You are on page 1of 15

General Axisymmetric Active Earth Pressure by Method

of Characteristics—Theory and Numerical Formulation


Y. M. Cheng1; Y. Y. Hu2; and W. B. Wei3
Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Active pressure for axisymmetric problem under general conditions is formulated and solved by an iterative finite difference
solution of the characteristic equations in the present paper. A general lateral stress coefficient is used instead of the Haar-von Karman
hypothesis and numerical modeling has also confirmed that Harr-von Karman hypothesis is a reasonable assumption for the present
problem if sufficient wall movement is allowed. The shape of the failure zone and variation of active pressure with depth and wall friction
are investigated. Some interesting results on the active pressures arising from the arch action are found. Principle of superposition of the
effect of soil weight, surcharge, and cohesive strength is discussed. Finally, the results in the present study are applied to bore pile
construction and it is demonstrated that no casing for construction is required under some conditions.
DOI: 10.1061/共ASCE兲1532-3641共2007兲7:1共1兲
CE Database subject headings: Axisymmetry; Earth pressure; Numerical models.

Introduction ciable. For circular excavation, active pressure has been studied
by Berezantzev 共1958兲 and Prater 共1977兲 using simplified slip line
The plane strain lateral earth pressure problem has been solved by solution and limit equilibrium method, respectively. Both of their
Sokolovskii 共1965兲, Kerisel and Absi 共1990兲, and Cheng 共2003兲 solutions are limited to c or ␾ soil only and the failure profile is
using the method of characteristics. In general, the active/passive assumed to be a cone with straight line boundary. Further, their
pressures by Chen 共1975兲 using limit analysis are smaller/greater solutions are also limited to excavation with level ground with
than the corresponding solutions by the method of characteristics zero wall friction. Mroz 共1967兲 has discussed the analysis of cor-
and Chen’s results are not commonly used for design. Extension ner and side regimes in axisymmetric plastic flow corresponding
of the lateral earth pressure solution to the earthquake problem to the edges and sides of the yield surface and has developed a
has been achieved by Kumar 共2001, 2002兲 for an inclined retain- graphical method for the solution of the axisymmetric flow prob-
ing wall with level backfill whereas Cheng 共2003兲 has developed lem. Houlsby and Wroth 共1982兲 have obtained lateral earth pres-
a rotation of axes technique for a general plane strain retaining sure for a simple smooth axisymmetric retaining wall with level
wall problem. ground for cohesive soil and cohesionless soil by the method of
Circular deep excavation is quite commonly used for ventila- characteristics. Drescher 共1983, 1986兲 has provided a detailed
tion shaft and access shaft. For circular foundation, circular exca- treatment of the stability of vertical circular cut by the method of
vation, or similar construction, the stress states are axisymmetric characteristics for level ground. Drescher 共1985兲 has also ex-
in nature. At present, some engineers still adopt plane strain ac- tended the works to the nonaxisymmetric problem for simple con-
tive pressure coefficient in the analysis and design but this ap- ditions. Jenike and Yen 共1962兲 have proposed solutions for the
proach is obviously not correct and conservative. It is well known ultimate active pressure for “rat-hole” which are extended by Hill
that due to the arch action developed within soil mass, active and Cox 共2000兲. With reference to the vertical force equilibrium
pressure for circular shaft will attain a maximum value with in- differential equation by Hill and Cox 共2000兲 which is actually
creasing depth which is greatly different from that in plane strain equal to Eq. 共1兲 in the present paper with the vertical stress ␴z
results. When the depth of excavation is great, the discrepancy assumed to be constant so that ␴z is absent in the equilibrium
between plane strain and axisymmetric results will be appre- differential equation. The solutions to the rat-hole by Jenike and
Yen 共1962兲 or Hill and Cox 共2000兲 represent only the ultimate
1
Associate Professor, Dept. of Civil and Structural Engineering, condition which is stabilized at a relatively great depth but the
Hong Kong Polytechnic Univ., Hunghom, Hong Kong. E-mail: pressure distribution at the top of the circular shaft is not deter-
ceymchen@polyu.edu.hk mined. It will also be demonstrated in a later part of this paper
2
Assistant Professor, Dept. of Civil Engineering, Zhejiang Univ., that if the angle of shearing resistance is low, the formation of
Zhejiang, China. E-mail: huyayuan@zju.edu.cn stable hole is not possible and the rat-hole solutions are not ap-
3
Ph.D. Student, Dept. of Civil and Structural Engineering, Hong plicable under such conditions.
Kong Polytechnic Univ., Hunghom, Hong Kong. E-mail: 05900931r@ For the more general case where the ground surface is not
polyu.edu.hk level and wall friction ␦ exists, a more rigorous solution of these
Note. Discussion open until July 1, 2007. Separate discussions must
general conditions is difficult and is still outstanding at present. In
be submitted for individual papers. To extend the closing date by one
month, a written request must be filed with the ASCE Managing Editor. particular, when the effects of friction angle, cohesive strength,
The manuscript for this paper was submitted for review and possible and surcharge are coupled, the individual result from these effects
publication on May 31, 2005; approved on April 7, 2006. This paper is is not strictly additive. The slip lines are also more complicated
part of the International Journal of Geomechanics, Vol. 7, No. 1, than the corresponding plane strain condition and have not been
February 1, 2007. ©ASCE, ISSN 1532-3641/2007/1-1–15/$25.00. studied in detail in the past. For axisymmetric lateral earth prob-

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007 / 1

Int. J. Geomech. 2007.7:1-15.


Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Cylindrical coordinate system and stress components

lem, a system of hyperbolic differential equations can be obtained


from equilibrium condition and Mohr-Coulomb yield conditions.
The families of characteristic equations for the axisymmetric
problem can be solved by an iterative finite difference method as
proposed in the present paper. Using the simplified assumptions
similar to that by Berezantzev 共1958兲, approximate analytical for-
mulas are also developed in the present study. From the results of
analysis, many interesting results for axisymmetric problems
which differ greatly with the plane strain solution are obtained in Fig. 2. 共a兲 Mohr circle under failure condition; 共b兲 sign convention
the present paper. and notation

Method of Characteristics for the Axisymmetric ⳵␴r ⳵␶rz ␴r − ␴␪ ⳵␶rz ⳵␴z ␶rz
Problem + + = 0, + + =␥ 共1兲
⳵r ⳵z r ⳵r ⳵z r

