You are on page 1of 13

SHEETPILE INTERLOCK TENSION

IN CELLULAR COFFERDAMS
By Mark P. Rossow, 1 A. M. ASCE

ABSTRACT: A method is presented for calculating the interlock tensions in the


sheetpile walls of a circular cellular cofferdam. A horizontal strip of unit width
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

of wall is considered and the governing equations are derived from equilibrium
and compatibility requirements for the strip. The pressure of the cell fill against
the sheetpile walls is assumed known. Numerical examples are presented in
which interlock tensions in the arc cell, main cell, and common-wall are cal-
culated for various pressures. Bulging of the cell walls and rotation of the legs
of the connecting Y are also found. The calculations provide theoretical justi-
fication for use of an equation for estimating the force in the common-wall and
lead to a generalization of that equation. The method may be applied to other
sheetpile wall configurations besides those of a circular cofferdam.

INTRODUCTION

General.—One of the many factors which must be considered in de-


signing a cellular cofferdam is the possibility of failure caused by exces-
sive interlock force in the sheetpile walls. In this paper, a method is
presented for the rational estimation of interlock forces, given the fill
pressure acting on the walls. After the various assumptions upon which
the method is based have been stated, the governing equations are de-
rived and the method is shown by numerical examples representative
of an actual circular cofferdam. Interlock force predictions obtained by
the method are compared to those obtained by several other methods,
and recommendations are made for design calculations. Although the
approach described in this paper is applied to a circular cellular coffer-
dam, it also applies with appropriate modifications to other configura-
tions such as diaphragm and cloverleaf cofferdams and circular bulkheads.
Relation to Previous Work.—A number of references are available in
which procedures for calculating interlock forces are examined. The scope
of these examinations is wider than that of this paper, since they treat
both: (1) How to estimate the maximum pressure of the fill bearing on
the cofferdam wall; and (2) how to calculate the interlock force, once the
pressure is known. This paper is concerned only with the second ques-
tion. Thus, the intent of this paper is to provide the designer with an
improved analysis tool for use after an estimate of the fill pressure has
been made. The intent is not to give a definitive answer to the difficult
soil-structure problem inherent in the analysis of a flexible soil-retaining
structure.
A typical layout of a cellular cofferdam is shown in Fig. 1. The inter-
lock force in the common-wall near the connecting Y pile has been of
special concern to designers. A long-established method of calculating
'Prof., Dept. of Civ. Engrg., Southern Illinois Univ. at Edwardsville, Edwards-
ville, 111. 62026.
Note.—Discussion open until March 1, 1985. To extend the closing date one
month, a written request must be filed with the ASCE Manager of Technical and
Professional Publications. The manuscript for this paper was submitted for re-
view and possible publication on September 12, 1983. This paper is part of the
Journal of Geotechnical Engineering, Vol. 110, No. 10, October, 1984. ©ASCE,
ISSN 0733-9410/84/0010-1446/$01.00. Paper No. 19210.
1446

J. Geotech. Engrg. 1984.110:1446-1458.


UNIT WIDTH

COMMON/WALL
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

u
PLAN VIEW ELEVATION VIEW

FIG. 1.—Region Selected for Analysis

this force is the TVA-secant formula (9). This formula, however, is ob-
jectionable for two reasons: (1) Its derivation is based on a balance of
forces in the inboard-to-outboard direction only (equilibrium is violated
in all other horizontal directions); and (2) the orientation of the Y pile
is assumed unchanged by cell filling even though field observations and
elementary considerations of statics (1,2,10) show that the Y must de-
form and rotate to equilibrate the forces from the three walls it joins.
Lacroix et al. (4) specifically criticize the TVA-secant formula and rec-
ommend against its use. These latter writers also cite an equation at-
tributed to Swatek (10) for the common-wall interlock force: t = pL, in
which p = the average fill pressure; and L = the average spacing, shown
in Fig. 2. No derivation of Swatek's equation appears to have been
published.
Sorota et al. (8), in their discussion of the Trident cofferdam, compare
interlock tensions calculated by various design procedures to interlock

