Professional Documents
Culture Documents
Employment of The P-Multiplier in Pile-Group Analysis
Employment of The P-Multiplier in Pile-Group Analysis
Abstract: This paper discusses the variation of the P-multiplier (Pm ) used with the p-y curve to assess the response of a pile group under
lateral loads, which is a crucial topic for the design of bridge pile foundations. Pm is influenced by the site geotechnical conditions (i.e., soil
profile, type and properties), pile front and side spacings, and pile-group deflection. The presented study shows the needs to incorporate these
factors with the recommended sets of Pm to avoid any compromise or uncertainty when Pm is treated as a single (unique) value based only on
pile spacings. The current study addresses these influential elements using the strain wedge (SW) model technique, suggested Pm values, and
data collected from full-scale pile-group load tests. The experimental results show that Pm is not unique and must be assessed based on the site
geotechnical conditions along with the pile-row front and side spacings. Because the employed Pm values must be a function of these in-
fluential factors, additional full- and model-scale load tests with different pile spacings and soil types might be required. The paper also
emphasizes that using other techniques, such as the SW model, in addition to the P-multiplier could increase the confidence in the predicted
pile-group lateral response. DOI: 10.1061/(ASCE)BE.1943-5592.0000196. © 2011 American Society of Civil Engineers.
CE Database subject headings: Pile foundations; Pile groups; Lateral loads; Strain; Deflection.
Author keywords: Pile foundation; Pile group; Lateral load; P-multiplier; Strain wedge model.
Introduction pile group if installed at different sites (i.e., different types or layer
thicknesses of the near-surface soils). This explains the significant
The design of a pile group under lateral loads is a very common variations in the Pm values recommended by different researchers
problem in foundation engineering. The most common method is and agencies (Fig. 2). At present, the use of the Pm technique in
the use of the P-multiplier (Pm ) with the p-y curves, as employed in pile-group design (with no data from a full-scale load test) relies
a number of computer programs. There are other applicable meth- only on the ratio of the pile-row spacing in the loading direction
ods, such as the group reduction factor method (Davison 1970) that to the pile diameter (SL =D). However, the suggested P-multipliers
is based on pile-group tests in sand and recommends a linear re- could be affected by ignoring the influence of (1) pile spacing in the
duction factor for the modulus of subgrade reaction [varies from same row (ST ), which is normal to the loading direction; (2) soil
25% for the three-diameter (3D) pile spacing to 100% for the type; (3) soil layer thickness and properties (soil profile); (4) level
8D pile spacing]. Ooi and Duncan (1994) developed a simplified of lateral loading; and (5) pile arrangements. For the same type of
group amplification procedure for the design of groups of piles and soil, the Pm values estimated based on the traditional p-y curves
drilled shafts based on the theories of Poulos (1971) and Focht and could be substantially different from those obtained from site-
Koch (1973). The amplifications factors are multiplied by the sin- specific p-y curves.
gle pile values of deflection and maximum moment. Comparisons In addition to the experimental data presented by Hughes et al.
with different methods used for pile-group analysis are presented (1978) for the horizontal stress distribution in the soil around lat-
by Ooi and Duncan (1994). erally loaded piles, Fig. 3 shows the distinctive variation of the hori-
The pioneering work of Brown et al. (1998) established the ap- zontal stress in the soil zone around the pile group based on the
proach of the P-multiplier and its employment in the analysis of finite-element analysis (Rao et al. 1998). The stress zone expands
pile groups as a constant reduction factor to reduce the p-value (soil horizontally as a function of pile arrangements and decreases with
resistance, F=L) of the isolated pile p-y curve (Fig. 1). The values of depth (Fig. 3). The leading work by Brown et al. (1988) established
Pm are obtained from full-scale load tests of pile groups with the concept of interaction among closely spaced piles in a pile
(mostly) three-pile diameter spacings in both directions and depend group (i.e., group action) based on the horizontal shear zone over-
on associated site geotechnical conditions. Pm values obtained from lap among the neighboring passive soil wedges in the group.
full-scale load tests are empirically determined through a number of The horizontal passive soil wedge overlap decreases with depth,
trials to match the calculated pile-group deflection with the mea- and the larger the pile-group deflection, the deeper and wider the
sured deflection. As a result, Pm is expected to change for the same zone of stress overlapping. Furthermore, the horizontal overlap of
stresses among the piles at a certain depth starts with a large value
1 of Pm (close to 1), where overlapped stresses are very small at that
Associate Professor, Dept. of Civil and Environmental Engineering,
Univ. of Alabama, Huntsville, AL 35806 (corresponding author). E-mail: particular depth, and then decreases with the increase of group de-
ashour@eng.uah.edu flection to some extent. As observed in a number of pile-group tests
2
Graduate Student, Research Assistant, Dept. of Civil and Environmen- (Ilyas et al. 2004; Brown and Reese 1985; Rollins et al. 2005), the
tal Engineering, Univ. of Alabama, Huntsville, AL 35806. pile-group interaction is very limited at the early stage of lateral
Note. This manuscript was submitted on April 14, 2010; approved on
loading (small deflection) (i.e., overlapping stresses in the sur-
November 6, 2010; published online on August 15, 2011. Discussion per-
iod open until February 1, 2012; separate discussions must be submitted for rounding soil are very small), and the individual piles in the group
individual papers. This paper is part of the Journal of Bridge Engineering, respond similar to a single pile. Increasing the pile spacing from
Vol. 16, No. 5, September 1, 2011. ©ASCE, ISSN 1084-0702/2011/5-612– 3D to 5D would extend the stage of pile-group deflection with
623/$25.00. Pavg: ¼ Psingle before the development of the group action and
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Stress Zone @ 10D Below
Stress Zone @ G.L.
