Settlement Ratio of Pile Groups in Sandy Soils From Field Load Tests
Settlement Ratio of Pile Groups in Sandy Soils From Field Load Tests
Settlement Ratio of Pile Groups in Sandy Soils From Field Load Tests
Abstract: This note studies settlement ratio, Rs, of pile groups in sandy soils, defined as the ratio of the settlement of a pile group to that
of a single pile at the same average load per pile. 31 cases of field pile-group load tests and the corresponding field single-pile load tests
were collected for this study. More than one-half of the cases consist of 3-diameter spaced, 9-pile groups. Based on the field test data,
statistical analyses of Rs at different load levels were conducted for pile groups with cap-ground contact 共PGCs兲 and pile groups with
freestanding caps 共PGFs兲, respectively. The mean of Rs decreases with the load level for both PGCs and PGFs, whereas the coefficient of
variation of Rs increases with the load level. The influence of cap-ground contact on Rs does not appear to be significant based on a
comparison of the mean Rs values of these PGCs and PGFs. In addition, a comparative study on Rs and group resistance ratio Rr, which
is defined as the ratio of the average resistance of a pile in a group to that of a single pile at the same settlement, was conducted to clarify
possible misunderstanding between Rs and pile group efficiency factor for driven pile groups in sandy soils. The value of Rs compares
settlement at the working load and is often larger than unity. The value of compares failure loads, which occur at different settlements
for pile groups and their respective single piles. is usually larger than unity due to soil densification and additional contributions from
the cap-ground contact for PGCs.
DOI: 10.1061/共ASCE兲1090-0241共2007兲133:8共1048兲
CE Database subject headings: Bearing capacity; Settlement; Pile foundations; Pile groups; Pile test; Serviceability.
1
Research Assistant, Dept. of Civil Engineering, The Hong Kong A data set of field static loading tests on groups of driven dis-
Univ. of Science and Technology, Clear Water Bay, Hong Kong. E-mail: placement piles and the corresponding single piles was collected
[email protected] from the literature 共Vesic 1969; Liu et al. 1984; Briaud et al.
2
Associate Professor, Dept. of Civil Engineering, The Hong Kong 1989兲, as shown in Table 2. The data set consists of 20 PGCs and
Univ. of Science and Technology, Clear Water Bay, Hong Kong 11 PGFs. The pile spacing varies from 2 to 4 diameters, with
共corresponding author兲. E-mail: [email protected] approximately one-half of the cases at a spacing of 3 diameters.
Note. Discussion open until January 1, 2008. Separate discussions The information of pile groups with a pile spacing of 6 diameters
must be submitted for individual papers. To extend the closing date by
or larger is excluded from Table 2 because those pile groups are
one month, a written request must be filed with the ASCE Managing
Editor. The manuscript for this technical note was submitted for review not commonly used in practice. The pile length in the analyzed
and possible publication on September 11, 2006; approved on December data set varies from 8 to 28 diameters with an average of 16.6
28, 2006. This technical note is part of the Journal of Geotechnical and diameters. Nine or fewer piles are present in each group, and
Geoenvironmental Engineering, Vol. 133, No. 8, August 1, 2007. more than one-half of the cases comprise 9-pile groups. Pile
©ASCE, ISSN 1090-0241/2007/8-1048–1054/$25.00. groups smaller than 4 piles are excluded because such groups do
Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Table 1. Empirical Formulas for Rs of Pile Groups in Sandy Soils
Fleming et al. 共1992兲
not have the necessary redundancy to maintain stability should 1. For both PGCs and PGFs, the mean of Rs decreases with the
any one pile fail 共Zhang et al. 2001; Paikowsky 2004兲. normalized load Q / Q f . No significant distinction exists be-
The driven pile groups analyzed are all founded in sandy soils. tween the mean values of Rs for PGCs and PGFs. Therefore,
The soil types involved include sandy silt and medium to dense the contribution of cap-ground contact to Rs is statistically
sands. Detailed soil profiles are reported in the corresponding not appreciable. This is consistent with the observation by
literature. Pile groups founded in clayey soils are not reported in Poulos 共1989兲 that the influence of cap-ground contact on
this note because the number of cases is insufficient for a mean- group settlement at working loads can be ignored for most
ingful statistical analysis. practical purposes.
