Professional Documents
Culture Documents
Behavior of Laterally Loaded Bored Piles in Cemented Sands
Behavior of Laterally Loaded Bored Piles in Cemented Sands
IN CEMENTED SANDS
Downloaded from ascelibrary.org by The University of Western Australia M209 on 06/11/19. Copyright ASCE. For personal use only; all rights reserved.
INTRODUCTION
Competent land deposits of cemented sands exist extensively in arid and
semiarid areas. In Kuwait cemented sands are located at ground level or
below a layer of windblown fine sand. The behavior of these deposits under
axial and lateral foundation loads has not been investigated by field loading
tests nor by analytical models. Recent work has dealt only with the strength
and consolidation characteristics of undisturbed soil samples in the labora-
tory, and indicates that cementation leads to the presence of a cohesion in-
tercept c and a slight increase in the angle of Shearing resistance <> | (Clough
et al. 1981; Ismael et al. 1986). The effect of artificial cementation on the
cone penetration resistance of sand was determined recently by laboratory
model tests (Rad and Tumay 1986).
In the absence of adequate information and reliable field test results on
cemented sands, the current local foundation design practice is based on
penetration tests, and the theories and empirical correlations developed for
noncemented cohesionless sands. This practice leads to uncertainties, and to
improper, generally conservative design with respect to both shallow and
deep foundations.
To examine the bearing capacity of shallow footings and the lateral ca-
pacity of bored piles in cemented sand, an extensive field testing program
was carried out recently in Kuwait. The program consisted of lateral load
tests on both short and relatively long bored piles. Two piles were instru-
mented with strain gages to measure bending strains during the tests. Sub-
sequently, plate loading tests were conducted on the ground surface using
the piles to provide reaction (Ismael 1989b).
Bored piles are commonly used in cemented sands due to several factors.
First, pile driving destroys cementation bonds and reduces skin friction along
the pile shaft (McClelland 1974; Murff 1987; Ismael 1989a). Second, in
'Assoc. Prof., Dept. of Civ. Engrg., Kuwait Univ., P.O. Box 5969, 13060 Safat,
Kuwait.
Note. Discussion open until April 1, 1991. 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 October 31,
1989. This paper is part of the Journal of Geotechnical Engineering, Vol. 116,
No. 11, November, 1990. ©ASCE, ISSN 0733-9410/90/0011-1678/$!.00 + $.15
per page. Paper No. 25206.
1678
45" 50
FIG. 1. Location of Test Site
1679
CEMENTED SILTY
X _0.3mx3m
SAND ( SM ) 26 3.8 1.87 26.0 •4.5 18 ~ Long Pile
,0.3m x5m
Long Pile
MEDIUM DENSE TO 60 3.1 1.90 None N.P
1000-
^ " " ^ - - ^ r ^ . ^ O O kPa
800
^ ^ ^ ^ ^ ^ £ 3 = 3 0 0 kPa
V> 600
i
Depth = 2m
I 1 1 1 1 1 1 1 1 1 1 1 1
0 2 4 6 8 10 12 14 16 18 20 22 24
Axial Strain £, °U
FIG. 3. Stress-Strain Curves and Failure Envelope from Drained Triaxial Tests
on Cemented Sand
in several heavy showers during the winter season. Excess evaporation over
rainfall leads to upward movement of ground water, resulting in precipitation
of carbonates and sulphates in the intergranular pore space and the formation
of crusts of cemented sands.