G.L.
D 3D
2D
2D
3D
7D
@ G.L.
10D S 2D
D S
3D
4D
6D 1.5D
8D 2D
Fig. 1. Current practice of P-multiplier (Pm ) in pile-group analysis:
(a) concept of the P-multiplier applied to a pile in a group, as suggested
by Brown et al. (1988); (b) piles in a group subjected to the P-multiplier Fig. 3. Stress zones in soil at different depths around rigid piles in-line
and side-by-side in pile groups (Rao et al. 1998, ASCE)
the degradation of Pm values (Ilyas et al. 2004). Once pile interac- with the progress of lateral loads, current practice applies Pm as a
tion and stress overlapping with front and side piles occurs, Pm de- single value along the whole length of the pile (Fig. 1). Conse-
velops and varies rapidly, especially in close-spaced pile groups. quently, the value of the Pm suggested for the same pile group based
However, the pile interaction could develop with the back rows ear- on pile spacings may change in response to the change of the
lier and faster than the front row, which is realistic and agrees with pile length to obtain a good agreement with the measured data
the test results presented by Rollins et al. (2005). (Rao et al. 1998).
Because the actual depth of the zone of interaction among the Currently, Pm is utilized as a function of pile-row spacing in
piles in a group (i.e., soil stress overlap) is unknown and increases the loading direction, SL [AASHTO 2007; U.S. Army Corps of
Engineers (USACE) 1993]. An interpolation between suggested
Pm values should be conducted to determine Pm for other pile spac-
1.2
ings, as is the case in AASHTO (2007), which provides Pm for only
3D and 5D pile-row spacings. Such suggestion may not apply to
1
one-row pile-group analysis loaded in the longitudinal direction
(side by side) where no pile-row spacings exist. Moreover, the
Pm value recommended for the side-by-side pile group must be ad-
0.8 justed when the one-row pile group is loaded in-line (Fig. 4). As
observed in experimental results, Gandhi and Selvam (1997) ad-
P-multiplier
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
The SW model calculates the ever-changing geometry of the
mobilized passive wedge of soil in front of the pile [Figs. 5(a)
and 5(d)] that varies according to the associated level of loading
and becomes deeper and wider (i.e., a larger fanning angle) with
Po increasing lateral loads and deflections. Compared with the elastic
theory [Fig. 5(c)], the stress distribution at a specific depth extends
Longitudinal Loading horizontally in a uniform fashion along the face of the mobilized
(Side-by-Side)
passive soil wedge. Therefore, no pile interaction is considered in
the SW model at a certain depth (i.e., Pm at that depth ¼ 1) before
Po the stresses in the surrounding soil, caused by neighboring piles,
start to overlap [Figs. 5(b) and 5(c)]. Such a pile interaction mecha-
nism is consistent with the results obtained from a number for
Transverse Loading pile-group tests (Ilyas et al. 2004; Brown and Reese 1985; Rollins
(In-line) et al. 2005) that exhibit no pile-group interaction at low levels of
loading. (i.e., small deflections). This characterization allows the
Fig. 4. Pile-group arrangement evaluation of the interaction among the neighboring piles in the
pile group at different depths and under increasing pile-head deflec-
Brown, McVay and their coworkers have significant contributions tion with no need for multipliers. In addition to the Pm technique,
to the improvement of the suggested Pm values. the SW model analysis can be used as an alternative method
The strain wedge (SW) model (Ashour et al. 2004) implements (AASHTO 2007) for the analysis of pile group to reduce the design
the interaction mechanism in the pile group. As shown in Fig. 5, the level of uncertainty.
passive soil wedges in front of the piles in the SW model develop as The interaction among the piles in a group decreases with depth
a function of the soil properties and the stress level in soil (i.e., pile [Fig. 5(d)], thus generating lower values of Pm near the ground sur-
deflection). This allows the assessment of the overlapping passive face (or pile head) and greater values at deeper points. Conse-
soil wedges of adjacent piles and the calculation of the additional quently, the values of Pm increase with depth in the same soil
stress-strain induced by pile interactions (Fig. 5). and become more complex in layered soils.