2. For both PGCs and PGFs, the COV of Rs appears to increase
with the normalized load Q / Q f . This is due to the fact that
Analysis of Pile-Group Settlement Ratio nonlinear soil behavior has a greater influence on pile settle-
ment as load increases, resulting in more scattered load-
For the convenience of comparison, the load-settlement curve of a settlement curves of pile groups under large loads as shown
single pile is represented by the relationship between the single in Figs. 2 and 3. The COV values are large because the
pile settlement and the normalized load, Q / Q f , defined as the information of pile groups of various sizes at several differ-
ratio of the applied load Q to the single-pile failure load Q f . Q f is ent sites is mixed in the statistical analysis; hence, the uncer-
determined by the Davisson failure criterion 共Davisson 1972兲, tainty from pile-group size, and the within-site and cross-site
which is commonly used for driven piles. The load-settlement uncertainties are compounded 共Zhang et al. 2004兲.
curve of a pile group is represented by the relationship between 3. The Rs value of a particular pile group at a selected load
the pile group settlement and the average load per pile normalized seems to be related to two factors. The first is the initial value
by the single-pile failure load Q f . Fig. 1 illustrates the normalized of Rs at small loads when the elasticity dominates the behav-
load-settlement curves of a single pile and a 3 ⫻ 3 pile group ior of both the pile group and the single pile, affected by the
reported by Liu et al. 共1984兲. The normalized load-settlement elastic pile-soil-pile interaction 共Poulos and Davis 1980兲.
curves of the PGCs and PGFs in Table 2 and their corresponding The second is how Rs decreases with load, significantly af-
single piles are plotted in Figs. 2 and 3, respectively. fected by the nonlinearity of the load-settlement behaviors of
Based on the normalized load-settlement curves in Figs. 2 and the pile group and the single pile at large loads.
3, Rs at various normalized load levels can be evaluated with Eq. 4. Note that more than one-half of the cases consist of
共1兲. Figs. 4共a and b兲 show the relationships between Rs and the 3-diameter spaced, 9-pile groups for both PGCs and PGFs
normalized load, in increments of 共1 / 6兲共Q / Q f 兲, for the PGCs and 共Table 2兲. Therefore, the trend curves in Fig. 5 are particu-
PGFs, respectively. It is important to note that the values of Rs at larly representative of this group size. The statistics of Rs for
small loads, Q / Q f = 1 / 6 or 1 / 3, are associated with relatively the evaluated data set may not be extrapolated to nonevalu-
large errors because of errors in transcribing the original load- ated conditions, such as groups with a large number of piles
settlement curves in the researched literature to the normalized or with very long piles, and groups underlain by highly com-
load-settlement curves in this note. Such errors become negligible pressible soil layers.
at large loads 关e.g., at the working load 共1 / 2兲Q f or larger兴.
In general, values of Rs of both PGCs and PGFs tend to de-
crease with load 共Fig. 4兲. The single pile settlement is generally
smaller than the corresponding pile group settlement at the same Group Settlement Ratio versus Group Resistance
average load per pile when the load is relatively small 关e.g., Ratio
smaller than 共2 / 3兲Q f 兴. The settlement of a single pile tends to
increase with load at a larger rate than that of the corresponding Values of Rs for driven pile groups in sandy soils are found to be
pile group. In other words, nonlinear behavior of the load- documented as larger than unity in the literature, as demonstrated
settlement response of a single pile is more pronounced than that by the test results in Figs. 4 and 5, the empirical equations in
of a pile group, due to the average capacity per pile of a driven Table 1, and the pile-interaction analysis by Polous and Davis
pile group in sandy soil being greater than the corresponding 共1980兲. Intuitively, one may infer that the corresponding values of
single pile capacity 共O’Neill 1983; Zhang et al. 2001兲. As a result, group efficiency factor should be smaller than unity. At first
the single pile settlement is close to, or larger than, the pile group glance, this appears to be contradictory to the common under-
settlement at large loads. Correspondingly, Rs can decrease to or standing that group efficiency factor of a driven pile group in
even become smaller than unity as load increases. sandy soils is usually larger than unity due to soil densification
The values of the mean and coefficient of variation 共COV兲 of and additional contributions from the cap-ground contact for
Rs for the PGCs and PGFs at six different normalized loads, Q / Q f PGCs 共Vesic 1969; Polous and Davis 1980; O’Neill 1983; Zhang
varying from 1/6 to 1, are calculated and presented in Fig. 5. The et al. 2001兲. For instance, the average values of 3-diameter
values of Rs in Figs. 4 and 5 present the following features: spaced PGCs and PGFs in sandy soils are 1.40 and 1.41, respec-
Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Table 2. Field Load Tests on Driven Pile Groups
1050 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007
Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Fig. 1. Illustration of normalized load-settlement curves 共Case No.
