You are on page 1of 9

Technical Note

Adaptation of Method of Fragments to Axisymmetric


Cofferdam Seepage Problem
Thushara Asela Madanayaka1 and Nagaratnam Sivakugan, Ph.D., P.E., F.ASCE2

Abstract: The method of fragments can be used to efficiently obtain solutions to a variety of two-dimensional seepage problems, especially
confined flow problems. With this method, the flow domain is divided into fragments that span zones, which are defined between certain criti-
Downloaded from ascelibrary.org by CASA Institution Identity on 01/12/24. Copyright ASCE. For personal use only; all rights reserved.

cal planes, based on the assumption that the equipotential lines through each such plane are coincident with the plane(s) in question. A dimen-
sionless form factor is then defined for each fragment, and the total flow rate is computed using these form factors in a straightforward manner.
An extension of the method of fragments to the axisymmetric cofferdam problem is presented in this paper. A range of numerical simulations
was used to generate design charts to obtain the required axisymmetric form factors and exit gradients. These were validated against detailed
numerical solutions and against analytical solutions reported in the literature. The design charts presented in this paper are considered to pro-
vide very reliable estimates of the seepage conditions and for a wide range of cofferdam geometries of practical interest. DOI: 10.1061/
(ASCE)GM.1943-5622.0000955. © 2017 American Society of Civil Engineers.
Author keywords: Method of fragments; Circular cofferdam; Axisymmetric; Flow rate; Exit gradient.

Introduction The circular cofferdam problem is symmetric about its vertical


axis, so an axisymmetric seepage analysis is appropriate. There is
Cofferdams are frequently used temporary structures in the con- little specific literature on seepage solution methods for the circular
struction industry. There are two common types, double walled and cofferdam. Neveu (1972) developed a graphical solution to estimate
circular. The flow rate into the excavation and the stability of the the flow rate by combining Green’s Theorem with the method of
excavated base against piping failure are equally important design images. However, it does not give a solution for the exit gradient,
considerations of cofferdams. According to Harza (1935) a factor of and its applicability is limited to isotropic conditions. Moreover,
safety (FS) against piping failure may be defined as drawing a flow net for circular cofferdams is not a straightforward
task, because it requires the consideration of the three-dimensional
Critical hydraulic gradient (3D) nature of a cofferdam. Also, not all engineers are familiar with
FS ¼ (1)
Maximum exit gradient the numerical simulations. Therefore, there is a perceived benefit in
having a simple yet accurate seepage solution method for the circu-
Hence, the flow rate and exit hydraulic gradient are generally lar cofferdams. An axisymmetric method of fragments appears to
very important concerns in cofferdam designs. be a promising method because it could provide efficient solutions
Most double-walled cofferdam seepage problems appearing in to the flow rate and exit gradient question and because it has been
the literature are solved as two-dimensional (2D) problems using shown that anisotropy can be incorporated in the method (Griffiths
one of three approaches: (1) the graphical method, which is based on 1984; Madanayaka and Sivakugan 2016). Further, the method of
a manually drawn flow net; (2) the finite-element, finite-difference, fragments solution can be made to be very usable by the average
or other numerical methods; and (3) analytical solutions. In addi- practitioners due to its simplicity.
tion, Griffiths (1984) proposed a simple solution method using the
method of fragments technique. More recently, Madanayaka and
Sivakugan (2016) further simplified the Griffiths (1984) solution
Method of Fragments: Application of Method of
by defining the expressions for estimating the form factors and
dimensionless exit gradient. These studies have shown that the Fragments to Axisymmetric Problems
method of fragments has its place as a tool for quick and reliable
estimates of flow rate and exit gradient. The method of fragments was originally proposed by Pavlovsky
(1956) for 2D problems, and Harr (1962, 1977) brought it to the
Western world. However, this is an approximate method because its
1
Ph.D. Candidate, College of Science and Engineering, James Cook accuracy relies on the assumption that the equipotential lines at the
Univ., Townsville, QLD 4811, Australia (corresponding author). ORCID: critical points (e.g., tip of sheet pile in cofferdams) are vertical.
https://orcid.org/0000-0002-3440-667X. E-mail: thushara.madanayaka@ Then flow region is divided into a few fragments along these equi-
my.jcu.edu.au potential lines, and a dimensionless form factor (Ui) is defined for
2
Associate Professor, College of Science and Engineering, James each fragment as (Sivakugan and Alaghbari 1993)
Cook Univ., Townsville, QLD 4811, Australia. E-mail: siva.sivakugan@
jcu.edu.au Nd
Note. This manuscript was submitted on May 9, 2016; approved on Ui ¼ (2)
Nf
March 8, 2017; published online on May 18, 2017. Discussion period
open until October 18, 2017; separate discussions must be submitted for
individual papers. This technical note is part of the International Journal where Nd = number of equipotential drops in the given fragment;
of Geomechanics, © ASCE, ISSN 1532-3641. and Nf = number of flow channels. Also, the flow rate through each

