Proceedings of the Institution of

Civil Engineers
Geotechnical Engineering 158
April 2005 Issue GE2
Pages 63–73
D. C. N. Nip C. W. W. Ng
Paper 13726 Graduate Geotechnical Professor, Department of
Engineer, Ove Arup & Civil Engineering, Hong Kong
Received 03/03/2004 Partners (HK) Ltd, Hong University of Science &
Accepted 19/07/2004 Kong Technology
Keywords:
design methods & aids/geotechnical
engineering/piles & piling

Back-analysis of laterally loaded bored piles

D. C. N. Nip MPhil and C. W. W. Ng MSc, PhD, CEng, MICE, MHKIE, MASCE

Field lateral load tests were conducted on two single design, because the very large deflection may not be tolerated
1.5 m diameter large bored piles in Hong Kong. The test by the structures. Despite the fact that the ultimate capacity of
piles are embedded in superficial deposits and weathered the soil around the pile may not be fully mobilised, the
rocks with various degrees of decomposition. In this behaviour of soil–pile interaction is non-linear. 1 For large-
paper, the pile load test results are interpreted and diameter bored piles, apart from the non-linear soil–pile
reported. A simple and rational method of back-analysis interaction behaviour, the concrete flexural behaviour may be
for long piles, using inclinometer data and assuming a also non-linear2,3 because concrete cracks when its relatively
fourth-order polynomial to represent the shape of soil small tensile strength is exceeded, resulting in a reduction of
reaction profile, is introduced. The method, requiring the pile’s flexural stiffness.
only simple spreadsheet-type operations, makes it
possible to incorporate non-linear concrete behaviour in Analysis of the behaviour of laterally loaded piles can usually
the back-analysis, and to deduce the p–y curves for be carried out using the semi-empirical p–y method. This
future design of piles in similar ground conditions. In method was developed from the subgrade reaction approach, 4
order to verify the rational method, an analysis is carried in which a pile is idealised as an elastic, transversely loaded
out using the back-analysed p–y curves, and the beam supported by a series of unconnected elastic springs
predictions are found to agree well with the measured representing the soil. The p–y method modified the subgrade
pile head deflections. This proposed method reaction approach by allowing the soil resistance, p, to vary
demonstrates how p–y curves are back-analysed non-linearly with the pile deflection, y. Shapes of p–y curves
rationally for long piles embedded in complex ground have been suggested by many researchers. For piles embedded
conditions, in contrast to traditional back-analyses using in sands, shapes of p–y curves have been suggested by
recommended p–y curves for a uniform soil stratum. McClelland and Focht, 1 Reese et al.,5 and Murchison and
O’Neill. 6 For piles embedded in clay, shapes of p–y curves have
NOTATION been recommended by Matlock, 7 Reese et al.8 and Reese and
a, b, c, d coefficients of fourth-order approximation Welch.9 Based on numerical analyses, this approach can
polynomial represent soil varying in any fashion with depth and under
D pile diameter static or cyclic loading conditions. However, unlike the elastic
EI pile flexural stiffness solutions, soil interactions are not taken into account. It is also
F0 applied lateral force at the ground level difficult to find relevant data from site investigations, other
M bending moment than back-analysing pile load test results, to determine the p–y
M0 applied bending moment at the ground level curves. Although Briaud 10 developed a method to derive p–y
N blow count from standard penetration test (SPT) curves from pressuremeter results, its application appears to be
P soil reaction force per unit length limited. In addition, the way p–y curves are affected by pile-
p soil resistance in kPa head fixity and the relative stiffness between pile and soil is
V shear force not fully understood. 11 For preliminary designs, it might be
y pile lateral displacement costly and time-consuming to perform detailed computer
z depth below the ground level or pile head analyses to develop p–y curves for specific or complex ground
z0 depth at which bending moment, shear force and conditions.
soil reaction force equal zero
˜z distance between two inclinometer data points This paper presents the results of two lateral load tests on
Ł pile rotation large-diameter single bored piles (P1 and P2) embedded in
superficial deposits, weathered rocks with various degrees of
1. INTRODUCTION decomposition, in Hong Kong. A back-analysis method using a
The design of laterally loaded piles is generally quite different fourth-order polynomial is introduced to interpret inclinometer
from the design of other foundations. Consideration of the data rationally such that the non-linear flexural concrete
maximum bending moments and deflections of piles rather behaviour is incorporated. Apart from the preliminary results
than the ultimate lateral resistance of the soil often governs the of rotation profiles as given by inclinometer measurements, the