F1
C
φ
Soil element considered τ τ
∆σh
m
φm
∆σh
as if in a conventional
τ triaxial test
p Pile p
φm Pile
Real stressed zone
(b)
Pile Pile
(d) (c)
Fig. 5. Characterization of the pile-group interaction and shear stress zones in the SW model analysis: (a) isolated pile; (b) pile initial interference and
soil wedge overlap; (c) three-dimensional passive soil wedge overlaps in a pile group in uniform soil; (d) uniform stress overlap in the SW model
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Strain Wedge Model with Pile Group σd E Eg
The main concept associated with the SW model is that traditional
one-dimensional beam on elastic foundation (BEF) pile response
(∆σh)g
parameters can be characterized in terms of three-dimensional ∆σhf SL=1
soil-pile interaction behavior. The SW model is based on the
∆σh
stress-strain and strength behavior of the soil as assessed in the tri-
axial test [Fig. 5(a)] and soil effective stress analysis (Norris 1986; ε εg ε
Ashour et al. 1998). The geometry and the size of the passive
soil wedge vary according to the stress-strain level in the soil in Triaxial Test at σ3 (= σvo)
front of the pile and the level of lateral load at the pile head
[Figs. 5(b) and 5(d)]. The major principal stress-strain (Δσh ε)
curve of the soil is determined at any depth (i.e., the ever-changing ϕm
Shear Stress
Young’s modulus, E ¼ SLΔσhf =ε). Based on the shape of the pas- ϕ (Pile in Group)
sive soil wedge and the pile properties (shape, stiffness, and head-
fixity), the modulus of subgrade reaction (E s ) (i.e., p-y curve) is ϕm
evaluated to create a full profile of E s values along the pile length (Isolated Pile)
at that particular step of lateral loading (Ashour and Norris 2000):
=
vo
2 σ3
Δσhf ¼ 2su ðclayÞ ð3Þ Fig. 6. Variation in soil stress-strain and Young’s modulus as result of
wedge overlap in a pile group
where SL = deviatoric stress level in soil; φ = full effective friction
angle; φm = mobilized effective friction angle; Su = undrained shear
strength; Δσhf = horizontal (deviatoric) stress change at failure; and deflects more under the same lateral load and the soil wedge gets
σ
vo = vertical effective stress. deeper. Figs. 5–7 describe the mechanism of the group action as it
As seen in Figs. 5(b) and 5(c), the piles (i.e., passive soil develops among neighboring piles and the associated reduction in
wedges) in the pile group interact horizontally in an amount that soil resistance. The induced εg and Eg (Fig. 6) are determined as
varies with depth. Therefore, the varying overlap of the wedges follows:
of neighboring piles in different sublayers over the depth of the
interference [Fig. 5(d)] and the associated increase in soil stress- ðεg Þi ¼ εi þ Δεi ð6Þ
strain can be determined as a function of the amount of overlap.
As pile lateral load increases, the wedges grow deeper and fan ðSLg Þi ðΔσhf Þi
ðE g Þi ¼ ð7Þ
out horizontally, thus causing a further change in overlap and group ðεg Þi
interference. The current average value of horizontal stress change
ðΔσh Þg and the associated stress level and strain (SLg and εg ) ac- The angles and dimensions of each passive wedge sublayer (φm ,
cumulated at the face of the passive wedge due to overlapping in a β m , and BC) are modified for the group effect according to the cal-
particular soil sublayer i (sand or clay) [Fig. 5(c)] are determined as culated value of SLg and εg . The new ðE s Þg of soil sublayer, i, is
follows: expressed as
ðΔσh Þg ¼ SLg Δσhf ð4Þ ðAg Þi Dðεg Þi ðE g Þi
½ðE s Þg i ¼ ð8Þ
δ i ðh xi Þ
X
ðSLg Þi ¼ SLi ð1 þ Rj Þ1:5 ≤ 1 ð5Þ Over the depth of interference (stress overlapping) among the piles
and as shown in Fig. 7, ðEs Þg of the pile in question will be less than
where j = number of neighboring passive wedges in soil layer that
overlap the wedge of the pile in question; and R (less than 1) = ratio
between the length of the overlapped portion of the face of the pas-
p
sive wedge and the total length of the face of the passive wedge Isolated Pile
(BC) and derives from all neighboring piles on both sides and
in front of the pile in question [Fig. 5(b)]. SLg and εg in each soil Es Pile in a Group
sublayer due to passive wedge overlapping are determined for each
pile in the group. Here εg is ≥ ε of the isolated pile and is deter-
mined based on the stress-strain relationship (Δσh versus ε). Com-
pared to the isolated pile, Fig. 6 explains the increase of SL and ε to (Es)g
SLg and εg in each soil sublayer as a result of soil wedge overlap.
The increase of stress-strain in soil is accompanied by an increase
in φm (i.e., larger fanning angle of the wedge) and a reduction in E y
and E s of the soil sublayer at that depth (Figs. 6 and 7). Because of
Fig. 7. Reduction in E s of an isolated pile due to group action
the soil wedge overlapping and the loss in soil resistance, the pile
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
E s of the isolated pile. The new profile of ðE s Þg will be used to Loading Direction
analyze the pile as a BEF (Fig. 8). The parameter Ag of sublayer
i at depth x is a function of the pile and passive soil wedge geometry Type 3 Type 4 Type 3
(including the wedge depth, h), the shear stress (τ ) at the pile side–
soil interface, and the deviatoric stress change, ðΔσh Þg : Leading Row
SL Type 1
BC i S1 2τ i S2 Type 1 Type 2
ðAg Þi ¼ þ ð9Þ
D ½ðΔσh Þg i Trailing Row
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
80 80
20 20
Load Load
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Deflection (mm) Deflection (mm)
80 80
Measured Pile Type 3 Measured Pile Type 4
Computed Pile Type 3 -SWM Computed Pile Type 4 -SWM
60 60
40 40
20 20
Load Load
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Deflection (mm) Deflection (mm)
Fig. 10. Lateral response of piles (by pile location) in a pile group using the SW model at the Treasure Island, California, test (Rollins et al. 2005,
ASCE)
Table 1. Soil Properties Employed in the Analysis of Treasure Island Test (mostly closely spaced pile groups of 3D), one might argue that
(Data from Rollins et al. 2005) the pile row side spacing effect is implicitly considered (embedded)
in the Pm values obtained from those tests. Nevertheless, changing
Soil layer Soil Unit weight, ε50 Φ
thickness (m) type γ ðkN=m3 Þ (%) (degree)
Su ðkN=mÞ2 the pile row side spacing would affect the overlap shear zone and
the induced stresses, as shown in Figs. 5(b) and 5(c).