© ASCE 06017012-1 Int. J. Geomech.

Int. J. Geomech., 2017, 17(9): 06017012


fragment is the same; hence, total flow rate per unit width can be The two fragments, separated along this joint, are shown in Fig. 2
defined as with the dimensions when the excavation depth is zero (a = 0). The
bases of the two fragments, the left boundary (i.e., vertical center-
kh1 kh2 khi kh line) of Fragment D and the sheet pile surfaces, are taken as imper-
q¼ ¼ ¼ ::: ¼ ¼P (3)
U1 U2 Ui U vious boundaries. The inner Fragment D has a radius of r. The outer

where h = head loss over the entire flow domain; hi = head loss r
within the ith fragment; and Ui = form factor for the ith fragment. P
Sivakugan and Rankine (2012) and Sivakugan et al. (2013) have
extended the method of fragments into 3D cases (hydraulic mine fill s
stopes) by defining 3D form factors. However, for the cofferdam
problems, the method of fragments presented in the literature is lim- T
ited to 2D problems. The work presented in this paper extends the
Downloaded from ascelibrary.org by CASA Institution Identity on 01/12/24. Copyright ASCE. For personal use only; all rights reserved.

method to axisymmetric situations. It requires proposing new axi- Fragment D


symmetric fragments and developing their form factors and dimen-
sionless exit hydraulic gradients.
(a)
Fig. 1 shows a sheet pile cofferdam enclosure, which is circular
in plan with a radius of r. The depth of the soil to the impervious re
stratum is T. The soil between the sheet piles is excavated to a depth r
of as, where a < 1. The head loss across the sheet pile is h. This is
an axisymmetric problem in which the flow rate is generally com-
puted per radian [shaded zone in Fig. 1(b)], which is multiplied by s
2p to obtain the total flow rate into the cofferdam. The arc length of
this shaded sector is r. The axisymmetric method of fragments T
approach is similar to the 2D Cartesian one. The equipotential line
at the tip of the cutoff wall is assumed to be vertical, and the axis of Fragment E
symmetry acts as an impermeable boundary. The flow domain con-
sists of an inner Fragment D and outer Fragment E, which are joined
(b)
at the vertical surface along perimeter of the cofferdam.
The joint, shown as a dashed vertical line in Fig. 1, is the equipo- Fig. 2. Axisymmetric fragment types: (a) Fragment D; (b) Fragment E
tential surface, and flow lines are assumed perpendicular to this.

r
Sheet pile
GL
h
s
s

T Frag. D Frag. E

q q

Equipotential line
Impervious stratum
(a)

Sheet pile r
radian

(b)

Fig. 1. Circular cofferdam in axisymmetric configuration: (a) elevation view; (b) plan view

© ASCE 06017012-2 Int. J. Geomech.

