You are on page 1of 5

HBRC Journal (2018) 14, 316–320

Housing and Building National Research Center

HBRC Journal

http://ees.elsevier.com/hbrcj

Winkler model for pile seismic analysis considering


end constraints effects
Ahmed M. Hassan

Civil Engineering Department, Minia University, Minia, Egypt

Received 24 February 2016; revised 3 November 2016; accepted 6 February 2017

KEYWORDS Abstract Beam on nonlinear Winkler foundation (BNWF) analysis of pile foundation subjected to
Pile foundation; vertically propagating earthquake ground motions is presented in this paper. The BNWF is consid-
Seismic analysis; ered a simplified approach that is capable to model nonlinear soil-pile-structure interaction. Soil
Winkler model; nonlinearity is simulated by p-y curves according to American Petroleum Institute (API) procedure.
p-y method The analyses are carried out using the open source finite element platform OpenSees. Results are
verified against centrifuge experiments and numerical finite element simulations published in the lit-
erature. In this paper pile tip and head constraints effects on bending moments and displacement
responses are investigated. Three types of pile tip constraints are applied: floating, end bearing,
and fixed supports. Pile top constraint is modeled as either free or fixed head. Three different earth-
quake ground motions are applied in the model. Effect of superstructure mass is investigated for the
three considered earthquake records and the different assumed pile end and head constraints. The
results indicate significant influence of pile tip and head constraints on pile bending moment and
displacement responses. A significant effect of superstructure mass on the analyses results is also
observed. The analyses show that earthquake record characteristics remarkably influence pile
response behavior. Different earthquake records scaled to the same peak ground acceleration give
different pile seismic responses.
Ó 2017 Housing and Building National Research Center. Production and hosting by Elsevier B.V. This is
an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction

Seismic analysis of pile foundation has gained growing interest


in the literature. The BNWF is considered a simplified
approach that is capable to model nonlinear soil-pile-
E-mail address: Ahmed.ismael@mu.edu.eg
structure interaction. Soil nonlinearity is modeled through a
Peer review under responsibility of Housing and Building National
Research Center.
series of p-y curves where p is soil resistance per unit length
of pile and y is pile lateral displacement [1,2]. The approach
has been adopted by many researchers [3–8]. Kimiaei et al.
[9] introduced a practical BNWF model for estimating the lat-
Production and hosting by Elsevier eral response of flexible piles embedded in layered soil deposits
http://dx.doi.org/10.1016/j.hbrcj.2017.02.001
1687-4048 Ó 2017 Housing and Building National Research Center. Production and hosting by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Winkler model for pile seismic analysis 317
 
subjected to seismic loading. They applied their approach to kz
p ¼ 0:9pu tanh y ð1Þ
offshore piles and found that it was reasonably able to model 0:90pu
recorded centrifuge responses specifically for moderate rates of
where p = lateral load, pu = ultimate bearing capacity,
peak base accelerations. Zhong and Huang [10] developed a
z = depth, k = initial modulus of subgrade reaction, and
simplified Winkler model for the lateral vibration of the com-
y = lateral deflection.
posite caisson-pile foundations. They demonstrated the relia-
The ultimate bearing capacity, pu, is determined according
bility of this model by conducting comparisons against
to API procedure which is implemented in the OpenSees by the
results of finite element simulations.
uniaxial material PySimple1 [18]. Required soil parameters for
Pile tip constraints’ influence is considered an important
PySimple1 command are pu and the displacement at which
factor in seismic pile analysis. Fernandez and Muria-Vila [11]
50% of pu is mobilized, y50 [19]. Procedures and graphs for cal-
used finite layer method for seismic analysis of piles with par-
culating pu and y50 are given in API [13]. The initial modulus
tially or fully constrained tips embedded in layered soil depos-
of subgrade reaction is determined as a function of angle of
its. They showed that pile internal forces depend on pile
internal friction / [13].
slenderness, soil-pile stiffness contact, soil deposit characteris-
To predict response of piles to seismic motion, the free-field
tics, and pile fixity conditions. The superstructure mass effect
soil displacement, ys, needs to be known first, and then
was studied by Hussien et al. [12] using three dimensional finite
imposed to the ends of Winkler soil springs. The pile response
element analysis.
is then given by the differential equation:
In this study, a BNWF model and soil nonlinearity were
simulated using the American Petroleum Institute [13] p-y d4 yp
curves. The model was programmed using the open source EI  pðyp  ys Þ ¼ 0 ð2Þ
dz4
software: Open System for Earthquake Engineering Simula-
tion ‘‘OpenSees” [14]. Proposed model was verified against where yp = pile lateral displacement.
published centrifuge results by Wu and Finn [15] and Hussien In this study solution is performed in two steps. Firstly, free
et al. [12]. The model was then used to evaluate influence of field response to input ground motion is determined by one-
pile tip and head constraints and superstructure mass. dimensional equivalent linear site response analysis using
Shake- 91 [20]. Secondly, free-field response is then applied
at soil springs ends as multi-support excitations as shown in
Model description
Fig. 1. Pile elements are modeled by the OpenSees elastic
beam-column elements. Interaction between the pile and soil
The proposed model is shown in Fig. 1 in which a single pile in was approximated by the zero-length spring element [19] to
a dry sand deposit underlain by a stiff layer is presented. The connect pile nodes to free-field soil nodes.
superstructure is represented by a lumped mass M. The BNWF
model simulates the laterally-loaded pile problem using Validation of proposed model
displacement-based beam elements to represent the pile and
a series of nonlinear springs to represent the soil. The soil
springs are generated using zero-length elements assigned sep- The proposed model was verified against published centrifuge
arate uniaxial material objects in the lateral directions [16]. In experimental results performed by Hussien et al. [12] at the
the API procedure, p-y curves for cohesionless soils are devel- Disaster Prevention Research Institute, Kyoto University,
oped based on the methods presented by Reese et al. [17]. The Japan. Details of the test can be found in Hussien et al. [12].
p-y curve is defined as follows: Applied base excitation was a sinusoidal motion with ampli-
tude 0.317 g and frequency 1.0 Hz. Pile in the prototype was

