Sindh Univ. Res. Jour. (Sci. Ser.) Vol.
45 (2) 301-304 (2013)
SINDH UNIVERSITY RESEARCH JOURNAL (SCIENCE SERIES)
3D Numerical Modelling of Liquefaction-induced Settlements and its Mitigation
Z. A. ALMANI++, A. A. MEMON, A. F. HABIB, K.. LAL*, S. F. SHAH**
Department of Civil Engineering, Mehran University of Engineering & Technology, Jamshoro, Pakistan
Received 26th March 2013 and Revised 2nd June 2013
Abstract: In this paper, the behaviour of a 3D soil-structure system of square footing pad constructed on the ground reinforced
with jet grouted columns was studied with 3D models using the commercially available FLAC 3D version 3.1 finite difference
Code. The isolated shallow square footing pad supporting a typical building was constructed on the ground susceptible to liquefaction and on the reinforced ground with stiff jet grouted column rows. The results showed that ground reinforced with one
row of continuously overlapping columns (forming a lattice) adjacent to the four sides of footing pad can be effectively applied
to reduce the liquefaction-induced large settlements to the tolerable limits.
Keywords: Liquefaction; Jet grouted columns; FLAC 3D, Model, Numerical Modelling
1.
INTRODUCTION
The buildings in earthquake-prone areas of
Pakistan are susceptible to damage due to liquefactionrelated large settlements. These large settlements are
due to bearing capacity failure (shear failure) as the
stiffness of soil is lost because of liquefaction. The
ground reinforcement with stiff high modulus jet
grouted/ deep mixing column rows which has performed well in earthquakes such as Kocaeli, Turkey,
1999 and Kobe, Japan, 1995 (Martin et al.,2004;
Mitrani et al., 2008) is cost effective and easy to install in the ground.
In the previous studies (Almani et al., 2012a) ,
the behaviour of shallow footing founded on the
ground reinforced with jet grouted columns was studied in 2D plane strain, which represents the long strip
footing pads in the direction perpendicular to the plane
of analysis. With this assumption, deformations
(strains) of the soil-structure system occur in the plane
of analysis, while the deformations (strains) in the
direction perpendicular to the plane of analysis were
assumed to be negligible. Therefore, the horizontal
boundaries are assumed to be fixed in the direction
perpendicular to plane of analysis. Furthermore, with
this assumption, the footing was considered to be a
strip of infinite length and the rows of columns were
supposed to be long and continuous (overlapping to
form walls) in the direction perpendicular to the plane
of analysis.
In this study, the behaviour of a 3D soilstructure system consisting of a shallow square footing
pad supported on the ground reinforced with jet
grouted columns was studied by constructing 3D mod
++
els using the commercially available FLAC 3D version
3.1 finite difference code (Itasca, 2009). In this way,
the effects due to the assumptions taken in the FLAC
2D plane strain model on the response could be eliminated which are more significant in the 3D parameters
such as geometry of treatment, structural foundations
and earthquakes.
2.
MATERIALS AND METHODS
The materials and methods are presented in
the following sections.
Model development
A three dimensional rectangular mesh was developed for this analysis as shown in (Fig.1). The mesh
consisted of eight-nodded brick elements, with every
node in the mesh connected to four other nodes, with
the exception of nodes at the boundaries of the mesh.
Fig. 1: Finite difference mesh for FLAC 3D analysis
Corresponding author: Email: [email protected], Tel. 2772253-7108
* Department of Civil Engineering, Mehran University of Engineering & Technology, Khairpur Campus, Pakistan
** Department of Basic Sciences and Related Studies, Mehran University of Engineering & Technology, Jamshoro
Z. A. ALMANI et al.,
Mesh density
In order to get optimum mesh density (making the mesh as fine as possible keeping the aspect ratio near to 1 and simulation time), three meshes of zone
size 0.25 m x 0.25 m x 0.25 m, 0.5 m x 0.5 m x 0.5 m
and 0.5 m x 0.5 m x 1 m size were initially tested. The
results showed that the response of the soil-structure
system was identical, particularly the value of the settlements and the pore pressures which are the variables
of primary interest for this study. Further, simulation
times of the finer mesh of 0.25 m x 0.25 m x 0.25 m
and 0.5 m x 0.5 m x 0.5 m were too long with this dynamic and coupled with a ground water flow problem
(due to the very small combined dynamic and flow
timestep). In view of the above, a mesh of 0.5 m x 0.5
m x 1 m zone size with an aspect ratio of 1 in the horizontal direction and 2 in the vertical direction, for
which a simulation time was about 72 hours, was taken
for this study.