Geotechnical Engineering 158 Issue GE2 Back-analysis of laterally loaded bored piles Nip • Ng 63
 proposed method makes it possible to deduce bending moment, values ranging between 35 and 60. The subsurface conditions
shear force and soil reaction profiles, together with p–y curves. at P2 consist of three layers: sandy fill, silty alluvium, and
In addition, in order to verify the proposed method of back- completely decomposed metasiltstone. The SPT N values are
analysis, the measured test results are compared with the considered to increase with depth, except in the sandy fill
predictions using the back-analysed p–y curves. This paper layers. From Fig. 1, it can be deduced that the soil at P1 is
aims at demonstrating how p–y curves are developed for any generally weaker than that at P2.
specific or complex ground conditions using simple
spreadsheet operations, rather than detailed computer analyses.
2.2. Method of pile construction
The test piles were designed to resist both vertical and lateral
2. LATERAL LOAD TESTS ON BORED PILES IN loads. The pile lengths for P1 and P2 are 28.0 m and 29.5 m
HONG KONG long respectively, and are governed mainly by the vertical
loads. The pile diameters for both test piles are 1.5 m. P1 was
2.1. Subsurface conditions founded on rock, whereas P2 was a floating pile. Similar
The two test piles are approximately 9.5 km apart. Prior to the construction methods were adopted for both test piles. A
test pile construction, one borehole was sunk to investigate the temporary casing was installed to the full length of the pile by
subsurface conditions for each test pile, and standard an oscillator, and excavation was carried out within the casing
penetration tests (SPT) were conducted in superficial deposits using a mechanical grab. On completion of the pile bore, the
and completely decomposed soils. 12 It is common practice in base was cleaned by air lifting to remove the sediments and
Hong Kong to use directly the field SPT N values, 13 without debris. Then reinforcement cages were lowered down one by
making any corrections to characterise soil properties, such as one inside the temporary casing to the base of the bore,
the angle of shearing resistance or the relative density. The followed by concreting (Grade 45) using tremie pipes with the
groundwater table was measured to be at about 1 metre below temporary casings being extracted at the same time. All the
ground level (mbgl) at P1, and 2 mbgl at P2. The subsurface reinforcements were hot-rolled high-yield steel bar (grade 460).
conditions at the test pile locations with SPT N plots are shown Concrete coring of each test pile was conducted after the load
in Fig. 1. test, and uniaxial compression tests of the concrete cylinders
were carried out. Based on the uniaxial compression test
The ground investigation at P1 revealed that the top few results, and taking into account the ageing effect, 14 the average
metres were covered by very loose fill, followed by alternate Young’s moduli of the concrete of P1 and P2 at the time of the
layers of sandy estuarine deposit 12 and clayey alluvium. They pile load tests were estimated to be 32 GPa and 31 GPa
were then underlain by a thick layer of silty to clayey karstic respectively.
surface deposit and a thin layer of moderately decomposed
marble, followed by cavity infill. 12 The SPT N values of the soil To facilitate the application of lateral loads, pile caps were
at P1 were generally less than 20, with some relatively large constructed at pile heads above the ground level. The square

Borehole 0 50 100 150 200

P1 P2
at
0
Fill
P1 P2
5 Estuarine deposit
5
Alluvium
Karstic surface deposit

10 10
Cavity Infill
Moderately decomposed
15 marble (Rock) 15
Depth: mbgl

P1 P2

Completely decomposed
metasiltstone
20 20

25 25

30 30
Rock head

35 35

Depth: mbgl (b)

(a)

Fig. 1. (a) Soil profiles; (b) SPT N values in superficial deposits and completely decomposed soils

64 Geotechnical Engineering 158 Issue GE2 Back-analysis of laterally loaded bored piles Nip • Ng
pile cap of P1 is 1.5 m thick and 1.8 m wide, whereas the
5000
square pile cap of P2 is 1.5 m thick and 2.5 m wide. The size of
the pile caps is considered to be comparable to the pile 4500
diameters, so it is considered that there was negligible
rotational restraint at the pile heads, and the test piles are 4000
considered to be free-headed in the analysis.
3500
2.3. Pile instrumentation
The typical instrumentation layout for laterally loaded test 3000

Lateral load: kN
piles is shown in Fig. 2. Both P1 and P2 were pushed against a
deadman (deadweight). Lateral loads were applied to the test 2500

piles using hydraulic jacks at approximately the mid-depth of

2000
the pile caps, and were monitored by a load cell. The rotations
along the length of each test pile were measured at 0.5 m depth
1500
intervals along two inclinometer tubes, which were installed in
reservation tubes opposite one another along the pile diameter 1000
and approximately aligned with the lateral loading direction.
Dial gauges were installed at the pile caps of both test piles to 500 P1 P2
obtain the pile cap deflection.
0
0 20 40 60 80 100 120 140
3. RESULTS OF PILE LOAD TESTS
Pile head deflection: mm