FIG. 2.—Geometry and Loading FIG. 3.—Forces Acting on Y


1447

J. Geotech. Engrg. 1984.110:1446-1458.


tensions measured in the field. They report that the TVA-secant formula
appears conservative.
Useful observations on interlock forces have also b e e n m a d e by
Schroeder a n d his co-workers (5,6,7), w h o , in addition, comment u p o n
the desirability of improving present approaches of accounting for the
effects of connecting arcs (6). This p a p e r is intended as a step in that
direction.
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

GOVERNING EQUATIONS

Derivation.—Consider a horizontal slice of unit w i d t h cut t h r o u g h a


portion of the cofferdam s h o w n in Fig. 1. If it is a s s u m e d that the cof-
ferdam has not yet b e e n dewatered, then the loading is symmetrical,
and only the cross-hatched region s h o w n in the sketch need be consid-
ered. The fundamental assumption u p o n which the m e t h o d is based is
that the sheetpile walls of the cofferdam behave as thin, inextensible
membranes (i.e., the walls offer n o resistance to bending) loaded by uni-
form pressure. From this assumption, it follows that a unit-width por-
tion of each wall assumes the shape of a circular arc a n d , furthermore,
the interlock forces in the walls are given b y the hoop-stress formula t
= pR, w h e r e p = the pressure acting o n the wall; and R = the radius of
the arc. Thus, a unit w i d t h of the cofferdam can be analyzed as three
intersecting cables loaded by k n o w n transverse pressures a n d with
boundary conditions dictated b y the conditions of symmetry. Fig. 2 de-
fines the geometry (radii R„, Rm a n d Rw; and subtending angles 0„, 9,„
and GOT) and the fill pressures (pa, p,„ a n d pw) associated with the prob-
lem. The subscripts "a," "m" a n d "w" indicate the arc cell, main cell
and common-wall, respectively. It should be pointed out that the com-
mon-wall pressure, pw, is treated here as an independent load parameter
representing the net pressure on the common-wall; pw need not equal
Pm ~~ Par since pm a n d pa are fill pressures acting at other locations. It
should also be emphasized that Ra, Rm , Rm, Q„, Qm and Qw describe the
geometry of the arc a n d main cells after the cell walls have been de-
formed and displaced by the pressure loads. In general, these values
differ from the "as-built" values existing after pile driving, b u t before
the cells have b e e n deformed by filling.
To formulate the governing equations, note that during cell filling, the
lengths of the cell walls (measured along the curve of the wall) remain
essentially unchanged, except for some relatively small extensions caused
by take-up of slack in the interlocks. If the (known) arc length of each
wall is denoted b y the letter "s" with the appropriate subscript, the
equations representing the condition that each wall length is fixed are

Rt9a = St (1)
RmQm = S„, (2)
R**w = sa (3)
Another geometric constraint follows from the condition that the cen-
terline-to-centerline distance in Fig. 2 is the k n o w n quantity L:
Ra sin (6 J + Rm sin (0,„) = L (4)

1448

J. Geotech. Engrg. 1984.110:1446-1458.


The set of governing equations may be completed by writing the equi-
librium equations for the walls
ta = PaRa (5)
t,„ = pmR„ . (6)
tw = pmRw (7)
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

in which i„, tm and tw = the interlock tensions; and by equating the


sums of the x and y force components acting on the Y in Fig. 3 to zero:
-ta cos (9„) + t,„ cos (9,„) - tw sin (Qw) = 0 (8)
ta sin (9„) + t„, sin (6,„) - tw cos (Qw) = 0 (9)
Solution Technique.—Eqs. 1-9 constitute a set of transcendental
equations, for which explicit expressions cannot be obtained for the nine
unknowns (R„, Rm, Rw, 9„, 0,„, Qw, ta, tm and tw) in terms of general
values of the pressures, arc lengths and length L. For specific numerical
values of these latter parameters, a numerical solution for the nine un-
knowns may be obtained as follows. First, define the nondimensional
variables

a=s
i < 10 >
m (11)

™=J (12)