0.5 Sand 19.5, 10 0.4 38 — Cox et al. (1984) presented the effect of the pile row side spac-
2.5 Sand 10.3 0.4 38 — ing on the group efficiency as monitored in a number of pile-group
1 Sand 10.3 0.5 36 — load tests to indicate the link between Pm and the pile side spacing.
2 Sand 10.3 0.8 33 — However, no Pm was developed to achieve the observed and sug-
1.5 Sand 10.3 0.7 34 — gested group efficiencies. The results presented by Cox et al. (1984)
1.75 Clay 9.5 1.0 0 20 prove the influence of the pile side spacing on the utilized Pm to
1 Sand 10.3 0.8 33 — satisfy the obtained group efficiency (Ge ).
1.7 Clay 9.5 1.0 0 20 The influence of the pile transverse spacing on group efficiency
and the value of Pm varies from one soil to another. The effect of the
pile transverse spacing on the group action and the Pm is almost
(Fig. 5), unlike in clay soil that is highly influenced by the pile front limited to 3D in soft clay and slightly expands with stiffer clay.
interference because of the smaller effective friction angle. Accord- The front interference among piles (in-line loading) influences
ing to the previous discussion on the decrease of the overlap shear the group efficiency more than the pile-side interference in clay
zone with depth, Pm in the same soil increases with depth to reach soils; this agrees with the experimental findings reported by
1.0 at the point where the wedge overlap diminishes. Cox et al. (1984) in soft clay. A similar pattern occurs with loose
and dense sands, for which the pile-side interference becomes more
Effect of Pile Row Side (Transverse, S T ) Spacing
influential in denser sands.
on P m
Table 3 shows the increase in the pile-group displacement stiff-
Very limited research has investigated the effect of the pile row side ness (K d ) in dense and loose sands. The 3 × 3 pile group maintains
spacing on group action. With most of full-scale pile-group load a constant pile-front spacing of 3D along with a pile side spacing
tests conducted on equal longitudinal and transverse spacings that varies from 2D to 6D. The piles and sands used in the
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Table 2. Pm Values Assessed by Different Research from Full- and Model-Scale Pile-Group Load Tests
Pm
Test type Pile group Soil type 1st row 2nd row 3rd row Reference
Full scale 3×3 Soft clay (Su ¼ 20 kPa) 0.65 0.50 0.48 Ilyas et al. (2004)
Full scale 2×2 Soft clay (Su ¼ 20 kPa) 0.9 0.5 — Meimon et al. (1986)
Full scale 3×3 Med. stiff clay (Su ¼ 48 kPa) 0.6 0.4 0.4 Rollins et al. (1998)
Full scale 3×3 Stiff clay (Su ¼ 72 kPa), for deflections 30 and 51 mm, respectively 0.7 0.6 0.5 Brown et al. (1988)
Centrifuge 3×3 Loose sand (Dr ¼ 33%) 0.65 0.45 0.35 McVay et al. (1995)
Centrifuge 3×3 Med. dense sand (Dr ¼ 55%) 0.8 0.45 0.3 McVay et al. (1995)
Full scale 3×3 Med. dense sand (Dr ¼ 50%) 0.8 0.45 0.3 Rollins et al. (2005)
Full scale 3×3 Dense sand (Dr ¼ 90%) 0.8 0.4 0.3 Brown et al. (1988)
Full scale 4×4 Silty sand and cemented sand 0.8 0.7 0.3 Ruesta and Townsend (1997)
y/D y/D
0 0.1 0.2 0.3 0 0.1 0.2 0.3
1 1
Longitudinal spacing = 3D Longitudinal spacing = 3D
Transverse spacing = 3D Transverse spacing = 3D
0.8 0.8
P-multiplier, Pm, at 1.8 m
Loose Sand
0.4 0.4
Dense Sand
0.2 0.2 Loose Sand
Dense Sand
0 0
0 2 4 6 8 10 0 2 4 6 8 10
Pile Deflection, y, cm Pile Deflection, y, cm
y/D y/D
0 0.1 0.2 0.3 0 0.1 0.2 0.3
1 1
Longitudinal spacing = 3D Longitudinal spacing = 3D
Transverse spacing = 3D Transverse spacing = 3D
0.8
P-multiplier, Pm, at 1.8 m
0.8
0.6
0.4
0.6
0.2
Lead Row Back Row
0.4 0
0 2 4 6 8 10 0 2 4 6 8 10
Pile Deflection, y, cm Pile Deflection, y, cm
Fig. 11. Variation of Pm with soil type (sand and clay) in a 3 × 3 pile group at 1.8 m depth
Table 3. Pile-Group Displacement Stiffness (K d ) at 2.54 mm (0.1 in.) comparison presented in Table 3 have the same properties as those
Deflection used in the previous section. As presented in Table 3, the changes in
K d (kN=mm) for different pile side spacings the pile-group side spacing resulted in a significant increase of the
Soil type 2D 3D 4D 5D 6D pile-group displacement stiffness (K d ) up to 34% and 20% in dense
and loose sand, respectively.