3.1. Observed pile load–deflection

The load is plotted against pile cap deflection from the dial Fig. 3. Lateral load against pile head deflection
gauge measurements for both test piles in Fig. 3. The response
of P2 shows that it is stiffer, and had less permanent deflection
(47 mm) after unloading from the maximum applied load. P1 is
less stiff, and underwent a rather larger permanent deformation around the pile, especially in the upper layers. It is estimated
of about 58 mm. This may probably be due to the fact that the from a 3D numerical analysis that the piles with previous
soil at P2 is generally stiffer according to the ground kentledge loading might have a lateral deflection 15% smaller
investigation results, as shown in Fig. 1. In addition, it should than the one without kentledge loading. 15 In view of the effects
be noted that the initial site conditions of the two test piles of kentledge on laterally loaded piles, it would be prudent to
were different. Before the lateral load test was conducted on consider any additional surcharge around piles in the analyses.
P2, a vertical pile load test, to the maximum test load of about
25 MN, was performed. The lateral pile load test results of P2 3.2. Observed pile rotation and deflection with depth
might be affected by the kentledge of about 30 MN, placed on The measured rotation and deduced deflection profiles from
two sides of the pile during the previous vertical pile load test. some selected inclinometer measurements of P1 and P2 are
There might have been some horizontal locked-in stresses that presented in Figs 4 and 5 respectively. The rotation profiles
developed in the soils, thus increasing the stiffness of the soil show that the rotations remain constant along the depth of the

Inclinometer

A
Hydraulic jack
Dial gauge and load cell

Pile cap
Dial gauge

Pile cap

Pile

Inclinometer

(a) (b)

Fig. 2. Instrumentation for laterally loaded piles: (a) plan; (b) section A–A

Geotechnical Engineering 158 Issue GE2 Back-analysis of laterally loaded bored piles Nip • Ng 65
 Rotation: radians Lateral deflection: mm

20·005 0 0·005 0·010 0·015 0·020 0·025 250 0 50 100 150 200
25 25

0 0

5 5
Depth: m below pile head

Depth: m below pile head

Cycle 1 (1335 kN)
Cycle 1 (1335 kN)
10 Cycle 2 (2000 kN)
Cycle 2 (2000 kN) 10
Cycle 3 (2745 kN)
Cycle 3 (2745 kN)
Cycle 4 (2955 kN)
Cycle 4 (2955 kN)
15 15

20 20

25 25

30 30
(a) (b)

Fig. 4. Profiles for P1 at peak loads of each cycle: (a) rotation; (b) lateral deflection

Rotation: radians Lateral deflection: mm

20·005 0 0·005 0·010 0·015 0·020 250 0 50 100 150
25 25

0 0

5 5
Depth: m below the pile head

Depth: m below pile head

Cycle 1 (983 kN) Cycle 1 (1335 kN)

Cycle 2 (1984 kN) Cycle 2 (2000 kN)
10 10
Cycle 3 (2745 kN) Cycle 3 (2745 kN)
Cycle 4 (3934 kN) Cycle 4 (2955 kN)
Cycle 5 (4939 kN) Cycle 5 (4939 kN)
15 15

20 20

25 25

30 30

(a) (b)

Fig. 5. Profiles for P2 at peak loads of each cycle: (a) rotation; (b) lateral deflection

pile caps, implying that both P1 and P2 can be reasonably deflected more than P2, illustrating that the soil at P2 was
idealised as free-head piles. Also, the deflected shapes show stiffer than that at P1, which is consistent with the dial gauge
that the induced deflection is confined to the upper part of the measurements. In addition, the deflections at the mid-depth of
pile and the lower parts of the piles remain vertical, thus the pile cap deduced from inclinometer measurements are also
indicating that they behaved as ‘flexible piles’. Generally, P1 comparable with the dial gauge measurements.

