Sei sulla pagina 1di 12

Numerical Analysis of Groundwater

Flow Through a Rectangular Cofferdam


Serdar Koltuk
University of RWTH Aachen, Department of Engineering Geology and
Hydrogeology, D-52064 Aachen, Germany
e-mail: serdarkoltuk@yahoo.de

Recep Iyisan
Istanbul Technical University, Civil Engineering Faculty
Ayazaga Campus 34646, Istanbul, Turkey
e-mail: iyisan@itu.edu.tr

ABSTRACT
In this study, finite element analysis of groundwater flow in a rectangular shaped cofferdam
model has been performed by taking into account of anisotropic permeability of soil layer.
The results of numerical analyses indicate that: (1) An increase in the horizontal permeability
to vertical permeability ratio k ⁄k (kh/kv) from 1 to 2 leads to an increase in the pore water
pressure on the sheet pile and particularly in the flow rate into the cofferdam. (2) Maximum
pore water pressure arises in the corner regions of the cofferdam. This can induce in seepage
failure analyses that safety factors obtained from three-dimensional models are overestimated.
(3) The values of pore water pressure and flow rate obtained from three-dimensional models
are higher than those in two-dimensional model. The difference between the values obtained
from 2D and 3D analyses decreases with increasing L/B ratio. The difference can be very
important in the seepage failure analyses and to estimate the required number and capacity of
pumps in square shaped cofferdams.
KEYWORDS: Cofferdam, groundwater flow, finite element analysis, anisotropic
permeability, 3D effect.

INTRODUCTION
A cofferdam is a temporary structure designed to keep water and soil out of the excavation
within water in order to build a permanent structure. When one is designing cofferdam, the main
problem is often dominated by the groundwater flow beneath the cofferdam. The flow rate must
be known to determine the required number and capacity of pumps, and the knowledge of the
pore water pressure distribution is necessary in order to estimate safety with respect to sheet pile
stability and seepage failure.
The term “seepage failure or hydraulic heave” is used to describe to unstable condition of the
excavation base which can most likely occur in non-cohesive soils when the upward seepage
force is greater than the gravity of soil at downstream side. As shown in the Figure 1a, the
amount of deformation at the upstream and downstream sides increases with increasing potential

- 2041 -
Vol. 18 [2013], Bund. K: Koltuk. S, Iyisan. R 2042

difference. Consequently, due to the loss of the soil at the base of sheet pile wall or due to the
reduction of passive pressure the collapse occurs. To estimate the safety against seepage failure in
non cohesive soil, different calculation methods have been developed. The most conventional
methods are presented by Terzaghi (Terzaghi, 1943; Terzaghi et. all., 1996). He found from
model test, that the heave zone, which is lifted by water, has the shape of a rectangular prism with
a width equal to half the wall penetration D/2 (see Fig. 1b).

Figure 1: Seepage Failure: a) Model experiment for seepage failure in cofferdam, b)


Terzaghi’s criteria against seepage failure

The factor of safety against seepage failure can be given by the following equation:

W
FS  (1)
U
where W is submerged weight of soil and U is average uplift force in the heave zone per unit
length of the sheet pile. Submerged weight of soil and average uplift force can be calculated by;

D
W   D. .(  sat   w ) (2)
2

D
U  D. .iav . w (3)
2
where sat and w are saturated unit weight of soil and unit weight of water, respectively and iav is
average hydraulic gradient in the heave zone due to groundwater flow as given in Equation 4.

h
iav  (4)
D
Vol. 18 [2013], Bund. K: Koltuk. S, Iyisan. R 2043

PORE WATER PRESSURE AND FLOW RATE DUE TO


GROUNDWATER FLOW
The concept of the groundwater flow is based on Laplace’s equation, which can be derived
from the flow continuity and Darcy’s law:

2h  2h 2h
kx .  ky.  kz . 0
x 2 y 2 z 2 (5)

where k x , k y and k z k , k the constants are coefficients of permeability of the soil, and
h h h
, , are hydraulic gradients in the x, y and z directions, respectively.
x y z
If the soil is homogeneous and isotropic soil permeability has no effect on the pore water
pressure distribution. However, many natural sedimentary soils are anisotropic. Three facts can
induce anisotropic permeability in sedimentary soils: Macro-stratification, micro-stratification
and flatness-orientation of grains. In cases of macro-stratification the thicknesses and
permeabilities of the soil layers are available. However, it is very difficult to quantify the
anisotropy of a soil layer which shows micro-stratification. In such soils, in situ pumping tests are
necessary to obtain information on the ratio of anisotropic permeability. From experimental and
theoretical investigation, it is found that, due to flatness and orientation of grains, the degree of
anisotropic permeability cannot be larger than kh/kv=2.5 k ⁄k 2.5even for very flat particles
and well pronounced orientation (Witt et al., 1983). In another study, a series of permeability tests
on high-quality undisturbed coarse grained soil samples was performed using a triaxial cell. The
undisturbed soil samples were recovered by the in-situ freezing sampling method. Based on the
test results, it is found that the coefficient of permeability in the horizontal direction is larger than
that in the vertical direction, and its difference is maximum 70% (Hatanaka et al., 1997, 2001).
Groundwater flow into the long excavation could be evaluated using a two-dimensional
Laplace’s equation:

2h 2h
 0 (6)
x 2 z 2
Equation 6 describes two sets of orthogonal curves in a vertical plane. The construction of the
flow lines and equipotential lines is called a flow net. Once a flow net has been drawn, quantities
such as flow rate and pore water pressure distribution may be evaluated. Intervals between
adjacent equipotential lines represent a constant difference in hydraulic head and intervals
between adjacent flow lines represent a constant flow rate. However, two-dimensional models
may not be safe solution to investigate the cofferdams with a small ratio of length to width (L/B),
such as square shaped cofferdams. Furthermore, the square shaped cofferdams having a large area
are required high capacity computers to be studied in three-dimensional. In this case, either the
model dimensions or the number of the node in the finite element model are reduced. However,
both options lead to faulty results.
Vol. 18 [2013], Bund. K: Koltuk. S, Iyisan. R 2044

NUMERICAL INVESTIGATION
In the present study, steady-state flow analyses are performed using the finite element
program ABAQUS 6.11 to investigate the effect of three dimensional groundwater flow on the
pore water pressure distribution and the flow rate taking account of anisotropic permeability of
soil layer. The cofferdam shape in the model is rectangular with a length L and a width B. The
width of the cofferdam in all cases is 4 m, and the length of the cofferdam is changed from B to
5B. The embedment length of the sheet pile D in all cases is 3 m. The water level on the upstream
side is 5 m higher than the ground surface, and the water level on downstream side is located on
the excavation base, as shown in the Fig. 2. The vertical permeability of the one-layered non-
cohesive soil medium is kvk 10-4 m/s and the horizontal permeability is varied as
khk -4 -4 -4
1.0x10 , 1.5x10 and 2.0x10 m/s.

Figure 2: Cofferdam in cross-section


The three-dimensional model considers only a quarter of the cofferdam and the two-
dimensional model considers the half of the cofferdam, taking advantage of symmetry as can be
seen in Figure 3. In the narrowest section of the three-dimensional model is used for two-
dimensional analyses, and only horizontal permeability is changed. The dimensions of the soil
layer are chosen such that the boundary effect on the numerical results is minimized. The one-
layered non-cohesive soil medium has a deep of 20 m and an area of 50 m x 50 m. 5 m potential
difference is replaced by a pore water pressure (POR) boundary condition of -50000 Pa on the
excavation base at the downstream side. Minus sign here represents the loss of the pore water
pressure due to groundwater flow. The pore water pressure on the ground surface at the upstream
side is set equal to zero. All vertical sides and the bottom side of the flow domain are
impermeable. The sheet piles are simulated using thin gaps whose surfaces are also impermeable.
The two-dimensional finite element mesh consisted of 1670 linear quadrilateral elements and
1770 nodes. The three-dimensional finite element mesh consisted linear hexahedral elements.
Element numbers in the case of L/B=1 and L/B=5 are 49514 and 76390, respectively. Node
numbers in the case of L/B=1 and L/B=5 are 53790 and 82350, respectively. It should be noted
that the mesh is finer near the sheet pile where flow gradients are concentrated.
Vol. 18 [2013], Bund. K: Koltuk. S, Iyisan. R 2045

Figure 3: Finite element model: a) three-dimensional mesh, b) two-dimensional mesh

