Vous êtes sur la page 1sur 4

Investigation of Pile Behaviour Using Embedded Piles

Etude du comportement des pieux en utilisant des éléments de pieux intégrés

H.K. Engin
Delft University of Technology, Delft, Netherlands
R.B.J. Brinkgreve
Delft University of Technology & PlaxisBV, Delft, Netherlands

ABSTRACT
The embedded pile model consisting of beam elements with non-linear skin and tip interfaces is developed in Plaxis 3D Foundation to
describe the pile-soil interaction in an efficient manner. The interaction between the pile and the surrounding soil at the pile shaft is
described by means of embedded interface elements. At the pile tip, the soil resistance against compression is represented by means of
embedded non-linear spring elements. There is no need for mesh refinement around piles as 3D mesh is not distorted by introducing
these elements which make embedded piles very efficient and time saving especially when a large group of piles is modelled. In this
paper a description of the model is given with preliminary validation results for single piles in compression and also in tension.
Performance of the model is investigated and validated for pile groups. Effectiveness of the model for bored piles is verified.
RÉSUMÉ
Un modèle de pieux intégrés formulé à partir d’éléments de poutre reliés aux éléments de sol par l’intermédiaire d’éléments
d’interface non linéaires en peau en en pointe a été mis au point dans le logiciel Plaxis 3D Foundation pour pouvoir décrire de
manière efficace l’interaction soil-pieu. L’interaction entre le pieu et le sol au niveau est prise en compte par l’intermédiaire
d’éléments d’interface. A la pointe du pieu, la résistance du sol vis-à-vis des efforts de compression est représentée par des éléments
de ressorts non linéaires. Il n’est pas nécessaire de considérer un raffinement du maillage au voisinage du pieu puisque le maillage
tridimensionnel n’est nullement affecté par l’introduction de ces éléments de pieux les rendant ainsi très efficaces et économiques du
point de vue temps de calcul tout particulièrement dans les cas où un grand nombre de pieux sont modélisés. Dans cet article, une
description du modèle est donnée ainsi que quelques résultats de validation préliminaires pour des pieux isolés en compression et en
traction. La performance du modèle est finalement évalué et validé pour un groupement de pieux. L’efficacité du modèle pour les
pieux forés est vérifiée.
Keywords : pile, finite element modelling, embedded pile

1 INTRODUCTION simple but efficient modification attached to the embedded pile


model the mesh dependency problem is solved.
The use of 3-dimensional finite element analyses in design of In the following parts, verification and validation examples
foundations has become more popular in last decades in are given in comparison with some field test results of single
accordance with the progress of PCs and more advanced FE bored piles (in compression and tension). Moreover, pile group
software. For a good level of accuracy finer meshes are generally behaviour is investigated numerically and validated by a case
required. More specifically, piled foundations generally involve a study.
large number of piles, which increase the analysis time. Besides,
there might be a considerable amount of effort required to create
the FE mesh. The situation is aggravated if there are inclined piles 2 EMBEDDED PILE MODEL
included in the foundation system.
The embedded pile model has been developed in the Plaxis The embedded pile model is a beam element type model, which
3D Foundation software, where the pile is modelled by means is connected to the (soil) continuum by embedded skin (shaft)
of beam elements which are connected with the continuum and foot (tip) interfaces. This model enables any type of
using special interface elements. As the pile is modelled as orientation for the pile (position and inclination) without
beam (line) it is easy to give any inclination in any direction. distortion of the mesh. At the points where pile intersects the
More importantly, it does not distort the mesh and introduce continuum (soil) elements additional pile nodes for the
very few nodes which decrease the computational effort formulation of interaction between pile and soil elements are
compared to regular volumetric pile modelling. introduced.
The pile - soil interaction is governed by relative movements The pile-soil interaction is represented by interfaces at the
between the (newly generated) pile nodes and the (existing) soil shaft and tip of the pile. The skin friction is defined by Eqn. 1.
nodes. In addition to the embedded beam model by Sadek &
Shahrour (2004), the interaction between the pile and the soil at Δt skin =T skin Δu rel with Δu rel =Δu p -Δu s (1)
the skin is established by means of line to volume interface
elements to represent the pile-soil contact at the skin and point where Δt is the traction increment at the integration points, Tskin
to volume interface element to represent the pile-soil contact at is the material stiffness matrix of the pseudo skin interface and
the base (Septanika 2005a&b). Δurel = (Δup – Δus) represents the relative displacement vector
Due to the fact that the volumetric pile is modelled as a line, between the soil and the pile. At the tip connection with the soil
a mesh dependent behaviour (not only in stiffness but also in is made by a spring in which the foot (tip) force is determined
capacity) is observed with the initial model description. By a by Eqn. 2.

