Академический Документы
Профессиональный Документы
Культура Документы
Civil Engineers
Geotechnical Engineering 162
June 2009 Issue GE3
Pages 151163
doi: 10.1680/geng.2009.162.3.151
Paper 800013
Received 15/02/2008
Accepted 23/12/2008
Keywords: foundations/
mathematical modelling/piles &
piling
M. M. Ahmadi
Assistant Professor,
Department of Civil
Engineering, Sharif University
of Technology, Tehran, Iran
S. Ahmari
PhD student, the University
of Arizona, US
PhD
and S. Ahmari
50
soil strain
soil strain at half of ultimate stress
factor relating soil elastic modulus to initial slope of
py curve
1 , 3 first and third principal stresses
1. INTRODUCTION
There are two general approaches to analyse laterally loaded
piles: simplified methods and continuum-based methods.
Simplified methods principally use the theory of a beam on an
elastic foundation. The so-called py curve method is one
such conventional and semi-empirical method. The assumption
of soil non-linear behaviour may be an advantage for the py
curve method, but the simulation of three-dimensional (3D)
pilesoil interaction by a one-dimensional spring element is a
disadvantage of this method.
There are two main continuum-based approaches for analysing
laterally loaded piles. The first approach 15 suggests that the
soil around the pile be treated as an elastic continuum. These
solutions are based on Mindlins solution for a point load in an
elastic half-space using superposition. In this approach the
appropriate elastic properties may be obtained by backanalysing experimental results, and hence most continuumbased methods need experimental information for calibration
of the required parameters. The major deficiency of these
elastic solutions is that they assume a constant elastic modulus
throughout the model, whereas in practice the soil close to the
pile shows a lower stiffness than the soil located further away.
This is because the soil close to the pile undergoes higher
strains, and so its stiffness decreases.
The second continuum-based approach applies non-linear
numerical methods to model the soilpile interaction. Because
of the computational difficulties of 3D modelling, twodimensional models have been used in many studies. Some
researchers 68 have demonstrated a 3D finite-element analysis
of laterally loaded piles in clay by using standard von Mises
constitutive law. Although they showed good trends in the
results of numerical analyses, they did not provide sufficient
field data for verification purposes. Comparison of soil ultimate
pressures predicted from finite-element analyses 6 with
experimental observations shows that the finite-element
analyses provide a stiffer response of the pile. It is argued 6 that
the lack of agreement between the predicted values of soil
ultimate pressure and field measurements is probably due to the
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
151
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
Pile
Soil
domain
Soil
domain
Pilesoil
interface
3. CONSTITUTIVE MODEL
The analyses performed in this study are meant to model
laterally loaded piles in clay. The finite-element procedure
consists of modelling pile, soil, and pilesoil interaction
(Figure 2); each is represented in the model by a different
constitutive law. An interface element is introduced to simulate
pilesoil interaction.
Geotechnical Engineering 162 Issue GE3
1 3
1=Ei 1=2C u Rf
1
50
1
1
Ei C u
2Rf
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
153
108 kPa and 2 3 107 kPa are used for steel and concrete
respectively. The elastic modulus of steel will increase in the
case of pipe piles, since they are modelled as solid piles. An
average value of 0.25 is assumed for Poissons ratio for both
materials. The Poissons ratio for concrete and steel materials
can be accurately specified, based on recommended values in
various codes. However, an error of the order of 0.1 would not
affect the analysis results.
4. NUMERICAL ANALYSIS
The finite-element mesh used in the analyses is shown in
Figure 3. The mesh is cylindrical in shape. The pile is modelled
as a solid cylinder inside the mesh. In the case of the pipe pile,
the pile is modelled as a solid cylinder. Therefore the elastic
modulus of the pile is increased proportionally in such a way
that the pile bending stiffness (EI ) remains constant. Owing to
the symmetrical nature of the loading direction, only half of
the model is used in the analysis. The curved boundary is
restrained in both the tangential and radial directions. The
surface of symmetry is restrained in the normal direction, and
for the bottom horizontal surface the nodes are restrained in
the vertical direction. The constructed model properties are
summarised in Table 1 (case 1).
The loading condition is simulated in two load steps. Initial
stresses are induced in the first step by applying the
gravitational body force. In the second step, a lateral load of
120 kN is applied at the pile head. The pile-head load is applied
in 20 increments, meaning that the load is applied in 20 steps
to the pile head. Using the Broms method, 18 the ultimate load
is calculated to be 195 kN. Thus the pile is loaded up to 61% of
its lateral capacity.
The at-rest condition is simulated by allowing the soil mass to
first settle under its own weight: thus the horizontal stresses
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
Case no.
