Professional Documents
Culture Documents
Volume 169 Issue GE2 Geotechnical Engineering 169 April 2016 Issue GE2
Pages 129–138 http://dx.doi.org/10.1680/jgeen.15.00025
DSSI analysis of pile foundations for an Paper 1500025
Received 05/02/2015 Accepted 08/10/2015
oil tank in Iraq
Published online 05/02/2016
Kumar and Choudhury Keywords: dynamics/geotechnical engineering/piles & piling
Seismic analysis and design of pile foundations in weak soil strata to support oil tanks are of great importance to
geotechnical practitioners. This paper presents an actual field study and dynamic soil–structure interaction (DSSI)
analysis for an oil tank foundation design in weak soil strata for the Kafza site in Iraq, which is located in a seismically
active region. For the particular in situ soil profile, a pile group foundation is proposed comprising a 24·1 m dia. by
1·5 m thick pile cap connecting together 89 piles, each pile being 800 mm in diameter and 26 m long. The foundation
system was modelled using the finite-difference-based geotechnical program Flac3D to carry out detailed DSSI
analysis. Field pile load test data for a static analysis were used for validation of the Flac3D model. The outcome of a
probabilistic seismic hazard assessment was used to estimate target input motions resulting in maximum bending
moments of 114 kN.m and maximum settlement of 21·9 mm for piles under dynamic conditions. The results obtained,
comprising total and differential settlements and bending moments, were used for the final design of the proposed
pile foundation and can be used for similar practical applications.
Notation 1. Introduction
Ap , A s base and surface areas of the pile Oil-storage tanks are important components of lifeline and
c cohesion at pile tip industrial facilities and are generally used at power plants, refi-
ci average cohesion of ith layer neries, airports and other places of national importance. In
E modulus of elasticity addition, a high percentage of these storage tanks are located
fck characteristic strength of concrete in seismically prone areas and hence are vulnerable to failure
G shear modulus in an earthquake event. Damage associated with such failures
Ki coefficient of earth pressure at ith layer is catastrophic, causing huge losses of both lives and property.
Kn , K s normal and shear stiffness at pile interface The seismic design of oil storage tanks is thus of paramount
L length of pile importance. During an earthquake event, failure modes such
Ms secant modulus as elephant’s foot buckling and excessive foundation settlement
N60 standard penetration test (SPT)-N value corrected are encountered in the piled foundations near the base of a
for 60% energy tank. These failures can be attributed to an increase in com-
Nc, Nq bearing capacity factors pressive load on the tank walls, excessive sloshing of the tank
PD effective overburden pressure at pile tip fluid and liquefaction of the underlying subsoil. Examples
PDi effective overburden pressure for ith layer include the 1964 Niigata, 1964 Alaska, 1971 San Fernando
Qu ultimate capacity of pile and 1999 Kocaeli earthquakes in which huge losses were
S factor defined in Equation 3 experienced when storage tanks containing combustible
αi adhesion factor for layer i materials failed, leading to extensive uncontrolled fires. With
γ shear strain amplitude the enormous extent of damage, earthquake engineers were
γp′ effective unit weight of pile compelled to incorporate the influence of seismic forces in the
ΔL spatial element size analysis and design of oil storage tanks located in seismically
Δεvd volumetric decrease vulnerable areas.
δ soil–pile friction angle
λ wavelength Local soil conditions, topography and characteristics of the
μ Poisson’s ratio earthquake input motion often dictate the selection of foun-
ϕ friction angle dation systems in seismically active areas. When the shallow
129
Downloaded by [ University of Liverpool] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering DSSI analysis of pile foundations for an
Volume 169 Issue GE2 oil tank in Iraq
Kumar and Choudhury
soil strata at a particular site have low bearing capacity and are penetration test (SPT)-N60 value of 48 overlying very soft to
located in a seismically vulnerable region, which is the case for soft clay with intermediate to high plasticity (plasticity index
the Khor Al Zubair port (Kafza) site in Iraq, a pile foundation of 15–30%) and an SPT-N60 value varying from 2–7, up to
is the preferred choice of foundation for oil tanks. The behav- 20 m depth. This is followed by a 4 m thick very stiff thinly
iour of pile foundations under the impact of seismic forces can bedded clay layer with an SPT-N60 value of 49 and then, at
be characterised as a complex soil–structure interaction, which greater depth, a dense to very dense sand layer.
has additional significant impact if the piles pass through soil
that can experience liquefaction or cyclic mobility. In such 4. Loading details of the oil tank
cases, liquefaction-induced lateral ground spreading is an foundation
important factor in failures of pile foundations in liquefaction- For the 15 m high and 23·15 m dia. oil tank, a pile foundation
prone areas (Abdoun and Dobry, 2002; Abdoun et al., 2003; was considered suitable for the ground conditions. The loading
Ishihara, 1997; Phanikanth et al., 2013; Tokimatsu et al., details are shown in Table 1.