66 Geotechnical Engineering 158 Issue GE2 Back-analysis of laterally loaded bored piles Nip • Ng
4. METHOD OF ANALYSIS It is shown in equation (2) that there is no constant term,
indicating that the boundary condition of zero soil reaction at
4.1. Polynomial approximation of flexural behaviour of pile head has been incorporated. By integrating the soil
laterally loaded piles reaction profile (i.e. equation (2)) with respect to depth once
Based on the ordinary beam theory, the differential equations and incorporating the boundary condition of applied shear
governing the behaviour of a pile subjected to lateral load can force at the pile head, the following equation for shear force
be derived as follows: profile can be derived:

dy d2 y d3 y d4 y az 5 bz 4 cz 3 dz 2
1 ¼ Łz ; EI 2 ¼ M z ; EI 3 ¼ V z ; EI 4 ¼ Pz 3 þ þ þ þ F0 ¼ V z
dz dz dz dz 5 4 3 2

where y is the lateral deflection, z is the depth below ground By integrating the soil reaction profile (i.e. equation (2)) with
level, Łz is the rotation, Mz is the bending moment, Vz is the respect to depth twice, and incorporating the boundary
shear force, Pz is the soil reaction force per unit length, and EI condition of calculated bending moment at pile head, the
is the flexural stiffness. following equation for bending moment profile can be derived.

The equations show that, when rotation measurements are

taken from inclinometers along the depth of a pile, integrating az 6 bz 5 cz 4 dz 3
4 þ þ þ þ F0 z þ M 0 ¼ M z
the rotation once as a function of depth z yields y, and then 30 20 12 6
differentiating once, twice and three times yields Mz , Vz and Pz
respectively. The procedures can then be used to deduce
By integrating the soil reaction profile (i.e. equation (2)) with
deflection, bending moment, shear force and soil reaction
respect to depth three times, and dividing by the flexural
profiles. The integration of rotation measurements provides a
stiffness (EI), the following equation for rotation profile can be
fairly reliable indication of deflection. However, direct
derived. The estimation of EI to account for the non-linear
derivations of bending moment, shear force and soil reaction
behaviour of concrete will be presented in the next section.
from rotation measurements amplify any measurement error,
resulting in bending moment, shear force and soil reaction

profiles that do not satisfy the boundary conditions at the pile X z
az6i bz5i cz4i dz3i ˜z
5 þ þ þ þ F0 z i þ M 0 3 ¼ Łz
head. Hence a method is introduced to rationally back-analyse z i ¼z0
30 20 12 6 (EI) zi
the bending moment, shear force and soil reaction profiles.

This proposed method involves rationally iterative procedures

to make a ‘best fit’ to the rotation measurements (i.e. where z0 is the depth where rotation, bending moment, shear
inclinometer data) by using a fourth-order polynomial force and soil reaction are all equal to zero. For the presented
representing the soil reaction profile as shown in the form of pile load test results, the adopted z0 values are found to be
equation (2) and then carrying out integration: approximately 2.5 m below the depth of zero rotation, which is
far away from the test pile toe. It should be noted that the test
piles are designed for both vertical and horizontal loads, and
2
ð az 4 þ bz 3 þ cz 2 þ dzÞ ¼ Pz the excessive pile length below z0 is required mainly to resist
large vertical load. The depth z0 defines an additional
It is implicitly assumed that the soil reaction profile could be boundary condition for long piles, which is defined to have
described by a polynomial, which is generally true for long zero/negligible rotation over a distance near the pile toe. This
piles in uniform soil. 16 A fourth-order polynomial is used additional boundary condition allows the coefficients of the
because it utilises four known boundary conditions at the pile polynomial as in equations (6) to (9) below to become solvable.
head and one boundary condition for the ‘flexible pile The effect of the selection of z0 on the back-analysed results
condition’. Any decrease in the order of the polynomial would will be discussed later. The term ˜z is the distance between two
result in some of the boundary conditions not being satisfied. measured inclinometer data points, which is equal to 0.5 m in
Any increase in the order would lead to the number of this case. In order to determine the coefficients a, b, c and d,
coefficients being greater than the number of boundary the following procedures are suggested:
conditions, which in turn would make the coefficient of the
polynomial unsolvable. The boundary conditions at the pile (a) Estimate the depth z0 and assume an initial distribution of
head are as follows: EI along the length of pile.
(b) Solve the simultaneous equations (6) to (9) to obtain the
(a) rotation, as measured by inclinometers (Ł0 ) coefficients a, b, c and d:
(b) shear force, as measured by load cells (F0 )
(c) bending moment as calculated by multiplying the  
6
az40 þ bz30 þ cz20 þ dz0 ¼ 0
measured lateral force by the eccentricity of the force
above the pile head for a free-head pile (M0 )
(d) zero soil reaction for cohesionless soil, which is often the az50 bz40 cz30 dz20
case in Hong Kong, where fill material is found near the 7 þ þ þ þ F0 ¼ 0
5 4 3 2
ground surface.