302
tion is based upon Mohr-Coulomb failure criteria in
conjunction with hysteric model as described in
Almani et al., 2012a. Soil properties are given in
(Table-1)
Table 1: Properties of soil layers
Soil layers
Medium
dense
layer
40 %
Dense Layer
80 %
Unit weight (KN/m3)
18.80
19.57
Porosity (void ratio)
0.47
(0.88)
0.42
(0.72)
2 x 10-7
1 x 10-7
Property
Relative density
Permeability (m/sec)
Peak friction angle
(Degree)
Pore pressure constants
Hardin-Drnevich damping
constant (ref)
Water Bulk modulus (kPa)
Water tension (kN/m)
Mechanical and hydraulic boundaries
The boundaries of the model were located in a
way that the responses of the structure soil systems in
the area of interest are not affected. For this purpose,
initially, the boundaries of the grid were located at a
distance of 2.5 times (50 m) of the layer thickness on
each side of the centre of the structure. To further optimise the mesh size, boundaries were located at a distance of 20 and 15 m from each side of the centre of
the structure. The results showed that the response was
not affected with closer boundaries of 15 m each side
of the centre of the structure. On the basis of this, a 30
m x 30 m x 20 m mesh was finally selected.
For static equilibrium of the model, fixed lateral and bottom boundaries were applied. For dynamic
simulation, free field boundaries were applied along
four lateral boundaries of the model. For hydraulic
boundaries, the pore pressures along the top boundary
of the model were set at zero for a free drainage surface.
Numerical modelling code selection and coupling of
modules
The computer code FLAC 3D Version 6.0 (for
Fast Lagrangian Analysis of Continua) was chosen for
numerical analysis. Liquefaction problem could be
modelled with this code by coupling the dynamic module with ground water flow module. For more details,
the FLAC Users Manuals Itasca Consulting Group,
2009.
Basic soil properties, soil liquefaction model and
damping
The Finn/Byrne liquefaction soil model in
FLAC 3D for modelling the phenomenon of liquefac-
32
48
C1=1.2;
C2=0.33
C1=0.43; C2=3.75
0.05
0.05
5 x 105
5 x 105
1 x 102
1 x 10
Water density (kg/m )
1000
1000
The Finn/Byrne soil model as described by
relationship-1 was calibrated by simulating the constant volume cyclic simple shear tests single element
test as described in detail in Almani et al., 2012a. The
model parameters C1 and C2 and the Hardin-Drnevich
hysteretic damping strain constant (ref) determined by
calibration are shown in (Table 2)
v 1
2
1
In the above relationship, v is volumetric
strain increment in each cycle, v is the accumulated
volumetric strain from previous cycle and is the shear
strain for the cycle, C1 and C2 are soil model constants.
Table 2 : Model Parameters
Model Parameters
Medium dense sand
layer
Dense sand
layer
C1
1.2
0.43
C2
0.33
3.75
ref
0.05
0.05
Ground reinforcement
The ground was reinforced with stiff jet
grouted/deep mixing circular column rows of 0.6 m diameter (or 0.5 m x 0.5 m square columns with crosssectional area equal to circular columns. The length of
columns was set as 11 m. The column jet grouted material (cemented sand) was represented with the MohrCoulomb soil model combined with HardinDrnevich
hysteretic dynamic damping model as described in
Almani et al., 2012b.
3D Numerical Modelling of Liquefaction-induced...
303
2.
RESULTS AND DISCUSSIONS
The square footing pad of 4 m x 4 m x 1 m
size founded on liquefiable ground at the depth of 1 m
from the ground surface as shown in (Fig. 2), was
modelled as 3D dynamic coupled with ground water
flow. This is a base (benchmark) case for comparison.
Fig. 4: Contours of vertical displacement-Benchmark case
Fig. 2 : The settlement of the footing-Benchmark case
The history and contours of the settlement at the centre
of the footing pad, as presented in Fig. 3 and Fig. ,
shows that the settlement of the footing pad is as large
as 0.7 m (vertical displacement of 70 cm) at the end of
10 seconds of cyclic loading. This behaviour of the
soil-structure system was identical to the predicted with
the FLAC 2D plane-strain model (Almani et al.
2012a), though the settlements observed in the 3D case
are smaller in magnitude than its counterpart in 2D due
to the assumption of strip long footing pad in the case
of the 2D plane-strain model (Almani et al., 2012a).