Figure 1 Schematic presentation of the applied BNWF model.


318 A.M. Hassan

11.6 m long and 11.0 m embedded depth. The outer pile diam- be exactly what is used in practice. The pile was extended
eter was 400 mm and wall thickness was 30 mm. Young’s mod- 0.60 m above ground level and two constraints were used for
ulus of pile material was 2.06  1010 kg/m2. The pile cap mass pile head: free and fixed. End bearing tip node was restrained
was 24,231 kg in the prototype and superstructure mass was against vertical displacement, whereas total fixation was
19,008 kg. The soil was dense sand with density = 1500 kg/ assigned for fixed pile tip. Floating tip was simulated by
m3, shear modulus = 5.1  106 kg/m2, Poisson’s ratio = 0.33, assuming a q-z spring at pile end node. This was implemented
and friction angle = 38°. Fig. 2 shows a comparison between in OpenSees by the command QzSimple1 [19]. Required
experimental results, computed BNWF by this study, and parameters to apply this command were qult = ultimate capac-
computed FEM values by Hussien et al. [12]. It may be noted ity of the spring material and z50 = displacement at which
that computed values by the proposed BNWF agree fairly well 50% of qult is mobilized. The two parameters were determined
with both experimental and numerical FEM results. by procedure outlined in Meyerhof [21].
Soil layer considered in the model was medium sand with
Model development properties shown in Table 1 which shows also properties of
the reinforced concrete pile and superstructure mass (M).
A reinforced concrete pile of diameter 0.6 m was used in the Three ground motions were applied: Imperial Valley 1979,
model as shown in Fig. 3. Three pile tip constraints were Kobe 1995, and Tottori 2000. Characteristics of these ground
assumed: floating, end-bearing, and fixed tip. It is worth noting motions are shown in Table 2 and their response spectra are
that considered pile tip constraints are idealized for the pur- shown in Fig. 4. The three ground motions were scaled to have
pose of studying their effects on pile behavior and may not an equal peak ground acceleration (PGA = 0.36 g). The
ground motion records were obtained from PEER NGA
Strong Motion Database (http://www.peer.berkeley.edu).
0
-2
Depth (m)

COMPUTED (BNWF) -4
Table 1 Soil and pile parameters.
MEASURED -6
-8 Property Value
COMPUTED (FEM)
-10 Soil density 1500 kg/m3
-12 Soil friction angle 30°
-200,000 -100,000 0 100,000 Pile diameter 0.60 m
Moment (kg.m) Pile Young’s modulus 2.5  109 kg/m2
Pile Poisson’s ratio 0.15
Figure 2 Comparison of Centrifuge, FEM (Hussien et al. [12], Superstructure mass (M) 100,000 kg
and BNWF results).

Figure 3 Considered pile tip constraints.

Table 2 Characteristics of the applied ground motions.