Shield 共1955兲 first solved the axisymmetric plastic flow for Tresca Hill and Cox 共2000兲 have assumed the condition of stable hole
共cohesive兲 material. Cox et al. 共1961兲 obtained the solutions of at great depth so that Eq. 共1兲 can be simplified by taking ␴z as
axisymmetric footing for a general c-␸ soil with no weight constant. The four stress components can be expressed in terms of
whereas Cox 共1962兲 included self-weight in the analysis for the the mean stress ␴ and the inclination angle ␺ which is formed by
circular footing problem. Based on the concept by Shield 共1955兲 extending the major principal stress to the r axis 关Fig. 2共b兲兴 as
and Cox et al. 共1961兲, Houlsby and Wroth 共1982兲 have obtained
␴r = ␴共1 + sin ␸ cos 2␺兲 − c cot ␸
lateral earth pressure for a simple smooth axisymmetric retaining
wall. Drescher 共1983, 1986兲 has provided a detailed treatment of ␴z = ␴共1 − sin ␸ cos 2␺兲 − c cot ␸ 共2a兲
the stability of vertical circular cut by the method of characteris-
tics for level ground but the effects of wall friction and inclined ␶rz = ␴ sin ␸ sin 2␺, ␴␪ = k␴1 = k␴共1 + sin ␸兲 − kc cot ␸
ground are not considered and the slip line profiles are also not
共2b兲
studied in detail. Bolton et al. 共1993兲 studied the vertical bearing
capacity factors for circular and strip footing and confirmed that where ␴ = 共␴1 + ␴3兲 / 2 + c cot ␸; c = cohesive strength; and k = ratio
Terzaghi’s approach to the superposition of bearing capacity between ␴␪ and ␴1. k has been taken as 1.0 by most researchers in
terms containing Nq, N␥, Nc is both safe and sufficiently accurate the past for simplicity, which is known as the Haar-von Karman
for circular footing. For axisymmetric active pressure determina- hypothesis, but this ratio should lie somewhere between 1 and
tion, Berezantzev 共1958兲 has adopted a simplified slip line active pressure coefficient ka theoretically 共to be derived later兲
analysis assuming straight line failure zone in determining axi- and should be taken as a value between k0 共at-rest兲 and 1.0 in
symmetric active pressure. All the previous studies on circular general. Prater 共1977兲 has considered both k = k0 and k = ka in the
footing or lateral earth pressure using the method of characteris- analysis with limit equilibrium method, where k0 = at-rest earth
tics are based on lateral stress coefficient k = 1, which will be pressure condition. Berezantzev has adopted the Harr-von Kar-
discussed later. In this paper, an axisymmetric problem will be man hypothesis in his simplified slip line solution of the axisym-
formulated on a more rigorous basis where k can be any arbitrary metric active pressure equation. Jenike and Yen’s approach 共1962兲
value. implicitly implied that the intermediate principal stress is equal to
The equilibrium equations for a toroidal element 共Fig. 1兲 can the minor principal stress whereas Hill and Cox have further con-
be written in cylindrical coordinates r, ␪, z as sidered the case of k = 1.0. Hill and Cox 共2000兲 also mentioned

2 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007

Int. J. Geomech. 2007.7:1-15.


that their solutions 共extensions of works by Jenike and Yen兲 are
also not applicable to those plastic regimes where k is not 1.0 or
intermediate principal stress is not equal to the minor principal
␺=

2
= const, m = 1, dZ = tan冉 3␲ ␸
4

2

dR 共8兲

stress. During shaft excavation, the radial stress will decrease,


Putting Eq. 共8兲 and m = 1 into Eq. 共4兲, we obtain
whereas the tangential stress will increase. The assumption of

冋 册
intermediate principal stress equal to the minor principal stress by
Jenike and Yen 共1962兲 or Hill and Cox 共2000兲 is not a reasonable d⍀ 共1 + sin ␸兲2 ⍀ C共1 − k兲 1 + sin ␸ 1
− k −1 − =−
assumption which tends to overestimate the ultimate active pres- dR cos ␸
2
R R sin ␸ cos ␸ cos ␸
sure. Drescher 共1986兲 has also obtained the solution using the 共9兲
Harr-von Karman hypothesis. The writers will demonstrate in a
Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

later section by numerical modeling that the Harr-von Karman


Let
hypothesis is a good assumption in general.

冉 冊
The characteristic lines for the solution are the ␣ and ␤ lines
on which the shear strength is fully mobilized. In literature they 共1 + sin ␸兲2 ␲ ␸
␩=k − 1 = k tan2 + −1
are often referred to slip lines, but their significance is related to cos2 ␸ 4 2
the equations of equilibrium instead of displacement. The geom-
etry dictates the slopes of the slip lines which can be written as ␰=
1−k

tan2
␲ ␸
+
4 2
冉 冊
+1 共10兲
dz
= tan共␺ + m␮兲 共3兲 Putting Eq. 共10兲 into Eq. 共9兲, the solution of differential equa-
dr tion 共9兲 is
where ␮ = ␲ / 4 − ␸ / 2 and m takes the value −1 for an ␣ line and
+1 for a ␤ line 关shown in Fig. 2共b兲兴. These equilibrium and yield R C共1 − k兲共1 + sin ␸兲
equations form a set of hyperbolic partial differential equations ⍀ = aRn + − 共11兲
共␩ − 1兲cos ␸ ␩ sin ␸ cos ␸
that will reduce to two ordinary differential equations expressing
the changes in stress along each characteristic line in terms of the In Eq. 共11兲, a is an integration constant which can be deter-
changing inclination ␺ and position 共r , z兲 mined from the boundary condition. Eqs. 共2a兲 and 共2b兲 are nor-
malized with the excavation radius r0 and reduce to the following
共1 − k − k sin ␸兲␴ − c共1 − k兲cot ␸ forms with Eq. 共8兲:
d␴ + m2␴ tan ␸ d␺ + dr
r
k共1 + sin ␸兲␴ + c共1 − k兲cot ␸ ⍀R = ⍀共1 − sin ␸兲 − C cot ␸, ⍀Z = ⍀共1 + sin ␸兲 − C cot ␸
+m tan ␩ dz 共12a兲
r
= ␥共m tan ␸ dr + dz兲 共4兲
⍀␪ = k⍀共1 + sin ␸兲 − kC cot ␸ 共12b兲
To simplify the study, the variables will be normalized with the
circular excavation radius r0, which take the form According to Eqs. 共12a兲, 共12b兲, and 共11兲, the components of
stress tensor can be expressed as follows:
␴ r z c q0
⍀= , R= , Z= , C= , Q0 = 共5兲
r 0␥ r0 r0 r 0␥ r 0␥ 共1 − sin ␸兲 共1 + ␩ − k兲
⍀R = a共1 − sin ␸兲R␩ + R − cot ␸ C
where r0 = radius of the circular excavation; z = vertical depth 共␩ − 1兲cos ␸ ␩
under consideration; ␥ = unit weight of soil; and q0 = external sur- 共13a兲
charge behind the circular excavation. Substituting Eq. 共5兲 into
Eqs. 共3兲 and 共4兲, we have
共1 + sin ␸兲
dZ ⍀Z = a共1 + sin ␸兲R␩ + R − ␰C cot ␸ 共13b兲
= tan共␺ + m␮兲 共6兲 共␩ − 1兲cos ␸
dR
where ␮ = ␲ / 4 − ␸ / 2
共1 + sin ␸兲
共1 − k − k sin ␸兲⍀ − C共1 − k兲cot ␸ ⍀␪ = ak共1 + sin ␸兲R␩ + kR − ␰kC cot ␸ 共13c兲
d⍀ + m2⍀ tan ␸ d␺ + dR 共␩ − 1兲cos ␸
R
k共1 + sin ␸兲⍀ + C共1 − k兲cot ␸ Here Rb denotes the intersect of the ␤ slip line which passes
+m tan ␸ dZ through the point 共1 , Z兲 with R axis, so
R
= m tan ␸ dR + dZ 共7兲
Rb = 1 + Z tan 冉 冊
␲ ␸

4 2
共14兲

Simplified Slip Line Solution Based Since 兩⍀z兩R=Rb = Q0, the integral constant a can be determined as
on Berezantzev’s Assumption
Q0 + ␰C cot ␸ 1
If we assume the slip lines to be straight lines in the R-Z plane, a= − 共15兲
R␩b 共1 + sin ␸兲 共␩ − 1兲cos ␸R␩−1
which are the assumptions as used by Berezantzev 共1958兲, the b

inclination of ␤ slip line is Put Eq. 共15兲 into Eqs. 共13a兲–共13c兲 and notice that

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007 / 3

Int. J. Geomech. 2007.7:1-15.


