Tunnelling and Underground Space Technology 25 (2010) 188–197

Contents lists available at ScienceDirect

Tunnelling and Underground Space Technology

journal homepage: www.elsevier.com/locate/tust

Review

Comparative evaluation of methods to determine the earth pressure distribution

on cylindrical shafts: A review
Tatiana Tobar, Mohamed A. Meguid *
Department of Civil Engineering and Applied Mechanics, McGill University, 817 Sherbrooke Street West, Montreal, Quebec, Canada H3A 2K6

a r t i c l e i n f o a b s t r a c t

Article history: Various methods used for calculating and measuring the earth pressure distribution on cylindrical shafts
Received 22 April 2009 constructed in sand are evaluated. Emphasis is placed on a comparison between the calculated earth
Received in revised form 15 October 2009 pressure using different methods for given sand and wall conditions. The effects of the assumptions made
Accepted 5 November 2009
in developing these solutions on the pressure distribution are discussed. Physical modeling techniques
Available online 4 December 2009
used to simulate the interaction between vertical shafts and the surrounding soil are presented. The earth
pressure measured and the wall movements required to establish active condition are assessed. Depend-
Keywords:
ing on the adopted method of analysis, the calculated earth pressure distribution on a vertical shaft lining
Cylindrical shafts
Earth pressure theory
may vary considerably. For shallow shafts, the theoretical solutions discussed in this study provide con-
Physical modeling sistent estimates of the active earth pressure. As the shaft depth exceeds its diameter, the solutions
Soil–structure interaction become more sensitive to the ratio between the vertical and horizontal arching and only a range of earth
pressure values can be obtained. No agreement has been reached among researchers as to the magnitude
of wall movement required to establish active conditions around shafts and further investigations are
therefore needed.
Ó 2009 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
2. Theoretical methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
2.1. Analytical solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
2.2. Limit equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
2.3. Slip line method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
2.4. Comparison between different theoretical solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
3. Experimental investigations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
3.1. Shaft sinking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
3.2. Temporary stabilization using fluid pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
3.3. Mechanically adjustable lining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
3.4. Discussion of experimental investigations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
4. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

1. Introduction (e.g. tunnels, pumping stations and hydroelectric projects). Deter-

mining the earth pressure acting on the shaft lining system is
Vertical shafts are widely used as temporary or permanent essential to a successful design. Classical earth pressure theories
earth retaining structures for different engineering applications developed by Coulomb (1776) and Rankine (1857) have been often
used to estimate earth pressure on shaft walls. These theories were
originally developed for infinitely long walls under plane strain
* Corresponding author. Tel.: +1 514 398 1537; fax: +1 514 398 7361.
E-mail addresses: tatiana.tobarvalencia@mail.mcgill.ca (T. Tobar), mohamed. conditions. Terzaghi (1920) investigated the effect of wall move-
meguid@mcgill.ca (M.A. Meguid). ment on the magnitude of earth pressure acting on a rigid retaining

0886-7798/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tust.2009.11.001
 T. Tobar, M.A. Meguid / Tunnelling and Underground Space Technology 25 (2010) 188–197 189

Nomenclature

a shaft radius q external surcharge

c soil cohesion r radial distance
Dr relative density S radial displacement at shaft wall or radial soil move-
F radial force ment at soil-wall interface
g gravitational constant of the earth T tangential force
Gs specific gravity W weight of the soil wedge
h excavation depth measured from ground surface
H shaft wall height Greek symbols
K coefficient of lateral earth pressure on circumferential a inclination of the failure surface
planes, K = rh/rv b angle between the reaction Q acting on the sliding body
Ka coefficient of earth pressure at active conditions, and the normal
Ka = tan2 ð45  /=2Þ d soil unit weight
Ko coefficient of earth pressure at rest / friction angle
Kr coefficient of earth pressure for cylindrical shafts k coefficient of lateral earth pressure on radial planes,
mr normalized earth pressure, mr = p/ca (k = rh/rv)
N = N/= tan2 ð45 þ /=2Þ r1, r2, r3 major, intermediate and minor principal stresses
n1 normalized extent of the yield zone, n1 = r/a rr radial stress
P=p lateral earth pressure rh tangential stress
pa active earth pressure rv vertical stress
Po lateral earth pressure at S = 0 mm / reduced angle of internal friction, / = / – 5°
P1 earth pressure force per unit length of the shaft
circumference