Earthquake Location Year Moment magnitude (MW) Recording station
Imperial Valley California, USA 1979 6.5 Cerro Prieto
Kobe Japan 1995 6.9 Nishi-Akashi
Tottori Japan 2000 6.6 OKY004
Winkler model for pile seismic analysis 319

10 2

Spectral Acceleraon (g)


0
1 -2
Imperial Valley

Depth (m)
FIXED TIP -4
Kobe 0.1 END BEARING -6
Toori
FLOATING TIP -8
0.01
-10
0.01 0.1 1 10
Period (s) -12
-0.001 0.001 0.003 0.005
Figure 4 Response spectra considered for earthquakes (damping Lateral Deflecon (m)
ratio = 5%).
Figure 7 Pile lateral deflection responses due to imperial valley
earthquake.

2
0
-2
Kobe, 1995

Depth (m)
-4
Imperial Valley, 1979
-6
Toori, 2000
-8
-10
Figure 5 Moment responses for imperial valley earthquake.
-12
-0.001 0.001 0.003 0.005
2 Lateral Deflecon (m)
0
-2 Figure 8 Lateral deflection due to considered earthquakes for
Depth (m)

Imperial Valley, 1979 -4 end-bearing piles.


-6
Kobe, 1995
-8
Toori, 2000 constraints had a significant influence on pile lateral deflection
-10
-12 seismic response.
-50,000 0 50,000 100,000 150,000
A comparison of displacement response for end bearing
Moment (kg.m)
piles subjected to different ground motions is shown in
Figure 6 End-bearing piles bending moment responses due to Fig. 8. In spite of the fact that the three earthquakes have
considered earthquakes.
the same peak ground acceleration (PGA = 0.36 g), their
responses are different which emphasized the influence of
Analyses, results, and discussions ground motion frequency content on seismic analysis.
Remarkable high deflections were obtained in Kobe earth-
Pile tip constraint effect quake compared to the other two earthquakes.

Pile head constraint effect


Analyses were performed by applying the three previously
mentioned ground motions for each pile tip constraints.
Fig. 5 shows bending moment responses for different pile tip In this section pile head constraint effect is presented. Fixed
constraints corresponding to Imperial Valley Earthquake. It versus free pile responses are compared for end-bearing piles.
may be noted that tip constraint conditions had a significant
effect on the resulting bending moment. Fixed-tip pile and
end-bearing piles have fairly bigger bending moment values 2
compared to floating-tip pile. 0
Fig. 6 is provided to compare the response of the earth- -2
Depth (m)

quake motion for end-bearing tip pile. It is remarkable that -4


FIXED HEAD
the bending moment response varies for each earthquake
-6
record although the three motions are scaled to same peak FREE HEAD
-8
ground acceleration. This emphasizes the effect of ground
-10
motion frequency content on seismic analysis.
-12
-150,000 -75,000 0 75,000 150,000
Pile lateral deflection response Moment (kg.m)

Fig. 7 shows pile displacement response for Imperial Valley Figure 9 Fixed versus free head pile response due to imperial
earthquake. As noted in the bending moment response, tip valley earthquake.
320 A.M. Hassan