Two borings were augered at the site to a depth of 5.5 m. Samples and
standard penetration tests (SPT) were carried out at 1-m intervals. Data from
the two boreholes were almost identical and indicated similarity of the soil
properties across the site. The soil conditions were described in connection
with the plate loading tests at the site (Ismael 1989b); however, a brief sum-
mary is shown in Fig. 2. The soil profile consists of a medium dense ce-
mented silty sand layer to a depth of 3 m. This is underlain by medium
1681
1682
1Am
®
•12 (®\~ p ~/®~ A (•)-> P —H®1
i . ®
BH2 BH1
lD 3.6 m i.5m -y-0- 3.6m
36mm threaded bar typical
1.8m 1-8m
oo
_L_. J 11 10 J — 3 ( ® ) - — P —<-(•)
®7
® — 4m
8m 8m
* Instrumented Piles
FIG. 4. Layout Plan of Test Piles
i
Downloaded from ascelibrary.org by The University of Western Australia M209 on 06/11/19. Copyright ASCE. For personal use only; all rights reserved.
100m
Leadwire
Terminal
Rubber
Protection
J 100 mm
1
100 mm
Hole.
_4l2mrTJ_
The installation of the piles was carried out under favorable ground con-
ditions. The holes were dry upon augering and no ground water was en-
countered. No casing was needed since no caving or collapse occurred within
a depth of 5 m. The steel reinforcing cage was lowered into position after
completion of the hole. Concrete was poured by free fall to the level of the
tip of the central rod which was then positioned in place, after which the
concrete pouring was continued to the top of the piles.
Instrumentation
For piles 5 and 12, each 5-m long, 12 electrical resistance strain gages
were attached to the inside of the reinforcing cage of each pile with two
gages placed diametrically at each of six levels. Fig. 5 shows strain gage
installation details. These gages were installed on separate steel bars, pro-
tected and checked, and affixed to the reinforcing bars of the steel cage at
a depth of 0.2 m, 1 m, 2 m, 3 m, 4 m, and 4.8 m below ground level. The
cage was then lifted and placed carefully in the hole. The strain gage wires
extending to ground level were connected to a strain indicator and read be-
fore and after the application of each load increment during the test. Piles
5 and 12 were instrumented in order to compare the behavior of a restrained
1684
TESTING PROGRAM
Nine lateral loading tests were carried out. For each four piles forming
the corners of a square, two tests were carried out along parallel edges such
as 10-11 and 9-12. This yielded results for four test piles. Then two more
tests were carried out on the same piles along edges 9-10 and 11-12 to ex-
amine the effect of testing the piles to failure in one direction on the load-
deflection behavior in a perpendicular direction. Thus four field tests were
carried out along the sides of each square plus one test between the two
small groups. Testing was carried out in March 1989, four weeks after pile
installation.
EQUIPMENT AND PROCEDURE OF FIELD TESTING
The test arrangement is shown in Fig. 6. The horizontal force to both piles
or pile groups was applied simultaneously by a 510-kN (52-ton) Holmatro
hydraulic jack having a long stroke of 250 mm. The jack was connected to
a hand-operated pump equipped with a calibrated pressure gage that read to
an accuracy of 10 kN (1 ton). The jack exerted its force on a reaction beam
placed between the piles. Short reaction beams were employed between the
pile groups [Fig. 6(c)]. Each beam employed had 25-mm thick plates welded
at its ends. The frictional resistance between the edge of this plate and ground
soils during lateral loading was considered negligible. The single piles were
poured such that their projections above ground level had parallel flats facing
each other. These flats served as bearing surfaces for the tests.
Horizontal displacements of each pile or pile cap were measured by dial
gages having a range of 51 mm. The gages were attached from reference
beams installed behind each pile. The beams were secured to steel angles
that were anchored well away from the piles.
To carry out a test, the upper 0.2 m of soil were removed between the
piles, and the ground was leveled prior to placement of the reaction beam
and jack. The load was applied in cumulative equal increments of 10 kN (1
ton) for the 3-m long piles, and 20 kN (2 ton) for the 5-m long piles. This
corresponds to one-tenth the estimated lateral capacity. Each load increment
was maintained for a time interval of not less than 15 minutes and until all
displacements had ceased. At each increment dial gage and strain gage read-
ings of instrumented piles were taken. The loads were applied until both
piles exhibited a total horizontal displacement exceeding 50 mm and con-
tinuous movement was occurring at no increase in load. At that point the
test was discontinued. All tests were static tests with no load cycling. It is
believed that load cycling will not affect the cementation bonds and cohesion
intercept at low stress levels corresonding to the design loads. However, it
may affect the ultimate lateral capacity as progressive failure is expected at
some points along the piles at high stress levels. The effect of load cycling
will be examined by a follow-up field testing program.
