Sunteți pe pagina 1din 7

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/273897883

Physical and numerical modelling of pile


foundations subjected to vertical and
horizontal loading in dry sand
CONFERENCE PAPER SEPTEMBER 2014

READS

67

5 AUTHORS, INCLUDING:
Y. S. nsever

Mehmet Yener zkan

Uludag University

Middle East Technical University

7 PUBLICATIONS 6 CITATIONS

8 PUBLICATIONS 67 CITATIONS

SEE PROFILE

SEE PROFILE

Tatsunori Matsumoto
Kanazawa University
55 PUBLICATIONS 319 CITATIONS
SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,


letting you access and read them immediately.

Available from: Y. S. nsever


Retrieved on: 06 April 2016

Physical and numerical modelling of pile foundations subjected to


vertical and horizontal loading in dry sand
Y.S. Unsever & M.Y. zkan
Department of Civil Engineering, Middle East Technical University, Ankara, Turkey

T. Matsumoto, S. Shimono & K. Esashi


Graduate School of Natural Science and Technology, Kanazawa University, Kanazawa, Japan

ABSTRACT: Vertical load test and cyclic horizontal load test on a 3-pile piled raft model were carried out in
dry sand model ground at 1-g field to investigate vertical and horizontal bearing mechanisms of the piled raft
model. Load tests on the raft and single pile were also carried out separately to investigate their interactions on
the piled raft. Numerical modelling of the load tests were carried out, aiming at obtaining more insight into
the behaviour of the piled raft subjected to vertical and horizontal loading. A FEM software, PLAXIS 3D, was
used for this purpose. The hardening soil model was employed for modelling the sand behaviour. The soil
parameters were estimated from a series of consolidated drained triaxial tests of the sand. It is shown from the
load tests that the behaviour of the piled raft is not mere summation of the components, but is largely
influenced by the interaction of the components as well as the level of the applied load.
1 INTRODUCTION
Application of piled raft foundation design is
increasing in the world as an economic foundation
system to reduce average and/or differential
settlement (e.g. Katzenbach & Leppla 2013;
Yamashita 2012). The international CPRF-Guideline
has been published by the International Society of
Soil Mechanics and Geotechnical Engineering
(ISSMGE 2012). In their life time, these foundations
may also be subjected to lateral loads such as winds
and earthquakes. Experimental studies on piled rafts
subjected to horizontal loading were carried out by
Horikoshi et al. (2003), Matsumoto et al. (2004),
Matsumoto et al. (2010) and so on. However, the
design framework of piled rafts subjected to lateral
loading has not been established (Matsumoto, 2013).
In order to understand the behaviour of piled raft
foundations under vertical and horizontal loading, a
series of vertical load test and cyclic horizontal load
test on a 3-pile piled raft model were carried out in
dry sand model ground at 1-g field. Load tests of the
components of the piled raft model, such as raft
alone and single pile, were also carried to investigate
their interactions in the piled raft. Numerical
modelling of the load tests were conducted, aiming
at obtaining more insight into the behaviour of the
piled raft subjected to vertical and horizontal
loading. A FEM software, PLAXIS 3D, was used for
this purpose. The hardening soil model was
employed for modelling the sand behaviour.

2 OUTLINE OF MODEL TESTS


In order to investigate the pile model foundations
behaviour, a series of load tests which includes static
vertical and horizontal loading of model foundations
(single pile, raft alone and piled raft) were considered in the scope of this paper.
2.1 Test set-up
Dry silica sand #6, having a relative density, Dr, of

about 70 % was used as a model ground. The


physical properties of the sand are summarised in
Table 1. The mechanical properties of the sand were
obtained from triaxial CD tests as described later.
Table 1. Properties of model ground.
Item
Density of soil particles, s (t/m3)
Maximum dry density, dmax (t/m3)
Minimum dry density, dmin (t/m3)
Maximum void ratio, emax
Minimum void ratio, emin
Median grain size, D50
Coefficient of uniformity, Uc

Value
2.66
1.542
1.280
1.079
0.725
0.423
1.880

Model ground was prepared in a laminar box that


had dimensions of 800 mm (in x-direction) x 500
mm (in y-direction) with a depth of 530 mm (Figure
1). The model ground was prepared by layers (10
layers of 50 mm and one layer of 30 mm) in order to
control the density of the model ground. Each layer
was tamped until Dr reached 70%.