wall. He concluded that for dense sand, a wall movement of about used to solve for the earth pressure. Conversely, the slip line meth-
0.1% of the wall height is necessary to reach the theoretical active od assumes the entire soil mass to be on the verge of failure, and
earth pressure. Following from the work of Terzaghi (1920), exten- the solution is obtained through a set of differential equations
sive earth pressure research has been conducted (Terzaghi, 1934, based on plastic equilibrium. Several attempts have been made
1953; Rowe, 1969; Bros, 1972; Sherif et al., 1982, 1984) to deter- to extend these methods to study the active earth pressure against
mine the wall displacement required for establishing the active cylindrical shafts in cohesionless media. Westergaard (1941) and
stress state under two-dimensional conditions. Terzaghi (1943), proposed analytical solutions; Prater (1977) used
Several theoretical methods have been proposed for the calcula- the limit equilibrium method; and Berezantzev (1958), Cheng and
tion of the active earth pressure on cylindrical retaining walls sup- Hu (2005), Cheng et al. (2007), Liu and Wang (2008), Liu et al.
porting granular material (e.g. Berezantzev, 1958; Prater, 1977). (2009) used the slip line method. In contrast to the classical earth
However, the earth pressure distribution obtained using these pressure theories, where the active earth pressure calculated using
methods was found to vary significantly. In addition, the required the Coulomb or Rankine method are essentially the same, the dis-
wall movement to reach the calculated pressures is yet to be tributions obtained for axisymmetric conditions may differ consid-
understood. erably depending on the chosen method of analysis, as discussed
Physical models have been used to measure the changes in below.
earth pressure due to the installation of model shafts in granular
material under normal gravity conditions or in a centrifuge. One
2.1. Analytical solutions
of the key challenges in developing a model shaft is to simulate
the radial movement of the supported soil during construction.
The earliest effort to investigate the state of stress around a
Researchers have developed different innovative techniques to
cylindrical opening in soil was made by Westergaard (1941),
capture these features either during or after the installation of an
who studied the stress conditions around small unlined drilled
instrumented lining.
holes, based on the equilibrium of a slipping soil wedge. Terzaghi
The objective of this study is to review some of the theoretical
(1943) extended Westergaard’s theory to large lined holes, thus
and experimental techniques to investigate the active earth pres-
proposed a method to calculate the minimum earth pressure ex-
sure on cylindrical shaft linings installed in cohesionless ground.
erted by cohesionless soil on vertical shafts liners. He determined
The assumptions made in developing different theoretical solu-
the equilibrium of the sliding soil mass assuming rh = rv = r1 and
tions and their effects on the calculated earth pressure are exam-
rr = r3 inside the elastic zone and employing the Mohr–Coulomb
ined. Finally, a comparison is presented between different
yield criterion. Terzaghi obtained Eqs. (1)–(3) below for the lateral
physical modeling techniques used to study the interaction be-
earth pressure on a shaft lining. As stated in Eq. (3), Terzaghi pro-
tween a shaft lining and the surrounding soil, and the measured
posed the use of a reduced angle of internal friction of the sand, /,
earth pressure distributions are reproduced.
to account for the effect of the nonzero shear stresses in the
solution.
2. Theoretical methods
h N/ þ 1 N/  ðN/  2Þn21
When Coulomb (1776) and Rankine (1857) developed their mr ¼ N þ1
ð1Þ
a 2N / N/ þ n1 /
two-dimensional earth pressure theories, they also established
N þ1
two simple methods of analysis: the limit equilibrium and the slip n21  1 2N/ a n1 /  1
tan / ¼  ð2Þ
line method. Both methods are based on plastic equilibrium, how- N
mr n1 / N/ þ 1 h N
n1 /
ever they differ in how the solution is obtained. The limit equilib- 
/ ¼ /  5 ð3Þ
rium method assumes a suitable failure surface, and basic statics is
190 T. Tobar, M.A. Meguid / Tunnelling and Underground Space Technology 25 (2010) 188–197