Pile properties Pile type
Diameter: m
Wall thickness: mm
EI: kN-m2
Embedment depth: m
Elastic modulus.
Poison ratio
e*: m
Soil properties Soil type
Cu : kPa
Total unit weight: KN/m3
50
Ei(i) : kPa
Ei(f) : kPa
K0
Rf
Cast-in-place pile
0.319
12.7
31280
12.8
2e8
0.25
0.0635
0.762
400,000
12.8
2.4e7
0.25
0.076
Over-consolidated clay
with secondary
structure
32
20
0.012
6400
2000
0.495
0.98
0.857
1
105
19.3
0.005
47250
395000
0.3
0.55-1
0.9
0.5
Pt: kN
80
60
40
20
0
0
20
40
Yt: mm
60
80
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
155
006
005
Displacement: m
004
On y-axis
003
On x-axis
002
001
000
0
001
Distance: m
M: kN m
0
50
100
150
Depth: m
10
12
14
and the other for a pile in stiff clay. The soil elastic modulus
and the soil shear strength parameters were back-calculated by
various trial analyses. Two separate models were constructed
for each case study in order to back-calculate the soil elastic
modulus and shear strength separately. The soil elastic modulus
is back-calculated with the assumption of soil elastic
behaviour. Equation 2 was used to estimate the soil elastic
modulus in the first trial. The elastic modulus was then
repeatedly changed so that the pile-head loaddeflection curve
had a good match with the initial portion of the field-measured
curve. Having done this, in another model, with the assumption
of elastic-hardening plastic behaviour of soil, the soil shear
156
The comparison between numerical predictions and fieldmeasured values is demonstrated in Figure 8 and later in
Figure 10 for each case study separately. Pile-head load
deflection, pile-head loadmaximum bending moment, and
bending moment diagram along the pile length are crosscompared for the two cases, and are discussed below.
6.1. Comparison for case 1
The comparisons in Figure 8 show a satisfying correspondence
between the numerical predictions and field measurements for
case 1. Figure 8(a) shows a small gap at higher loads. Figure
8(c) shows that the predicted bending moment values along the
pile length agree reasonably well with the measured field data
down to a depth of 4.5 m. Below this depth, the two diagrams
deviate slightly from each other.
The numerical analysis carried out in this study gives an elastic
modulus of 2000 kPa. This value is around one third of the
elastic modulus estimated by Equation 2. The back-calculated
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
185
0
6
8
10
190
195
200
205
0
2
4
Depth: m
60
Depth: m
Depth: m
Water content: %
20
40
6
8
10
10
12
14
Case 1
12
Case 1
Case 2
14
Case 2 used
in analysis
14
10
20
Cu: kPa
30
40
50
(c)
60
0
50
Cu: kPa
100
150
200
6
8
Measured
strength
10
Depth: m
Depth: m
12
(b)
(a)
Case 2
Measured
strength
8
Backcalculated
strength
10
12
Backcalculated
strength
14
16
12
14
(d)
(e)
Figure 7. Soil properties variation with depth for cases 1 and 2: (a) water content; (b) total unit weight profile and the assumed
profile in the analysis; (c) variation of 50 (strain at half of ultimate stress) for case 2; (d) measured vane strength and backcalculated strength profile for case 1; (e) measured UU triaxial strength and back-calculated strength profile for case 2
2
D
C uv C uh
D2 H
3H
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
157
140
120
100
Pt: kN
80
Field-measured
60
FEM
40
20
0
0
20
40
60
80
Yt: mm
(a)
140
120
Pt: kN
100
80
60
Field-measured
40
FEM
20
0
0
50
100
150
200
Mmax: kN m
(b)
Mmax: KN m
0
50
100
150
Depth: m
8
Field-measured
10
FEM
12
K0
0
02
04
06
08
10
14
(c)
05
10
15
Depth: m
20
25
30
35
40
45
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
6.3. Discussion
The difference between the numerically predicted and fieldmeasured bending moments, despite the good agreement for
displacements, may be attributed to inaccuracies in simulation
of the initial stresses in the model, to the constitutive law
applied for the soil behaviour, or to the assumed variation of
shear strength in the back-calculation procedure.