Int. J. Geomech., 2017, 17(9): 06017012


 
Fragment E begins at the radius of r and extends to re, which is very kh s rR
large. The sheet pile is driven to a depth s into the soil bed. The b ¼ ¼ f fs; r; T; Rg ¼ f ; (4)
q T T
thickness of the soil bed is T, where the excavation within the cof-
pffiffiffiffiffiffiffiffiffiffiffiffiffi
ferdam is to a depth of as. R ¼ kV =kH , where kV and kH = vertical and horizontal permeabil-
Similar to the dimensionless form factors used in the 2D method ities, respectively. According to the Eq. (3), the flow rate through the
of fragments, a form factor b is used to represent the fragments, in two fragments should be same; therefore, Eq. (4) can be rearranged as
this case of axisymmetric flow.
The main attributes of an overall form factor b must be q¼
khD khE
¼ ¼
kh
(5)
1. It should be a dimensionless number. bD bE bD þ bE
2.
Numerical Procedure
b cofferdam ¼ b D þ b E
Through a series of runs using finite-element software RS2 V9.011,
where b D and b E are the form factors of Fragments D and E,
Downloaded from ascelibrary.org by CASA Institution Identity on 01/12/24. Copyright ASCE. For personal use only; all rights reserved.

flow rate q was computed for each fragment for different combinations
respectively. of s/T and rR/T keeping the soil isotropic (R = 1). The study covers the
3. The ratio of head losses (hi) and form factors b i are the same s/T ratio ranging from 0.1 to 0.9 and rR/T from 0.1 to 1. In practical
for all fragments, i.e. applications, it is unlikely to see ratios of s/T and rR/T less than 0.1,
and it is uneconomical to have a s/T ratio greater than 0.9. Noting that
hD hE h the flow rate q is proportional to k and h, the values of k and h were
¼ ¼
bD bE bD þ bE kept constant in all runs. They have no influence on b , which is only
sensitive to the two dimensionless parameters s/T and rR/T. Also, each
where h = hD þ hE = total head loss over the entire cofferdam, run was considered as a flow-only problem, and four-nodded quadri-
and hD and hE = head losses within Fragments D and E. lateral elements were used to perform the simulations.
4. When the geometry is scaled by x times, the form factor should When modeling the semi-infinite domain of Fragment E, radius
remain the same. of the right boundary re was maintained at least as six times the ra-
In circular cofferdams, when the geometry is scaled by x times, the dius of cofferdam r [Fig. 2(b)] because Pavlovsky (1922) [see
total flow is also scaled by x times, whereas the flow rate per unit Neveu (1972)] showed that when the ratio r/re is about 6, the veloc-
length of the perimeter remains the same at a constant head loss h. ity potential distribution and seepage quantity is approximately
Therefore, to ensure that the form factor of any fragment b i remains identical to the case of infinite extension. Although Fragments D
the same when the geometry is scaled up, the flow rate per unit and E are different in geometry, both geometries become similar
length of the perimeter is used in defining the form factor b . Here, when r equals infinity, and situations can be simulated by a single
b is a function only of the geometry of the given flow fragment sheet pile wall in the 2D Cartesian system. Hence, Fragment D and
described by embedded depth (s), radius (r), thickness of the soil E form factors are the same when r equals infinity.
(T), and anisotropy of permeability (R). Then form factor b can be Madanayaka and Sivakugan (2016) noted that when increas-
written as ing the fragment width b (b = width of the corresponding plane

3.0 0.8
2D FF at b = ∞ 2D EG at b = ∞
Axy. FF - Frag. D Axy. EG - Frag. D
Axy. FF - Frag. E
2.5
0.7
Exit Gradient (EG) -iEs/hD

2.0 s/T
Form Factor (FF)

s/T 0.7 0.1


0.8
1.5
0.4

0.6
1.0
0.4

0.6
0.5 0.1

0.8
0.0 0.5
0.0 3.0 6.0 9.0 12.0 0.0 3.0 6.0 9.0 12.0
rR/T rR/T
(a) (b)

Fig. 3. Convergence of axisymmetric form factor and dimensionless exit gradient values to the 2D values at the infinite extent of fragment width b:
(a) form factors; (b) exit gradient

© ASCE 06017012-3 Int. J. Geomech.