VDG-L

In horizontal loading test of the piled raft,


vertical load was applied by placing five lead plates
(497 N in total, Figure 1) prior to starting of the
horizontal loading, and then horizontal load was
applied by means of rotating wooden rods and wires
in a displacement-controlled manner.
The detailed description of the tests is given by
Unsever et al. (2014).

VDG-R

400
150

150
100

LC-L

LC-R
HDG

30

240
255

530

P2

P1

P3

2.2 Triaxial CD tests of the sand and modelling

20

800

Figure 1. Test setup for horizontal loading of piled raft.

Table 2. Properties of the model pile.


Item
Outer diameter, D (mm)
Wall thickness, t (mm)
Length, L (mm)
Cross sectional area, A (mm2)
2nd moment of inertia, I (mm4)
Youngs modulus, E (N/mm2)
Poissons ratio,

30

80

SH1-2

shear strain
gauges

Value
20.00
1.1
255
65.31
2926.2
64000
0.31

40

80

40

80

20

SG1

40

80

A series of triaxial CD tests of the sand having Dr =


70 % were carried out under different confining
pressures, p0 (p0 = 11, 50, 100 and 150 kPa). In
addition, a cyclic CD was conducted under p0 = 100
kPa. The soil specimens had a height of 100 mm
with a radius of 50 mm.
The test results, deviatoric stress, q, versus axial
strain, a, and volumetric strain, vol, versus a, are
shown in Figures 3 and 4, respectively. It is seen that
the initial stiffness increases with increasing p0 and
stress-strain relations exhibit non-linearity (Figure 3),
and that a relatively large dilatancy occurs (Figure 4).
In order to model numerically the above observed
behaviour of the sand, the Hardening Soil model
(HS model) (Schanz et al., 1999) was employed.
Principle parameters of HS model are summarised in
Table 3. The parameters except for Eoed, m and Rf
were estimated from the cyclic CD test
with p0 p0ref 100 kPa .

240

SG2

40

Silica sand #6

30

SG3

800

80

40

vertical strai n
gauges

255

SG5

40
SG6

35

1.1

15

20

mm

600
500
400

Figure 2a shows the single pile model (SP) made


of aluminium, properties of which are summarised in
Table 2. The pile was instrumented with strain
gauges at six levels to obtain axial forces, bending
moments and shear forces induced in the pile during
loading tests. Piled raft (PR) model was composed
of three model piles and a rectangular raft of
stainless steel having dimensions of 240 x 80 mm
with a thickness of 30 mm (Figure 2b). Centre-tocentre pile spacing, s, was 80 mm, 4 times the pile
diameter, D = 20 mm (s/D = 4). The 3 piles were
rigidly connected to the raft. Note that the sand
particles were adhered on the pile shaft and the raft
base to increase their friction resistance.

Experiments

300
200
100

20

(a) Single pile


(b) Piled raft
Figure 2. Single pile model and piled raft model.

p0=150 kPa
p0=100 kPa (cyclic)
p0=100 kPa
p0= 50 kPa
p0= 11 kPa

Hardening soil

700

10

Axial strain, a (%)

Figure 3. Deviatoric stress q versus axial strain a.


Silica sand #6
-7

Volumetric strain, vol (%)

80

SG4

Deviatoric stress, q (kPa)

40

255

p0=50 kPa

-6

p0=150 kPa
p0=100 kPa (cyclic)
p0=100 kPa
p0= 50 kPa
p0= 11 kPa

p0=11 kPa

-5
-4

p0=100 kPa

-3
-2

Experiments
p0=150 kPa

-1
0
1
2

Hardening soil
0

10

Axial strain, a (%)

Figure 4. Volumetric strain vol versus axial strain a.