Dense sand ðφ ¼ 40°Þ 103 114 128 135 138
Fig. 12 shows the effect of the pile row side spacing on the Pm
Loose sand ðφ ¼ 30°Þ 55 62 64 65 66
determined at 1.5 m below the ground surface (pile head) in loose
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
y/D y/D
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
1 1
φ = 40o φ = 40o
Longitudinal spacing = 3D Longitudinal spacing = 3D
P-multiplier, P m, at 1.5 m
0.8
Lead Row Back Row
Load 0.6 Load
0.6 Transverse (Side) Spacing
3D Transverse (Side) Spacing
4D 0.4 3D
5D 4D
5D
0.4
0.2
0.2 0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Pile Deflection, y, cm Pile Deflection, y, cm
y/D y/D
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
1 1
φ = 30o φ = 30o
Longitudinal spacing = 3D Longitudinal spacing = 3D
0.8
P-multiplier, P m, at 1.5 m
P-multiplier, Pm, at 1.5 m
0.8
Lead Row Back Row
Load 0.6 Load
Transverse (Side)Spacing
3D Transverse (Side)Spacing
4D 0.4 3D
5D 4D
0.6 5D
0.2
0.4 0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Pile Deflection, y, cm Pile Deflection, y, cm
Fig. 12. Effect of pile row side (transverse) spacing on Pm at 1.5 m depth
and medium-dense sands. The pile front spacing employed in lateral load (i.e., deflection) on the pile group is associated with
Fig. 12 remains constant at 3D center-to-center. As presented in the development of interaction among adjacent piles and the deg-
Fig. 12, Pm increases considerably in response to the increase of radation in Pm values. Once the horizontal stress in the overlap
the pile side spacing, especially with the leading pile row. In reality, shear zone of soil at that depth reaches its ultimate value (at larger
the pile spacing, level of deflection, and local geotechnical condi- deflection), the Pm sustains a constant value (flat curve). The values
tions govern the depth of interference among adjacent piles (i.e., the of Pm presented in Table 2 were back-calculated over a certain
horizontal shear zone of overlap). As seen in Fig. 12, a number of range of pile-group deflection, as indicated in related references.
the Pm curves maintained a constant value of 1.0 at small pile de- The constant value of Pm that is applied along the pile is not esti-
flection as an indication of no pile interference (soil sublayer over- mated based on the pile-group response (i.e., pile rows) at the early
lapping) at that depth and up to that level of pile deflection. stage of loading (small group deflection) when the group efficiency
Increasing the pile deflection is accompanied by a decrease in is close to 100% or when the group action starts to develop and
Pm as a result of soil wedge overlap at that depth. The deeper
degrades over a short range of deflection, especially with the
the soil sublayer in the dominant near-surface soil, the larger the
closely spaced pile groups, as seen in the results presented by
pile-head deflection before soil sublayer overlapping occurs.
Rollins et al. (1998, 2005). Also, the pile bending stiffness could
influence the rate of pile interactions and how Pm develops. Such an
P m Variation with the Pile-Group Level of Loading effect, which is not discussed in this paper, was addressed by
The level of interaction among adjacent piles in a group varies with Brown et al. (1988) and is considered in the SW model analysis.
the level of lateral loading. Such a behavior occurs during full-scale The shape of the passive wedge of soil [Fig. 5(a)] varies accord-
load tests when the individual piles in the group act as isolated piles ing to the associated level of loading and becomes deeper and wider
at lower levels of loads (especially with larger pile spacings) fol- (larger fanning angle) with the increase of lateral load and deflec-
lowed by a softer response (i.e., group action effect) with the in- tion. This characterization allows the evaluation of the interaction
crease of lateral loads. Fig. 12 of Pm at 1.5 m depth versus pile among neighboring piles in the pile group at different depths and
deflection exhibits no pile interaction (Pm ¼ 1) at that depth up under increasing pile-group deflection in the near-surface lay-
to a particular value of pile deflection. However, increasing the ered soils.
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Table 4. Variation of Group Efficiency (Ge ) with Clay Type
Soft clay (Su ¼ 20 kPa) Medium-stiff clay (Su ¼ 48 kPa)
Longitudinal loading Transverse loading Longitudinal loading Transverse loading
Pile ratio (L=D) Pg (kN) Ge Pg (kN) Ge Pg (kN) Ge Pg (kN) Ge
10 16.7 0.98 11.1 0.68 45.1 0.86 31.5 0.62
20 26.8 0.94 18.2 0.63 81.0 0.88 64.2 0.68
30 44.1 0.97 32.8 0.73 96.8 0.94 56.1 0.68
Note: Pile-group lateral resistance is calculated at 25.4 mm (1.0 in.) cap deflection.