1 − sin ␸
1 + sin ␸
= tan2
␲ ␸

4 2
, 冉 冊 1 − sin ␸
cos ␸
= tan
␲ ␸

4 2
冉 冊 Eq. 共20兲 is the same as Berezantzev’s 共1958兲 original formula
for axisymmetric active pressure. Eqs. 共17兲 and 共18兲 are more
general in that k is now a variable. The simplified slip line solu-
Eq. 共13兲 will be simplified to tion actually violates the yield and limiting equilibrium condition

冉 冊
and the stress field is hence not statically admissible. Another
␲ ␸ major limitation of Eq. 共20兲 is that superposition of the effects
冋 冉冊 册 冉冊 冉 冊
tan − ␩−1 ␩
4 2 R R ␲ ␸ from friction angle, cohesive strength, and surcharge is assumed
⍀R = R 1− + Q0 tan2 −
␩−1 Rb Rb 4 2 which is, however, not strictly correct. When the depth is small so

冋 冉 冊 冉 冊册 ␩
that the problem is not far from plane strain condition, the error
1−k+␩ R ␲ ␸ by Eq. 共20兲 is, however, relatively small as compared with the
Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

−C −␰ tan2 − cot ␸ 共16a兲


␩ Rb 4 2 more rigorous numerical analysis of the axisymmetric character-
istic equation and can be viewed as a conservative approximation

冉 冊
␲ ␸ of the active earth pressure. When the depth is small, Eq. 共20兲 will

冋 冉冊 册 冉冊
tan + ␩−1 ␩ also reduce to the classical plane strain active pressure equation.
4 2 R R
⍀Z = R 1− + ⍀0 As axisymmetric active earth pressure must be smaller than or
␩−1 Rb Rb equal to plane strain active pressure, for the first term of Eq. 共18兲,

+ C␰ 冋冉 冊 册
R
Rb

− 1 cot ␸ 共16b兲
the quantity ␩ must satisfy

␩−1艌−1 共21兲
␲ ␸
冉 冊
冋 冉冊 册 冉冊
tan + ␩−1 ␩ Putting Eq. 共21兲 in Eq. 共10兲, we obtain
4 2 R R
⍀␪ = kR 1− + k⍀0
␩−1 Rb Rb

+ k␰C 冋冉 冊 册
R
Rb

− 1 cot ␸ 共16c兲 冉
k 艌 tan2 45 −

2

= Ka 共22兲

In Eq. 共16a兲, put R = 1, the normalized active earth pressure is


given by Hence the lateral stress ratio k will lie between 1.0 and Ka. Using

冉 冊
Eqs. 共17兲 and 共18兲, the upper and lower estimates of the active
␲ ␸

冉 冊 冉 冊
tan − pressure can be determined by using k = 1.0 and Ka.
4 2 1 1 ␲ ␸ According to the first term of Eq. 共19兲 and Eq. 共10兲, it can be
tan2
concluded that if ␸ ⬎ 2 arctan冑2 / k − ␲ / 2 at specified value k,
Pa = 1− + Q0 −
␩−1 R␩−1 R␩b 4 2

冋 冉 冊册
b
the active earth pressure induced by soil weight tends to a limiting
1−k+␩ ␰ ␲ ␸
−C − ␩ tan2 − cot ␸ 共17兲 value as z tends to infinity. If ␸ 艋 2 arctan冑2 / k − ␲ / 2, the active
␩ Rb 4 2 pressure will tend to infinity as z tends to infinity. When k = 1,
The actual active pressure can then be obtained by Eq. 共5兲 as the critical angle ␸ is approximate 19.5° whereas the general
critical angle ␸ will increase with decreasing k in accordance with

冉 冊
␲ ␸ ␸ = 2 arctan冑2 / k − ␲ / 2.

冉 冊 冉 冊
tan −
4 2 1 1 ␲ ␸
pa = ␥r0 1− + q0 tan2 −
␩−1 R␩−1 R␩b 4 2

冋 冉 冊册
b
Numerical Solution of Axisymmetric Slip
1−k+␩ ␰ ␲ ␸ Line Equations
+c − ␩ tan2 − cot ␸ 共18兲
␩ Rb 4 2
where The limitations of the simplified Eqs. 共17兲 and 共18兲 are
1. Level backfill is assumed,

冉 冊 ␲ ␸ 2. Wall friction is neglected,

冉 冊 冉 冊
r0 + z tan − 3. Slip lines are straight lines, and
␲ ␸ z ␲ ␸ 4 2
Rb = 1 + Z tan − = 1 + tan − = 4. Coupling effects between weight of soil, external surcharge,
4 2 r0 4 2 r0 and cohesive strength of soil are neglected.
When k = 1, Eq. 共10兲 becomes Houlsby and Wroth 共1982兲 and Drescher 共1983, 1986兲 have

冉 冊 冉 冊
adopted the method of characteristics for the solution of the
␲ ␸ ␲ ␸ simple axisymmetric problem but the effects of inclined ground
␩ = tan2 + − 1 = 2 tan ␸ tan + , ␰=1 共19兲 and wall friction are not considered. The limitations of the previ-
4 2 4 2
ous works by various researchers can be overcome by the use of
and Eq. 共18兲 becomes a more rigorous solution of Eqs. 共6兲 and 共7兲 developed in the

冉 冊
␲ ␸
present paper which can be written in finite difference form for

冉 冊 冉 冊
tan − iterative analysis. For a point W 共RW , ZW , ⍀W , ␺W兲 as shown in
4 2 1 1 ␲ ␸ Fig. 3 共a typical Riemann type problem兲, which is the intersection
pa = ␥r0 1− + q0 tan2 −
␩−1 R␩−1 R␩b 4 2 point of an ␣ line which passed through a known “point”

冋 冉 冊 册
b
P共R P , Z P , ⍀ P , ␺ P兲 and a ␤ line which passed through another
1 ␲ ␸
+c tan2 − − 1 cot ␸ 共20兲 known point Q 共RQ , ZQ , ⍀Q , ⌿Q兲, the following substitutions can
R␩b 4 2 be derived:

4 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007

Int. J. Geomech. 2007.7:1-15.


A = 共␺ p − ␺Q兲tan ␸ D = ␺ P + ␺Q

+ 冋 k共sin ␸ + 1兲tan ␸共Zw − ZQ兲 + 共1 − k − k sin ␸兲共RW − RQ兲


RW + RQ
− 冋 k共sin ␸ + 1兲tan ␸共ZW − ZQ兲 + 共1 − k − k sin ␸兲共RW − RQ兲
共RW + RQ兲tan ␸


k共sin ␸ + 1兲tan ␸共ZW − Z P兲 − 共1 − k − k sin ␸兲共RW − R P兲
RW + R P
册 −
k共sin ␸ + 1兲tan ␸共ZW − Z P兲 − 共1 − k − k sin ␸兲共RW − R P兲
共RW + R P兲tan ␸

共23a兲 共23d兲

B = A*共⍀ P + ⍀Q兲 + tan ␸共RQ − R P兲 − 共2ZW − Z P − ZQ兲 − 2C共1 − k兲 E = Z P − ZQ + tan ␸共2RW − R P − RQ兲 − 2C共1 − k兲
Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

⫻ 冋 RW − R P
RW + R P
cot ␸ +
ZW − Z P RW − RQ

RW + R P RW + RQ
cot ␸ +
ZW − ZQ
RW + RQ
册 ⫻ 冋 共RW − R P兲cot ␸ + ZW − Z P 共RW − RQ兲cot ␸ − ZW + ZQ
RW + R P
+
RW + RQ

共23b兲 共23e兲


The finite difference equations for point W can be written as
C1 = 共A − 2兲*⍀*p⍀Q + ⍀Q tan ␾共RW − R P兲 − 共ZW − Z P兲 − 2C共1 − k兲

␺W + ␺ P

− ␮ − RQ tan 冉
␺W + ␺Q
+␮ 冊
冉 冊册 冋
ZQ − Z P + R P tan
2 2

冉 冊 冉 冊
RW − R P ZW − Z P RW =
⫻ cot ␾ + − ⍀ P tan ␸共RW − RQ兲 ␺W + ␺ P ␺W + ␺Q
RW + R P RW + R P tan − ␮ − tan +␮

冉 冊册
2 2
RW − RQ ZW − ZQ
+ 共ZW − ZQ兲 + 2C共1 − k兲 cot ␸ − 共23c兲 共24a兲
RW + RQ RW + RQ