(13)
*=f
r=P~ (14)
Pm
Next, Eqs. 1-3 and 5-7 can be used to eliminate the radii and tensions
from Eqs. 4, 8 and 9. The resulting equations can then be simplified by
making use of Eqs. 10-14, thus yielding

sin (6.) + l^-j sin (9m) - 1 = 0 (15)

aq\ (m\ (wr\


1
cos (6.) + I— I cos (6,„) - 1 — 1 sin (6„) = 0 (16)
laq\ (m\ (wr\
\^j sin (6.) + ^—J sin (9,„) - (^—J cos (6„) = 0 (17)
For small values of net common-wall pressure, pw, the parameter r of
Eq. 14 is small. Calculations then show that the subtending angle %w is
also small, thus leading the last term in both of Eqs. 16 and 17 to behave
1449

J. Geotech. Engrg. 1984.110:1446-1458.


as 0/0. To avoid the resulting numerical instability in the solution al-
gorithm, for r < 0.15, Eq. 17 should be rewritten as

^ - j sin (6.) + (~^j sin (6ra) - (wr) cos (e„) = 0 (18)

and Eq. 16 is rewritten as


i
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

cos (eJ + cos (0m) [6 (9J2] = (19)


\?) (f) "w " °
in which sin (Qw) has been approximated by the first two terms in its
series expansion, 6^ — {[(6H))3]/6}.
Eqs. 15-17 (or, in the case of r small, Eqs. 15, 18 and 19) were solved
by the Newton-Raphson iterative method, using the as-built values of
the angles as the initial estimates of the equations's roots. It was found
necessary, however, to limit the size of the Newton-Raphson increments
during the first 10 iterations in the following manner. The usual New-
ton-Raphson formula (3) for the (i + l)th approximation to the solution
of an equation/(x) = 0 was changed to xi+1 = x{ - [£,/;/(/'),], in which
k( = z/10 for i = 1, 2, ..., 10; and kt = 1.0 for i > 10. Through use of
this device, the algorithm was prevented from straying far from the ini-
tial estimate of the roots during the early iterations. Convergence was
obtained for all examples considered.

EXAMPLES

Various Loadings of Completely Filled Cells.—To illustrate the ap-


plication of the equations derived in the previous section, an example
was solved based on the following parameter values (which correspond
approximately to an actual cofferdam):
a = 0.576 (20a)
tn = 0.766 (20b)
w = 0.383 (20c)
L = 43.2 ft (13.2 m) (20d)
The as-built values of the subtending angles are

K =\ (21«)

e« = ~ (2ib)

eB = 7 (2ic)
6
Corresponding as-built values of the radii R„, R,„ and Rw may be found
through use of Eqs. 1-3, 10-12, 20-21.
Results of the analysis of the example are given in Figs. 4-9. In Fig.
1450

J. Geotech. Engrg. 1984.110:1446-1458.


PRESENT METHOD
• TVA SECANT FORMULA
• SWATEK'S EQUATION

' Pa '

6.5 -Pm" IS" 1 ' 4 \ _


•1.5 KIPS/SQ FT I
_ _
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

6.0 '%
^
5.5
0.8\
Rfl =
0.6s

4.5 :
4(1 i i i
0.15 0.30 0.45 0.60 1.25 1.50 1.75 2.00 2.25 2.50
COMMON-WALL PRESSURE, MAIN-CELL PRESSURE, P m ,
P„,, IN KIPS/SQ FT IN KIPS/SQ FT

FIG. 4.—Insensitivity of Common-Wall FIG. 5.—Variation of Common-Wall


Tension to pw (1 kip/in. = 0.175 kN/mm; Tension With Pressure in Main Cell (1
1 kip/sq ft = 47.8 kN/m2) kip/in. = 0.175 kN/mm; 1 kip/sq ft =
47.8 kN/m2)