Effect of Pile Arrangement in the Group and Lateral pile-group load test to assess appropriate Pm values at the site in
Load Direction question. The SW model results presented in this study were ob-
tained from the computer program SWM6.2 (Ashour and Norris
The laterally loaded pile-group analysis is a soil-pile interaction 2007). The current SW model deals with a rectangular matrix
problem that is governed by the soil and pile properties. Hence, for pile arrangement (no staggered distribution) with free- or
the direction of lateral load (i.e., the arrangement of piles) should fixed-head pile conditions. The failure to achieve the appropriate
affect the response of the individual piles in the group. Such an
stability convergence with any of the four pile types in the SW
effect can be clearly observed with a one-row pile/shaft group
model leads to the drop of that load increment and moving to
loaded first in the longitudinal direction (side-by-side) and then
the next one.
in the transverse direction (in-line) (Fig. 4). Rao et al. (1996,
1998) experimentally showed the effect of pile arrangement on Chaiyi, Taiwan, Pile-Group Load Test in Sand/Clay
pile-group resistance that proved the need for different values of Layered Soil
Pm according the pile arrangement in the group. This pattern of
pile-group behavior agrees with the results shown in Table 4 using The Chaiyi full-scale load test (Brown et al. 2001) was conducted
a 3 × 1 pile group of steel pipe piles (D ¼ 0:324 m and on 3 × 2 fixed-head bored piles, which had piles 1.5 m in diameter,
EI ¼ 28:6 MN · m2 ) embedded in different types of soil. Currently, 34.9 m in length, and 3D pile spacing center-to-center in both di-
there is no clear guidance on how to assume Pm for a one-row pile rections. The bored pile initial bending stiffness (EI) was
group. With the pile-group interference mechanism described in 6:8 GN · m2 . Because of the degradation in the pile’s EI value
Fig. 5, the front and side interference among the piles in one- as a result of concrete cracking, the nonlinear material modeling
row pile group can be determined. Reese et al. (2006) discussed of reinforced concrete adopted in the SW model analysis is utilized
the group efficiency (reduction factor) for longitudinal and trans- (Ashour et al. 2001). The base of the 2-m-thick pile cap is located at
verse loading (side-by-side and line-by-line). However, no Pm was the ground surface. To obtain a good match with the measured pile-
developed to achieve the observed and suggested group efficien- group deflection (Fig. 13), Brown et al. (2001) suggested Pm values
cies. The results presented by Reese et al. (2006) proves the sig- of 0.5, 0.4, and 0.3 with the first (lead), second, and third pile rows,
nificant effect of the transverse (side) pile spacing on the utilized respectively, with adjusted p-y curves. Table 5 presents the soil
Pm to satisfy the obtained group efficiency (Ge ). Table 4 presents properties used in both SW model and presented by Brown et al.
the variation in the group efficiency (i.e., employed Pm ) of the same (2001). AASHTO (2007) provides general P-multipliers of 0.7,
pile group installed in different soils. The group efficiency also 0.5, and 0.35 for the same sequence and distribution of pile rows.
changes with the pile length and the Pm employed along the whole Therefore, the pile-group response calculated according to the Pm
pile (as implemented in current practice) and may change with values recommended by AASHTO will come out stiffer than the
L=D. Table 4 shows the diversity of the calculated group efficiency measured results if the same adjusted p-y curves are used. Using
(i.e., Pm ) in the same soil that accompanies the variation of the pile the traditional p-y curves would require a different set of Pm to
side spacing. Cox et al. (1984) presented experimentally group ef- match the measured response.
ficiency for longitudinal loading larger than that of the transverse
loading with pile spacings that range from 1:5D to 4D in soft clay. Houston Pile-Group Load Test in Sand
Rao et al. (1998) presented different pile-group behavior in soft
clay and underlined the effect of pile length (L=D of short and long Brown et al. (1988) conducted a full-scale lateral load test on a free-
piles) on group efficiency (i.e., Pm ). Despite the limited changes in head 3 × 3 pile group. The 3D center-to-center pile group was em-
the group efficiency as a result of the pile length (L=D) variation, Ge bedded in approximately 3 m of dense to very dense uniform sand
in Table 4 for the pile group under transverse (in-line) loading is overlying overconsolidated clay. The piles consisted of steel pipe
significantly less than the longitudinal loading in both types of clay. with an outside diameter of 0.275 m, a wall thickness of 9.3 mm,
Therefore, Pm in the same soil should account for the pile-group a 13-m embedded length, and a bending stiffness (EI) of 1:9×
arrangement. 104 kN · m2 . The same soil properties suggested by Morrison
and Reese (1986) are used in the SW model analysis to obtain
the pile-group response presented in Fig. 14. Brown et al. (1988)
Effect of Site Geotechnical Conditions used Pm values of 0.8, 0.4, and 0.3 with the first (lead), second, and
third pile rows, respectively, as shown in Fig. 14. AASHTO rec-
A number of case studies are presented to exhibit the capability of ommends different P-multipliers of 0.7, 0.5, and 0.35 for the same
the technique to assess the pile-group response with and without pile group based on the pile-row front spacing. Fig. 15 displays the
Pm . Pm values employed in pile-group analysis using other com- variation of Pm with pile deflection at a depth of 0.9 m below the
puter programs were obtained through a number of trials (runs with pile head for the leading and back rows, as determined from the SW
different Pm values) to reach a reasonable agreement with measured model. Pm decreases with increasing deflection at that depth until it
results. This is consistent with the need to conduct a full-scale approximately levels.