ZW =
ZQ + Z P + 共RW − R P兲tan 冉 ␺W + ␺ P
2

− ␮ + 共RW − RQ兲tan
2

␺W + ␺Q
+␮ 冊 共24b兲
2

B 冑B2 − 4共A + 2兲C1 found from S and K1, K3 is found from K1 and K; L is found
⍀W = + 共24c兲 from K3 and K2. S, K2, and L become the known boundary
2A + 4 2兩A + 2兩
conditions for the fan zone.
2. Fan zone—Node S can be viewed as a degenerate ␤ charac-
D 共⍀ P − ⍀Q兲共A − 2兲 + 2E teristic with unique R and Z, but varying ␺ and ⍀. The first
␺W = + 共24d兲
2 2共2⍀W + ⍀ P + ⍀Q兲tan ␸ two terms of Eq. 共4兲 demonstrate that the significant effect in
the fan zone is that stress increases exponentially with rota-
Refer to Fig. 3 where points P and Q are known and W is to be
tion ⌬␺ of the ␤ line from K2 to L1. For example, L1 is
determined. RW and ZW can be found from Eqs. 共24a兲 and 共24b兲.
found from S and K2. Similarly, L2 is found from L1 and L.
⍀W and ␺W can then be calculated from Eqs. 共24c兲 and 共24d兲 by
S, L3, and M now become the known boundary conditions
using an initial trial of ␺W = 共␺ P + ␺Q兲 / 2 共assuming characteristics’
for the passive zone.
lines are straight兲. However, in general, the characteristics’ lines
3. Passive zone—On the side of the circular wall, R is known
are curved and the solution can be improved by updating ␺W
and the value of ␺, being a function of mobilized friction on
calculated from Eq. 共24d兲 and repeating the iteration analysis
the excavation wall, can be determined from the wall condi-
until convergence is achieved 共within a specified tolerance of
tion. The mobilization of friction depends on the wall friction
0.1% by the writers兲.
angle which is assumed to be known. The solutions for ⍀
For the general problem as shown in Fig. 4, the ground line KS
and Z at M1 can therefore be determined from the boundary
makes an angle ␣ with horizontal direction. Proceeding from the
condition at L3. M2 can then be found from M1 and M.
known boundary condition 共external normal surcharge q0兲 on the
Finally, ⍀ and Z can be found at T using M2 together with
free surface KS 共Fig. 4兲, the entire stress field within KLMT can
the known values of R and ␺ at T. At this stage, all the
be determined. First, the boundary KS is subdivided to a set of
variables on the circular wall are known. The active pressure
equally spaced points. The solution then marches in toward the
at excavation wall can then be determined according to the
wall boundary. The whole calculation iterates until an acceptable
following procedure:
accuracy is achieved. A very coarse mesh applied to a circular
excavation for calculating active pressure case is shown in Fig. 4, sin ␦ sin共⌬ − ␦兲
which will be used to illustrate the marching scheme used in the ⌬ = arcsin , P=⍀ 共25兲
sin ␸ sin ⌬
computer program WELLKA developed in this paper.
Steps for solution: where ␦ = friction angle between soil and the wall; and
1. Active zone—This is a Cauchy type boundary condition and ⍀⫽mean stress at the wall. The dimensionless active earth
the boundary conditions at S, K1, and K are known. K2 is pressure Pa and the actual active earth pressure pa are

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007 / 5

Int. J. Geomech. 2007.7:1-15.


Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Computation scheme of new point W from known points


P and Q

Pa = 冑共P cos ␦ − C cot ␾兲2 + 共P sin ␦兲2, Pa = r0␥Pa 共26兲


Using the technique as developed, the effect of friction angle,
wall friction and cohesive strength, backfill angle as well as ex-
ternal surcharge can all be determined for general problems,
which is a major extension of the works by Houlsby and Wroth
共1982兲. The applicability of the present technique will be illus-
trated by a comparison with the results from simplified slip line Fig. 5. 共a兲 Variation of active pressure pa with depth z for different
analysis. radii; 共b兲 variation of dimensionless active pressure Pa with
If the soil follows associated flow rule, the stress characteristic dimensionless depth Z for different radii
lines, velocity characteristic lines, and slip lines will coincide and
the present method will fully satisfy the complete failure require-
ment. The iterative technique as presented in this paper considers ics aspects of the failure mechanism, which are clearly the limi-
only the stress characteristic but not the flow rule or the kinemat- tations of the present technique. The present technique is hence
strictly limited to soil obeying associated flow rule.

Numerical Solution of Active Earth Pressure


for Circular Excavation Wall

The above-derived method of characteristics allows the solution


of active pressure for circular excavation taking surcharge stress
and soil weight simultaneously into account and the analysis is
coded into a program WELLKA by the writers. In order to con-
firm that the set of variables as discussed before can be normal-
ized against radius r0, active earth pressure is first calculated with
r and z and then normalized by radius r0. The results of analysis
with different radii are shown in Fig. 5共a兲. After normalization,
the results are shown in Fig. 5共b兲. Obviously, the results
in Fig. 5共a兲 can be normalized to a single curve as shown in
Fig. 5共b兲 and the normalization as used in Eq. 共5兲 is validated.
Fig. 6 gives ␣ and ␤ characteristics’ loci for a circular wall
with horizontal backfill, ␸ = 40° and ␦ = 0° when k = 1. The trend
of the characteristic loci for k = 0.5 is basically similar to the case
of k = 1.0 and is hence not shown. The ␣ and ␤ characteristics’
lines are approximately straight at shallow depth, which are
similar to that in plane strain condition. In fact, the writers have
compared the loci of these two lines under plane strain and axi-
symmetric conditions and the angle of the exit of ␤ lines are
Fig. 4. A typical stress characteristic mesh practically the same at the top 共very minor difference comes from

6 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007

Int. J. Geomech. 2007.7:1-15.


Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Variation of Pa with depth for different k at ␸ = 40°

depth whereas the horizontal extent of the slip lines due to sur-
Fig. 6. Slip lines for ␸ = 40°, ␦ = 0°, and k = 1
charge maintain a relatively constant value over a major depth.
In order to assess the applicability of the simplified analytical
truncation errors in analysis兲. This result is expected because solutions, results from WELLKA are compared with those from
when depth is shallow, the problem will reduce to a plane strain Eq. 共17兲 for ␸ = 40° and the comparison is shown in Fig. 7. When
problem. The value of ka␥ obtained from the present analysis is dimensionless depth Z is small and ␤ characteristics are approxi-
also found to be independent of k and is equal to the plane strain mate straight lines, the numerical solution should be almost equal
active pressure. Such results have partially demonstrated the ap- to analytical solution and this is confirmed in Fig. 7. If dimen-
plicability of the present numerical technique in axisymmetric sionless depth Z becomes greater, ␤ characteristics will not be a
problem. straight line and the analytic value will deviate from the more
When the depth is great, the ␣ and ␤ characteristics’ lines accurate slip line solution. The slip line solution is always smaller
are curved and the exact loci are controlled by both the unit than the analytical value 共on the safe side兲 and the differences
weight of soil and the dimensionless radius. Berezentzev 共1958兲 become more important if dimensionless depth Z is large. Since
has assumed the characteristics’ lines to be straight lines but it the differences between the analytical solution and the slip line
is clear that the complete characteristics’ lines for axisymmetric solution are relatively small if Z is not great, ka␥ by Eq. 共17兲 can
active pressure involve both concave and convex curves, which be adopted for general design whereas the more rigorous solution
is different from the corresponding plane strain solution. From should be adopted if the depth of excavation is great.
Figs. 6共a and b兲, it is noticed that there are some minor differ-
ences in the curvature of the slip lines. The horizontal extent of
the slip lines due to self-weight of soil decreases gradually with Variations of Active Pressure Coefficients

The axisymmetric active pressure coefficient ka␥ for self-weight


of soil can be considered by assuming q0 = 0 and c = 0 in the

Fig. 7. Comparison between slip line and analytical solutions for


␸ = 40°. 共Note: 1,2 denote numerical solution by WELLKA and that
by Berezantzev for k = 1.0; 3,4 denote numerical solution by Fig. 9. Variation of active pressure coefficient ka␥ with dimensionless
WELLKA and that by Berezantzev for k = 0.5.兲 depth Z for different backfill inclination angle ␣ at ␸ = 40°

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007 / 7

Int. J. Geomech. 2007.7:1-15.


Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Variation of ratio of active earth pressure coefficients Fig. 12. Variation of dimensionless Pa with dimensionless depth Z
⬘ with dimensionless depth Z at ␸ = 40°
ka␥ / ka␥ for different soil friction angle 共␸ ⬍ 20° 兲, k = 1.0

analysis. In Fig. 8, the relations of Pa with depth Z at ␸ = 40° for ⬘ with dimensionless depth Z
In Fig. 10, the variation of ka␥ / ka␥
k ranging from Ka to 1.0 are shown. It should be pointed out that is plotted at ␸ = 40° for various backfill angle ␣. As shown in
the lower bound Ka is actually not possible as k should be greater Figs. 9 and 10, when Z → 0, the ratio of the coefficient of active
than k0. Furthermore, k equal to k0 is equivalent to yielding in ⬘ is 1
earth pressure for circular excavation to plane strain ka␥ / ka␥
transverse direction, which is not possible. The lower bound Ka is and is independent of k, which is the expected result. The results
adopted for k as this bound is adopted by Prater 共1977兲. It is as shown has also demonstrated the accuracy of the program
concluded that the active earth pressure becomes greater as k is WELLKA. The effect of ␣ is more pronounced at the top of the
smaller. k is hence an important factor in the determination of excavation but will gradually be masked out by the circular effect.
axisymmetric active pressure. The coefficient of active earth pres- k has also played a very important role if k is large. The effect of
sure for the self-weight of soil under circular excavation can be ␣ will be negligible if the dimensionless depth Z is large. If k is
defined as small, the effect of ␣ will be important over a much greater depth.
In fact, the effect of k is greater than that for ␣.
pa Pa In Figs. 11 and 12, the dimensionless pressures Pa are plotted
ka␥ = = 共27兲 against depth Z for friction angle ␸ ranging from 15 to 40° at
z␥ Z
k = 1. The results are plotted in Figs. 11 and 12 as there are major
In Fig. 9 the coefficient of active earth pressure ka␥ is plotted differences between the results when ␸ is greater than or smaller
against dimensionless depth Z at ␸ = 40° for backfill inclination than 20°. In Fig. 11 where ␸ ⬎ 20°, when Z tends to infinity, Pa
angle ␣ = – 30°, respectively. The coefficient of active earth pres- tends to a maximum constant at a relatively small Z. The Z value
sure ka␥ for plane strain at c = 0, q0 = 0 is where Pa approaches a constant maximum magnitude decreases
rapidly with increasing ␸. When the soil friction angle is smaller
cos2 ␸

冋 冑 册
⬘ =
ka␥ 2 共28兲 than approximate 19.5° as shown in Fig. 12, Pa will diverge with
sin ␸ sin共␸ − ␣兲 Z. In Figs. 13 and 14, the active earth pressure coefficients ka␥
1+
cos ␣ instead of Pa are also plotted against depth Z for different soil

Fig. 13. Variation of active pressure coefficient ka␥ with


Fig. 11. Variation of dimensionless Pa with dimensionless depth Z dimensionless depth Z for different soil friction angle 共␸ ⬎ 20° 兲,
for different soil friction angle 共␸ ⬎ 20° 兲, k = 1.0 k = 1.0

8 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007

Int. J. Geomech. 2007.7:1-15.


Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 14. Variation of active pressure coefficient ka␥ with Fig. 16. Variation of active pressure coefficient kaq with
dimensionless depth Z for different soil friction angle 共␸ ⬍ 20° 兲, dimensionless depth Z for different soil friction angle, k = 1.0
k = 1.0

friction angle. When Z tends to infinity, ka␥ will tend to zero. The active pressure coefficients are required to be incorporated into
product of ka␥␥z will attain a maximum constant value Pa if ␸ is computer analysis.
greater than 19.5° but will diverge if ␸ is smaller than this value. To consider the effect of surcharge, assume a problem where
When Z is large and the soil friction angle is smaller than 19.5°, ␥ = 0 and c 共or C兲 = 0. In Fig. 16, the coefficient kaq of active earth
Pa is approximately proportional to Z␤, where ␤ is less than 1.0 pressure is plotted against depth Z for different soil friction angle.
and is a small value if soil friction angle is great. If soil friction When the dimensionless depth Z tends to infinity, the coefficient
angle is greater than 19.5° so that Pa attains a maximum constant kaq of the active earth pressure induced by the surcharge tends to
with increasing Z, ka␥ can be approximated by the following func- zero, which is drastically different from that in plane strain con-
tion 共for k = 1兲: dition. kaq can also be fitted by Eq. 共30兲 as 共for k = 1兲

1 + a␥1Z + a␥2Z2 . . . Pa 1 + aq1Z + aq2Z2 . . .


ka␥ ⬇ Ka␥ 共29兲 = kaq ⬇ Kaq 共30兲
1 + b␥1Z + b␥2Z2 + b␥3Z3 . . . Q0 1 + bq1Z + bq2Z2 + bq3Z3 . . .

where Ka␥ = plane strain active earth pressure coefficient. Eq. 共29兲 where Kaq = active pressure coefficient under plane strain condi-
is a good fit for the active pressure coefficient determined by tion. Eq. 共30兲 will tend to plane strain result if Z tends to 0 or
slip line solution, which is demonstrated by the comparisons in limZ→0 kaq = Kaq = tan2共45− ␸ / 2兲. As shown in Fig. 17, Eq. 共30兲 is
Fig. 15. When radius r0 of well is infinity, the dimensionless also a very good fit to the more rigorous results from the method
depth Z is zero and the axisymmetric problem will degenerate to of characteristics and the errors are not more than 3% over a wide
plane strain result. Under this condition, limZ→0 ka␥ = Ka␥ range of Z. The coefficients aq1, bq1, bq2, bq3 are shown in Table 1
= tan2共45− ␾ / 2兲, which is also predicted by the use of Eq. 共29兲. and may be useful for numerical integration purposes.
As shown in Fig. 15, the differences between results from the To consider the effect of cohesive strength, set Q0 = 0 and
method of characteristics and Eq. 共29兲 are not greater than 3% ␥ = 0 in the analysis. In Fig. 18, the coefficients kac of active earth
over a wide range of Z, which are even better than the simplified pressure induced by dimensionless soil cohesion C are plotted
analytical Eq. 共20兲. Eq. 共29兲 will be useful if the total active against depth Z at different friction angle. When the dimen-
pressure force is to be determined by numerical integration or the sionless depth Z tends to infinity, the coefficient kac of active

Fig. 15. Comparison between results from method of characteristics Fig. 17. Comparison between results from method of characteristics
and curve fitting Eq. 共29兲 for ka␥, k = 1.0 and curve fitting Eq. 共30兲 of kaq, k = 1.0

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007 / 9

Int. J. Geomech. 2007.7:1-15.