4, a key quantity calculated in the analysis—the common-wall tension


tw—is shown to be insensitive to what estimate is made for the com-
mon-wall pressure. Thus, even though only a crude estimate for the
common-wall pressure may be available, the model still can produce useful
information about the common-wall tension. The figure also shows that,
as would be expected, tw increases significantly as the values assumed
for the pressure in the arc cell are increased (with pm held fixed).
Fig. 5 shows the increase in tw occurring with increasing main-cell
pressure and for three different assumptions about the ratio of arc-cell
to main-cell pressure. For comparison purposes, results calculated from
the following generalized version of Swatek's equation
tw = PaRa sin (60) + pmR,„ sin (6,„) (22)
have also been plotted. As-built values of 6„, 6,„, R„ and Rm are used in
Eq. 22. This equation is seen to give results close to those obtained by
solving Eqs. 1-9. If the assumption is made that pa/pm = 1/ the results
obtained from Eq. 22 agree so closely with those obtained by solving
Eqs. 1-9 that only one curve has been plotted in Fig. 5.
Fig. 5 also shows the common-wall tension calculated from the TVA-
secant formula (9), which, for the present example, is (p,„ must be as-
sumed equal to pa)

tw= (pmL) sec (23)

The formula is seen to predict common-wall tension even greater than


that obtained from Eqs. 1-9 under the conservative assumption that the
arc-cell pressure is 1.3 times the main-cell pressure.
1451

J. Geotech. Engrg. 1984.110:1446-1458.


= 1 ,U 1 1 1 1 1 6.0
CL PRESENT METHOD /
_ AS-BUILT GEOMETRY s'.
M 6.0
* 3*--0.15 , / '
._eE5.0 - p
m . / ^
2*

% 4 "°
2 ^,**
,''
J ^
^^^1.3
^tt^^"^
_
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

•" 3.0 s ^ ^ - ^ ^1.0 _


Li) ^f^ j**"*"""'*^ _^-—"""""""^

? 2.0 ^ ^ ^ - o , J -
^ - ^ i" ^
i— i i i
m i
5 1.0
1.0 1.25 1.50 1.75 2.00 2.25 2.50 1.0 1.25 1.50 1.75 2.00 2.25 2.50
MAIN-CELL PRESSURE, MAIN-CELL PRESSURE,
P m , IN KIPS/SQ FT

FIG. 6.—Variation of Main-Cell Tension FIG. 7.—Variation of Arc-Cell Tension


With Pressure in Main Cell (1 kip/in. = With Pressure in Main Cell (1 kip/in. =
0.175 kN/mm; 1 kip/sq ft = 47.8 kN/m2) 0.175 kN/mm; 1 kip/sq ft = 47.8 kN/m2)

Fig. 6 shows the variation of tension in the main cell as a function of


main-cell pressure, for the same three pressure ratios p„/p m used pre-
viously. Also plotted are the results obtained by using the as-built value
of Rm in Eq. 6. The use of the as-built value is seen to lead to a signif-
icantly higher value of the tension than is found by solving the complete
set of governing equations, Eqs. 1-9. In contrast, in Fig. 7 use of the as-
built value of Ra in Eq. 5 to calculate ta leads to a lower value of the
tension.
Comparing Figs. 5-7 shows that the tension in the common-wall is
significantly higher than that in the arc and main cells. This finding agrees
with finite element analyses and extensive field measurements of a re-
cently-built cofferdam which the writer has reviewed. For the Trident
cofferdam, Sorota et al. (8) reported that measured interlock tensions in
the common-wall near the Y were from 0-20% higher than those in the
main cell. Based on field and laboratory observations, Schroeder and
Maitland (6) stated that the common-wall tension is equal to or less than
the main-cell tension. The reasons for this disagreement about the com-
mon-wall tension are not clear, but are perhaps related to the wide range
of possible loads which the fill may deliver to the cell walls, given vari-
ations in fill material, filling technique, cell geometry and the existence
of arching.
Fig. 8 shows the changes in orientation (the absolute value of the dif-
ference between the as-built angle and its equilibrium value) which the
legs of the Y pile must undergo to achieve equilibrium, for various as-
sumptions about the common-wall pressure. The angle changes all cor-
respond to counterclockwise rotations for the legs of the Y in Fig. 3. The
most significant effect shown in Fig. 8 is the large rotation occurring in
the common-wall leg. It should be emphasized that these results are
based on the assumption that the Y has no bending stiffness, and in-
stead consists of three tensile members meeting at a joint.
The re-orientation of the Y can be seen similarly in Fig. 9, which also
1452