RESULTS OF ANALYSES
The results of three-dimensional finite element analyses, whose comparison and discussion
with two-dimensional numerical analyses are given below. In the comparison, pore water
pressure on the sheet pile, safety factor against seepage failure according to Terzaghi, and flow
rate into the cofferdam are taken into account. Fig. 4 and 5a show the contours of the loss of pore
water pressure in two and three-dimensional models for anisotropic permeability ratio
kh/kv=1k ⁄k 1. The loss of the pore water pressure on the horizontal cross section at the level
of the sheet pile base at the downstream side is shown in Fig. 5b. In the evaluation of the pore
water pressure on the sheet pile, the values along two axes are used. One of these is the values in
the middle of length (L) and is named middle; the other one is the values in the corner of the
cofferdam and is named corner. The area shown with the red rectangle in Fig. 5b is the base of
Terzaghi’s heave zone. The pore water pressure in the middle of this area is used as uplift
pressure in the evaluation of seepage failure. The flow rates obtained from 3D numerical
analyses are for a quarter of the cofferdam so that the flow rates obtained from numerical
analyses must be multiplied by 4. The flow rates obtained from 2D numerical analyses are for the
half part and per unit length of the cofferdam perpendicular to the page plane. Therefore, the
flow rates in 2D numerical analyses are multiplied by 2 and changing L (L= 4, 8, 12, 16, 20 m).
Vol. 18 [2013], Bund. K: Koltuk. S, Iyisan. R 2046

Figure 4: Contours of the loss of water pressure in two-dimensional flow for kh/kv=1
k ⁄k 1

Figure 5: Contours of the loss of pore water pressure in three-dimensional flow for
kh/kv=1 and L/B=5: a) entire model, b) horizontal cross section at the level of sheet pile
base at the downstream side.
Vol. 18 [2013], Bund. K: Koltuk. S, Iyisan. R 2047

EFFECT OF ANISOTROPIC PERMEABILITY


Fig. 6 shows the effect of anisotropic permeability ratio (kh/kv) on the pore water pressures in
the middle of the cofferdam at the upstream and downstream sides, in the cases of the length to
width ratio (L/B) 1 and 5. The water pressure on the wall increases with increasing anisotropy
ratio at the upstream and downstream sides. When the ratio of horizontal permeability to vertical
permeability increases from 1 to 2, the pore water force per unit length of the sheet pile
perpendicular to the page plane at the upstream side increases by about 1%, in the case of L/B=1.
At the downstream side, this increase is 4.5%. In the case of L/B=5, the pore water force per unit
length of the cofferdam at the upstream side increases by about 2% and the increase at the
downstream side is 5.5%.

3.0
kh/kv=1 (L/B=1)
Distance  from wall  base (m)    b

2.5 kh/kv=1.5 (L/B=1)
kh/kv=2 (L/B=1)
2.0 kh/kv=1 (L/B=5)
kh/kv=1.5 (L/B=5)
kh/kv=2 (L/B=5)
1.5

1.0

0.5 (a)

0.0
50 55 60 65 70 75 80
Pore water pressure (kPa)
3.0
kh/kv=1 (L/B=1)
Distance from  wall  base (m)  b

2.5 kh/kv=1.5 (L/B=1)
kh/kv=2 (L/B=1)
2.0 kh/kv=1 (L/B=5)
kh/kv=1.5 (L/B=5)
1.5 kh/kv=2 (L/B=5)

1.0

0.5 (b)

0.0
0 10 20 30 40 50 60 70 80
Pore water pressure (kPa)

Figure 6: Effect of anisotropic permeability ratio on the pore water pressure on the sheet
pile for L/B=1 and L/B=5 a) at the upstream side, b) at the downstream side
Vol. 18 [2013], Bund. K: Koltuk. S, Iyisan. R 2048

The safety factors against the seepage failure depending on anisotropy ratio and L/B are
given in the following figure. The unit weight of the non-cohesive soil is selected as 20 kN/m3.
When the ratio of horizontal permeability to vertical permeability increases from 1 to 2, the safety
factor against seepage failure decreases by about 5%, regardless of L/B.

0.96

0.94
Safety Factor

0.92

0.9
kh/kv=1

0.88 kh/kv=1.5

kh/kv=2

0.86
1 2 3 4 5 6
Ratio of length to width of the cofferdam

Figure 7: Effect of anisotropic permeability ratio on the safety factor


against seepage failure

Figure 8 shows the change of the flow rate taking account of anisotropic permeability and
L/B. When the ratio of horizontal permeability to vertical permeability increases from 1 to 2, the
flow rate into the cofferdam increases by about 10%, in the case of L/B=1 and 12% in the case of
L/B=5.