[2] C.L. Reese, W.F. Van Impe, Single Pile and Pile Groups Under
Lateral Loading, 2nd ed., CRC Press, Balkema, 2011.
[3] M.H. El-Naggar, K.J. Bentley, Dynamic analysis for laterally
loaded piles and dynamic p-y curves, Canad. Geotechn. J. 37
(2000) 1166–1183.
[4] S.J. Brandenberg, P. Singh, R.W. Boulanger, B.L. Kutter,
Behavior of piles in laterally spreading ground during
earthquakes, in: Proceedings of the 6th Caltrans Seismic
Research Workshop, CA, 2001, pp. 2–106.
[5] W.D.L. Finn, N. Fujita, Piles in liquefiable soils: seismic analysis
and design issues, Soil Dyn. Earthq. Eng. 22 (2002) 721–742.
[6] N. Allotey, M.H. El Naggar, Generalized dynamic Winkler
model for nonlinear soil-structure interaction analysis, Canad.
Figure 10 Superstructure mass effect on imperial valley earth- Geotechn. J. 45 (4) (2008) 560–573.
quake moment responses. [7] F. Castelli, M. Maugeri, Simplified approach for the seismic
response of pile foundation, Am. Soc. Civil Eng., ASCE J.
The applied earthquake was Imperial Valley. The influence of Geotechn. Geoenviron. Eng. 135 (10) (2009) 1440–1451.
pile head constraint conditions is significant on bending [8] R.J. Armstrong, R.W. Boulanger, M.H. Beaty, Equivalent static
moment values and pattern as clearly indicated in Fig. 9. analysis of piled bridge abutments affected by earthquake-
induced liquefaction, ASCE J. Geotechn. Geoenviron. Eng. 140
(8) (2014) 1–10.
Superstructure mass effect
[9] M. Kimiaei, M.A. Shayanfar, M.H. El Naggar, A.A.
Aghakouchak, Nonlinear response analysis of offshore piles
Effect of superstructure on pile seismic response is taken into under seismic loads, in: Proceedings of 13th World Conference
account by considering different superstructure masses (0, on Earthquake Engineering, Vancouver, B.C., Canada, August
50  103 kg, and 100  103 kg) applied to end-bearing piles 1–16, 2004, Paper No. 3056.
for Imperial Valley earthquake. A significant effect of super- [10] R. Zhong, M. Huang, Winkler model for dynamic response of
structure mass on bending moment values was observed. Pile composite caisson-piles foundations: lateral response, J. Soil
bending moment values increased by increasing superstructure Dyn. Earthq. Eng. 55 (4) (2013) 182–194.
[11] L. Fernandez, J. Aviles, D. Muria-Vila, Fully and partially toe-
mass as indicated in Fig. 10.
restrained piles subjected to ground motion excitation, J. Soil
Dyn. Earthq. Eng. 39 (2012) 1–10.
Summary and conclusions [12] M.N. Hussien, T. Tobita, S. Iai, Experimental and FE analysis
of seismic soil-pile-superstructure interaction in sand, in:
The simplified BNWF method was used to investigate pile tip Annuals of Disaster Prevention Research Institute, Kyoto
University, Japan, 2010, No 53B:299–306.
and head constraints and superstructure mass effects on seis-
[13] American Petroleum Institute Recommended practice for
mic pile responses. Pile tips were assigned three constraints:
planning, design, and constructing fixed offshore platforms,
floating, end-bearing, and fixed. Pile head was assumed either API RP 2A-WSD, Washington, DC, 2007.
free or fixed. The superstructure is represented by a lumped [14] F. McKenna, G.L. Fenves, M.H. Scott, B. Jerermic, Open
mass at the pile top. Superstructure effect was accounted for System for Earthquake Engineering Simulation. <http://
by considering three mass values: 0, 50  103, and opensees.berkely.edu>, 2000.
100  103 kg. Three earthquakes were applied in the analyses: [15] G. Wu, W.D.L. Finn, Dynamic nonlinear analysis of pile
Imperial Valley, Kobe, and Tottori. foundations using finite element method in the time domain,
For the cases, soil parameters, and earthquake motions Cand. Geotechn. J. 34 (1997) 44–52.
considered in this study, the following conclusions may be [16] C. McGann, P. Arduino, OpenSees examples, University of
Washington, <http://opensees.berkely.edu>, 2006.
drawn:
[17] L.C. Reese, W.R. Cox, F.D. Koop, Analysis of laterally loaded
piles in sand, in: Proceedings of the 6th Offshore Technology
 BNWF model proved to be efficient method of modeling Conference, Houston, Texas, 1974, Paper 2080, pp. 473–483.
pile seismic analysis. [18] R.W. Boulanger, The PySimple1 material, 2003, <http://
 Pile bending moment and lateral deflection shape and val- opensees.berkeley.edu>.
ues are remarkably influenced by pile ends restrain and [19] Mazzoni, F. MacKenna, M.H. Scott, G.L. Fenves, et al.
superstructure mass. OpenSees command language manual, Open System for
 Pile seismic responses are dependent on the earthquake Earthquake Engineering Simulation, 2006, <http://opensees.
applied in the analysis. Different responses may be obtained berkely.edu>.
by applying different ground motions although scaled to [20] I.M. Idris, J.I. Sun, User’s manual for SHAKE91, a computer
program for equivalent linear seismic response analysis of
the same PGA.
horizontally layered soil deposits, in: Center for Geotechnical
Modeling Department of Civil and Environmental Engineering,
University of California, Davis, CA, 1992.
References [21] G.G. Meyerhof, Bearing capacity and settlement of pile
foundations, Am. Soc. Civil Eng., ASCE J. Geotech. Eng.
[1] D.S. Liyanapathirana, H.G. Poulos, Seismic lateral response of Division 102 (3) (1976) 195–228.
piles in liquefying soil, Am. Soc. Civil Eng. ASCE J. Geotechn.
Eng. 131 (12) (2005) 1466–1479.

You might also like