1998). However, Bhattacharya (2003), Dash et al. (2010) and
Chatterjee et al. (2015) identified buckling instability as 5. Static design of the pile foundation
another possible failure mechanism, especially when piles are The static design of the pile foundation was carried out as per
subjected to vertical load combined with a lateral load on the IS 2911 (Part 1/Sec 4) (BIS, 2010). Shaft resistance for the fill
pile head. Various researchers have studied and analysed the layer was ignored and negative skin friction (down-drag) was
behaviour of pile foundations in liquefiable soil using exper- considered for the soft clay layer. Through an iterative pro-
imental, analytical and numerical procedures. Bhattacharya cedure, a pile of length 26 m and diameter 800 mm was
et al. (2004) proposed an alternative mechanism based on pile obtained. The factored ultimate vertical load carrying capacity
buckling for predicting the failure of piles in liquefiable soil of each pile was found to be 2052 kN considering shaft resist-
deposits during an earthquake. It is clear that analysis of such ance for the remaining two layers and end bearing, as per the
foundations for oil tanks to ensure safety in seismically active expression for ultimate bearing capacity
regions is a major concern for geotechnical practitioners.
X
Qu ¼ ½ðαi ci þ Ki PDi tan δÞAs þ ðcNc þ PD Nq ÞAp
In the work reported here, a dynamic soil–structure interaction 1:
Ap Lγ′p
(DSSI) analysis was carried out in order to understand the be-
haviour of an oil tank foundation under static and seismic
loading conditions using the finite-difference based computer in which Qu is the ultimate bearing capacity of the pile, αi is
program Flac3D (ICG, 2009). The analyses were based on the the adhesion factor for the ith layer, δ is the soil–pile friction
results of a probabilistic seismic hazard assessment (PSHA) angle, ci is the average cohesion of the ith layer, c is the cohe-
and geotechnical investigations of the Kafza site in Iraq. sion at the pile tip, Ki is the coefficient of earth pressure at the
ith layer, PDi is the effective overburden pressure for the ith
2. Site location layer, PD is the effective overburden pressure at the pile tip, γp′
The site is located approximately 40 km south of Basrah in the is the effective unit weight of pile, L is the length of the pile,
free trade zone at Kafza in Iraq (latitude 47°92′94″N and Ap and As are the base and surface areas of the pile, respect-
longitude 30°03′42″E). Kafza is Iraq’s third largest port in ively, and Nc and Nq are bearing capacity factors.
terms of size and goods shipped to the port of Basrah. SKA
Energy planned to build a fuel oil storage tank capable of A group interaction factor of 0·65 was selected on the basis of
storing and distributing 20 000 t of fuel. The oil tank foun- the degree of soil homogeneity, the slenderness ratio and pile
dation has to be constructed in the terminal plots 2–3–4 next spacing as per Randolph and Wroth (1979). Thus, a factored
to existing plot 1 near Um Qasr port on a tributary of ultimate load of 1333 kN was obtained. Considering the
the Shatt-el-Arab River in South Iraq, as shown in Figure 1. importance of the structure, moments induced due to wind
The site is located on the Arabian tectonic plate, which is loading and the seismic vulnerability of the Kafza region, the
subducted below the Eurasian plate, resulting in the ultimate load was further reduced by 50% (a safety factor of
formation of the Zagros mountain range in Iran (Agard et al., 2·0). Thus, a net allowable load of 665 kN was assumed for the
2011). The thickness of the earth’s crust is about 45 km in the pile foundation, and this was the deciding factor in determining
region. the numbers of piles required to support the tank.