Int. J. Geomech., 2017, 17(9): 06017012


strain fragment of axisymmetric fragment D) beyond two times Fragment D; and s = cutoff wall height of Fragment D. Similar
the permeable layer thickness T in the 2D Cartesian system, the to the form factors estimated earlier, dimensionless exit gradient
increase in the flow rate is insignificant, and it can simulate the values when r tends to the infinity were also estimated by using
situation of infinite extent (b = 1). Therefore, 2D fragment geo- the 2D fragment geometry with the ratio of fragment width to
metries with the fragment width equal two times the permeable the permeable layer thickness equal to two. Fig. 3(b) shows the
layer thickness were used to estimate the form factor of both convergence of dimensionless exit gradient values of Fragment
Fragments D and E when r equals infinity. The convergence of D when increasing the rR/T ratio to the 2D exit gradient values
Fragments D and E form factor values to the 2D form factor val- at infinite width.
ues when increasing the rR/T is shown in Fig. 3(a). The computed form factor values for both fragments are pre-
Also, the exit gradient value (iE) of each run for Fragment D sented in the form of design charts that can be used to determine the
was computed at the downstream end, next to the sheet pile flow rate per unit length along the perimeter q. The form factor
[Point P in Fig. 2(a)]. The computed values are presented in charts for Fragments D and E are shown in Figs. 4 and 5, respec-
dimensionless form ðiE s=hD Þ, where hD = head loss through tively. The dimensionless exit gradient chart for Fragment D is
Downloaded from ascelibrary.org by CASA Institution Identity on 01/12/24. Copyright ASCE. For personal use only; all rights reserved.

20.00
rR/T

0.1
18.00

16.00

14.00

0.15
12.00

β 10.00
0.2

8.00 0.25

0.3
6.00 0.35
0.4

0.5
4.00 0.6
0.7
0.8
1.0
2.0
2.00 ∞

0.00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
s/T
Fig. 4. Form factor values for Fragment D (from finite-element simulation)

© ASCE 06017012-4 Int. J. Geomech.

Int. J. Geomech., 2017, 17(9): 06017012


shown in Fig. 6. The effect of the soil anisotropy, R, is directly condition (R = 1) of three different radii (rR/T = 0.1, 0.4, and 0.8) is
incorporated in these charts. shown in Figs. 7 and 8, respectively. However, for the comparison,
an extensive number of cofferdam geometries (108) were studied
for s/T ratios ranging from 0.1 to 0.9 and rR/T varied from 0.1 to 1.
Validation of Method of Fragments for a
The worst cases were encountered when the embedded depth of cut-
Circular Cofferdam
off walls were relatively short (i.e., s=T  0:2), which includes 24
cofferdam geometries. Here, the proposed method of fragments pre-
The flow rate and exit gradient predicted using the proposed method
dicted higher values, and their average deviations for the seepage
of fragments were compared against full numerical solutions and
quantity and exit gradient are 11 and 16%, respectively. For other
against analytical solutions reported in the literature.
situations (i.e., 0:2 < s=T  0:9 and 0:1  rR=T  1), 84 geome-
tries were studied, and the proposed method tended to slightly over-
Comparison with Full Numerical Solutions
estimate the seepage quantity and exit gradient for all the cases with
The comparison of the seepage quantity and exit gradient values average deviations of 3 and 4%, respectively. In summary, increas-
Downloaded from ascelibrary.org by CASA Institution Identity on 01/12/24. Copyright ASCE. For personal use only; all rights reserved.

when the excavation depth equals zero (a = 0) for the isotropic ing the depth of the cutoff wall (s) and/or radius of the cofferdam (r)

rR/T


2.00

2.0

1.75
1.0
0.8
0.7
0.6
1.50 0.5
0.4
0.35
0.3

1.25 0.25

0.2

β 0.15
1.00

0.1

0.75

0.50

0.25

0.00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
s/T
Fig. 5. Form factor values for Fragment E (from finite-element simulation)

© ASCE 06017012-5 Int. J. Geomech.