The history of pore pressure at the 4 m depth
under the centre of the footing pad, as presented in
Fig., shows that the pore pressure increases from the
initial hydrostatic value of 43x103 Pa to the peak value
of 138x103 Pa in the initial cycles of cyclic loading,
which soon decreases to 60x103 Pa due to the effect of
dilation (expansion) caused by monotonic shearing under the influence of the structure. This pattern of pore
pressure variation is the same as predicted with the 2D
plane-strain model (Almani et al, 2012a).
Fig. 5. Pore pressure vs. time under the structure-Benchmark
In the other study, isolated shallow square
spread footing 4 m x 4 m x 1m size was founded on the
ground reinforced with continuously overlapping column rows forming one lattice beneath and two lattices
at 0.5 m spacing around the footing pad in both directions, as shown in (Fig. 6).
Fig. 3: The settlement of footing vs. time-Benchmark case
The histories of the settlement at the centre of
the footing pad, as presented in (Fig. 7) shows that the
settlement of the footing is within the tolerable limits
of 0.04 m (4 cm) when the dynamic excitation is
applied for 10 second duration as against the 0.7 m
(70 cm) in the benchmark case. However, there is
slight heave or rebound of footing due to transmission
of motion through the stiff the columns to the base of
footing from the base of the model in this reinforced
case.
Z. A. ALMANI et al.,
304
5.
CONCLUSION
The following conclusion can be drawn from
the above studies:
Fig.2: Ground reinforced with rows of jet grouted columns
Fig. 3: Settlement of footing vs. time-Reinforced case
The history of pore pressure at 4 m depth under the centre of the footing pad, as shown in (Fig. 8),
indicates that at the centre of the pad in the untreated
soil zone, the pore pressure increases from the hydrostatic value of 45x103 Pa to the peak value of 100x103
Pa during initial the cycles of the dynamic event (as
against its peak value of 138x103 Pa in the unreinforced benchmark case). This pattern of pore pressure
variation under the structure is the same as observed in
2D in the initial cycles of loading, but in later cycles
pore pressure in the 3D model is relatively higher in
magnitude than the counterpart 2D reinforced model
case (Almani et al., 2012b).
3D shallow square footing settles by 70 cm in
the punching type bearing capacity (shear) failures
when the foundation soil liquefies when pore pressures
under the footing quickly rise in the few cycles.
By reinforcing the ground with jet grouted
columns under the structure, pore pressures somewhat
decrease, but are still high enough to cause significant
liquefaction under the footing.
Ground reinforced with one row of continuously overlapping columns (forming a lattice) adjacent
to the four sides of footing pad and two rows of continuously overlapping columns in both direction (one
lattice) beneath the footing pad, or rows of continuously overlapping columns forming one lattice beneath
and one lattice adjacent to the footing pad, can be effectively applied to meet the tolerable settlement limits.
ACKNOWLEDGEMENTS
The research presented in this paper was
carried out as part of PhD studies at University of
Nottingham. The authors wish to acknowledge the
support received from University of Nottingham, UK
and Mehran University of Engineering & Technology
Jamshoro, Pakistan.
REFERENCES:
Almani, Z.A., K. Ansari, and A.A. Memon (2012a)
Mechanism of liquefaction-induced large settlements
of buildings. Mehran University Research Journal of
Engineering & Technology, Jamshoro, Sindh, vol..
(31): 4, 635-650.
Almani, Z. A., N.A. Memon, A.A. Memon (2012b)
Stiff Columns as Liquefaction Mitigation Measure for
Retrofit of Existing Buildings. Mehran University Research Journal of Engineering & Technology, vol. (31):
4, 659- 668.
Itasca Consulting Group, Inc., Itasca FLAC3D V3.1
(2009) Fast Lagrangian Analysis of Continua, user
Manuals, Minneapolis, USA.
Fig. 4: Pore pressure vs. time under footing-Reinforced case
These results confirm the 2D results that when
ground is reinforced at certain spacing with rows of jet
grouted columns, the peak pore pressure in the initial
cycles decreases by 30% as compared to base benchmark. This decrease of pore pressure is not significant
that liquefaction does not occur and settlements could
be prevented.
Martin, J. R., C.G. Olgun, J. K. Mitchell, and
H.T.Durgunoglu (2004) High modulus columns for
liquefaction mitigation. Journal of Geotechnical and
Geoenvironmental Engineering, ASCE, vol. (130): 6,
561-571.
Mitrani, H., and S.P.G. Madabhushi (2008) Centrifuge
modelling of inclined micro-piles for liquefaction
remediation for existing buildings. Geomechanics and
Geo engineering-An International Journal, vol. (3): 4,
245-256.