Geotechnical Engineering 158 Issue GE2 Back-analysis of laterally loaded bored piles Nip • Ng 67
 other laterally loaded piles embedded in similar soil profiles
az60 bz50 cz40 dz30 can be predicted using the p–y method.
8 þ þ þ þ F0 z0 þ M 0 ¼ 0
30 20 12 6
Brown et al. have proposed a method to interpret inclinometer

 data for the determination of p–y curves. 17 However, the back-
X 0
az6i bz5i cz4i dz3i ˜z
9 þ þ þ þ F0 z i þ M 0 3 ¼ Ł0 analysis method proposed in this paper is different from that
z i ¼z0
30 20 12 6 ð EI Þ z i suggested by Brown et al. The main difference is that Brown et
al.’s method assumes the shape of the p–y curves along the
length of a pile, whereas the proposed method assumes the
(c) With the calculated values of the four coefficients in step shape of a soil reaction profile (i.e. the fourth-order
(b), compute the bending moment profile using equation polynomial). For both methods, if concrete non-linear
(4). By using a non-linear relationship between flexural behaviour is to be taken into account, a computer program is
stiffness and bending moment (i.e. EI and M), which is needed to establish the relationship between flexural stiffness
discussed in the following section, determine EI at each and bending moment. Apart from this, the proposed method
data point along the length of the pile. requires only spreadsheet operations to implement the
(d) With the calculated values of EI in step (c), analyse the rationally iterative procedures. Owing to the preselected p–y
rotation profile by using equation (5). relationship, Brown et al.’s method requires a computer
(e) If the calculated rotation profile, i.e. equation (5), closely program (e.g. COMP624P) in the iterative procedures to
approximates the measured rotation profile, the values of compare the calculated and measured pile deflection. In
the coefficients are acceptable; otherwise, try another addition, the assumption of using a polynomial in the proposed
distribution of EI along the length of the pile and repeat method enables the study of shapes of p–y curves and a better
steps (b), (c) and (d). For a calculated rotation smaller than choice of p–y curves for further design calculations. It might
the measured value, it can be seen from equation (5) that be easy to preselect a particular shape of p–y curve for sand or
the calculated EI value, obtained from the non-linear clay in Brown et al.’s method, but for complex ground
relationship between EI and M, is too large. This implies conditions such as silty soils or clayey soils with sand lenses, it
that the calculated M value is too small. As the M value is might be difficult to determine the shape of the p–y curve and
calculated by solving the simultaneous equations based on how it varies with the length of a pile. However, the proposed
equation (1), a larger assumed EI value will then be method is limited to long piles, as the depth z0 (at which
required to give a larger M value for a given curvature. rotation, bending moment, shear force and soil reaction equal
zero) does not exist for short piles.

Generally, the proposed method assumes a fourth-order 4.2. Incorporation of flexural stiffness of concrete piles in
polynomial to model the soil pressure profile. For simplicity, a the analysis
polynomial, which is an easy mathematical function to Concrete behaves non-linearly, even at low tensile stress
understand, is chosen. By iterative procedures of curve-fitting levels, 2,3 as a result of cracking, resulting in a reduction of the
with inclinometer data, the profiles of soil pressure (Pz ) and flexural stiffness (EI) of the concrete, which depends on the
lateral deflection (y z ) can be deduced. The proposed method is bending moment of the pile developed along the pile shaft. In
rational, in that the iterative procedures involve changes of EI some routine designs the value of EI for a cracked section is
along the pile only. usually taken as half the value for an uncracked section, for
simplified design. However, this might overpredict the pile
This simple rational method is particularly useful as deflection and underestimate the maximum bending moment
inclinometers are often the only measurements along the of the pile. Hence a more rational approach is to incorporate
lengths of piles in
commercial lateral load tests,
12000
and it is necessary to deduce
bending moment, shear force
10000
and soil reaction profiles
from these measurements.
Pile stiffness, EI: MN m2

From the calculated bending 8000

moment profiles, the
magnitude and location of 6000
the maximum bending
P1 P2
moment can be obtained. In 4000
addition, by studying the
deduced deflection profile
2000
from rotation data and the
back-analysed soil reaction
0
profile, this technique allows 0 2 4 6 8 10 12
evaluation of the p–y curves, Bending moment, M: MN m
which is discussed later.
Based on the results of a pile Fig. 6. Computed relationship between pile flexural stiffness and bending moment
load test, the behaviour of

68 Geotechnical Engineering 158 Issue GE2 Back-analysis of laterally loaded bored piles Nip • Ng
 Rotation, è: radians Bending moment, M: kN m
20·005 0 0·005 0·010 0·015 0·020 0·025 0 5000 10000 15 000
0 0

5 5

10 10
Depth: m below pile head

Depth: m below pile head

Measured
15 Back-analysed 15
Back-analysed

20 20

25 25

30 30
(a) (b)