J. Geotech. Engrg. 1984.110:1446-1458.


0.7 FT
J2.9FT
$
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

o
z \ \ / AS--BUILT
<
COMMON-WALL -0.3
5.0 . MAIN CELL /! I m
ARC CELL
/ ,! Pw
-0.7
0.0
: i
0.0 0.1 0.2 0.3 0.4 0.5 0.6
RATIO OF COMMON-WALL TO
MAIN-CELL PRESSURE, P w /P m 5.4 FT

FIG. 8.—Rotation of Y for Various Val- FIG. 9.—Equilibrium Configurations for


ues of Common-Wall Pressure Two Different Loadings. pa/p„, = 1.0 (1
ft = 0.305 m)

shows the displacements (bulging) of the cell walls. (Recall that the bulg-
ing has been calculated under the assumption that the lengths of the
cell walls do not change during filling.) In general, the higher tension
in the common-wall causes it to displace and straighten out more than
the other two walls. Compared to the main-cell as-built diameter of 63.2
ft (19.3 m), the wall displacements shown in Fig. 9 are unrealistically
large, but the figure clearly indicates how the walls rearrange themselves
to carry the load. It is interesting to note that the wall displacements
depend on pressure ratios, rather than on absolute magnitudes of indi-
vidual pressures.
Simulated Arc-Cell Filling.—To illustrate the application of the method
to another situation, an example involving simulated arc-cell filling was
solved. Fig. 10 shows the situation studied. Two adjacent main cells are

FIG. 10.—Region Selected for Analy- FIG. 11.—Geometry and Loading for
sis: Arc-Cell Filling Arc-Cell Filling
1453

J. Geotech. Engrg. 1984.110:1446-1458.


assumed to be already filled, while the three associated arc cells are ini-
tially empty. Fill is then gradually placed in the middle arc-cell. These
assumptions imply symmetry about the centerline of the middle arc-cell,
and thus only the cross-hatched region shown need be analyzed. Fig.
11 shows the geometry and loading for this region. Rotations at the Y
and interlock tensions are to be calculated at various stages of filling.
The arc-cell wall on the right in Fig. 10 has been omitted from Fig. 11,
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

since the arc cell is not filled and thus the wall carries no tension.
The governing equations for this example are the same as for the pre-
vious example, Eqs. 1-9, except that 0„, must be replaced by 0m - (ir/
2), and the compatibility relation, Eq. 4, must be replaced by
Rw sin (e„) = Rm sin (9m) (24)
The values of a and w and the as-built values of 0„ and 0ro are the same
as in the previous example. The as-built value of 0„, is now taken to be
5 IT/6 radii and m is equal to 1.920.
As in the previous example, prediction of the net common-wall pres-
sure, pw, is problematic. Clearly, before the arc cell is filled pw is equal
to the main-cell pressure, pm. Subsequent filling of the arc cell then re-
duces p„, to some fraction of pm. Based on these observations, a rea-
sonable expression for pw is

P« = P,„ - (^~) (25)

This equation was used in the numerical calculations to be described


next.
Fig. 12 shows the variation of the interlock forces as the arc-cell pres-
sure approaches that of the main cell. As would be expected, the ten-
sions in both the arc cell and common-wall increase. The rate of increase
(slope) of the common-wall tension, however, is less than that of the

UJ 3 0 . 0

COMMON-WALL
MAIN CELL

LU 15.0 " .•''

I
< 10.0 - y
S 5.0 - ./•' P w ' P m - 1 ~ a/SfPa'Pm* -

5 o.o v — i 1 1 1 J
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
RATIO OF ARC-CELL TO RATIO OF ARC-CELL TO MAIN-CELL
MAIN-CELL PRESSURE, Pa/Pm PRESSURE, P a / P m

