sand, with D1050.17mm and
f]c<9Q>* D5050.34 mm, was used for all the calibration chamber model
ATo}FL 2 pile tests. The test sand is classified as poorly graded ~SP! in the
$%B5$+ Unified Soil Classification System, so the maximum dry density
@$Yb#$/ of the sand is near the low end of the typical range for sands. The
(p^S~Ax maximum and minimum dry unit weights of the sand were 15.89
JXL'\De ; and 13.04 kN/m3, respectively.
!1("(Eb A series of laboratory tests were conducted to characterize the
:fhB*SYK sand. The results from these tests are summarized in Table 1. The
$<:'!#% internal friction angle of the sand and the interface friction angle
Jlz9E|*qV between the sand and steel were measured from direct shear tests
RTZ:U@
under normal stresses of 40–240 kPa. The peak friction angles of
(,shiK[5f the sand with relative densities of 23, 56, and 90% were 34.8,
/Ad6+cY 38.2, and 43.4°, respectively, and the critical-state friction angle
f
P+QxOz was 33.7°. The peak interface friction angles between the pile and
9+t=| the sand were 17.0, 17.5, and 18.4° for DR523, 56, and 90%,
[Q|M/|mnR1 respectively, and the critical-state interface friction angle was
b##1hm~+9 16.7°. This angle is lower than commonly reported values because
SijS5irfk the test pile was made of stainless steel pipe with a very
EPv%LX_j smooth surface.
'\
XsTs#L Calibration Chamber and Sample Preparation
sx:Hv1d All model pile tests were conducted in soil samples prepared
3Mur*tj# within a calibration chamber with a diameter of 775 and a height
@\!ww/QT of 1250 mm. In order to simulate various field stress conditions,
RU7!U mf two rubber membranes, which can be controlled independently,
CGkI\E were installed on the bottom and inside the lateral walls of the
vsc&Ju%k calibration chamber. The consolidation pressure applied to the
*N`;I@Q"[ two rubber membranes was maintained constant by a regulator
~+=E"9Oo panel throughout each pile test.
UP?D@ogl< The soil samples were prepared by the raining method with a
tR5tPPw constant fall height. The falling soil particles passed through a
6A.P6DW sand diffuser composed of No. 8 and No. 10 sieves in order to
>r=6A
control flow uniformity and fall velocity. The soil samples had
[#>{4qY2 DR523, 56, and 90%. After sample preparation, the samples
(m/aV were consolidated to the desired stress state during approximately
w1cw1xX* 30 h by compressed air transferred to the rubber membranes.
=E!x~S;N Measurements made in calibration chambers are subject to
g9`[Y~ chamber size effects. Many researchers have attempted to estimate
YroNpu]s the chamber size needed for boundary effects on pile bearing
s+'XQs^{aj capacity or cone resistance to become negligible. Parkin and
,&[7u9@ Lunne ~1982! suggested 50 times the cone diameter as the minimum
HZ{n&iJ chamber diameter for chamber size effect on cone penetration
:,47rN,qa resistance to become acceptably small. Salgado et al. ~1998!,
]H>+m
9 based on cavity expansion analyses, found that 100 times the cone
cFDxjX?~ diameter was the minimum chamber diameter to reduce chamber
?|lI Xz size effects on cone resistance to negligible levels. Diameters of
LZ4xfB( the chamber and test pile used in this study are 775 and 42.7 mm,
l0. FiO@_Q respectively. The lateral and bottom boundaries are located at a
&u=8r* distance equal to 18.2 pile radii from the pile axis and 23.0 pile
8ZW?|-i radii below the maximum depth reached by the pile base, respectively.
/7x\;&bc Considering the results of the research on chamber size
z,avQR& effects mentioned above, the size of the chamber used in this
nGns}\!7' study is not sufficiently large for chamber size effects on pile
/h7.oD8CU bearing capacity to be neglected. The flexible boundary causes
.$P|^Zx, lower radial stresses than those that would exist in the field. Accordingly,
=},{8fZ4 the chamber tests done as part of this study produce
zA,/@/'( lower pile load capacities than those that would be observed in
l=xt;c! the field. A correction for chamber size effects is then necessary.
*<xrp*O It is discussed in a later section.
_0.pvQ Model Piles and Test Program
Fe5jdV< Model Pile
Ch7Egzl7? An open-ended pile is generally driven into sands in a partially
(_U^ plugged mode, and its bearing capacity is composed of plug load
05"qi6tncz capacity, annulus load capacity, and shaft load capacity. In order
c_Tzyh7l4 to separate pile load capacity into its components, an instrumented
L{<7.?{Y double-walled pile was used in the testing. A schematic
QkL@JF]Re diagram of the pile is shown in Fig. 2. The model pile was made
<}]{~y of two very smooth stainless steel pipes with different diameters.
S~> 5INud It had an outside diameter of 42.7 mm, inside diameter of 36.5
9qre|AA mm, and length of 908 mm.
|AC6sfA+ The wall thickness of the test piles used in this study is larger
KJdzv!l= than those of piles typically used in practice. Szechy ~1959!
GQ[pG{_+ showed that the degree of soil plugging and bearing capacity of
a*s\Em7f two piles with different wall thicknesses do not differ in a significant
kN.B/itvA way ~with bearing capacity increasing only slightly with increasing
aHC%19UN wall thickness!; only driving resistance depends significantly
rH.gF43O: upon the wall thickness. So the load capacity of the test
gB >pd?d piles reported in this paper are probably larger, but only slightly
V_f`0\[x so, than what would be observed in the field.