Shear force, V: kN Soil reaction, P: kN/m

24000 22000 0 2000 4000 21500 21000 2500 0 500 1000 1500
0 0

5 5
Depth: m below pile head

10 10
Depth: m below pile head

15 15

Back-analysed Back-analysed

20 20

25 25

30 30

(c) (d)

Fig. 7. Back-analysed profiles for P1 at maximum load (2955 kN): (a) rotation; (b) bending moment; (c) shear force; (d) soil reaction

the non-linearity of concrete behaviour such that EI varies strain compatibility and the equilibrium of forces. The details
with depth by evaluating the variation of EI with bending are presented in reference 15. The variation of EI with bending
moments for different states of the concrete, including moment for all test piles is shown in Fig. 6. The figure
uncracked, partially cracked and fully cracked sections. 18 The indicates that flexural stiffness starts to reduce at an
curve of EI against bending moment for a reinforced concrete approximate bending moment of 1.5 MN m, and decreases
section of a bored pile is established from the requirements of gradually to about 40% of the maximum value for an

Geotechnical Engineering 158 Issue GE2 Back-analysis of laterally loaded bored piles Nip • Ng 69
 uncracked section. This figure is used for calculating the (EI)zi there is reasonable agreement between the measured and back-
to estimate the rotation profiles, as shown in equation (5). analysed rotation profiles. They also indicate that the
maximum shear force occurs at the ground surface, and the
5. DEDUCTION OF p – y CURVES maximum bending moments are found to be at depths of about
The rational back-analysis of inclinometer data yields the 5 m (3D) and 3 m (2D) for P1 and P2 respectively (where D is
back-analysed rotation, bending moment, shear force and soil the pile diameter).
reaction profiles of P1 and P2 at the maximum loads, which
are shown in Figs 7 and 8 respectively. The figures show that Analysis of the behaviour of laterally loaded piles can be

Rotation, è: radians Bending moment, M: kNm

20·005 0 0·005 0·010 0·015 0·020 25000 0 5000 10000 15000
0 0

5 5

10 10
Depth: m below pile head

Depth: m below pile head

Measured Back-analysed
15 Back-analysed 15

20 20

25 25

30 30
(a) (b)

Shear force, V: kN Soil reaction, P: kN/m

24000 22000 0 2000 4000 6000 21000 0 1000 2000 3000
0 0

5 5

10 10
Depth: m below pile head

Depth: m below pile head

15 15

Back-analysed Back-analysed

20 20

25 25

30 30

(c) (d)

Fig. 8. Back-analysed profiles for P2 at maximum load (4939 kN): (a) rotation; (b) bending moment; (c) shear force; (d) soil reaction

70 Geotechnical Engineering 158 Issue GE2 Back-analysis of laterally loaded bored piles Nip • Ng
usually done by the p–y method. This method is developed maximum difference between the measured and back-
mainly for sand or clay. However, as noted in Section 2.1, the analysed pile head deflection is about 10%.
particles of soils in Hong Kong are often silt sizes or even a (b) The back-analysed soil reaction profiles near the pile head
mixture of sand, silt and clay (such as sandy clayey silt), which for all the analyses are similar, and the difference between
are especially common for the superficial deposits and the back-analysed maximum soil reaction values ranges
decomposed rocks with various degrees of decomposition. By from 17% to 26%.
using the proposed rational back-analysis method to generate
p–y curves, it is easier to study the shape of the p–y curves for Based on the above findings, it can be said that the potential
complex ground conditions. Based on the deduced soil pressure errors in the back-analysed p–y curves near the pile head,
and deflection profiles, p–y curves can be generated. They are which are critical in laterally loaded pile analyses, using
presented in Fig. 9. At a greater depth, the pile did not move different z0 values would be acceptable for engineering
much and very small soil pressure was developed, so only p–y purposes.
curves at shallow depth are shown. The figure indicates that
the behaviour of the soil is non-linear, and it undergoes
6. COMPARISON BETWEEN MEASUREMENTS AND
progressive plastic yielding.
PREDICTIONS
In order to validate the proposed method of back-analysis, an
A sensitivity analysis was carried out to study the effect of
analysis is carried out using these back-analysed p–y curves.
different z0 values on the back-analysed p–y curves for the test
The back-analysed p–y curves are implemented through the
pile P1 at the maximum applied load (2955 kN). Based on the
OASYS computer program ALP 20 to calculate pile head
subgrade reaction method, Reese and Matlock 19 suggested a
deflections. The non-linear concrete behaviour is taken into
series of deflection, rotation, bending moment, shear force and
account by adjusting EI along the length of pile based on the
soil reaction profiles for laterally loaded free-headed piles in
relationship of flexural stiffness and bending moment shown in
soil with linearly increasing soil modulus with depth. For the
Fig. 6. The comparison between predicted and measured load-
case of this sensitivity analysis, according to Reese and
deflection behaviour is shown in Fig. 11. The figure shows that
Matlock, the range of z0 , where bending moment, shear force
that the results using the back-analysed p–y curves have a
and soil reaction are negligible, would be between 13.75 m and
good agreement with the measurements for P1 and P2,
15.75 m. Three back-analyses were therefore carried out for
verifying the back-analysed p–y curves from the rational
z0 ¼ 13.75 m, z0 ¼ 14.75 m and z0 ¼ 15.75 m. The back-
method.
analysed deflection (y) and soil reaction (P) profiles are shown
in Fig. 10. The following summarises the findings:
7. SUMMARY AND CONCLUSIONS
Two lateral load tests were conducted on two 1.5 m diameter
large single bored piles embedded in superficial deposits,
(a) The back-analysed deflection profiles for all the analyses weathered rock with various degrees of decomposition in Hong
are comparable to the measured deflection profiles, and the Kong, and the measured results were interpreted and discussed.