Table 3. Material properties of the sand.


Item
Secant stiffness, E50 (kPa)*
One-dimensional stiffness, Eoed (kPa)*
Unloading/reloading stiffness, Eur (kPa)*
Stress dependency parameter for stiffness, m
Non-linear factor, Rf
Poisson's ratio,
Internal friction angle, ' (deg.)
Dilatancy angle, (deg.)

Value
29.56103
23.65103
99.59103
0.5
0.75
0.19
43.2
15.8

* values for a reference stress, p0ref = 100 kPa


It was assumed that Eoed = E50/1.2. The values of
m and Rf listed in Table 3 were determined so that
the results of numerical simulations fit to the
measured results.
The results of the numerical simulations of the
CD tests are compared with the measured results in
Figures 3 and 4. The numerical modelling fairly
simulate the measured results, although the postpeak softening behaviours measured in the CD tests
are not simulated using the numerical modelling.
3 NUMERICAL MODELLING OF LOAD TESTS
The numerical study was carried out by three
dimensional finite element program, PLAXIS 3D.
Figure 5 shows the finite element mesh of the
vertical loading of the piled raft model, for an
example. Half of the model foundation and the
model ground were modelled due to symmetric
conditions. Interface elements were arranged along
the pile shafts and at the raft base. The finite element
mesh consists of 10-noded 29570 triangular
elements for the piled raft model. Similar FEM
modelling was adopted for the cases of single pile
and raft alone. In the case of the single pile, only the
centre pile was modelled without the raft and two
edge piles. Only the raft was modelled without the
three piles for the case of the raft alone.
The model sand ground was modelled throughout
using the HS model having the parameters listed in
Table 3. The interface friction angle for the pile shaft
800

and the raft base was estimated as 31.1 degrees from


the horizontal loading test of the raft alone model on
the model ground.
The raft and the piles were assumed to be linearly
elastic. Although the model piles were hollow
cylinders with an end plate, they were modelled by
combination of beam elements surrounded by solid
elements, following Kimura & Zhang (2000). In the
hybrid modelling of pile, a large portion of the
bending stiffness, EI, and axial stiffness, EA, of the
pile are shared by the beam elements, still keeping
large enough stiffness of solid pile elements
compared to the stiffness of the surrounding ground.
The big advantage of the hybrid modelling of pile is
that axial forces, bending moments and shear forces
of the pile can be estimated easily from the factored
values of those of the beam elements.
The following FEM analysis procedure was
adopted:
Step 1: K0 consolidation of the ground alone.
Step 2: Setting the foundation in the model ground
and gravity calculation.
Step 3: Calculation of loading process (placing of
weight plates is included, if present).
In the analyses, loading of the model was
performed by displacement-controlled manner. The
displacement of the model foundation was increased
by 0.2 mm increment until 2.00 mm displacement,
then the interval was increased to 0.4 mm and
analyses was continued until failure occurred.
4

EXPERIMENTAL AND ANALYTICAL


RESULTS

4.1 Vertical load test of the single pile model


Figure 6 shows the measured and calculated
relationships of the pile head load and the pile head
displacement, w. Both results have similar initial
stiffness which is an important criteria in design.
Although the calculated maximum capacity is about
67 % of the measured one, the calculated loadsettlement relation is very close to the measured
curve until this load.
Vertical load, V (N)

(mm)

240

100

200

300

400

500

40
250

530

Settlement, w (mm)

Measured
Calculated
1

Figure 5. Finite element mesh of the piled raft model.

Figure 6. Vertical load-pile head settlement relationship of the


single pile.