FIG. 12.—Variation of Interlock Force FIG. 13.—Rotation of Y During Arc-Cell


During Arc-Cell Filling Filling
1454

J. Geotech. Engrg. 1984.110:1446-1458.


arc-cell tension, thus indicating that the common-wall is more efficient
than the arc cell in picking up the increases in load. The source of this
greater efficiency lies in the common-wall's significantly larger change
in shape, as shown in the plot of changes in subtending angles in Fig.
13. Since the subtending angles are inversely proportional to the radii
through Eqs. 5-7, Fig. 13 also implies that the radius of the common-
wall increases significantly (i.e., the wall flattens out) as the arc cell is
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

filled. The angle changes plotted in the figure correspond to counter-


clockwise rotations of the arc-cell and common-wall legs and a clockwise
rotation of the main-cell leg of the Y.
A final observation on the arc-cell filling example relates to the main-
cell tension plotted in Fig. 12. It seems unrealistic that this tension re-
mains constant during arc-cell filling, since filling the arc will affect the
stress distribution in the main-cell fill, especially near the Y, and thus
change the loading on the main-cell wall. The present model, however,
requires a uniform pressure distribution over the entire wall and thus
cannot represent a change in pressure along a wall. This and other
shortcomings of the method of this paper are examined further in the
next section.

CRITIQUE OF METHOD

The approach defined by Eqs. 1-9 is based on two significant as-


sumptions: (1) The sheetpile walls have no resistance to bending about
a vertical axis; and (2) the sole action of the fill against each wall is a
uniform pressure of known magnitude.
The first assumption appears to be a reasonable approximation of the
actual wall behavior since the sheetpiles are free to rotate relative to each
other in their interlocks, and also since the resistance of the web to bend-
ing about a vertical axis is small. In general, to the extent that fabrication
and installation errors lead to bending, such bending should be pri-
marily a localized effect having relatively little influence on the overall
transmission of membrane forces throughout the sheetpile structure.
The second assumption is decidedly a simplification of a complex
physical situation. In general, the distribution of fill pressure against a
wall is not uniform, but varies with distance along the wall. Further-
more, friction effects will also be present. Soil-wall interactions near the
connecting Y are especially difficult to describe quantitatively and con-
stitute a challenge even for finite-element analysis. Indeed, one simply
cannot escape the fact that for the present method, as well as for pre-
vious methods of calculating interlock forces, a good estimate of the av-
erage soil pressure bearing on the wall in the loaded and deformed state
is essential if useful results are to be obtained.

PRACTICAL APPLICATIONS

Based on the results which have been presented, the following rec-
ommendations are made for design calculations.

1. Of the existing procedures for calculating the interlock force in the


1455

J. Geotech. Engrg. 1984.110:1446-1458.


common-wall, Swatek's formula, t = pL, strikes a balance between ac-
curacy and conservatism, and its use is recommended if no information
is available to distinguish between the pressure of the fill in the arc cell
and in the main cell.
2. If information about the pressures in both the arc and main cells is
available, then the generalized Swatek equation, Eq. 22, is recommended.
3. If information about the pressures on all three walls is available or
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

if greater accuracy is desired (e.g., a design parameter is near a critical


value) or if it is desired to estimate interlock forces during filling of the
arc-cell, then the method described in this paper is recommended. The
method also can be used to predict the interlock force in the arc-cell and
main-cell walls, in addition to the common-wall. This feature may be
especially useful if different types of piling, e.g., high-strength and reg-
ular, are to be used in the different parts of the cofferdam.
4. The present method may also be used, with appropriate modifi-
cations, for other sheetpile wall configurations.

SUMMARY AND CONCLUSIONS

An analytical method has been presented for the calculation of the


interlock tension in the sheetpile walls of a cellular cofferdam. The method
is based on the assumption that the walls act as inextensible membranes
(offering no resistance to bending about a vertical axis). A set of gov-
erning equations has been derived by stating the conditions of equilib-
rium and compatibility. Examination of the calculations for an example
problem reveals that the method provides theoretical backing for Swa-
tek's equation for the force in the common-wall. The calculations also
suggest how Swatek's equation may be generalized to allow the value
of the fill pressure in the arc-cell to differ from that in the main-cell. The
calculations in the example also confirm the previously observed im-
portance of the rotation of the Y connection in carrying the loads.
Although the method has been presented here for the particular sheet-
pile wall configuration corresponding to a circular cofferdam, the basic
approach (inextensible membranes, equilibrium and compatibility) also
applies to other sheetpile structures. However, since the method's pre-
dictions of interlock forces depend crucially on the specification of the
fill pressure acting on a wall, the confidence to be placed in the predic-
tions is directly related to the confidence with which a single number
can be used to represent the complex soil-structure interaction which
occurs in a flexible earth-retaining structure.