2000 2000

Back-analysed by the rational method Back-analysed by the rational method

1600 1600

z 5 2·5 m
Soil pressure, p: kPa
Soil pressure, p: kPa

1200 1200
and
z 5 6·25 m 4·0 m

z 5 4·75 m
800 800

z 5 1·0 m

400 z 5 3·25 m 400

z 5 1·75 m
0 0
0 0·01 0·02 0·03 0·04 0·05 0·06 0 0·01 0·02 0·03 0·04 0·05 0·06

Lateral deflection to pile diameter ratio, y/D Lateral deflection to pile diameter ratio, y/D

(a) (b)

Fig. 9. Back-analysed p–y curves for: (a) P1; (b) P2

Geotechnical Engineering 158 Issue GE2 Back-analysis of laterally loaded bored piles Nip • Ng 71
 Deflection, y: mm Soil reaction, P: kN/m

250 0 50 100 150 2900 2600 2300 0 300 600 900 1200
0 0

5 5
Depth: m below pile head

Depth: m below pile head

10 10

z0 5 13·75 m z0 5 13·75m
15 15
z0 5 14·75 m
z0 5 14·75m
z0 5 15·75 m
Measured z0 5 15·75m
20 20

25 25

30 30
(a) (b)

Fig. 10. Back-analysed profiles for different z0 values of P1 at maximum load (2955 kN): (a) lateral deflection; (b) soil reaction

consisting of different soil strata and non-uniform particle size

3000 distribution, it is difficult to select appropriate
2500 recommendations for typical p–y curves, which are for
uniform soils. A rational method was introduced to back-
Lateral load: kN

2000
analyse the p–y curves, using only spreadsheet operations.
1500 Measured

1000 Back-analysed The rational method for long piles assumes a fourth-order
polynomial to represent the shape of soil reaction profile. The
500 iterative procedures of the proposed method incorporate
0 variation of the flexural stiffness distribution along the length
0 2 4 6 8 10 of a pile to best-fit the measured inclinometer data. Bending
Pile cap deflection/pile diameter: %
moments, shear forces and soil reaction along the pile length,
(a)
together with p–y curves, are back-analysed. In order to
6000 validate the proposed back-analysis, the predictions using the
5000 back-analysed p–y curves are compared with the
measurements. They show reasonable agreement. This
Lateral load: kN

4000
illustrates that the proposed back-analysis method can be
3000 adopted to deduce useful p–y curves in complex ground
Measured
2000 conditions from inclinometer data for further pile design
Back-analysed
calculations.
1000

0
0 1 2 3 4 5 6 7
Pile cap deflection/pile diameter: %
8. ACKNOWLEDGEMENTS
(b)
This research project was supported by a research grant CRC96/
99.EG04 from the Cooperative Research Centre scheme of the
Fig. 11. Comparison between predicted and measured load–
Research Grant Council of the Hong Kong Government of the
deflection behaviour: (a) P1; (b) P2
Special Administrative Region (HKSAR). The authors would
like to acknowledge the contributions provided by the
Kowloon-Canton Railway Corporation (KCRC) and Ove Arup
and Partners Hong Kong Ltd (OAP). In particular, the help of
The piles were instrumented with inclinometers. The Messrs D. E. Gerken, I. Thoms, S. C. Lee, C. Calton and J. Blake
inclinometer data revealed that the deflection is confined to the of KCRC, Drs S. Hill, B. Littlechild and G. Plumbridge of OAP,
upper part of the piles, which are considered to be ‘flexible’. As and Professors W.H. Tang and C. K. Shen of the Hong Kong
the piles were embedded in complex ground conditions University of Science and Technology is highly appreciated.