130
Downloaded by [ University of Liverpool] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering DSSI analysis of pile foundations for an
Volume 169 Issue GE2 oil tank in Iraq
Kumar and Choudhury
47°52’58·8”E 47°53’2·4”E
30°10’55·2”N
30°10’55·2”N
N
4
Plot 2
1m
29 68
30°10’51·6”N
30°10’51·6”N
Warehouse
Mets
Plot 2 Plot 3
12 000 m2 16 000 m2
Plot 1
2
8m
20 05
47°52’58·8”E 47°53’2·4”E
on
ati
oc
el
Sit
131
Downloaded by [ University of Liverpool] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering DSSI analysis of pile foundations for an
Volume 169 Issue GE2 oil tank in Iraq
Kumar and Choudhury
60 m
12
0m
Z m
Y 120
X
Layer Depth: m Soil type Soil unit weight: Cohesion: Friction angle, ϕ: Modulus Poisson’s Permeability:
γ, kN/m3 c, kPa degrees of elasticity, ratio, μ m/s
E: kPa
interface cohesion and friction angle. The cohesion and friction at the interfaces were calculated as per Timoshenko and
angle at the interface was taken as per IS 2911 (Part 1/Sec 4) Goodier (2002). To include end bearing effects between the
(BIS, 2010). The normal stiffness (Kn) and shear stiffness (Ks) pile tip and the soil grid below the pile, a normal yield spring
132
Downloaded by [ University of Liverpool] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering DSSI analysis of pile foundations for an
Volume 169 Issue GE2 oil tank in Iraq
Kumar and Choudhury
Material Concrete grade Young’s modulus: kPaa Poisson’s ratio, ν Diameter: m Length or
thickness: m
a
= 5000( fck)1/2, as per IS 456 (BIS, 2000)
in the axial direction was defined and attachment conditions Vertical load: kN
were set. The compressive strength of the spring was equal to 0 500 1000 1500 2000 2500 3000 3500
the end bearing resistance of the pile, calculated as per IS 2911 0
(Part 1/Sec 4) (BIS, 2010). The pile cap was modelled with Field results
2
shell elements, which are triangles of uniform thickness lying Flac3D results
between three nodal points. Each pile and shell element 4
Vertical settlement: mm
133
Downloaded by [ University of Liverpool] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering DSSI analysis of pile foundations for an
Volume 169 Issue GE2 oil tank in Iraq
Kumar and Choudhury
Lateral load: kN The loads given in Table 1 were considered and the piles were
0 50 100 150 200 ‘wished’ in place.
0
134
Downloaded by [ University of Liverpool] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering DSSI analysis of pile foundations for an
Volume 169 Issue GE2 oil tank in Iraq
Kumar and Choudhury
Contour of Z-displacement: m
–3·3626 × 10–4
0·0000
2·0000 × 10–3
4·0000 × 10–3
6·0000 × 10–3
8·0000 × 10–3
1·0000 × 10–2
1·2000 × 10–2
1·4000 × 10–2
1·6000 × 10–2
1·8000 × 10–2
2·0000 × 10–2
2·2000 × 10–2
2·3388 × 10–2
Z
Y
X
slope at both low and high strain. Thus, the secant shear 0
modulus Ms is given by 0·0001 0·001 0·01 0·1 1
Cyclic shear strain: %
2: Ms ¼ S 2 ð3 2SÞ 0S1
Figure 7. Modulus reduction curves chosen for soil models at
different depths in Flac3D
and
L2 L
3: S¼
L2 L1 9.3 Soil constitutive model for dynamic modelling
Flac3D contains a built-in constitutive model, called the Finn
The parameters L1 and L2 are extreme values of logarithmic model (Byrne, 1991), which incorporates the formula of Byrne
shear strain (log10 γ) – that is, values at which the tangent of (1991) into the Mohr–Coulomb plasticity model formula for
the slope becomes zero. Thus, setting L1 = − 3 and L2 = 1 volumetric straining under the application of cyclic loading in
means that the S-shaped curve will extend from a lower cyclic saturated conditions in which all soil layers are considered to
shear strain of 0·001% (10−3) to an upper cyclic strain of 10% be undrained. This model captures the basic mechanism that
(101). Based on these formulations, numerical fits were chosen can lead to liquefaction and cyclic mobility in sand and clay
for the input parameters in Flac3D and fitted curves for like materials, respectively. The empirical relation shown in
dynamic soil properties were obtained, as shown in Figure 7. Equation 4 (Byrne, 1991) associates the increment of volume
135
Downloaded by [ University of Liverpool] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering DSSI analysis of pile foundations for an
Volume 169 Issue GE2 oil tank in Iraq
Kumar and Choudhury
decrease Δεvd with the cyclic shear-strain amplitude γ 9.6 Free-field analysis of soil model
The soil model was analysed under the chosen acceleration–
Δεvd εvd time history, which was applied at its base in order to capture
4: ¼ C1 exp C2
γ γ the response of soil under cyclic mobility in the absence of the
foundation. The pore water pressure ratios at different depths
in the clay layer (i.e. 5 m, 10 m and 15 m below the normal
in which
ground level) were obtained. It was found that the soil at 5 m
5: C1 ¼ 05C1c depth had a pore water pressure ratio of 0·85 at the end of the
54 s long input earthquake motion. This value being fairly
close to 1 (pore pressure ratio = 1) means full cyclic mobility
was achieved and this suggests that the soil may experience
moderate cyclic mobility at shallow depth. At depths of 10 m
6: C1c ¼ 87ðN60 Þ125
and 15 m from the normal ground level, the pore pressure
ratios were 0·65 and 0·55, respectively, decreasing with an
increase in depth. This result suggests that the soil is safe
7: C2 ¼ C2c against cyclic mobility for the chosen acceleration–time history
at depths greater than 5 m.