The inclination of the 5-m long instrumented pile at ground level was
measured at selected load increments by means of a plumb bob attached to
a string that was fastened to the threaded reaction rod.
ANALYSIS OF TEST RESULTS
The field tests conducted fell into one of the following three cases:
1685
••'••• \
/ % — - . - • •
'i »'
(a)
I I -*> ,
>x ••).
• A - ' - .•>:
.w
rw
FIG. 6. (a) Lateral Load Test Setup; (b) Strain Gage Readings During Test on 5-
m Long Pile; (c) Lateral Testing of Group of Two Piles
1686
2. Free-headed 5-m long single piles with L/D = 16.67 located in two layers,
a surface layer of cemented sand underlain by very dense sand. The strength
parameters are 20 kPa and 35° from 0 to 3.5 m, and 0 and 43° from 3.5 m to
5 m.
3. Restrained piles 5 m long in a group of two with L/D = 16.67 in similar
soil conditions as case 2.
For the analysis of the test results, several methods were considered. These
included the linear subgrade reaction theory (Broms 1964), nonlinear (p-y)
analysis, elastic continuum analysis (Budhu and Davies 1988), and finite
element analysis (Poulos 1982). The approach adopted here is the nonlinear
(p-y) analysis, in which the soil is considered to be replaced by a series of
discrete springs to approximate the local resistance of the soil continuum to
lateral pile movement. The pile is considered as a beam-column. In this case
the governing differential equation is
EId4y Qd\
+ _ p _ w = o p= -Esy (1)
dx dx
where Q = axial load on the pile; y = lateral deflection of the pile at a point
x along the length of the pile; p = horizontal soil reaction per unit length;
EI = flexural rigidity of the pile; W = distributed load along the length of
the pile; and Es = kx the soil modulus.
If Q = 0 and W = 0, as in the present tests, Eq. 1 reduces to
EId4y
—V +Esy = 0 (2)
dx
The solution of Eq. 1 or 2 can readily be obtained by the use of a computer
program that employs the finite difference method, e.g., the program entitled
LPILE1 (Reese 1985). However, to solve the problem, it is necessary to
establish a set of p-y curves along the pile. Experimental p-y curves have
been determined by others for several types of soil conditions based on full-
scale tests on instrumented piles. These conditions include soft clays (Mat-
lock 1970), stiff clays with the presence of free water (Reese et al. 1975),
stiff clays with no free water present (Reese and Welch 1975), and sub-
merged cohesionless sand (Reese et al. 1974). For soft offshore carbonate
rocks with unconfined strength ranging from 0.5 M N / m 2 to 5 M N / m 2 , Abbs
(1983) proposed hybrid p-y curves, which are a combination of the estab-
lished curves for stiff clays and sands. However, for land deposits of ce-
mented sands possessing both strength components c and (j> and behaving
under drained conditions, no such curves have been determined or proposed.
The load versus pile head deflection data were plotted for the test piles.
Fig. 7 shows the data points for the 5-m long single piles and the average
curve for the four test piles. The response is nonlinear at all load levels.
The response of pile 10 was softer than the other three piles at loads ex-
ceeding 100 kN, indicating early yielding or progressive failure. The re-
sponse of the pile groups was very consistent at all load increments, as shown
in Fig. 8.