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
12000
4.5 m
3 x2 Pile Group
Fixed head
4000
Measured (Brown et al. 2001)
SW Model (No P-multipliers)
Loading Direction Suggeted from the Test
(a) (Pm = 0.5, 0.4, 0.3)
0
0 10 20 30 40
Group Deflection, Yg, mm
(b)
Fig. 13. Measured and computed pile-group response of the Chaiyi, Taiwan, load test: (a) six 1.5-m-diameter bored pile group (fixed head); (b) lateral
response of bored single pile and pile group
Table 5. Soil Properties Employed in the SW Model Analysis for the Chaiyi, Taiwan, Test (Data from Brown et al. 2001)
Soil layer thickness (m) Soil type Unit weight, γ ðkN=m3 Þ ε50 (%)a φ (degree) Su ðkN=m2 Þa
3 Fine sandy silt (SM) 19 0.5 35 —
5 Silty fine medium dense (SM) 9.2 0.5 35 —
4 Silty clay (CL) 9.2 0.7 — 60
5 Silty fine sand (SM) 9.4 0.5 34 —
8 Clayey silt with medium dense sandy silt (ML=CL) 9.2 0.5 34 —
7 Silty clay with silty fine sand (CL) and (SM=CL) 9.2 0.5 — 115
11 Silty clay with silty fine sand (SM=CL) 9.2 0.5 — 121
a
The SW model uses ε50 for sands (Ashour et al. 1998).
100 1
Trailing Row
Average Load per Pile, Pave,kN
Leading Row
0.8
75
P-multiplier
0.6
P-multiplier at 0.9 m below pile head
50
0.4
Measured (Compression Stroke)
25 Measured (Tension Stroke)
SW Model
0.2
Pm = 0.8, 0.4, 0.3
0 0
0 20 40 60 0 25 50 75
Group Deflection, Yg , mm. Pile Deflection, y, mm
Fig. 14. Measured and computed pile-group response of the Houston Fig. 15. Variation of Pm at 0.9 m depth as calculated from the SW
load test model (Houston test)
Hong Kong Load Test in Sandy and Silty Sand Soils interpretation provided by Ooi et al. (2004) that addressed the
Ng et al. (2001) conducted a full-scale 2 × 1 shaft group load test in pile-cap rotation and the increase of the pile-cap-joint slope with
Hong Kong. The two 1.5-m diameter drilled reinforced concrete increasing loads (i.e., shaft deviation from fixed-head conditions).
shafts were installed at 6D center-to-center shaft spacing. The Based on blow counts of the standard penetration test (SPT-N) val-
21-m long drilled shafts were embedded in a 2-m-thick cap with ues, Ng et al. (2001) estimated a friction angle of 39° and effective
plan dimensions of 2:5 × 11:5 m and a base located at the ground unit weight of 11 kN=m3 .
surface. Such a cap thickness and shaft head embedment in the cap Fig. 16 depicts the shaft group analyzed using the SW model
cannot provide adequate restraint against the pile-head rotation to with (1) a constant EI ¼ 4 GN · m2 for the shaft’s cracked section
maintain full pile-head fixity, therefore, the group could be ana- (linear analysis) and (2) nonlinear analysis (Ashour et al. 2001) us-
lyzed as a free-head shaft group. This is in agreement with the ing degrading EI from the moment-curvature relationship. The
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
6000 layer. However, one test is not enough to support such behavior,
and additional full-scale load tests are needed.
Fig. 17 shows the response of the pile groups spaced at 3:7D and
Group Lateral Load, kN
5D in both directions using the SW model analysis and recom-
4000 mended Pm values. The Pm values estimated by using the Rollins
et al. (1998) recommendations provided reasonable agreement with
the measured pile-group response (0.71, 0.58, 0.58 and 0.9, 0.86,
0.86 for the 3:7D and 5D pile spacings, respectively) at larger de-
2000 flections. Considering the Huang et al. (2001) suggestion to ac-
count for the effect of pile installation, Ooi et al. (2004) used a
Measured (Ng et al. 2001) second set of Pm values 20% less than those recommended by
SW Model (Linear analysis)
SW Model (Non-linear analysis) Rollins et al. (1998). Unlike the first set of Pm that provided better
predictions at larger pile-head deflections, the second set of Pm
0
0 20 40 60 80 gave better results at lesser deflections. The adjustment of the
Group Deflection, Y g, mm p-y curves used in the group analysis apparently had a significant
impact on the results of the computations and the obtained instal-
Fig. 16. Measured and computed pile-group load-deflection curves of lation effect ratio (Huang et al. 2001)
the Hong Kong load test
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Briaud, J. L., Smith, T., and Mayer, B. (1984). “Laterally loaded piles Morrison, C., and Reese, L. C. (1986). “Lateral-load test of a full-scale pile
and the pressuremeter: Comparison of existing methods.” Laterally group in sand.” U.S. Dept. of Interior and Federal Highway
loaded deep foundations: Analysis and performance, ASTM, West Administration, Reston, VA, and Washington, DC.
Conshohoken, PA, 97–111. Ng, C. W. W., Zhang, L., and Nip, D. C. N. (2001). “Response of laboratory
Brown, D. A., Morrison, C., and Reese, L. C. (1988). “Lateral load behav- loaded large-diameter bored pile groups.” J. Geotech. Geoenviron.
ior of pile group in sand.” J. Geotech. Eng., 114(11), 1261–1276. Eng., 127(8), 658–669.