λ
9: ΔL
10
Pile X-component force: N
Considering this criterion, the element size was considered –3·6267 × 105
–2·0000 × 105
small enough to allow seismic wave propagation in the present 0·0000
2·0000 × 105
study. 4·0000 × 105
6·0000 × 105
8·0000 × 105
1·0000 × 106
1·2000 × 106
9.5 Boundary conditions for dynamic analysis 1·4000 × 106
1·6000 × 106
Many geotechnical problems can be idealised by assuming that 1·6895 × 106
136
Downloaded by [ University of Liverpool] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering DSSI analysis of pile foundations for an
Volume 169 Issue GE2 oil tank in Iraq
Kumar and Choudhury
at the periphery and loads reduce gradually from the edge to ASTM (1995b) D 3966-90: Standard test method for pile under
the centre, with a minimum value of 619 kN at the centre. The lateral load. ASTM International, West Conshohocken,
maximum bending moment observed at the head of the pile PA, USA.
was 114 kN.m. Bhattacharya S (2003) Pile Instability During Earthquake
Liquefaction. PhD thesis, University of Cambridge,
10. Conclusions Cambridge, UK.
Three-dimensional non-linear finite-difference analyses that Bhattacharya S, Madabhushi SPG and Bolton MD (2004)
incorporate the dynamic soil–structure interaction of a pile An alternative mechanism of pile failure in liquefiable
group for an oil tank foundation were carried out using soil deposit during earthquake. Géotechnique 54(3):
Flac3D. The accuracy of the developed numerical model was 203–313.
ensured by comparing the model results with field pile load BIS (Bureau of Indian Standards) (2000) IS 456: Plain and
test results – these validated the numerical model results both reinforced concrete – code of practice, 4th Revision.
qualitatively and quantitatively. For the free-field analysis, the Bureau of Indian Standards, New Delhi, India.
variation of pore water pressure ratio with depth below the BIS (2010) IS 2911: Design and construction of pile
surface under the chosen input motion indicated the occur- foundations – code of practice, 2nd revision. Bureau of
rence of partial cyclic mobility at shallow depth. Piles near the Indian Standards, New Delhi, India.
periphery were subjected to maximum axial loads of 697 kN Bowles JE (1997) Foundation Analysis and Design, 5th edn.
and 1690 kN under static and seismic conditions, respectively. McGraw-Hill, New York, NY, USA.
It is to be noted that the factored ultimate load carrying Byrne P (1991) A cyclic shear-volume coupling and pore-
capacity of a single pile was 2052 kN. The pile group was sub- pressure model for sand. Proceedings of the 2nd
jected to a maximum rotational displacement of 5·81 10−4, International Conference on Recent Advances in
which is lower than the permissible limit for oil storage tanks. Geotechnical Earthquake Engineering and Soil Dynamics,
The axial loads and rotational displacements under static and St. Louis, MO, USA, pp. 47–55.
seismic conditions were below the permissible limits, indicating Chatterjee K, Choudhury D and Poulos HG (2015) Seismic
satisfactory service performance. analysis of laterally loaded pile under influence of vertical
loading using finite element method. Computers and
This study highlights the importance of numerical modelling Geotechnics 67: 172–186.
techniques in understanding the behaviour of pile group foun- Dash SR, Bhattacharya S and Blakeborough A (2010) Bending–
dations involving complex soil–structure interactions under buckling interaction as a failure mechanism of piles in
static as well as dynamic loading conditions. The developed liquefiable soils. Soil Dynamics and Earthquake
model can be used for designing foundations for oil-storage Engineering 30(1–2): 32–39.
tanks under similar conditions. Desai SS and Choudhury D (2015) Site-specific seismic ground
response study for nuclear power plants and ports in
Mumbai. Natural Hazards Review ASCE 16(4): 04015002.