Figure 8: Effect of anisotropic permeability ratio on the flow rate


Vol. 18 [2013], Bund. K: Koltuk. S, Iyisan. R 2049

COMPARISON AND DISCUSSION OF THE TWO AND


THREE DIMENSIONAL RESULTS
In this section, the results of the two- and three- dimensional FE analyses are compared for
isotropic soil condition (kh/kv=1). The results of these comparisons will be also valid for
anisotropic conditions.
The cases of L/B=1 and L/B=5 are used in order to compare the pore water pressures acting
on the sheet pile at the downstream and upstream sides. As it can be seen from Fig. 9 the pore
water pressure values obtained from three-dimensional analyses are higher than the values
obtained from two-dimensional analyses.

(a)

(b)

Figure 9: Comparison of the 2D and 3D numerical analyses with respect to pore water
pressure on the sheet pile: a) at the upstream side, b) at the downstream side
Vol. 18 [2013], Bund. K: Koltuk. S, Iyisan. R 2050

When the pore water force per unit length of the wall obtained from 3D analysis in the middle
of the cofferdam at the upstream side is compared with that obtained from 2D analysis it can be
seen that 3D analysis leads to an increase of 5.5% and 1% in the cases of L/B=1 and L/B=5,
respectively. The increase at the downstream side is 12% for L/B=1, and 1.5% for L/B=5. If this
comparison is made in the corner of the cofferdam at the upstream side it can be seen that 3D
analysis leads to an increase of 7% and 5.5%, in the cases of L/B=1 and L/B=5, respectively. The
increase in the corner of the cofferdam at the downstream side is 15% for L/B=1, and 12% for
L/B=5. It should be mentioned that the horizontal pressure acting on the wall consists of the pore
water pressure and the effective soil pressure. As an increase of pore water pressure will lead to a
decrease of effective soil pressure, the change ratio of total pressure to each other at upstream and
downstream sides is important for sheet pile stability.
The safety factors against the seepage failure obtained from 2D and 3D analyses depending
on varying of L/B ratio are given in Fig.10. The unit weight of non cohesive soil used in the
seepage failure analyses is selected as 20 kN/m3. As can be seen from the figure, two-
dimensional model yields a higher safety factor compared to three-dimensional model, so that it
remains on the unsafe side. In the case of L/B=1, 3D analysis causes a decrease of 12% on the
safety factor, and the effect of the increasing L/B ratio on the decrease of safety factor is
negligible. It should be noted that Terzaghi’s failure zone was obtained from two-dimensional
model experiments. Namely, the validity of Terzaghi’s heave zone such as rectangular prism for
3D analyses is uncertain. Since maximum pore water pressure arises in the corner regions of the
cofferdam, the values of safety factor obtained for the cofferdam model used in this study may be
greater than that in fact.

Figure 10: Comparison of the 2D and 3D numerical analyses with respect to safety factor
against seepage failure

The flow rate values obtained three-dimensional cases also are higher than those obtained
two-dimensional cases, as can be seen in Fig. 11. For instance, 3D analysis induces an increase of
29% and 8% in the flow rates into the cofferdam in the cases of L/B=1 and L/B=5, respectively.
Vol. 18 [2013], Bund. K: Koltuk. S, Iyisan. R 2051

Figure 11: Comparison of the 2D and 3D numerical analyses with respect to flow rate