Int. J. Geomech., 2017, 17(9): 06017012


1.00
rR/T
0.1
0.15
0.95
0.2
0.25

0.90 0.3

0.35

0.4
0.85
Downloaded from ascelibrary.org by CASA Institution Identity on 01/12/24. Copyright ASCE. For personal use only; all rights reserved.

0.5

0.80

0.6

0.75
0.7

iEs/hD 0.70 0.8

0.65 1.0

0.60

0.55

2.0

0.50

0.45 ∞

0.40
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
s/T
Fig. 6. Exit gradient values for Fragment D (from finite-element simulation)

makes the seepage quantity and exit gradient predicted by the pro- 11% for the low depth of cutoff wall (s = 0.2 T), and for the other
posed method of fragments converge to the actual values computed two cases (s = 0.5 T and s = 0.8 T) it is about 3%. For the excavation
from the full numerical solutions. However, all the figures of seep- depth (a) is equal to 0.505, and a significant average deviation of
age quantity and exit gradient predicted by the proposed method are 18% is observed for s = 0.2 T. This is due to further lowering of the
on the conservative side (i.e., overestimate). s/T ratio of Fragment D when increasing a [i.e., s(1 − a)/(T − as)],
and in this case it is 0.11 while the s/T ratio of Fragment E
Comparison with Early Research Findings remains 0.2. For the same excavation depth, the average devia-
tion is 6% for the other two cases (s = 0.5 T and s = 0.8 T). When
The seepage quantity estimated by the proposed method is com- a = 0.808, the average deviation is about 12% for s/T = 0.5 and
pared here with the solutions by the Neveu (1972) for the isotropic 0.8. Also for all these situations, the values given by the pro-
condition (R = 1). The comparison result is shown in Fig. 9. When posed method are conservative (i.e., overestimate). In summary,
the excavation depth is zero (a = 0), the average deviation is about the proposed method of fragments solutions are in good agreement

© ASCE 06017012-6 Int. J. Geomech.

Int. J. Geomech., 2017, 17(9): 06017012


1.0
r Full NM
0.9 Proposed

0.8

h 0.7
αs
s 0.6

q/kh
0.5
T
0.4
Downloaded from ascelibrary.org by CASA Institution Identity on 01/12/24. Copyright ASCE. For personal use only; all rights reserved.

0.3
rR/T
0.2 0.8
Impervious stratum 0.4
0.1
0.1
0.0
0 0.2 0.4 0.6 0.8 1
s/T

Fig. 7. Comparison of the proposed fragments solutions with full numerical model (NM) solution for seepage quantity

h = 10m 4.0
r Full NM
T = 20m
3.5 Proposed

3.0
h
αs 2.5
s
iE iE 2.0

T
1.5

1.0
rR/T
Impervious stratum 0.5 0.1
0.4
0.8
0.0
0 0.2 0.4 0.6 0.8 1
s/T
Fig. 8. Comparison of the proposed fragments solution with full numerical model (NM) solution for exit gradients

with the analytical solutions of Neveu (1972). When increasing the of fragments. Three design charts have been developed: two for
excavation depth, the level of discrepancy increases; however, this estimating the form factor for each of the axisymmetric fragments
deviation is on the conservative side, overestimating the seepage (defined in the paper as D and E), and one for the dimensionless
quantity. exit gradient of Fragment D. These form factor and exit gradient
charts represent an efficient means of obtaining both the quantity
of seepage and the exit gradient and for virtually any circular cof-
Summary and Conclusion ferdam geometry of practical interest. The outcomes of the adap-
tation of the method of fragments to the problem described in this
A simple method for solving seepage problems pertaining to the paper were compared with detailed numerical solutions and to the
circular cofferdam has been presented. It is based on the method analytical work of Neveu (1972) for this same class of problem. It

© ASCE 06017012-7 Int. J. Geomech.