where mr = p/ca, normalized earth pressure; n1 = r/a, normalized circumference) P1 is expressed by Eq. (4), where Kr is the coefficient
extent of the yield zone; and N/ = tan2 (45 + / /2); a = shaft radius; of earth pressure for cylindrical shafts given by Eq. (5).
h = excavation depth; r = radial distance.
2
Fig. 1a shows the values of normalized earth pressure, mr, ver- P1 ¼ 0:5K r ch ð4Þ
 
sus normalized depth, h/a, originally computed by Terzaghi (1943) h 1 a k
for / = 40° and those computed by the authors for / = 40° and 41°. Kr ¼ tanða þ bÞ   ð5Þ
a tan a 3 tan a h 3
For / = 40° the difference between the calculated and original data
is rather small. Increasing the friction angle from 40° to 41° causes where a = shaft radius; h = excavation depth; a = inclination of fail-
a reduction in the normalized earth pressure by approximately 9%. ure surface; = angle between the reaction Q acting on the sliding
These numerical values were obtained using the reduced friction body and the normal (b=/ for active condition); k = coefficient of
angle given by Eq. (3). The values of mr are computed from Eq. lateral earth pressure on radial planes.
(2) for different assumed values of n1 and the corresponding values The earth pressure on the shaft is computed as follows. First the
of h/a are then computed from Eq. (1). The procedure is detailed by earth pressure, P1, is computed at various depths, i.e. by incremen-
Terzaghi (1943). tally increasing the depth. Second, the difference in force between
Fig. 2 shows, among other solutions, the calculated earth pres- successive increments is divided by the depth increment to obtain
sure distribution with depth using the above equations for a shaft the average earth pressure for the increment. This average pressure
lining of radius, a, and height, h, installed in cohesionless soil with is plotted versus depth as shown in Fig. 2. The values used in plot-
/ = 41°. The pressure generally increases with depth and reaches a ting Prater’s solution in Fig. 2 have been calculated using the above
normalized value, p/ca, of 0.25 at a depth of approximately 5 times procedure (/ = 41° and k = Ko) in conjunction with the graphs pre-
the shaft radius. For h/a greater that 5, the pressure increase is less sented by Prater (1977) to obtain the values of the coefficient Kr.
significant and reaches a normalized value, p/ca, of 0.30 at a depth As shown in Fig. 2, Prater’s method predicts a zero earth pres-
of approximately 15 times the shaft radius. sure at some depth below the surface; however, Prater recom-
mended that the maximum earth pressure value should be used
for design purposes.
2.2. Limit equilibrium

Prater (1977) adapted Coulomb wedge theory for axisymmetric 2.3. Slip line method
conditions assuming a conical failure surface. He introduced into
the analysis tangential and radial forces, T and F (see Fig. 1c). The Berezantzev (1958) extended the slip line method to calculate
force T is a function of the earth pressure coefficient on radial the earth pressure acting on cylindrical walls with horizontal back-
planes, k, which is defined by the stress ratio rh/rv. Prater argued fill and uniform surcharge as shown in Fig. 1b. To solve the equilib-
that k is a decisive parameter whose value should range between rium equations under axisymmetric conditions Berezantzev
Ka and Ko and not equal to unity as was implicitly assumed by Ter- introduced into the analysis the Haar–Von Karman hypothesis
zaghi (1943). The earth pressure (force per unit length of the shaft which states that the hoop stress is equal to either the major or