An analysis was first carried out that ignored the presence of the cohesion
of the upper sands, on the assumption that its effect is not significant. The
1687
FIG. 7. Pile Head Load versus Deflection for 5-m Long Single Piles
480
320
0 240 -
160 -
12 16 24
Deflection, mm
FIG. 8. Pile Group Load versus Cap Deflection for 5-m Long Restrained Piles
p-y curves developed by Reese et al. (1974) were employed. The coefficient
k, which defines the initial slope of the p-y curve, was taken as 24.4 MN/
m and 61 mN/m 3 for the upper and lower layers, respectively, as recom-
mended by Reese (1985). Fig. 9 shows the load versus pile head deflection
1688
Deflection , mm
Deflection,mm
FIG. 10. Comparison between Measured and Predicted Load Deflection Re-
sponse for 5-m Long Single Pile
1689
effect of soil layering was considered using the proposal of Georgiadis (1983).
The method is based on the concept of an equivalent thickness of the layer
existing below the upper layer.
A parametric analysis was carried out to examine the influence of the
parameter k on the load-deflection behavior. The curves shown by dotted
lines above and below the analysis 1 curve in Figs. 9 and 10 correspond
respectively to values of k twice and half the presented values. The mag-
nitude of this parameter has evidently little effect on the predicted behavior
for the cases considered herein.
As shown in Figs. 9 and 10, the given analyses furnished predictions that
are much softer, and therefore unsatisfactory, compared to the measured pile
response. This points out the importance of soil resistance at and near the
ground surface and that satisfactory prediction of the lateral response of bored
piles in cemented sand requires consideration of the cohesion component of
soil resistance.
It is instructive to note herein that the linear subgrade reaction theory (Broms
1964) is not the most suitable for prediction of the load-deflection response.
The method overestimates the deflection at working loads in both sand and
clay deposits (Ismael and Klym 1981, 1978), based on measured values of
the coefficient of subgrade reaction. Figs. 9 and 10 show the linear predic-
tions for the 3-m and 5-m piles employing the subgrade reaction values rec-
ommended by Terzaghi (1955). These values are 6,600 and 17,600 k N / m 3
for a strip with a width of unity at a depth of unity in medium and dense
sands, respectively. Both predictions are considered unsatisfactory.
A second analysis was employed using the research findings of J. Brinch-
Hansen (1961), which implies that the ultimate lateral load capacity of a
narrow anchor block or a pile in cohesionless soils is greater than would be
calculated using conventional passive earth pressure theory. On this basis
the ultimate soil resistance along the piles Pu is expressed by
Pu = Cp<Tpd (3)
where Cp = correction factor for small width of pile; <T„ = passive earth
pressure = Kp •CT„;KP = coefficient of lateral earth pressure; a„ = effective
vertical pressure; and d = pile diameter.
For soils having both components of strength c and <j>, Duncan (1974)
suggested using Eq. 3 to calculate the ultimate soil resistance, provided that
Cp is given by
Cp= 1.5, for 0 S <)> S 15° (4a)
<f>
C„ = —, for <|) > 15° (4b)
P
10
Employing the Rankine earth pressure theory in the analysis, and taking
the cohesion of the cemented sand into account, <rp in Eq. 3 is given by
45 + cf>\ 2 / 4 5 + <j>
arp = 2c tan I I + cr„ tan I I (5)
From Eqs. 3 - 5 , Pu can be calculated. The p-y curves have been approxi-
mated by a cubic parabola (Matlock 1970) with the equation
1690
QI . I i I i I i I i I i I
0 4 8 12 16 20 24
D e f l e c t i o n , mm
FIG. 11. Comparison between Measured and Predicted Pile Load versus Deflec-
tion for 5-m Long Restrained Pile
1600
1
1400
1
1200
1
1000
800 /
/ ^^LPILEI Generated p-y Curves
600
\ v - n
i i i i i i i
15 20 25 30 35 40
y, mm •
FIG. 12. Input and LPILE1 Generated p-y Curves Employed in Analysis 2
analytical predictions and the large ultimate soil resistance obtained via the
strain gage measurements. An angle of wall friction 8 = 10° or ~ 0.3 $
was considered. The coefficient of lateral earth pressure Kp employed was
4.84 using a curved failure surface (Shields and Tolunay 1973), which is
1.29 times the corresponding Rankine coefficient of 3.69 for 4> = 35°. The
cohesion c was not included in obtaining the Kp-values. Analysis 3 was
carried out using the larger coefficient and p-y curves determined from Eq.