Acknowledgements
EPRI (Electric Power Research Institute) (1993). Guidelines for
The authors acknowledge Dr Jaykumar Shukla and
site specific ground motions. Electric Power Research
Prof. D. L. Shah of University of Baroda, India for their con-
Institute, Palo Alto, CA, USA, TR-102293.
tributions to the static design of the pile group for this industry
ICG (Itasca Consulting Group (2009) FLAC3D: Fast Lagrangian
project, funded by Chemie Tech. India Ltd.
Analysis of Continua Version 4.0. Itasca Consulting Group,
Minneapolis, MN, USA.
Ishihara K (1997) Geotechnical aspects of the 1995 Kobe
REFERENCES earthquake: Terzaghi oration. Proceedings of the 14th
Abdoun T and Dobry R (2002) Evaluation of pile foundation International Conference on Soil Mechanics and Foundation
response to lateral spreading. Soil Dynamics and Engineering, Hamburg, Germany, pp. 2047–2073.
Earthquake Engineering 22(9–12): 1051–1058. Kuhlemeyer RL and Lysmer J (1973) Finite element
Abdoun T, Dobry R, O’Rouke TD and Goh SH (2003) Pile method accuracy for wave propagation problems. Journal
response to lateral spreads: centrifuge modeling. Journal of of Soil Mechanics & Foundations ASCE 99(SM5):
Geotechnical and Geoenvironmental Engineering ASCE 421–427.
129(10): 869–878. Ordonez GA (2012) RspMatchETD – A Pre-Processor and
Agard P, Omrani J, Jolivet L et al. (2011) Zagros orogeny: a Post-Processor for RspMatch 2005 and RspMatch 2009.
subduction-dominated process. Geological Magazine Geomotions, LLC, Lacey, WA, USA.
148(5–6): 692–725. PEER (Pacific Earthquake Engineering Research Center) (2010)
ASTM (1995a) D 1143-81: Standard test method for pile under PEER Ground Motion Database. Pacific Earthquake
axial compression load. ASTM International, West Engineering Research Center, University of California,
Conshohocken, PA, USA. Berkeley, CA, USA.
137
Downloaded by [ University of Liverpool] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.
Geotechnical Engineering DSSI analysis of pile foundations for an
Volume 169 Issue GE2 oil tank in Iraq
Kumar and Choudhury
Phanikanth VS, Choudhury D and Reddy GR (2013) Behavior of Research Center, University of California, Berkeley, CA,
single pile in liquefied deposits during earthquakes. USA, Report No. UCB/EERC-88/15.
International Journal of Geomechanics ASCE 13(4): Timoshenko SP and Goodier JN (2002) Theory of
454–462. Elasticity, 3rd edn. Tata McGraw-Hill Education,
Randolph MF and Wroth CP (1979) An analysis of the Maidenhead, UK.
vertical deformation of pile groups. Géotechnique 29(4): Tokimatsu K, Oh-oka H, Satake K, Shamoto Y and Asaka Y
423–439. (1998) Effects of lateral ground movements on failure
Seed HB and Idriss IM (1970) Soil moduli and damping factors patterns of piles in the 1995 Hyogoken-Nambu
for dynamic response analysis. Earthquake Engineering earthquake. In Geotechnical Earthquake Engineering and
Research Center, University of California Berkeley, Soil Dynamics III (Dakoulas P, Yegian M and Holtz RD
Report No. EERC 70-10. (eds)). American Society of Civil Engineers, Reston, VA,
Sun JI, Golesorkhi R and Seed HB (1988). Dynamic moduli and USA, Geotechnical Special Publication No. 75,
damping ratios for cohesive soils. Earthquake Engineering pp. 1175–1186.
138
Downloaded by [ University of Liverpool] on [17/09/16]. Copyright © ICE Publishing, all rights reserved.