72 Geotechnical Engineering 158 Issue GE2 Back-analysis of laterally loaded bored piles Nip • Ng
REFERENCES 10. BRIAUD J. L. SALLOP: simple approach for lateral loads on
1. MCCLELLAND B. and FOCHT J. A. Jr. Soil modulus for piles. Journal of Geotechnical and Geoenvironmental
laterally loaded piles. Transactions of the ASCE, 1958, 123, Engineering, ASCE, 1997, 123, No. 10, 958–964.
1049–1063. 11. BUDHU M. and DAVIES T. G. Analysis of laterally loaded
2. NG C. W. W., LINGS M. L. and NASH D. F. T. Back-analysing piles in soft clays. Journal of Geotechnical Engineering,
the bending moment in a concrete diaphragm wall. The ASCE, 1988, 114, No. 1, 21–39.
Structural Engineer, 1992, 70, No. 23/24, 421–426. 12. GEOTECHNICAL CONTROL OFFICE. Guide to Rock and Soil
3. VITHARANA N. D. Rational prediction of lateral behaviour of Descriptions, Geoguide 3. GCO, Hong Kong, 1988.
concrete piles incorporating pile (concrete) non-linearity. 13. BRITISH STANDARDS INSTITUTION. Methods of Test for Soils for
Proceedings of the 14th International Conference on Soil Civil Engineering Purposes. In-situ Tests. BSI, Milton
Mechanics and Foundation Engineering, Hamburg, 1997, Keynes, 1990, BS 1377 part 9.
915–920. 14. BRITISH STANDARDS INSTITUTION. Structural Use of Concrete.
4. HETENYI M. Beams on Elastic Foundations. University of BSI, Milton Keynes, 1985, BS 8110 Parts 1 and 2.
Michigan Press, Ann Arbor, MI, 1946. 15. NIP D. C. N. Analysis of Laterally Loaded Large-diameter
5. REESE L. C., COX W. R. and KOOP F. B. Analysis of laterally Bored Piles in Hong Kong. MPhil thesis, Hong Kong
loaded piles in sand. Proceedings of the Offshore University of Science and Technology, 2000.
Technology Conference, Houston, TX, 1974, paper No. OTC 16. BROMS B. B. Lateral resistance of piles in cohesionless soils.
2080. Journal of the Soil Mechanics Division, ASCE, 1964, 90,
6. MURCHISON J. M. and O’NEILL M. W. Evaluation of p–y No. SM3, 123–156.
relationships in cohesionless soils. Proceedings of a 17. BROWN D. A., HIDDEN S. A. and ZHANG S. Determination of
Symposium on Analysis and Design of Pile Foundations, p–y curves using inclinometer data. Geotechnical Testing
San Francisco, ASCE, 1984, pp. 174–191. Journal, ASTM, 1994, 17, No. 2, 150–158.
7. MATLOCK H. Correlation for design of laterally loaded piles 18. NG C. W. W. An Evaluation of Soil–Structure Interaction
in soft clay. Proceedings of 2nd Offshore Technology Associated with Deep Excavation. PhD thesis, University of
Conference, Dallas, TX, 1970, 577–594. Bristol, UK, 1992.
8. REESE L. C., COX W. R. and KOOP F. D. Field testing and 19. REESE L. C. and MATLOCK H. Non-dimensional solutions for
analysis of laterally loaded piles in stiff clay. Proceedings laterally loaded piles with soil modulus assumed
of 7th Offshore Technology Conference, Dallas, TX, 1975, proportional to depth. Proceedings of the 8th Texas
671–690. Conference on Soil Mechanics and Foundation Engineering,
9. REESE L. C. and WELCH R. C. Lateral loading of deep Austin, TX, 1956, 1–41.
foundation in stiff clay. Journal of Geotechnical 20. ARUP. Oasys ALP Laterally Loaded Pile Analysis Manual.
Engineering Division, ASCE, 1975, 101, No. 7, 633–679. Ove Arup & Partners, London, 1997.

What do you think?

To comment on this paper, please email up to 500 words to the editor at [email protected]
Proceedings journals rely entirely on contributions sent in by civil engineers and related professionals, academics and students. Papers
should be 2–5000 words long, with adequate illustrations and references. Please visit www.thomastelford.com/journals for author
guidelines and further details.

Geotechnical Engineering 158 Issue GE2 Back-analysis of laterally loaded bored piles Nip • Ng 73