Fig. 1. (a) Earth pressure distributions using Terzaghi (1943) for / = 40° and 41°. (b)
Earth pressure acting on a cylindrical retaining wall as studied by Berezantzev Fig. 2. Earth pressure distributions a shaft of depth h, radius a in cohesionless soil
(1958). (c) Failure surface assumed by Prater (1977). and no surcharge.
 T. Tobar, M.A. Meguid / Tunnelling and Underground Space Technology 25 (2010) 188–197 191

pffiffiffiffiffiffi ! " #
the minor principal stress (Yu, 2006). Thus, under active conditions Ka a aK a 1  k þ g ae
Berezantzev assumed that inside the plastic zone the tangential P a ¼ ac 1  g1 þq g  cot /  g Ka c ð7Þ
g1 r b
rb g rb
and radial stresses are equal to the major and minor principal
stresses, respectively (rh ¼ rv ¼ r1 and rr ¼ r3 ). Thus k = rh/
pffiffiffiffiffiffi
rv = 1. To simplify the calculations the slip lines were approxi- where g ¼ k tan2 ð45 þ /=2Þ  1; r b ¼ a þ h K a ;e ¼ ð1  kÞg1 tan2
mated to straight lines in the vertical direction and the Mohr–Cou- ð45 þ /=2Þ þ 1; 0 < g < 1 and / – 0.
lomb failure criterion was adopted. The governing equations took Cheng and Hu (2005) found that the case of k = 1 produces the
the form of two partial differential equations that he solved using lowest lateral pressure and therefore a value of k ¼ K o ¼ ð1  sin /Þ
the Sokolovski step-by-step computation method. Eq. (6) gives the was suggested for engineering applications. The upper and lower
simplified form of the solution that evaluates the earth pressure on bounds of the lateral earth pressure can then be obtained using
the shaft wall as reported by Fujii et al. (1994). k = Ko and k = 1, respectively, as shown in Fig. 2 (c = 0 and q = 0).
Cheng et al. (2007) and Liu and Wang (2008) introduced addi-
pffiffiffiffiffiffi " # ! " ! #
tional parameters into the analysis including wall friction, backfill
Ka a aK a a
Pa ¼ ac 1  g1 þ q þ cot / Ka  1 c ð6Þ slope, surcharge loads and soil cohesion. Solution of the character-
g1 rb rgb rgb
istic equations was obtained numerically leading to a lengthy set of
expressions that are omitted in this review. The results indicated
pffiffiffiffiffiffi
where g ¼ 2 tanð/Þ tanð45 þ /=2Þ; r b ¼ a þ h K a ; K a ¼ tan2 ð45 that the pressure distribution is consistently smaller than the
/=2Þ; q = external surcharge; c = soil cohesion; a = shaft radius; one obtained using the simplified solution of Cheng and Hu
h = excavation depth. (2005). Liu and Wang (2008) examined the effect of wall inclina-
As shown in Fig. 2, for a shaft of radius, a, in cohesionless soil tion and developed a solution that was essentially similar to that
and no external surcharge, q, the earth pressure distribution based obtained by Cheng and Hu (2005) simplified solution. They con-
on Berezantzev is similar to that calculated by Terzaghi, however cluded that the analytical solution presented by Cheng and Hu pro-
the maximum pressure is smaller by approximately 40%. vides a reasonable estimate of the active pressure on a vertical
Cheng and Hu (2005) extended Berezantzev’s theory by modify- shaft for horizontal backfill material and zero wall friction.
ing the Haar–Von Karman hypothesis, i.e. k = 1, to develop a more Liu et al. (2009) further extended Berezantzev’s theory by
general solution considering a variable earth pressure coefficient, k. assuming a linearly varying k such that it decreases across the plas-
An expression for the active earth pressure was proposed as given tic zone from unity at the shaft circumference to Ko at the elasto-
below. plastic interface. The results obtained based on this method were