Table 1. a␥1, b␥1, b␥2 and aq1, bq1, bq2, bq3 Values

共deg兲 a␥1 b␥1 b␥2 aq1 bq1 bq2 bq3
40 0.3941 1.1133 0.5451 0 1.8496 −0.046 0.9701
37.5 0.3105 0.9918 0.3767 0 1.6390 0.1391 0.5505
35 0.2399 0.8779 0.2528 0 1.4499 0.2403 0.2890
32.5 0.1813 0.7714 0.1634 0 1.2809 0.2816 0.133
30 0.1319 0.6716 0.1009 0 1.1293 0.2795 0.0473
28 0.0998 0.5959 0.0653 0 1.0227 0.2456 0.0145
Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

26 0.0716 0.5227 0.0392 0 0.9265 0.1921 0.0012


24 0.0392 0.4436 0.0173 0.0213 0.8604 0.1522 0
22 0.0446 0.3954 0.0141 0.0512 0.8066 0.1256 0
20 0.0535 0.3647 0.0128 0.1088 0.7821 0.1246 0

earth pressure tends to a finite negative constant which is drasti- Cost = kaq − kac tan ␸ 共33兲
cally different from kaq. When the well radius r0 is infinity so
that the circular case degenerates to plane strain case, In Fig. 19 the variable Cost is plotted against the dimensionless
kac = Kac limZ→0 h共Z兲 = limZ→0共Pa / C兲 = −2 tan共45− ␸ / 2兲, which is factor Z for soil angle ␸ = 20– 40°. The upper line in Fig. 19 is
the corresponding coefficient under plane strain condition. Cost for the axisymmetric condition under k = 1 as well as plane
When k = 1, we can define an equivalent hydrostatic pressure strain condition 共practically coincide兲. The values of Cost as
␴e = c / tan ␸ acting on the soil surface and within the whole soil obtained from method of characteristics are practically 1.0
mass and the present problem may then be viewed as a simple and the maximum deviation is less than 1.5%. The minor devia-
constant ␸ case with the origin shifted. The total active earth tion from 1.0 may be due to rounding off error in numerical
pressure Pa due to cohesion would then be equal to kaq␴e, of computation. Eq. 共31兲 is applicable under the condition of k = 1. If
which ␴e is already acting. The extra contribution is therefore k ⫽ 1, Eq. 共31兲 is invalid, which is shown in Fig. 19.
共kaq − 1兲␴e. Using the principle of superposition similar to that in
bearing capacity problem, we can write
Investigation of Stress Factor k
Pa Pa
kac = = cot ␾ = 共kaq − 1兲cot ␾ = 关kaqg共Z兲 − 1兴cot ␾ 共31兲 The stress factor k is an important parameter in the axisymmetric
C ␴e
lateral earth pressure problem and it has been demonstrated pre-
so viously that it cannot be less than the plane strain active pressure
coefficient Ka. During shaft excavation, the radial stress will de-
kac 关kaqg共Z兲 − 1兴cot ␾ crease whereas the tangential stress will increase and the tangen-
h共Z兲 = = tial stress will be a function of the initial condition. k should
Kac − 2 tan共45 − ␾/2兲
hence be greater than k0 and less than or equal to 1.0. In fact, k
关cot共45 − ␾/2兲 − tan共45 − ␾/2兲g共Z兲兴cot ␾ should be an indeterminate value as k should be controlled by the
= 共32兲
2 amount of wall movement. To assess the value of k, it can be
estimated by establishing a self-equilibrium stress state due to
kac in Eq. 共31兲 takes a form very similar to the Nc factor in gravity with a prescribed at-rest earth pressure coefficient and
bearing capacity problem. To verify Eqs. 共31兲 and 共32兲, define excavation is performed by using the program FLAC in the

Fig. 18. Variation of active pressure coefficient kac with Fig. 19. Variation of Cost with dimensionless depth Z for different
dimensionless depth Z for different soil friction angle, k = 1.0 soil friction angle for axisymmetric and plane strain condition

10 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007

Int. J. Geomech. 2007.7:1-15.


Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 21. Effects of ␻ on kaq␥

lim f共z,␥,r0,k兲 = 1, lim g共z,␥,r0,k兲 = 1, lim h共z,␥,r0,k兲 = 1


z→0 z→0 z→0

共35兲
If the variables are normalized according to a scale length of
Fig. 20. Active pressure for different friction angles with sufficient radius r0, Eq. 共35兲 will then be expressed as
wall movement 共radius of shaft= 2 m, k0 = 1 − sin ␾, density of
soil= 20 kN/ m3兲 Pa = Ka␥ f共Z,k兲Z + Kaqg共Z,q0,k兲Q0 + Kach共Z,c,k兲C 共36兲
where
present study. From parametric study, it is found that the active
pressure and k are practically independent of the elastic modulus z c q0
Z= , C= , Q0 =
of soil. Results of numerical modeling for a 2 m radius shaft r0 r 0␥ r 0␥
excavation in cohesionless soil are shown in Fig. 20. The results It will be shown in a later section that q and c have only a small
as shown in Fig. 20 have also demonstrated that below a friction effect on the functions g and h, hence Eq. 共36兲 can be written as
angle of 20°, the overburden stress and active pressure cannot
achieve ultimate constant values which are similar to the results Pa ⬇ ka␥ f共Z,k兲Z + kaqg共Z,k兲Q0 + kach共Z,k兲C 共37兲
from previous study. These numerical results have demonstrated
the effectiveness of the numerical algorithm as developed in the Finally, a more rigorous check on the superposition assump-
present paper. Besides that, the stress factor k is found to be tion will be carried out. When ␸ ⫽ 0°, q0 ⫽ 0, ␥ ⫽ 0, c ⫽ 0,
extremely close to 1.0 共less than 1% difference兲 for all friction Eqs. 共34兲 and 共36兲, which are based on the principle of super-
angles and at different depths of excavation. The writers have position, are going to be verified. According to Eqs. 共31兲 and 共33兲,
carried out numerous numerical modeling assuming normal con- when k = 1, the active earth pressure by soil cohesion can be
ditions where the initial shear stress is zero and found that k is regarded as a surcharge stress, so we need to consider the super-
practically equal to 1.0. However, it should be pointed out that if position of weight and surcharge stress only. Define a dimension-
the wall movement allowed is small, k is actually less than 1.0. In less parameter ␻ as the superposition factor
the extreme case where the wall movement is practically 0, k is Q0
also practically equal to k0, which is less than 1.0. ␻= 共38兲
Z
␻ will be used as an indication of the relative importance of
Principle of Superposition surcharge and self-weight effects. The total active pressure Pa
will be related to a combined active pressure factor given by
The principle of superposition has been used by Terzaghi 共1943兲
for bearing capacity factors in the bearing capacity problem. Chen Pa
kaq␥ = 共39兲
共1975兲 and Kerisel and Absi 共1990兲 have also applied a similar Z + Q0
approach in the determination of plane strain lateral earth pres-
Fig. 21 shows the relation of kaq␥ with ␻ for a smooth circular
sure. The applicability of this practice under axisymmetric condi-
well at ␸ = 40°. It can be seen that kaq␥ remains practically con-
tion will be examined in this section. According to the principle
stant for ␻ ⬎ 102 or ␻ ⬍ 10−3. These two limits correspond to
of superposition, the total active pressure pa will be a combination
surcharge being either small or large. When ␻ is smaller than 10−3
of the effects due to weight of soil ka␥, external surcharge kaq, and
and greater than 102, kaq␥ is practically not affected by ␻, which
cohesive strength of soil kac, which can be expressed as
indicates that the principle of superposition is applicable.
pa = ka␥z␥ + kaqq0 + kacc = Ka␥ f共z,␥,r0,k兲z␥ + Kaqg共z,q,r0,k兲q0 As shown in Fig. 22, when k = 1, the results from a combined
analysis obtained from the coupling analysis are nearly the same
+ Kach共z,c,r0,k兲c 共34兲
as those obtained from direct superposition. When k ⬍ 1, results
where f, g, and h = ratios between axisymmetric and plane strain from the principle of superposition are greater than the results
active pressure coefficients and from a combined analysis, which means that superposition is a

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007 / 11

Int. J. Geomech. 2007.7:1-15.


Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 22. Superposition effects with z for different k at ␸ = 35°, Fig. 24. Variation of coefficient kaq with dimensionless depth Z for
c = 20 kPa, and q = 10 kPa different wall friction, k = 1.0

safe practice to be used for design. It is interesting to note that the


differences between the earth pressure from combined analysis of the resultant active pressure occurs at ␦ = 20°. The results as
and superposition increase with decreasing k. shown in Figs. 23–25 hence comply with the common under-
standing that wall friction reduces active pressure except that we
should use the normal component of the active pressure in
ka, ka␥, and kac for Smooth and Rough Walls consideration.