Proceedings of the 17th International Conference on Soil Mechanics and Geotechnical Engineering 1189
M. Hamza et al. (Eds.)
© 2009 IOS Press.
doi:10.3233/978-1-60750-031-5-1189
1190 H.K. Engin and R.B.J. Brinkgreve / Investigation of Pile Behaviour Using Embedded Piles

re l with Δ u rel = Δ u p -Δ u s
Δ F fo o t = D fo o t Δ u fo ot fo ot (2)
3500
PILE CAPACITY
foot foot
where ΔF is the force increment, D represents the material 3000

stiffness matrix of the pseudo spring element at the foot, Δurelfoot Total Capacity Skin Friction
2500
represents the relative displacement vector between the soil and
the pile at the foot. 2000

Load (kN)
The element stiffness matrix representing the pile-soil
1500
interaction at the mantle has been derived based on the internal Base Resistance
virtual work consideration (for detailed explanations refer to 1000

Septanika 2005a). The capacity of the embedded pile is


500 Test Result
described according to elasto-plastic formulations at the skin
Embedded Pile
and tip interfaces. The elasto-plastic formulation enables the 0
control on the pile capacity (both on skin and tip capacities). 0 5 10 15 20
Settlement (mm)
25 30 35 40

Therefore a skin-traction model has to be defined according to:


Figure 2. Load-displacement curves of Alzey Bridge pile load test
(a) constant/linear traction-depth model, (b) multi-linear model, presented together with Plaxis 3D Foundation (Embedded pile) results
or (c) layer-dependent model which relates the allowable (after Engin et al. 2007)
traction to the strength of the adjacent soil layer. For the
ultimate tip (foot) resistance representing the base failure (due The load - displacement behaviour of embedded pile models
to penetration or tension) at the pile tip, a simplified criterion is are presented together with South Surra pile test results in
utilized (Eqn.3). Figure 3. Results obtained for the embedded pile model of 5.3
m long test pile is in good agreement with the pile load test
ΔFaxialfoot ≤ ΔFmaxfoot (compression) & ΔFaxialfoot =0 (tension) (3) results. Similarly, 3.3 m long test pile results for embedded pile
shows that the embedded pile is able to catch the real behaviour.
where ΔFaxialfoot is the axial tip. 500

Using aforementioned formulations alone result in mesh- Pile Capacity (L=5.5m)

dependent behaviour. This has been improved by the 400


application of an elastic region approach for the soil region
within the pile (Figure 1). This approach appears to be sufficient Pile Capacity (L=3.3m)
Load (kN)

300
for reducing the undesirable mesh-dependent effects (Engin
2006 and Engin et al. 2007).
200
2·R Test Pile (L=5.5m)

Actual pile region Embedded Pile (L=5.5 m)


100
Test Pile (L=3.3m)
Embedded Pile (L=3.3 m)
0
0 5 10 15 20 25 30 35 40 45
Embedded pile Settlement (mm)

Figure 3. Load - displacement behaviour of South Surra test piles and


Plaxis 3D Foundation embedded pile models (After Engin et al. 2008b).
Figure 1. Sketch of elastic region approach (Engin et al. 2007).
Moment (MNm/m)
3.2 Pile Group Behaviour 1,2

3 VALIDATION OF THE MODEL 1,0


3.2.1 Poulos Hypothetical Example0,8

In this part work on validation of embedded pile model is given. Embedded pile efficiency in pile0,6group analysis was previously
The single and group pile behaviours are given separately. The investigated (Engin et al. 2008a) for the hypothetical example of FE Ta & Small

FE + BE Sinha
Plate (GASP)
Plaxis 3D Fnd

Strip (GASP)