Int. J. Geomech., 2017, 17(9): 06017012


0.8
α=0 Proposed
0.7
Neveu (1972)
0.6

0.5 s/T = 0.2

q/kh
s/T = 0.5
0.4

0.3 s/T = 0.8

0.2

0.1
Downloaded from ascelibrary.org by CASA Institution Identity on 01/12/24. Copyright ASCE. For personal use only; all rights reserved.

0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
rR/T
(a)

0.8
α = 0.505 s/T = 0.2
Proposed
0.7
Neveu (1972)
0.6
s/T = 0.5
0.5
q/kh

0.4
s/T = 0.8
0.3

0.2

0.1

0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
rR/T
(b)

0.8
α = 0.808 Proposed
0.7 Neveu (1972)
s/T = 0.5
0.6

0.5
q/kh

0.4 s/T = 0.8

0.3

0.2

0.1

0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
rR/T
(c)

Fig. 9. Comparison of the proposed fragments solution with analytical solution by Neveu (1972) for seepage quantity: (a) a = 0; (b) a = 0.505;
(c) a = 0.808

was found that the proposed solutions are in very good agreement There are, however, limits to the applicability of the charts that
with the outcomes of the other two methods of solution men- have been presented. The geometry of the circular cofferdam must
tioned previously. be such that

© ASCE 06017012-8 Int. J. Geomech.

Int. J. Geomech., 2017, 17(9): 06017012


s Harza, L. (1935). “Uplift and seepage under dams on sand.” Trans. ASCE,
0:1   0:9
T 100(1), 1352–1385.
Madanayaka, T. A., and Sivakugan, N. (2016). “Approximate equations for
and the method of fragment.” Int. J. Geotech. Eng., 10(3), 297–303.
Neveu, G. (1972). “Axisymmetrical seepage flow through a circular sheet
rR pile cofferdam.” M.Eng. thesis, McGill Univ., Montreal.
0:1   1:0
T Pavlovsky, N. (1922). The theory of movement of ground water under hy-
draulic structures and its main applications, Pertogrod, USSR.
It is felt that these limits still cover most situations of engineer- Pavlovsky, N. N. (1956). Collected works, Doklady Akademii Nauk USSR,
ing interest. For extreme cases in which rR/T > 1.0, the design Leningrad, USSR.
charts still provide approximate answers that are also conservative RS2 V9.011 [Computer software]. Rocscience, Toronto.
but would not have very good levels of accuracy. Sivakugan, N., and Alaghbari, M. Y. S. (1993). “Method of fragments–
quick solutions to seepage problems.” Proc., Environmental
Management: Geo-Water and Engineering Aspects, A.A. Balkema,
References
Downloaded from ascelibrary.org by CASA Institution Identity on 01/12/24. Copyright ASCE. For personal use only; all rights reserved.

Amsterdam, Netherlands, 491–496.


Sivakugan, N., and Rankine, K. (2012). “Three-dimensional method of
Griffiths, D. V. (1984). “Rationalized charts for the method of fragments fragments to study drainage through hydraulic fill stopes.” Int. J.
applied to confined seepage.” Geotechnique, 34(2), 229–238. Geomech., 10.1061/(ASCE)GM.1943-5622.0000158, 612–615.
Harr, M. E. (1962). Groundwater and seepage, McGraw-Hill, New York. Sivakugan, N., Rankine, K., Lovisa, J., and Hall, W. (2013). “Flow rate
Harr, M. E. (1977). Mechanics of particulate media, McGraw-Hill, New computations in hydraulic fill mine stopes.” Indian Geotech. J., 43(3),
York. 195–202.

© ASCE 06017012-9 Int. J. Geomech.

Int. J. Geomech., 2017, 17(9): 06017012

You might also like