In previous sections, the results are confined to the case of a


smooth wall. In this section we consider the effect of the wall Circular Excavation without Bracing or Casing
friction angle on the values for ka␥, kaq, and kac. In Figs. 23–25,
the coefficients of active earth pressure with dimensionless In this section we shall apply some of the previous results in the
depth Z for ka␥, kaq, and kac for wall friction angle ␦ = 0 – 30° at study of circular excavation. In previous sections, it is demon-
␸ = 40° are shown. From Figs. 23–25, it is clear that the effects of strated that active earth pressure induced by soil cohesion starts at
wall friction on ka␥, kaq, and kac are small in the circular case, a negative value and ends at a negative constant when the dimen-
which is similar to the corresponding situation in plane strain sionless depth Z tends to infinity. For the active earth pressure
condition. In general, the effect of wall friction has greater influ- induced by the self-weight of soil, it starts at zero on ground
ence on kac but lesser effect on ka␥ and kaq. It should be noted that level. If soil friction angle is greater than approximately 19.5° at
for ka␥ and kaq, the coefficients reduce with ␦ initially but increase k = 1, the active earth pressure Pa induced by self-weight of soil
again after ␦ is greater than 20°. These two coefficients take the will tend to a constant as Z tends to infinity. In stiff soil like
maximum value at ␦ = 40°, which appears to contradict the com- hardened clay or soil derived from weathering of rock, the theo-
mon understanding that wall friction reduces active pressure. If retical active earth pressure can be smaller than zero, which
we examine the normal component of the active pressure, we will means that support to the circular excavation will not be neces-
find that the normal component of the active pressure always sary. This phenomenon is well known in bore pile construction in
decreases with increasing ␦. However, the tangential component stiff clay and dense sand. The axisymmetric active pressure as
of the active pressure increases with ␦ so that the minimum value

Fig. 23. Variation of coefficient ka␥ with dimensionless depth Z for Fig. 25. Variation of coefficient kac with dimensionless depth Z for
different wall friction, k = 1.0 different wall friction, k = 1.0

12 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007

Int. J. Geomech. 2007.7:1-15.


Table 2. Limit Minimal Dimensionless Cohesion Cmin 共at k = 1兲 1 + a␥1Z
f共Z兲 ⬇ ,
␸ Cmin Cmin / tan共␸兲 ␸ Cmin Cmin / tan共␸兲 1 + b␥1Z + b␥2Z2
共deg兲 共deg兲 共41兲

冋 冉 冊
tan ␸ K␣␥共a␥1 / b␥2兲 tan ␸ K␣␥共a␥1 / b␥2兲
1 ␸
h共Z兲 ⬇ cot 45 −
40 0.1572 1.026 28 0.5518 1.011 2 2
37.5
35
32.5
0.2005
0.2572
0.334
1.025
1.024
1.020
26
24
22
0.7132
0.9556
1.4391
1.004
1.000
1.000

1
2
冉 冊
tan 45 −
␸ 1 + aq1Z
2 1 + bq1Z + bq2Z2 + bq3Z3

cot ␸

30 0.4357 1.017 20 2.0493 1.000


Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

Replace Eq. 共40兲 with Eq. 共41兲 and write

determined in the present study can be used for the study of this
Z + a␥1Z2
condition. For simplification, safety factor is not considered in the Pa = Ka␥
following discussion. 1 + b␥1Z + b␥2Z2
According to Eq. 共36兲, the active earth pressure for a well
without support can be expressed as 共assuming k = 1 and no
surcharge兲

+ C Kay
1 + aq1Z
1 + bq1Z + bq2Z2 + bq3Z3

− 1 cot ␸ 艋 0 共42兲

When C is known, we can calculate with Eq. 共42兲 the limit


Pa = ka␥ f共Z兲 + kach共Z兲C 艋 0 共40兲
maximum depth without support. From Eq. 共42兲, a circular exca-
When soil friction angle is greater than 19.5°, according to vation can be carried out without support if the dimensionless soil
Table 1 and Eq. 共29兲, cohesion C satisfies

共Z + a␥1Z2兲共1 + bq1Z + bq2Z2 + bq3Z3兲


C 艌 tan共␸兲 共43兲
共1 + bq1Z + bq2Z2 + bq3Z3兲共1 + b␥1Z + b␥2Z2兲/Ka␥ − 共1 + b␥1Z + b␥2Z2兲共1 + aq1Z兲
The limit minimal soil dimensionless cohesion Cmin can be obtained from Eq. 共41兲 as

冦 冧
Z + 共a␥1 + bq1兲Z2 + 共a␥1bq1 + bq2兲Z3 + 共a␥1bq2 + bq3兲Z4 + bq3Z5
Cmin = tan共␸兲Max 共44兲
共1 + bq1Z + bq2Z2 + bq3Z3兲共1 + b␥1Z + b␥2Z2兲
− 共1 + b␥1Z + b␥2Z2兲共1 + aq1Z兲
Ka␥

The dimensionless cohesion C is computed with Eq. 共44兲 and tive earth pressure in bored pile construction supported by slurry.
Table 1, which gives Eq. 共45兲 and Table 2 There are two forces acting on the well surface: Active earth
pressure pa, which causes the hole to collapse, and the hydrostatic

Cmin ⬇ 共1.026 – 1.000兲tan共␸兲tan2 45 − 冉 冊 ␸ a␥1


2 b␥2
共45兲
pressure from slurry which can be calculated from the equation
ps = ␥sz due to slurry. Summing up these two forces, the net pres-
sure pz is
Eq. 共45兲 is valid only when ␸ is greater than 20°. When soil
friction angle is smaller than 19.5°, since the active earth pressure
induced by weight itself tends to infinity whereas the active pres-
sure induced by cohesion tends to a finite limit at infinite dimen-
sionless depth, the limit minimum soil cohesion for circular
excavation without support does not exist.
Fig. 26 shows the variable pa with depth z at saturated weight
␥⬘ = 10 kN/ m3, ␸ = 30° for three cases: 共1兲 c = 5 kN/ m2,
q = 17 kN/ m2; 共2兲 c = 3.75 kN/ m2, q = 0 kN/ m2; and 共3兲
c = 3.75 kN/ m2, q = 17 kN/ m2. For simplification, the safety fac-
tor is not considered in the analysis. For case 1, pa is always
smaller than zero, which means that, theoretically, support for the
circular excavation will not be necessary at all depths. For the
second case where c = 3.75 kN/ m2, q = 0 kN/ m2, the variable pa is
negative above z = 20 m, which means that support for circular
excavation ranging from 0 to 20 m is not necessary. For the third
case where c = 3.75 kN/ m2, q = 17 kN/ m2, pa is smaller than zero
for z = 1.5– 17.5 m. Hence support is theoretically necessary for
the top 1.5 m but is not required for a depth between 1.5 and
17.5 m. Fig. 26. Variable of pa with depth z for different cohesion at
Another interesting application of the present study is the ac- ␸ = 30° and ␥⬘ = 10 kN/ m3

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007 / 13

Int. J. Geomech. 2007.7:1-15.


strain analysis. The failure surface is similar to plane strain at the
top but arch action has caused a change of the shape at greater
depth. Even more interesting is that the backfill angle of inclina-
tion ␣ affects only the active pressure coefficient at the top of the
excavation. If the depth is great, the circular effect will mask out
this effect so that the backfill angle has only a small effect on the
active pressure coefficient if the depth is great. This new result
has demonstrated that the arch action is more important than the
backfill angle in the active pressure determination when the depth
is great.
Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