0,4
work on previous papers is summarized in the first part. Poulos for 9 pile case (2001).
0,2
Secondly, pile group behaviour is validated by a hypothetical
0,0
example by Poulos and a case study.
Average Settlement (mm) Differential Settlement (mm)
3.1 Single Pile Tests 50,0 10,0
45,0 9,0
40,0 8,0
The behaviour of embedded pile in compression is previously 35,0 7,0
30,0 6,0
validated (Engin et al. 2007) for a bored pile case based on the 25,0 5,0
Alzey Bridge pile load test data (Sommer and Hambach 1974).
FE Ta & Small

FE Ta & Small
Poulos & Davis

20,0 4,0
FE + BE Sinha
Plate (GASP)

FE + BE Sinha
Plate (GASP)
Plaxis 3D Fnd
Plaxis 3D Fnd

Strip (GASP)

Strip (GASP)

15,0 3,0
The results showed that the model can be efficiently used for
Randolph

10,0 2,0

the bored pile case (Figure 2). 5,0


0,0
1,0
0,0
Secondly, tension tests on bored piles (Ismael et al., 1994),
which were carried out in Kuwait (test sites South Surra and Figure 4. Poulos hypothetical example and comparison of embedded
Umr Gudayr) is used for validation of embedded piles in tensile pile model analysis results (After Engin et al. 2008b).
loading (Engin 2007a&b, Engin et al. 2008b). The first site
(South Surra) has a profile of medium-dense and very-dense In Figure 4, the analysis results of the embedded pile model
weekly cemented calcareous sand. Two short bored piles, which are given. It can be seen that embedded pile model gives
were 0.3 m diameter and having lengths of 3.3 m and 5.3 m, relatively small maximum moment and average settlement
were tested in axial tension to failure. One pile (5.3 m long pile) compared to other methods. On the other hand, the highest
was instrumented with strain gauges to measure the axial load differential settlement value is obtained for the embedded pile
distribution at all load increments. model. It can be seen that methods of Poulos - Davis and
Randolph only give information for the average settlement.
H.K. Engin and R.B.J. Brinkgreve / Investigation of Pile Behaviour Using Embedded Piles 1191

3.2.2 Messe-Torhaus Building Piled Raft Case Table 1. Model parameters used in the analyses (Messe-Torhaus)
The 130 m Messe-Torhaus building in Frankfurt (Figure 5a), Property Unit L1** L2**
which was constructed between 1983 and 1986, was the first
Material model - HS* HS*
building in Germany designed with a piled-raft foundation. A Unit weights, γ’ / γdry kN/m3 11/19 10/20
total number of 84 bored piles with a length of 20 m and Reference Secant stiffness, E50 ref
kN/m2 3.0×104 3.5×104
diameter of 0.9 m are located under two 17.5 m x 24.5 m large Ref. oedometer stiffness, Eoedref kN/m2 3.0×104 4.28×104
rafts. The distance between the two rafts is 10 m. Ref. unloading-reload. stiffness, Eurref kN/m2 7.5×104 1.05×105
Stress dependency power, m - 0.5 1.0
Poisson’s ratio, νur
130m
- 0.2 0.2
Cohesion, c’ kN/m2 0.0001 20
o
Internal friction, ø 35 20
100m
Dilatancy angle, ψ o
0 0
At rest lat. press. coeff. for NC, KoNC - 0. 426 0.8
Overconsolidation ratio, OCR - 1.0 1.0
Past overburden press., POP kN/m2 0 0
Interface stiffness ratio, Rint - 1.0 1.0
* HS: Hardening Soil Model
** L1: Quaternary sand and gravel, L2: Frankfurt Clay