4.3 Vertical load test of the piled raft model

100

The measured and calculated relationships of the


vertical load and the settlement of the piled raft are
shown in Figure 9. The measured and calculated
loads supported by the 3 piles and the raft in the
piled raft are also shown in the figure. It was
observed in the experiment that the raft load attains
the peak value of 2684 kN at w = 8.5 mm then
slightly decreases with increasing w. The behaviour
of the raft in the piled raft is comparable with that of
the measured in the vertical load test of the raft alone
in Figure 8. In contrast, the behaviour of the piles in
the piled raft is totally different from that of the
single pile. The load of the 3 piles in the piled raft is
1692 N at w = 3 mm and 3184 N at w = 12 mm. The
maximum load of the single pile was 460 N at w = 3
mm (see Fig. 6) and 633 N at w = 12 mm. That is,
the load of the 3 piles in the piled raft is much higher
than 3 times the load of the single pile. This
tendency becomes significant for larger settlements.
It is thought that the load transfer from the raft base
to the ground enhances the pile resistance in the
piled raft due to the resulting increase in the stiffness
and the strength of the soil surrounding the piles in
the piled raft. The calculation results fairly simulate
the above-mentioned experimental results, although
the calculated responses of the total load, the raft
load and the 3 piles load exhibit bi-linear (elasticperfectly plastic) response.
The proportion of vertical load carried by the 3
piles and the raft with the change in normalised
settlement, w/D, are given in Figure 10. As it is
measured, the 3 piles take about 70 % of the applied
load at very early stage of loading, then the load
proportion taken by the 3 piles decreases with
increasing w/D and levels off at a value of 52 %
when w/D is 0.15. The load proportion taken by the
raft, of course, has the counter trend. The calculation
simulates well the measured behaviour, although the
calculated proportion of the load by the 3 piles is
smaller than the measurement.

200

300

50
100
150
200
250
w/D 0.01 0.02 0.03 0.04 0.05
Mes.
Cal.

Figure 7. Distribution of axial forces along the pile shaft.

The measured and calculated distributions of the


axial forces along the pile shaft at normalised
displacements, w/D = 0.01, 0.02, 0.03, 0.04 and 0.05
are is given in Figure 7. There is reasonable
agreement between the measurements and the
calculations, especially at small displacements.
4.2 Vertical load test of the raft alone model
Figure 8 compares the measured and calculated loadsettlement relationships of the raft alone model. In
the experiment, the maximum load of 3200 N was
attained at w = 8.5 mm. Post-peak softening
behaviour was observed and the residual ultimate
load of 2915 N was obtained.
It is seen from Figure 8 that initial stiffness of the
raft is simulated very well until w reaches 1.5 mm.
After that settlement, the calculated stiffness of the
raft is larger than the measurement. The calculated
maximum load of 2972 N is comparable with the
measured residual load, although the analysis does
not express the measured softening behaviour.
Vertical load, V (N)

Settlement, w (mm)

0
0
1
2
3
4
5
6
7
8
9
10
11
12

500

1000

1500

2000

2500

3000

3500

Vertical loading of full raft


Measured
Calculated

Figure 8. Load- settlement relationships of the raft alone.

Vertical load, V (N)

Verical displacement, w (mm)

Depth from G.L., z (mm)

Axial force, Fa (N)

1000

2000

3000

4000

5000

6000

7000

0
1
2
3
4
5
6
7
8
9
10
11
12

Total (Mes.)
3 piles (Mes.)
Raft (Mes.)

Total (Cal.)
3 piles (Cal.)
Raft (Cal.)

Figure 9. Vertical load- settlement relationship of the piled raft.

Proportion of load carried


by each component (%)

100

3 piles (Mes.)
3 piles (Cal.)

80

60

40

Raft (Mes.)
Raft (Cal.)

20

0
0.00

0.05

0.10

0.15

0.20

displacement, u/D, are shown in Figure 12. The


measured and calculated resistances of the 3 piles
and the raft in the piled raft are also indicated in the
figure. The horizontal load carried by the 3 piles
continues to increase with increasing u/D, whereas
the raft resistance tends to level off after u/D exceeds
about 0.05. When measurements and calculations are
compared, it is seen that calculation overestimates
the experiment results. However, the calculated
trends of the total load, the 3-piles load and the raft
load comparable with the measurements.