ACKNOWLEDGMENTS

This work was done in part while the writer served as a consultant to
the St. Louis District of the U.S. Army Corps of Engineers, and helpful
discussions with many people in the District are gratefully acknowl-
edged. D. Dressier of the Office of the Chief of Engineers, Washington,
D.C., provided encouragement to pursue the study. E. Demsky of the
St. Louis District pointed out the relation of Swatek's equation to the
method presented in this paper.
1456

J. Geotech. Engrg. 1984.110:1446-1458.


All opinions expressed in this p a p e r are solely the writer's a n d should
not be interpreted as representing Corps policy or practice.

APPENDIX I.—REFERENCES

1. Dismuke, T. D., "Cellular Structures and Braced Excavations," Foundation


Engineering Handbook, H. F. Winterkorn and H. Fang, eds., Van Nostrand
Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

Reinhold Co., New York, N.Y., 1975, pp. 451-452.


2. Dismuke, T. D., "Stress Analysis of Sheet Piling in Cellular Structures," Pro-
ceedings of the Conference on Design and Installation of Pile Foundations and Cel-
lular Structures, Envo Publishing Co., Inc., Pennsylvania, 1970, pp. 339-366.
3. Henrici, P., Elements of Numerical Analysis, John Wiley and Sons, New York,
N.Y., 1964, pp. 105-107.
4. Lacroix, Y., Esrig, M. I., and Lusher, U., "Design, Construction and Per-
formance of Cellular Cofferdams," Lateral Stresses in the Ground and Earth Re-
taining Structures, ASCE, June, 1970, pp. 271-328.
5. Maitland, J. K., and Schroeder, W. L., "Behavior of Model Circular Sheetpile
Cells," Journal of the Geotechnical Engineering Division, ASCE, Vol. 105, No.
GT7, Proc. Paper 14712, July, 1979, pp. 805-821.
6. Schroeder, W. L., and Maitland, J. K., "Cellular Bulkheads and Coffer-
dams," Journal of the Geotechnical Engineering Division, ASCE, Vol. 105, No.
GT7, Proc. Paper 14713, July, 1979, pp. 823-838.
7. Schroeder, W. L., Marker, D. K., and Khuayjarernpanishk, T., "Performance
of a Cellular Wharf," Journal of the Geotechnical Engineering Division, ASCE,
Vol. 103, No. GT3, Proc. Paper No. 12790, Mar., 1977, pp. 153-168.
8. Sorota, M. D., Kinner, E. B., and Haley, M. X., "Cellular Cofferdam for
Trident Drydock: Performance," Journal of the Geotechnical Engineering Divi-
sion, ASCE, Vol. 107, No. GT12, Proc. Paper 16733, Dec, 1981, pp. 1657-
1676.
9. "Steel Sheet Piling Cellular Cofferdams on Rock," U.S. Tennessee Valley,
TVA Technical Monograph No. 75, Vol. 1, Dec, 1957.
10. Swatek, E. P., "Cellular Structure Design and Installation," Proceedings of the
Conference on Design and Installation of Pile Foundations and Cellular Structures,
Envo Publishing Co., Inc., Pennsylvania, 1970, pp. 413-423.

APPENDIX II.—NOTATION

The following symbols are used in this paper:

a = nondimensional length of arc-cell wall;


/ = function u^ed in stating Newton-I^aphson formula;
k = factor limiting Newton-Raphson step size;
L = distance between symmetry boundaries;
m = nondimensional length of main-cell wall;
p = net pressure acting on wall;
q = ratio of arc-cell to main-cell pressure;
R = radius;
r = ratio of common-wall to main-cell pressure;
s = arc length;
t = tension in wall;
w = nondimensional length of common-wall;
x = coordinate direction parallel to cofferdam centerline;
y = coordinate direction transverse to cofferdam centerline; a n d
0 = angle subtended by portion of cell wall.
1457

J. Geotech. Engrg. 1984.110:1446-1458.


Downloaded from ascelibrary.org by University of Leeds on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.

Subscripts
a = arc cell;

w = common-wall.
m = main cell; and
i = iteration counter;

1458

J. Geotech. Engrg. 1984.110:1446-1458.

You might also like