Liu_CG1, adapted from Liu et al. 1984兲
tively, based on results of field and model load tests 共Zhang et al.
2001兲.
This section attempts to explain why both Rs and of a pile
group in sandy soils can be larger than unity. To clarify the pos-
sible confusion with Rs and , the concept of the group efficiency
factor is examined here. The group efficiency factor, , is defined
as the ratio of the pile group capacity, Qgf , to the sum of the
capacities of the individual piles, Q f . The pile group capacity is
Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
rg
Rr = 共2兲
rs
Based on the normalized load-normalized settlement curves for
the two cases shown in Fig. 7共a兲, the Rs values at different nor-
malized loads and the Rr values at different normalized settle-
ments are calculated by Eqs. 共1兲 and 共2兲, respectively, and plotted
in Figs. 7共b and c兲. Rs decreases with load, whereas Rr increases
with settlement. At small loads or small settlements 共on the initial
portion of the load-settlement curves兲, Rs is larger than unity,
whereas Rr is smaller than unity. At large loads or large settle-
ments 共on the tail portion of the load-settlement curves兲, Rs ap-
proaches or becomes smaller than unity, whereas Rr approaches
or becomes larger than unity.
The Rs values of the two cases are calculated with the empiri-
cal formulas summarized in Table 1. It is noted that e is assumed
to be 0.5 in Method I by Fleming et al. The calculated Rs values
are considered at 共1 / 2兲Q f , conventionally defined as the working
load 共i.e., Q / Q f = 0.5兲. At the working load level, the measured Rs
values fall within the range of the calculated Rs. In other words,
these empirical formulas for estimating Rs lead to good agreement
with the measured Rs values at the working load. The measured
Rs decreases with load so that the empirical formulas become
inappropriate at larger loads since the influence of load level or
nonlinear behavior of the pile load-settlement response is ignored
in these formulas. Mandolini and Viggiani 共1997兲 proposed that
the group settlement-interaction factor should only be applied to
Fig. 4. Relationships between Rs and normalized load the elastic component of settlement of an adjacent pile, as the
plastic component of settlement is due to a localized phenomenon
and is not transmitted to the adjacent piles. Therefore, the effects
of group interaction on settlement of pile groups become less
significant as load increases, particularly at very large loads under
which nonlinear behavior of the pile load-settlement response
dominates.
The values of the two cases are calculated based on the Qgf
values determined by the slope tangent method 共1兲 and the ref-
erence displacement method 共2兲: 1 = 0.97 and 2 = 1.10 for case
Briaud_FG; 1 = 0.85 and 2 = 1.34 for case Liu_CG2. The re-
ported values are 0.99 for case Briaud_FG by Briaud et al.
共1989兲 and 1.38 for case Liu_CG2 by Liu et al. 共1984兲. It is found
that the values from the reference displacement method are
closer to the reported values. Fig. 7共c兲 shows the values
calculated using the reference displacement method and reported
by Briaud et al. 共1989兲 and Liu et al. 共1984兲.
The variations of Rs with load and Rr or with settlement in
Figs. 7共b and c兲 demonstrate that
1. The value of Rs compares settlement at the working load and
Fig. 5. Variations of the mean and COV of Rs of pile groups with
is often larger than unity.
normalized load
2. The value of compares failure loads, which occur at dif-
ferent settlements for pile groups and their respective single
piles.
Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Fig. 7. Case study: 共a兲 relationships between normalized load and
normalized settlement for single piles and pile groups; 共b兲 relation-
ships between Rs and normalized load; and 共c兲 relationships between
Rr or and normalized settlement
Fig. 6. Failure criteria for interpreting pile group capacity: 共a兲 slope localized nature of the plastic component of settlement,
tangent method; 共b兲 limiting displacement rate method; and 共c兲 which is not transmitted to the adjacent piles.
reference displacement method. Q=applied load; S=pile group 3. The variations of Rs and Rr with load or settlement are com-
settlement; Sr=reference failure settlement; ⌬S=pile group settlement pared to clarify possible misunderstanding between Rs and
increment at a load step; and t=loading time at a load step. of driven pile groups in sandy soils. The value of Rs com-
pares settlement at the working load and is often larger than
unity. The value of compares failure loads, which occur at
3. is usually larger than unity due to soil densification and different settlements for pile groups and their respective
additional contributions from the cap-ground contact for single piles. is usually larger than unity due to soil densi-
PGCs. fication and additional contributions from the cap-ground
contact for PGCs.
Conclusions
Acknowledgments
The settlement ratio Rs of driven pile groups in sandy soils has
been studied based on results of field load tests on 31 pile groups This research was substantially supported by the Research Grants
and their corresponding single piles. The following conclusions Council of the Hong Kong SAR 共Project No. HKUST6126/03E兲.
can be drawn:
1. For both PGCs and PGFs, the mean of Rs decreases with the
load level, whereas the COV of Rs increases with the load References
level. The contribution of cap-ground contact to Rs is statis-
tically not significant. Briaud, J. L., Tucker, L. M., and Ng, E. 共1989兲. “Axially loaded 5 pile
2. The decrease of Rs with load is significantly affected by the group and single pile in sand.” Proc., 12th Int. Conf. on Soil Mechan-
Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
ics and Foundations Engineering, Vol. 2, Rio de Janeiro,1121–1124. deep foundations.” NCHRP Rep. No. 507, Transportation Research
Davisson, M. T. 共1972兲. “High capacity piles.” Proc., Soil Mechanics Board of the National Academies, Washington, D.C.
Lecture Series on Innovations in Foundation Construction, ASCE Il- Poulos, H. G. 共1989兲. “Pile behavior: theory and application.” Geotech-
linois Section, Chicago, 81–112. nique, 39共3兲, 365–415.
Fleming, W. G. K., Weltman, A. J., Randolph, M. F., and Elson, W. K. Poulos, H. G. 共2000兲. “Analysis and design of pile foundations.” Con-
共1992兲. Piling engineering, Wiley, New York. tinuing professional development course, The Hong Kong University
Ismael, N. F. 共2001兲. “Axial load tests on bored piles and pile groups in of Science and Technology, Hong Kong, 8–9.
cemented sands.” J. Geotech. Geoenviron. Eng., 127共9兲, 766–773. Poulos, H. G., and Davis, E. H. 共1980兲. Pile foundation analysis and
Liu, J. L., Yuan, Z. L., and Zhang, G. P. 共1984兲. “Research on working design, Wiley, New York.
mechanism and capacity of pile group.” Rep. No. 2–1, China Acad- Skempton, A. W. 共1953兲. “Discussion: Piles and pile foundations.” Proc.,
emy of Building Research, Beijing. 3rd ICSMFE, Vol. 3.
Mandolini, A., and Viggiani, C. 共1997兲. “Settlement of piled founda- Vesic, A. S. 共1969兲. “Experiments with instrumented pile groups in sand.”
tions.” Geotechnique, 47共4兲, 791–816. Performance of deep foundations, ASTM STP 444, ASTM, West Con-
McCabe, B. A., and Lehane, B. M. 共2006兲. “Behavior of axially loaded shohocken, Pa., 177–222.
pile groups driven in clayey silt.” J. Geotech. Geoenviron. Eng., Zhang, L. M., Tang, W. H., and Ng, C. W. W. 共2001兲. “Reliability of
132共3兲, 401–410. axially loaded driven pile groups.” J. Geotech. Geoenviron. Eng.,
O’Neill, M. W. 共1983兲. “Group action in offshore piles.” Geotechnical 127共12兲, 1051–1060.
practice in offshore engineering, S. G. Wright, ed., ASCE, New York, Zhang, L. M., Tang, W. H., Zhang, L. L., and Zheng, J. G. 共2004兲.
25–64. “Reducing uncertainty of prediction from empirical correlations.” J.
Paikowsky, S. G. 共2004兲. “Load and resistance factor design 共LRFD兲 for Geotech. Geoenviron. Eng., 130共5兲, 526–534.
Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org