500
450
400
Pt: kN
350
300
Field-measured
250
200
150
FEM without reduction
100
50
0
0
10
15
20
Yt: mm
(a)
25
30
35
500
450
400
Pt: kN
350
300
250
Measured
200
FEM
150
500
100
450
50
400
0
0
200
400
600
800
Pt: kN
Mmax: kN m
(b)
50
0
350
1000
M: kN m
450
300
250
200
Without initial stresses
150
950
100
50
0
0
10
15
20
Yt: mm
(a)
25
30
35
40
Depth: m
4
50
0
M: kN m
150
350
550
750
950
8
2
Measured
10
14
Depth: m
FEM
12
(c)
Figure 10. Comparison between FEM results and fieldmeasured results, case 2: (a) pile-head loaddeflection curve;
(b) pile-head loadmaximum bending moment curve along
pile length; (c) bending moment distribution along pile length
for lateral load of 445 kN
10
With initial
stresses
12
Without initial
stresses
14
(b)
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
159
60
50
40
p: kN/m
30
Depth 0
1B
20
2B
4B
10
0
0
002
004
006
008
y: m
E pymax Ei
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
40
45
35
40
35
25
30
20
FEM
15
Matlock20
10
Wu et al.9
p: kN/m
p: kN/m
30
FEM
25
20
Matlock20
15
Wu et al.9
10
5
0
0
0
005
010
015
020
005
y: m
(a)
010
015
020
y: m
(b)
90
80
60
70
50
p: kN/m
60
p: kN/m
40
30
FEM
20
50
FEM
40
Matlock20
30
Matlock20
Wu et al.9
20
Wu et al.9
10
10
0
0
005
010
015
020
005
y: m
(c)
010
015
020
y: m
(d)
Figure 13. Comparison of numerically predicted py curves in this study with curves suggested by Matlock 20 and Wu et al. 9 at
various depths (B diameter): (a) at ground level; (b) at depth of B; (c) at depth of 2B; (d) at depth of 4B
140
60
120
50
100
Pt: kN
Pult: KN/m
40
30
FEM
20
80
a 20
60
a 30
40
a 40
Matlock20
20
Wu et al.9
10
0
0
0
0
40
60
80
Yt: mm
20
Depth/diameter
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
161
140
120
120
100
100
80
N4
60
N5
40
N8
Pt: kN
Pt: kN
140
Ca 10 kPa
60
40
N9
20
Ca 0 kPa
80
Ca 30 kPa
20
0
0
0
20
40
Yt: mm
60
80
20
40
60
80
100
Yt: mm
140
120
Pt: kN
100
80
FKN 1
60
FKN 5
40
FKN 15
20
FKN 40
0
0
20
40
Yt: mm
60
80
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
11.
12.
13.
A sensitivity analysis was carried out on the model dimensions,
mesh fineness, contact stiffness, and pilesoil adhesion. The
outcome of the analysis shows that the optimum soil domain
dimension is 40 times the pile diameter. Meshing as fine as
N 8 (number of soil domain divisions along loading
direction in front of the pile) is sufficient. Soil contact stiffness
has no effect on the pile response. The study of pilesoil
adhesion effect on pile-head deflection shows that the rate of
variation in pile-head deflection is much less than the rate of
variation in pilesoil adhesion.
REFERENCES
1. DOUGLAS D. J. and DAVIS T. G. The movement of buried
footing due to moment and horizontal load and the
movement of anchor plates. Geotechnique, 1964, 14, No. 2,
115132.
2. SPILLER W. R. and STOLL R. D. Lateral response of piles.
Journal of the Soil Mechanics and Foundations Division,
ASCE, 1964, 90, No. 6, 19.
3. MAURICE J. and MADIGNIER F. Pieu vertical sollicite
horizontalement. Annales des Ponts et Chaussees, 1968,
No. 6, 337383.
4. MATHEWSON C. D. The Elastic Behavior of Laterally Loaded
Piles. PhD thesis, University of Canterbury, Christchurch,
New Zealand, 1969.
5. POULOS H. G. Behaviour of laterally loaded piles: IIPile
groups. Journal of the Soil Mechanics and Foundations
Division, ASCE, 1971, 97, No. 5, 733751.
6. BROWN D. A. and KUMAR M. py curves for laterally loaded
piles derived from three-dimensional finite element model.
Proceedings of the 3rd International Symposium on
Numerical Methods in Geomechanics, Niagara Falls, 1989,
683690.
7. BROWN D. A. and SHIE C. F. Three-dimensional finite
element model of laterally loaded piles. Computers and
Geotechnics, 1990, 10, No. 3, 5979.
8. ARISTONOUS M., TROCHANIS J. B. and PAUL C. H. Threedimensional non-linear study of piles. Journal of
Geotechnical Engineering, 1991, 117, No. 3, 429447.
9. WU D., BROMS B. B. and CHOA V. Design of laterally loaded
piles in cohesive soils using py curves. Soils and
Foundations, 1998, 38, No. 2, 1726.
10. MARSLAND A. Large in-situ test to measure the properties of
stiff fissured clay. Proceedings of the 1st AustraliaNew
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
Downloaded by [ Universidade de Brasilia] on [10/12/15]. Copyright ICE Publishing, all rights reserved.
Ahmadi Ahmari
163