CONCLUSIONS
Groundwater flow beneath a cofferdam has been usually evaluated using flow net. Once a
flow net has been drawn, quantities such as flow rate into the excavation and pore water pressure
distribution throughout the soil layer may be determined. However, this method does not give
safe results for the cofferdams having the small ratio of length to width (L/B).
The present research deals with the finite element analyses to investigate the effect of the
three-dimensional groundwater flow on the pore water pressure and the flow rate beneath
rectangular shaped cofferdam models taking account of anisotropic permeability. From various
series of numerical calculations, it is found that:
(1) The anisotropic permeability ratio of the soil layer (kh/kv) causes an increase of the pore water
pressure on the sheet pile, and this increase especially appears at the downstream side.
However, an increase of the pore water pressure induces a decrease of effective pressure.
Therefore, the change ratio of total pressure to each other at upstream and downstream sides
plays an important role in sheet pile stability analyses.
When the anisotropic permeability ratio increases from 1 to 2, the safety factor against
seepage failure decreases by about 5%, regardless of L/B ratio. Anisotropic permeability has
particularly an effect on the flow rate into the cofferdam. When the ratio of kh/kv
k ⁄k increases from 1 to 2, flow rates show an increase of 10% and 12%, in the cases of
L/B=1 and L/B=5, respectively.
(2) In three dimensional models, maximum pore water pressure arises in the corner regions of the
cofferdam. In 3D analyses, this can lead to lower safety factor compared with that obtained
according to Terzaghi since Terzaghi’s heave zone is based on 2D model tests.
(3) The values of pore water pressure and flow rate in three-dimensional cases are higher than
those in two-dimensional case. The difference between the values obtained from 2D and 3D
analyses decreases with increasing L/B ratio. When the pore water forces per unit length of
the wall obtained from 3D analysis is compared with that obtained from 2D analysis, 3D
analysis causes an increase of 7% and 5.5% on the pore water pressure in the corner of the
Vol. 18 [2013], Bund. K: Koltuk. S, Iyisan. R 2052

cofferdam at upstream side, in the cases of L/B=1 and L/B=5, respectively. The increase at the
downstream side is 15% for L/B=1, and 12% for L/B=5.
With respect to safety factor against seepage failure, two-dimensional model yields a higher
safety factor compared to three-dimensional model, so that it remains on the unsafe side. 3D
analysis induces a decrease of 12% in the safety factor in the case of L/B=1 and the effect of
increasing L/B ratio on the decrease of safety factor is negligible. With respect to flow rate into
the cofferdam, 3D analysis leads to an increase of 29% and 8%, in the cases of L/B=1 and L/B=5,
respectively.

REFERENCES
1. ABAQUS (2011) “Abaqus/standard user`s manual, Version 6.11,” Hibbitt, and Sorenson, Inc.
2. Das, B.M. (2002) “Principles of Geotechnical Engineering,” Brooks/Cole Thomson Learning.
3. Harr, M.E. (1962) “Ground Water and Seepage,” McGraw-Hill, New York.
4. Hatanaka, M., Uchida, A., Takehara, N. (1997) “Permeability Characteristics of High-Quality
Undisturbed Sands Measured in a Triaxial Cell,” Soils and Foundations, Vol. 37, No. 3, pp.
129-135.
5. Hatanaka, M., Uchida, A., Takehara, N. (2001) “Permeability Characteristics of High-Quality
Undisturbed Gravelly Soils Measured in Laboratory Tests,” Soils and Foundations, Vol. 41,
No. 3, pp. 45-55.
6. Kaiser, P. K., Hewitt, K. J. (1982) “The Effect Of Groundwater Flow on the Stability and
Design of Retained Excavations,” Can. Geotech. J., Vol. 19 (2), pp. 139-153.
7. King, G.W. (1990) “Design Charts for Long Cofferdams,” Geotechnique 40, No. 4, pp. 647-
650.
8. Puller, M. (2003) “Deep Excavations: A practical Manuel,” Thomas Telford, London.
9. Soubra, A., Kastner, R., Benmansour, A. (1999) “Passive Earth Pressures in the Presence of
Hydraulic Gradients,” Geotechnique 49, No. 3, pp. 319-330
10. Tanaka, T., Kusaka, T., Nagai, S., Hirose, D. (2009) “Characteristics of Seepage Failure of
Soil Under Various Flow Conditions,” Proc. of the 19th International Offshore and Polar
Engineering Conference, Osaka, Japan, pp. 90-95.
11. Tanaka,T., Song, S., Shiba, Y., Kusumi, S., Inoue, K. (2012) “Seepage Failure of Sand in
Three Dimensions-Experiments and Numerical Analyses,” 6th International Conference on
Scour and Erosion, Paris, pp. 221-228.
12. Terzaghi, K. (1943) “Theoretical Soil Mechanics,” Wiley, New York.
13. Terzaghi, K., Peck, R. B., Mesri, G. (1996) “Soil Mechanics In Engineering Practice,” John
Wiley & Sons.
14. Witt, K.J., Brauns, J. (1983) “Permeability-Anisotropy due to Particle Shape,” ASCE, Vol.
109, No.9, pp.1181-1187.

© 2013, EJGE

Potrebbero piacerti anche