Table 2. Embedded pile properties


Property Unit Value
Pile diameter, D m 0.9
Pile length, L m 20
Unit weight, γ kN/m3 15
-3m
Æ Quaternary Layer
Moment of inertia, I2=I3 m4 0.032
Poisson’s ratio, νur - 0.2
-23m Æ Frankfurt Clay Young’s modulus, E kN/m2 2.35×107
Max. skin friction:
Ttopmax kN/m 453
Figure 5. Messe-Torhaus Building (a) cross-sectional view soil profile, Tbottommax kN/m 453
(b) piled raft layout and instrumentation (after Reul & Randolph 2003 Max. tip resistance kN 1200
and Chow 2007).
Each piled raft foundation is loaded to total design load of
The bottom of the 2.5 m thick raft lies 3 m below ground the structure (200 MN). Load settlement behaviour is given in
level (Figure 5(a)). The subsoil profile is composed of Figure 7. The raft and pile shares are also given in this figure.
quaternary sand and gravel up to 2.5 m below the bottom of
the rafts, followed by the Frankfurt clay. The groundwater Load (MN)
0 50 100 150 200 250
level lies below the rafts. For the geotechnical monitoring of 0
Measured-Total
the piled raft foundation, six instrumented piles, eleven Measured-Piles
10
contact pressure cells and three multi-point borehole Measured-Raft
Plaxis 3D Fnd - Embedded Pile - Total
extensometers were used. The position of the measurement 20 Plaxis 3D Fnd - Embedded Pile - Piles
Settlement (mm)

devices is shown in Figure 5(b). 30


Plaxis 3D Fnd - Embedded Pile - Raft

The design of the piled raft foundation was based on a


conventional approach where the piles are assumed to be loaded 40

to their ultimate bearing capacities and raft assumed to transmit 50


the remaining part of the load (Katzenbach et al. 2000).
60
Plaxis 3D Foundation is used to model this particular case
to validate the embedded piles for the group behaviour. The 70
finite element mesh (36456 quadratic elements) is given in 80
Figure 6.
Figure 7. Load settlement of the northern piled-raft of Messe Torhaus
Building (after Katzenbach et al. 2000).
8000
Measured
Reul
7000 APRILS
Plaxis 3D Foundation
6000

5000
Load (kN)

4000

3000

Figure 6. Finite Element model of Messe-Torhaus piled-raft foundation. 2000

1000

The soil parameters used in the finite element analysis are 0


given in Table 1. For the raft a unit weight, γraft=25 kN/m3, a 1 2 3 4 5 6
Pile Number
Young’s modulus of E=3.7×107 kN/m2 and a Poisson’s ratio
Figure 8. Comparison of measured and calculated pile loads under the
of ν=0.2 is used. A reduced stiffness was used during raft
northern raft (after Chow 2007).
installation to ensure a proper transfer of the raft load on to
the underlain soil. Embedded pile properties are given in
The comparison of pile load distributions is given in
Table 2. All piles are given the same pile resistances (tip and
Figure 8. Furthermore, the measured and calculated axial
skin).
load and skin friction distributions along test piles TP1 and
TP5 are given in Figure 9.
1192 H.K. Engin and R.B.J. Brinkgreve / Investigation of Pile Behaviour Using Embedded Piles

Axial Force Distributions (kN) 5 CONCLUSIONS


0 2000 4000 6000 8000
0 In this paper, the embedded pile model is explained. The main
advantage of the embedded pile model is its efficient
5
formulation, using line elements in combination with elasto-
plastic line-to-volume and point-to-volume interfaces.
In order to validate the applicability in practical cases,
several validation examples have been demonstrated for single
Depth (m)

10

pile tests (both in for compression and tension behaviour). In


15
this contribution, attention is focused on pile groups.
In addition to a hypothetical model according to Poulos, the
Measured-TP1 Messe-Torhaus piled raft foundation is modelled for the
20 Measured-TP5 validation of the embedded piles with respect to pile groups.
TP1 - Plaxis 3D Foundation The results show a reasonable agreement with the measured
TP5 - Plaxis 3D Foundation data and clearly indicate pile group effects.
25
It is concluded that the embedded pile has a great potential
Skin Friction Distributions (kN/m2)
for modelling piles and piled-raft foundations. It will become
-50 0 50 100 150 200
0
particularly advantageous for piled foundations with a large
number of piles or in situations with inclined piles.