6. The results are plotted in Figs. 9-11, which show a very close agreement
with the load-deflection response, particularly for the 3-m long pile and the
5-m long restrained pile.
However, there is a consistent deviation at the higher loads with stiffer
predictions than the actual pile response prior to the soil yielding and pile
failure under lateral loading. This may be explained in part by noting that
as soil starts to fail progressively, cementation bonds are broken at some
points, leading to a reduction in soil mass strength. A close examination of
Figs. 9-11 and 13 reveals that analysis 3 yielded substantially greater pile
capacity than was observed. The calculated ultimate soil resistance Pu also
exceeded the experimental values except at ground level (Fig. 13). Because
of this, and in view of the possible ground disturbance resulting from poor
construction practice, this analysis is not recommended for the design of
drilled shafts under lateral loading. Employing the experimental p-y curves,
1693
£ , I , I , I . I . I . I . I , I
0 5 10 15 20 25 30 35 40
y , mm
FIG. 13. Comparison between Experimental and Fitted p-y Curves of Cemented
Sands
the calculated pile head load-deflection response was almost identical to the
observed behavior.
A comparison between the measured bending moments along the 5-m long
piles and analyses 2 and 3 is shown in Fig. 14 for a lateral loading of 50
kN. As can be seen, close agreement exists between both analyses and the
actual measurements. The only significant deviation occurs at a depth of 0.2
m for the restrained pile, where the measured moment is nearly half the
analysis values. For the free-headed pile, the moment reached a maximum
value at a depth of 1 m and decreased to zero at a depth of 3 m at all lateral
load increments. The applied lateral load versus the maximum measured mo-
ment is plotted in Fig. 15 for the 5-m long single pile with the analysis
predictions. As shown, analyses 2 and 3 plot very close to the actual mea-
surements.
-1
-3
Strain
Gage
proximately the same maximum loads were reached in both cases. The small
effect of retesting in the perpendicular direction may be explained by noting
that the theoretical interference between the passive shear wedges in two
perpendicular directions around the pile is small (see Fig. 16). Therefore the
effect during retesting in the perpendicular direction is likewise expected to
be small.
0 3
. ^'^~
120 - /zs"
-<2
- Measured^>//
80- •^Analysis 1
O
O
.
•D
&
Z. 40
a.
< - //
l 1 1 1 l I 1
20 40 60 80 100 120 140
Max. Moment, kN. m
FIG. 15. Pile Head Load versus Maximum Moment for 5-m Long Single Pile
1695
Overlapping Shear
Zones
Pile
- Initial Tests
FIG. 16. Effect of Lateral Loading in One Direction on Load Deflection Response in Perpendicular Direction
EFFECT OF PILE CAP RESTRAINT
A comparison between the load-deflection response of the 5-m long single
Downloaded from ascelibrary.org by The University of Western Australia M209 on 06/11/19. Copyright ASCE. For personal use only; all rights reserved.
piles and the restrained piles is shown in Fig. 17. In the case of the restrained
piles, it was assumed that the load is shared equally between the two piles
in the group. The effect of shadowing, if any, was not considered or ex-
amined herein (Brown et al. 1988). From Fig. 17 it may be noted that the
pile cap significantly restrained the piles, making the slope at ground level
equal to zero and the maximum load per pile larger than that of the single
piles. In terms of the maximum loads reached, which is indicative of ulti-
mate soil failure, the ratio of maximum loads was 215/150 = 1.43. This
emphasizes the importance of the presence of a rigid pile cap. It is note-
worthy that the maximum measured moments along the restrained pile (Fig.