G5;V.#"Z[ Eighteen strain gauges were attached to the outside surface of
*i@T!O(1)M the inner pipe at nine different levels in order to measure the base
-bm,:Iy! load capacity ~summation of plug and annulus load capacities!
+sRP<as Fig. 1. Definition of incremental filling ratio and plug length ratio
7`dY 1.rq Table 1. Soil Properties of Test Sand
l])Q.m Property Value
Z%e|*GS{ Coefficient of uniformity Cu 2.21
lLMPw}r< Coefficient of gradation Cc 1.23
/BKtw8 Maximum void ratio emax 0.986
x6%#wsvS Minimum void ratio emin 0.629
!k-` eJ| Minimum dry density gd,min 13.04 kN/m3
~&KX-AC@ Maximum dry density gd,max 15.89 kN/m3
rVcBl4&1*g Specific gravity Gs 2.64
q]XHa ," Peak friction angle fpeak 34.8–43.4°
-njQc:4W,- Critical-state friction angle fc 33.7°
;s}3e#$L Peak interface friction angle d 17.0–18.4°
$rB6< Critical-state interface friction angle dc 16.7°
3S;N(A4 JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 2003 / 47
lQL:3U0DjU from the load transfer curve along the inner pipe. Two strain
+Vy_9I(4Z gauges were also attached to the outside surface of the outer pipe
:XYy7xz< in order to measure shaft load capacity. A gap of 4 mm between
a:b^!H># the outer pipe and the pile toe, which was sealed with silicone,
aq kix"J prevented the base load from being transferred to the outer pipe.
n_9x"m$ The outer pipe, therefore, experienced only the shaft load.
r.<JDdj Many researchers have relied on linear extrapolation to separate
HY*\ k# the base load capacity into plug and annulus capacities ~Paik
<xqba4O and Lee 1993; Choi and O’Neill 1997; Lehane and Gavin 2001!.
hfv%,,e Linear extrapolation would apply strictly only if the inside unit
0D~=SekQ9 friction between the pile and soil plug were constant between the
@RVOXkVo second lowest strain gauge and the pile base, as shown in Fig. 3.
5r7h=[N In reality, the inside unit friction between the soil plug and the test
)$_,?*fq: pile increases dramatically near the pile base. Use of linear extrapolation,
(tKMBxQo8 therefore, leads to an overestimation of annular resistance.
L
{qJ-ln: This overestimation increases as the distance between the
o%qkq K1 lowest strain gauge and the pile base increases. In part to avoid
hDvpOIUL1 this uncertainty, in this paper we use the base load capacity to
4|f}F analyze the test results instead of the plug and annulus load capacities
,ux+Qz5( separately. The base load capacity of the test pile was
A?,A(-0C obtained from the upper strain gauges located on the inner pipe,
bjzx!OCpV for which the measured vertical loads reached a limit value ~Fig.
n| C|& 3!.
TY6
rwU Test Program
SQE`
U Seven model pile tests were performed in dry soil samples with
z6cYC, three different relative densities and five different stress states.
Y`^o7'Z2^P Each test is identified by a symbol with three letters ~H high, M
O]ZC+]}/ medium, L low!, signifying the levels of the relative density, vertical
0H+c4IW and horizontal stresses of the sample, respectively. A summary
$"fzBM?5 of all model pile tests is presented in Table 2. Five model
FWY[=S pile tests were conducted in dense samples with DR590% and
8W,*eke? five different stress states. Two model pile tests were conducted in
Noz&noq loose and medium samples consolidated to a vertical stress of
9|3o< 98.1 kPa and horizontal stress of 39.2 kPa. The model piles were
*~;8N|4< driven by a 39.2 N hammer falling from a height of 500 mm.
3+9
U1:1[. During pile driving, the soil plug length and the pile penetration
On%,l depth were measured at about 40 mm intervals, corresponding to
s.rT] 94% of the pile diameter, in order to calculate the IFR. The
-)RJ\V^{9 change in soil plug length during pile driving was measured using
n_P(k-^U* a ruler introduced through an opening at the top plate of the pile
iRs V#s ~see Fig. 2!. In order to measure the soil plug length, driving
^1VbH3M operations were suspended for no more than a minute each time.
1R^4C8*B Static pile load tests were performed when the pile base was
nq@5j0fK located at depths of 250, 420, 590, and 760 mm. The pile load
I.a0[E/, tests were continued until the pile settlement reached about 19
oyW00]ka mm ~44% of the pile diameter!, at which point all the test piles
2fbU-9Rfn had reached a plunging limit state ~Fig. 4!. The ultimate load of
uP6-cs each test pile is defined as the load at a settlement of 4.27 mm,
>BJ}U_ck corresponding to 10% of the pile diameter. The total load applied
OZT^\Ky_l to the pile head was measured by a load cell, and settlement of the
Sn ^Aud pile head was measured by two dial gauges. Details of the model
)Mi'(C; pile, sample preparation, and test program have been described by
rS,j;8D- Paik and Lee ~1993!.
&CUC{t$VHX Model Pile Test Results
F.0d4:A+ Pile Drivability
N&x:K+Zm. Fig. 5~a! shows pile penetration depth versus hammer blow count
Pi){ h~B> for all the test piles. As shown in the figure, the hammer blow
?K<ZkYw? count per unit length of penetration increases as pile penetration
ETm]o
depth increases, since the penetration resistances acting on the
5~[N/Gl base and shaft of the piles during driving generally increase with
B{PLIisc Fig. 2. Schematic of model pile
(#z;(EN0t Fig. 3. Determination of plug and annulus loads
Qi:j)uDW Table 2. Summary of Model Pile Test Program
Snx<