Brown, D. A., O’Neill, M. W., McVay, M., El Naggar, M. H., and Norris, G. M. (1986). “Theoretically based BEF laterally loaded pile analy-
Chakraborty, S. (2001). “Static and dynamic lateral loading of pile sis.” Proc., 3rd Int. Conf. on Numerical Methods in Offshore Piling,
groups.” Rep. 461, National Cooperative Highway Research Program, Institute Francais Du Petrole, Nantes, France, 361–386.
Washington, DC. Ooi, P. S. K., Chang, B. K. F., and Wang, S. (2004). “Simplified lateral
Brown, D. A., and Reese, L. C. (1985). “Behavior of a large-scale analyses of fixed head piles and pile groups.” J. Geotech. Geoenviron.
pile group subjected to cyclic lateral loading.” U.S. Dept. of Interior Eng., 130(11), 1140–1151.
and Federal Highway Administration, Reston, VA, and Washington, Ooi, P. S. K., and Duncan, J. M. (1994). “Lateral load analysis of groups
DC. of piles and drilled shafts.” J. Geotech. Geoenviron. Eng., 120(6),
Cox, W. R., Reese, L. C., and Crubbs, B. S. (1984). “Lateral load tests of 1034–1050.
25.4 mm diameter piles in very soft clay in side-by-side and in-line Poulos, H. G. (1971). “Behavior of laterally loaded piles: II—Pile groups.”
groups.” Laterally loaded deep foundations: Analysis and performance, J. Soil Mech. and Found. Div., 97(5), 733–751.
ASTM, West Conshohoken, PA, 122–139. Rao, S. N., Ramakrishna, V. G., and Raju, G. B. (1996). “Behavior of pile-
Davison, M. T. (1970). “Lateral load capacity of piles.” Transportation
supported dolphins in marine clay under lateral loading.” J. Geotech.
Research Record 333, Transportation Research Board, Washington,
Eng., 122(8), 607–612.
DC, 104–112.
Rao, S. N., Ramakrishna, V. G. S. T., and Rao, M. B. (1998). “Influence
Focht, J. A., and Koch, K. J. (1973). “Rational analysis of the lateral per-
of rigidity on laterally loaded pile groups in marine clay.” J. Geotech.
formance of offshore pile groups.” Proc., 5th Offshore Technology
Geoenviron. Eng., 124(6), 542–549.
Conf., Vol. 2, Richardson, TX, 701–708.
Reese, L. C., Isenhower, W. M., and Wang, S. T. (2006). Shallow and deep
Gandhi, S. R., and Selvam, G. J. (1997). “Group effect on driven
foundations, Wiley, New York, 441–536.
piles under lateral load.” J. Geotech. Geoenviron. Eng., 123(8),
702–709. Reese, L. C., Wang, S. T., Arrellaga, J. A., and Hendrix, J. (1996). “A pro-
Huang, A. B., Hsueh, C. K., O’Neill, M. W., Chern, S., and Chen, C. gram for the analysis of a group of piles subjected to axial and lateral
(2001). “Effects of construction on laterally loaded pile groups.” J. Geo- loading.” GROUP Version 4.0, ENSOFT, Inc., Austin, TX.
tech. Geoenviron. Eng., 127(5), 385–397. Rollins, K. M., Johnson, S. R., Petersen, K. T., and Weaver, T. J. (2003).
Hughes, J. M. O., Goldsmith, P. R., and Fendall, H. D. W. (1978). “The “Static and dynamic lateral load behavior of pile groups based on
behavior of piles to lateral loads.” Rep. No. 178, Civil Engineering full-scale testing.” 3rd Int. Conf. on Offshore and Polar Engineering,
Dept., Univ. of Auckland, Auckland, New Zealand. Int. Society for Offshore and Polar Engineering, Cupertino, CA, 8.
Illyas, T., Leung, C. F., Chow, Y. K., and Budi, S. S. (2004). “Centrifuge Rollins, K. M., Lane, J. D., and Gerber, T. M. (2005). “Measured and com-
model study of laterally loaded pile groups in clay.” J. Geotech. puted lateral response of a pile group in sand.” J. Geotech. Geoenviron.
Geoenviron. Eng., 130(3), 274–283. Eng., 131(1), 103–114.
Kim, J. B., and Brungraber, R. J. (1976). “Full-scale lateral load tests of pile Rollins, K. M., Peterson, K. T., and Weaver, T. J. (1998). “Lateral load
groups.” J. Geotech. Eng. Div., 102(1), 87–105. behavior of full-scale pile group in clay.” J. Geotech. Geoenviron.
McVay, M., Casper, R., and Shang, T. I. (1995). “Lateral response of Eng., 124(6), 468–478.
three-row groups in loose to dense sands at 3D and 5D pile spacing.” Ruesta, P. F., and Townsend, F. C. (1997). “Evaluation of laterally loaded
J. Geotech. Geoenviron. Eng., 121(5), 436–441. pile group at Roosevelt bridge.” J. Geotech. Geoenviron. Eng., 123(12),
Meimon, Y., Baguelin, F., and Jezequel, J. F. (1986). “Pile group behavior 1153–1161.
under long term lateral monotonic and cyclic loading.” Proc., 3rd Int. U.S. Army Corps of Engineers (USACE). (1993). “Design of pile founda-
Conf. on Numerical Methods in Offshore Piling, Institute Francais Du tions.” Technical engineering and design guides No. 1, Washington,
Petrole, Nantes, France, 286–302. DC.
Downloaded 15 Jun 2012 to 111.68.97.27. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org