14) were also smaller than the values measured along the single pile.
ADDITIONAL COMMENTS
Finally, it was previously established (Ismael 1989b) that saturation or
wetting of cemented sands leads to partial loss of the cohesion strength com-
ponent c due to softening or loss of cementation bonds. For a long pile in
cemented sand subjected to lateral loading, the wetting effect will be re-
stricted to a thin surface zone below ground level. Therefore, it may not
significantly affect the pile response. However, for a short pile, extensive
wetting can lead to a softer pile response under lateral loading. This effect
may disappear gradually if dry conditions prevail for a long period of time,
leading to subsequent gain of cementation bonds. For this and other reasons
such as poor construction practices, which may result in improper installa-
220 -
200
s^® Pile in a Group of Two
180
160
140
z
•* 120 J3^^ Free Headed Pile
1 100
80
60 S i t e : South Surra
Piles: 0.3m Diam,
40
5 m Long
20
n 1 1 . 1 . 1 . I . 1
12 16 20 24
Deflection, mm
FIG. 17. Comparison between Lateral Load Deflection Response for Free-Headed
and Restrained Piles
1697
1. Cemented soils possess both components of shear strength c and t|>, and it
is not approriate to ignore the cohesion c in the analysis of the pile response
under lateral loading.
2. The measured response of the bored piles under lateral loads is nonlinear
and indicates the need to use nonlinear p-y curves to predict behavior along the
pile.
3. Parabolic p-y curves were obtained from the instrumentation data for the
cemented sand deposit. Similar curves were employed in the analysis using con-
ventional earth pressure theories.
4. The use of analysis 2, employed herein, is recommended because it is rea-
sonably conservative and yields good predictions of the measured pile response
under lateral loading.
5. For short piles, extensive wetting and ground disturbance will cause soft-
ening and partial loss of cementation in the top surface layer. In this case, a
reduction of 30%-40% in the magnitude of the cohesion is recommended in the
analysis.
6. The effect of lateral loading in one direction on the load-deflection response
in a perpendicular direction is to cause a softer pile response. However, the effect
is small and nearly the same maximum loads were reached in both cases.
7. The restraining effect due to the presence of a rigid cap is significant. The
maximum load on a restrained pile increased by 40%, and deflections at working
loads decreased to half the values corresponding to a similar free-headed pile.
ACKNOWLEDGMENTS
The investigation reported herein was supported by the Kuwait University
Research Administration under research project No. RMU 043. Thanks to
M. A. Mollah of the Ministry of Public Works, Kuwait, for providing as-
sistance in the field and laboratory tests for this research.
APPENDIX. REFERENCES
Abbs, A. F. (1983). "Lateral pile analysis in weak carbonate rocks." Geotechnical
practice in offshore engineering, ASCE, Austin, Tex.
Brinch-Hansen, J. (1961). "The ultimate resistance of rigid piles against transversal
forces." Bulletin No. 12, Danish Geotechnical Institute, Copenhagen, Denmark.
Broms, B. B. (1964). "Lateral resistance of piles in cohesionless soils." J. Soil Mech.
and Found. Div., ASCE, 90(3), 123-156.
Brown, D. A., Morrison, C , and Reese, L. C. (1988). "Lateral load behavior of
pile group in sand." J. Geotech. Engrg., ASCE, 114(11), 1261-1276.
Budhu, M., and Davies, T. (1988). "Analysis of laterally loaded piles in soft clays."
J. Geotech. Engrg., ASCE, 114(1), 21-39.
Clough, G. W., et al. (1981). "Cemented sands under static loading." J. Geotech.
Engrg., ASCE, 107(6), 799-817.
Georgiadis, M. (1983). "Development of p-y curves for layered soils." Proc, Geo-
tech. Practice in Offshore Engrg., ASCE, New York, N.Y., 536-545.
1698
1699