ACKNOWLEDGEMENTS
10
Depth (m)

The authors of this paper would like to thank to Mr. Richard Witasse
for the translation of the paper title and the abstract into French.
15

REFERENCES
20

Chow, H.S.W. 2007. Analysis of piled raft foundations with piles of


25 different lengths and diameters. PhD diss., University of Sydney.
Engin, H.K. 2006. A report on embedded piles. Plaxis internal report.
Figure 9. Comparison of measured and calculated distributions of pile Engin, H.K., Septanika, E.G. & Brinkgreve, R.B.J. 2007. Improved
(a) axial load and (b) skin friction for Test Piles 1&5 (after Katzenbach embedded beam elements. In Pande & Pietruszczak (eds),
et al. 2000). Numerical Models in Geomechanics; Proceedings of the 10th
International NUMOG Symposium, Rhodes, Greece, April 25-27,
2007:475-480. Rotterdam: Balkema.
Engin, H.K. 2007a. Tension piles report. Plaxis internal report.
Engin, H.K. 2007b. Çekme kazıklarının “gömülü kazık” elemanlarıyla
4 DISCUSSION OF RESULTS modellenmesi. In Özkan (ed), 2. Geoteknik Sempozyumu Bildiriler
El Kitabı, Adana, Turkey, November, 22-23, 2007. Adana:Artı
Considering the Messe-Torhaus case study, the load-settlement Ofset.
behaviour is in reasonable agreement with the measured Engin, H.K., Septanika, E.G., Brinkgreve, R.B.J.& Bonnier, P.G. 2008a.
behaviour. However, the pile load share is over-predicted in the Modeling piled foundation by means of embedded piles. In
finite element model. The measurements show that the piles Karstunen & Leoni (eds), Geotechnics of Soft Soils: Focus on
carry ~80% of the total load whereas the FE model predicts Ground Improvement; Proceedings of the 2nd International IWGSS
~92%. This can be explained by looking at the individual pile Workshop, Glasgow, Scotland, September 3-5, 2008. London,
Taylor&Francis.
contributions. In general, the tip resistance and skin friction are
Engin, H.K., E.G. Septanika & R.B.J. Brinkgreve 2008b. Estimation of
reasonably well predicted (considering the limited data points of pile group behaviour using embedded piles. In Singh (ed),
the measurements), except for the lower part of the inner pile. International Association for Computer Methods and Advances in
Here, the calculated skin friction distribution is larger than Geomechanics; 12th International IACMAG Conference, Goa, India,
measured, which leads to the over-predicted pile contribution October, 1-6, 2008: 3231-3238
(~50%). Ismael, N.F, Al-Sanad, H.A. & Al-Otaibi, F. 1994. Tension tests on
It should be noted that the top portions of the pile show a bored piles in cemented desert sands. Canadian Geotechnical
negative skin friction. This can be attributed to the Journal 31: 4: 597-603.
settlement of the upper layer during raft installation. Katzenbach, R., Arslan, U. & Moorman, C. 2000. Piled Raft Projects in
Germany. In Hemsley (ed), Design Application of Piled Raft
Although this is not shown in the measurements, it seems
Foundations, 323-391. London: Thomas Telford.
realistic in this situation. Poulos, H.G. 2001. Piled raft foundations: design and applications.
Considering Figure 8, it can be seen that pile group Geotechnique 51: 2: 95-113.
effects are considered in the model to some extent. The load Reul, O. & Randolph, M. F. 2003. Piled rafts in overconsolidated clay:
on corner piles is higher than on centre piles. The former is comparison of in situ measurements and numerical analyses.
well predicted (TP3 & TP5). For the other instrumented Geotechnique 53: 3: 301–315.
piles, the FE model over-predicts the pile loads (except for Sadek, M. & Shahrour, I. 2004. A three dimensional embedded beam
outer edge pile, TP2). There is a deviation from element for reinforced geomaterials. International journal for
measurements as the capacity is defined the same for all numerical and analytical methods in geomechanics 28: 931–946.
Septanika, E. G. 2005a. A finite element description of the embedded
piles (no difference with respect to the particular pile
pile model. Plaxis internal report.
location). Moreover, the influence of stress increase due to Septanika, E. G. 2005b. Validation testing embedded pile in Plaxis 3D
raft loading on the pile resistance is not considered in the Foundation. Plaxis internal report.
model. This could be improved by using the layer-dependent Sommer, H. & Hambach, P. 1974. Groβpfalversuche im Ton für die
resistance model (model c in Section 2). Gründung der Talbrücke Alzey. Der Bauingenieur 49: 310-317.