The writers have developed curve fitting equations that can


provide very good results of the active pressure without using the
method of characteristics. Such equations are useful for numerical
integration of the total active pressure or computer implementa-
tion. The active pressure coefficients due to self-weight of soil
and surcharge are found to decrease with depth. The actual active
pressure due to self-weight may increase indefinitely with depth
Fig. 27. Variation of dimensionless Pz, Pa, and Ps with 共at a slow rate兲 or attain a maximum value depending on the
dimensionless Z lateral stress coefficient k and the friction angle of soil. The actual
active pressure due to surcharge will decrease with depth, which
pZ = pa − ps = ka␥ f共Z兲z␥ + kaqg共Z兲q0 + kach共Z兲c − ␥wz 共46兲 means that the circular effect 共arch action兲 has masked the effect
of surcharge if the depth is great. The behavior of the active
Define a set of dimensionless variables as pressure due to self-weight of soil and external surcharge under
axisymmetric condition is drastically different from the corre-
pz ps ␥w
PZ = , Ps = , W= 共47兲 sponding plane strain results.
r 0␥ r 0␥ ␥ For the cohesive strength of soil, its effect on the active pres-
Eq. 共46兲 will become sure increases 共not decreases as compared with self-weight of soil
and external surcharge兲 with depth and attains a maximum value
PZ = Pa − Ps = ka␥ f共Z兲Z + kaqh共Z兲Q0 + kach共Z兲C − WZ 共48兲 if the depth is great. The maximum value can be attained by any
If PZ ⬍ 0 at depth Z, the well surface at this point is stable and value of lateral stress coefficient k, which is also different from
the slurry pressure is supporting the excavation. The variation of the effect of self-weight of soil. The writers have also demon-
dimensionless Pz, Pa, and Ps with Z at ␸ = 8°, C = 0, Q0 = 0.55, strated that the plane strain relation kac = 关kaq − 1兴cot ␸ is also gen-
W = 0.638 is shown in Fig. 27. When Z ⬎ 3.777, Pz is negative and erally applicable for the axisymmetric problem.
the well is stable. For Z ⬍ 3.777, Pz is positive and support in The writers have demonstrated that if k is equal to 1.0, the
form of casing will be necessary. differences between the results from combined analysis and the
results from principle of superposition are very small and can be
neglected. As the numerical results have demonstrated that k = 1 is
Discussion and Conclusion a reasonable assumption, the principle of superposition can gen-
erally be used for analysis and design without major problem.
In the present study, the writers have developed the simplified and The results in the present paper can be applied to the analysis
more rigorous methods in determining axisymmetric active pres- of bore pile and circular cut construction. In general, if c is
sure under general conditions. It is found that the lateral earth greater than a minimal value as given by Table 2, construction can
pressure coefficient k has a significant effect on the results of be carried without support. The case studies have clearly demon-
analysis. Based on numerical modeling, it is established that the strated that the present studies can be used for evaluating the
assumption of k = 1, which is commonly used, is reasonable and needs for support during circular cut.
adequate in most cases if the soil is allowed to be moved suffi- The present work represents extension of the previous works
ciently. For cases where the wall movement may not be sufficient on axisymmetric solution of the slip line equations and many
to mobilize the ultimate limit state, the solutions provided in the interesting results have been found from the present study. The
present paper provide the bounds to the design active pressure by limitation of the present study is the requirement of associated
using k equal to k0 and 1.0. flow rule, otherwise, only the stress characteristic is satisfied ac-
From simplified slip line solution, an analytical formula has cording to the present formulation. The several interesting phe-
been developed for active pressure determination when wall fric- nomena of arch action as observed from the present study are,
tion and the backfill slope angle are both 0. For the general prob- however, still valid even if the nonassociated flow rule is adopted
lem, the more rigorous solution as developed in the present paper in the numerical modeling, and this is established from parametric
can be adopted for analysis. The analytical formula from the study by the writers. The effect of dilation angle on the active
simplified slip line analysis, which does not satisfy yield and pressure is generally small when friction angle is small and is
equilibrium condition rigorously, has been shown to be a good noticeable when the friction angle is high 共the effect on the nu-
approximation of the active pressure 共always underestimate as merical values is, however, not significant兲.
compared with more rigorous analysis兲 if the depth of excavation
is low. When the depth of excavation is great, the results from the References
more rigorous analysis should be adopted.
It is found that the failure surface from iterative finite differ- Berezantzev, V. G. 共1958兲. “Earth pressure on the cylindrical retaining
ence solution has an interesting shape different from that in plane wall.” Proc., Brussels Conf. on Earth Pressure Problems, Vol. 2,

14 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007

Int. J. Geomech. 2007.7:1-15.


21–27. problems in soils by the method of characteristics.” Proc., 4th Int.
Bolton, M. D., and Lau, C. K. 共1993兲. “Vertical bearing capacity for Conf. on Numerical Methods in Geomechancis, Edmonton, Alta.,
circular and strip footings on Mohr-Coulomb soil.” Can. Geotech. J., Canada, Vol. 3, 1059–1071.
30, 1024–1033. Jenike, A. J., and Yen, B. C. 共1962兲. “Slope stability in axial symmetry.”
Chen, W. F. 共1975兲. Limit analysis and soil plasticity, Elsevier, New York. Proc., 5th Symp. on Rock Mechanics, Univ. of Minnesota, Pergamon,
Cheng, Y. M. 共2003兲. “Seismic lateral earth pressure coefficients by slip New York, 689–711.
line method.” Comput. Geotech., 30共8兲, 661–670. Kerisel, J., and Absi, E. 共1990兲. Active and passive earth pressure tables,
Cox, A. D. 共1962兲. “Axially symmetry plastic deformation in soils. II. Balkema, Rotterdam, The Netherlands.
Kumar, J. 共2001兲. “Seismic passive earth pressure coefficients for sands.”
Indentation of ponderable soils.” Int. J. Mech. Sci., 4, 371–380.
Can. Geotech. J., 38, 876–881.
Cox, A. D., Eason, G., and Hopkins, H. G. 共1961兲. “Axially symmetric
Downloaded from ascelibrary.org by UNICAMP - Universidade Estadual De Campinas on 12/22/14. Copyright ASCE. For personal use only; all rights reserved.

Kumar, J., and Chitikela, S. 共2002兲. “Seismic passive earth pressure co-
plastic deformation in soils.” Philos. Trans. R. Soc. London, Ser. A, efficients using the method of characteristics.” Can. Geotech. J., 39,
254, 1–45. 463–471.
Drescher, A. 共1983兲. “Limit plasticity approach to piping in bins.” ASME Mroz, Z. 共1967兲. “Graphically solution of axially symmetric problems of
Trans. J. Appl. Mech., 50, 549–553. plastic flow.” J. Appl. Math., 18, 219.
Drescher, A. 共1986兲. “Kinematics of axisymmetric vertical slopes at col- Prater, E. G. 共1977兲. “An examination of some theories of earth pressure
lapse.” Int. J. Numer. Analyt. Meth. Geomech., 10, 431–444. on shaft lining.” Can. Geotech. J., 14共1兲, 91–106.
Drescher, A., and Vgenopoulou, I. 共1985兲. “A theoretical analysis of Shield, R. T. 共1955兲. “On the plastic flow of metals under conditions
channeling in bins and hoppers.” Powder Technol., 42, 181–191. of axial symmetry.” Philos. Trans. R. Soc. London, Ser. A, 233,
Hill, J. M., and Cox, G. M. 共2000兲. “Cylindrical cavities and classical 267–287.
rat-hole theory occurring in bulk materials.” Int. J. Numer. Analyt. Sokolovskii, V. V. 共1965兲. Statics of granular media, Pergamon,
Meth. Geomech., 24, 971–990. New York.
Houlsby, G. T., and Wroth, C. P. 共1982兲. “Direct solution of plasticity Terzaghi, K. 共1943兲. Theoretical soil mechanics, Wiley, New York.

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / JANUARY/FEBRUARY 2007 / 15

Int. J. Geomech. 2007.7:1-15.

You might also like