0.25

800

Nomarised settlement of piled raft, w/D

700

Horizontal load, H (N)

Figure 10. Load sharing of 3 piles and the raft.


1200

Pile load (N)

1000

P1&P3 (Mes.)
P2 (Mes.)
P1&P3 (Cal.)
P2 (Cal.)

800

Pile 2
(Centre pile)

600
500
400
300
200
100

600

0
0.00

400

0.05

0.10

0.15

0.20

0.25

Normalised horizontal displacement, u/D

Figure 12. Horizontal load-normalised horizontal displacement


relationship of the piled raft.

Piles 1 & 3
(Edge piles)
0.05

0.10

0.15

0.20

0.25

0.30

Normalised settlement of piled raft, w/D

Figure 11. Vertical loads at edge and centre piles.

The vertical loads taken by the edge piles (P1 and


P3) and the centre pile (P2) are plotted against w/D
in Figure 11. It was measured that the edge piles (P1
and P3) support slightly larger load than the centre
pile (P2) until w/D reaches 0.1, and thereafter P2
supports larger load than the edge piles. The above
mentioned response of the piles at smaller w/D is
similar to that of a piled raft having a rigid raft in a
uniform elastic ground. The larger load of P2 at
larger w/D is thought to be due to a larger stress
level in the soil surrounding P2 caused by load
transfer from the raft base to the soil.
Although higher initial stiffness of the piles
measured in the experiment is not simulated well in
the calculation, the calculation results in Figure 11
simulate reasonably well the measured trend
qualitatively.
4.4 Horizontal load test of the piled raft model
In the cyclic horizontal load test of the piled raft
model, a vertical load of 497 N in total was applied
on the raft top prior to the start of horizontal loading.
Although cyclic horizontal loading was applied to
the model, the behaviour in the 1st loading in
positive direction is focused in this particular paper.
The measured and calculated relationships of
horizontal load, H, and normalised horizontal

Figure 13 shows the change in the percentage of


horizontal load carried by the raft against u/D. At the
beginning of the test, the horizontal load carried by
the raft is about 85%, and it decreases with
increasing u/D and finally becomes constant around
25%. The calculation simulates well the measured
behaviour qualitatively and quantitatively.
Perc. of horz. load carried by the raft (%)

200

0
0.00

Meas. Calc.
Total
3 Piles
Raft

100
80

Measured
Calculated

60
40
20
0
0.00

0.05

0.10

0.15

0.20

0.25

Normalised horizontal displacement, u/D

Figure 13. Percentage of horizontal load carried by the raft.

The axial load carried by each pile of the piled


raft is given in Figure 14. Before applying the
horizontal load, edge piles (P1 and P3) carry the
same amount of load, which is smaller than that of
the centre pile (P2). After starting of the horizontal
loading, the vertical load on the front pile (P3) starts
to increase, while the vertical load on the rear pile
(P1) starts to decrease, which suggests that the
contact pressure at the raft base increases around P3
and decreases around P1. The calculation simulates
the measured trend of the changes of pile loads.

400

Pile 1 (Rear pile)


Pile 2 (Centre pile)
Pile 3 (Front pile)

300
250

Pile axial load (N)

5 CONCLUDING REMARKS

Calc. Meas

350

200
150
100
50
0
-50
-100
0.00

0.05

0.10

0.15

0.20

0.25

Normalised horizontal displacement, u/D

Figure 14. Axial load-displacement relationships.

Depth from G.L., z (mm)

0
50

P2 in PR

100
150
200

Meas. Calc.
u/D= 0.02
u/D= 0.04
u/D= 0.06
u/D= 0.08
u/D= 0.10

REFERENCES

250
-10

-8

-6

-4

-2

Bending Moment, M (N.m)


Figure 15. Bending moments along the pile shaft for P2.

Depth from G.L., z (mm)

0
50

P3 in PR

100
150
200

Meas. Calc.
u/D= 0.02
u/D= 0.04
u/D= 0.06
u/D= 0.08
u/D= 0.10

250
-10

-8

A series of vertical and horizontal loading tests on


the single pile, the raft alone and the piled raft
models were carried out in dry sand to investigate
the resistance mechanisms of the piled raft. FEM
modelling of the experiments were conducted to
obtain more insight into the mechanisms and to
explore a possible design approach.
It was found from the load tests on the model
foundations that the behaviour of the piled raft is not
mere summation of the components, but is largely
influenced by the interaction of the components as
well as the level of the applied load.
The numerical modelling in which soil model
capable of considering stress dependent behaviour
such as the Hardening soil model, was able to
simulate quantitatively well the experimental results
of the vertical load tests on the single pile and the 3pile piled raft. However, the numerical modelling
simulated qualitatively the horizontal load test on the
piled raft. Consideration of anisotropic nature of the
sand, for example, would be needed in future study.

-6

-4

-2

Bending Moment, M (N.m)


Figure 16. Bending moments along the pile shaft for P3.

The bending moments along the pile shaft for P2


and P3 are given in Figures 15 and 16, respectively.
The maximum negative bending moment occurs at
the top of each pile. Maximum positive bending
moment occurs at depths of 0.55-0.60 of the pile
length. It is seen from comparison of Figures 15 and
16 that bending moments generated in the front pile
are larger than that of the centre pile. The same
behaviours were also observed in the experimental
study of Horikoshi et al. (2003). Although the
calculation overestimates the measured bending
moments, the calculation simulates well the trends of
measured distributions in both piles.

Horikoshi, K., Matsumoto, T., Hashizume, Y., Watanabe, T. &


Fukuyama, H. 2003. Performance of piled raft foundations
subjected to static horizontal loads. IJPMG-International
Journal of Physical Modelling in Geotechnics 3(2): 37-50.
International Society of Soil Mechanics and Geotechnical
Engineering. 2012. ISSMGE Combined Pile-Raft
Foundation Guideline (Ed. Katzenbach, R., Choudhury, D.).
Katzenbach, R. & Leppla, S. 2013. Economic solutions for
geotechnical challenges like super high-rise buildings and
urban tunnelling. Int. Conf. State of the art of pile
foundation and pile case histories, Indonesia, A1-1-A1-12.
Kimura, M. & Zhang, F. 2000. Seismic evaluations of pile
foundations with three different methods based on threedimensional elasto-plastic finite element analysis. Soils and
Foundations 40(5): 113-132.
Matsumoto, T., Fukumura, K., Kitiyodom, P., Oki, A. &
Horikoshi, K. 2004. Experimental and analytical study on
behaviour of model piled rafts in sand subjected to
horizontal and moment loading, Int. Journal of Physical
Modelling in Geotechnics, 4(3): 1-19.
Matsumoto, T., Fujita, M., Mikami, H., Yaegashi, K., Arai, T.
& Kitiyodom, P. 2010. Load tests of piled raft models with
different pile head connection conditions and their analyses.
Soils and Foundations 50(1): 63-81.
Matsumoto, T. 2013. Implications for design of piled raft
foundations subjected lateral loading. Proc. Int. Symp. on
Advances in Foundation Engineering, Singapore, 113-136.
Plaxis BV, Netherlands User Manuals, Plaxis 3D . 2013.
Schanz, T., Vermeer, P.A. & Bonnier P.G. 1999. The
hardening soil model: Formulation and verification. In
R.B.J. Brinkgreve., Beyond 2000 in Computational
Geotechnics. Rotterdam: Balkema. 281-290.
Unsever, Y.S., Matsumoto, T., Shimono, S. & Ozkan, M.Y.
2014. Static cyclic load tests on model foundations in dry
sand. Geotech. Eng. Journal of SEAGS & AGSSEA 45(2)
(accepted for publication).
Yamashita, K. 2012. Field measurements on piled raft
foundations in Japan". Proc. of Int. Conf. on Testing and
Design Methods for Deep Foundations, Kanazawa, 79-96.

S-ar putea să vă placă și