Fig. 3. Model shaft used in the shaft-sinking method and the measured earth pressure distribution (adapted from Walz (1973)).
192 T. Tobar, M.A. Meguid / Tunnelling and Underground Space Technology 25 (2010) 188–197

found to agree with those previously reported by Cheng et al. ting was directed through the vertical tube out of the container.
(2007). Each lining ring was divided into three equal segments that were
Based on the above studies it can be concluded that, for axisym- kept in position using z-shaped aluminum arms attached to a cen-
metric excavations under active conditions, there exist two coeffi- tral piece as shown in Fig. 3a. These z-shaped pieces were equipped
cients of lateral earth pressure: one defined as the ratio of radial with strain gauges, and the entire system was calibrated to directly
stresses acting on circumferential planes, K = rr/rv; and the second read the earth pressure acting against the lining. The normalized
defined as the ratio of tangential stresses acting on radial planes,
k = rh/rv. In other words, during shaft construction the initial stres-
ses redistribute such that the value of K decreases until it reaches
Ka, while the value of k increases such that Ka < Ko < k. Therefore,
the coefficient k provides a measure of the horizontal arching that
has occurred in the soil adjoining the excavation.

2.4. Comparison between different theoretical solutions

A summary of the earth pressure distribution calculated using

some of the above methods for a given shaft geometry (height, h
and radius, a) and soil property (/) is presented in Fig. 2. Although
all methods predict pressures that are less than the at-rest and ac-
tive values, the distributions of earth pressure with depth notably
differ. The Terzaghi and Berezantzev methods implicitly assume k
equals unity, leading to a minimum value of the active earth pres-
sure. This is consistent with the results of the plastic equilibrium
and slip line methods. Both solutions result in pressure distribu- Fig. 4. Test setup used in the fluid pressure technique (adapted from Lade et al.
(1981)).
tions that ultimately reach a constant earth pressure at some depth
below surface. As discussed earlier, Prater’s method predicts a dif-
ferent pressure distribution that can be characterized (for the same
shaft geometry and soil conditions) by a rapid increase in pressure
up to a depth of about 4.5 times the shaft radius and then a de-
crease to zero at a depth of 8.5 times the shaft radius. The solution
of Cheng and Hu provided the lower and upper bounds of the lat-
eral earth pressure as given by k = 1 and k = Ko, respectively. For
k = 1 the earth pressure is the same as that calculated using the
Berezantzev method. Fig. 2 shows that for shallow shafts, where
the shaft height ranges from 1 to 2 times the shaft radius, the dif-
ference between the above theoretical methods is insignificant.

3. Experimental investigations

Several studies have been conducted to measure the earth pres-

sure distribution due to the installation of a model shaft in granular
material. To simulate the lining installation and the radial soil
movement during construction, different techniques have been
developed that can be grouped into three main categories: (a) shaft
sinking; (b) temporary stabilization of the excavation using fluid
pressure (liquid or gas); and (c) the use of a mechanically adjust-
able lining. These techniques are briefly described and samples of
the experimental results are presented.

3.1. Shaft sinking

The sinking technique consists of advancing a small model cais-

son equipped with a cutting edge at a recess distance, S, from the
lining surface. This recess S is used to simulate the induced soil
movement during construction. Walz (1973) investigated the lat-
eral earth pressure against circular shafts using the above tech-
nique. The shaft lining consisted of a 105 mm diameter and
630 mm deep tube composed of 12 steel rings and a cutting edge
ring equipped with recess, S, ranging from 0 to 5 mm, as shown
in Fig. 3a. The soil container used was a cylindrical tub of 1 m
diameter and 1 m deep filled with dry sand. Prior to the filling pro-
cess, a hollow tube of small diameter was installed vertically across
the container. This tube was attached to the cutting edge ring at
the soil surface, and then pulled down using a motor to sink the Fig. 5. Radial strains and the corresponding earth pressure distributions (adapted
model shaft. As the shaft was advanced into the soil, the soil cut- from Lade et al. (1981)).
 T. Tobar, M.A. Meguid / Tunnelling and Underground Space Technology 25 (2010) 188–197 193

earth pressure distribution versus normalized depth for S = 0 and stresses around the shaft and the surface settlement, respectively.
2 mm are shown in Fig. 3b. The introduction of the 2 mm recess An overview of the test setup is shown in Fig. 4. The radial strains
has lead to a significant decrease in the measured earth pressure in the lining and the normalized earth pressures versus normalized
along the lining, with a maximum reduction of about 75% at h/ depth are reproduced in Fig. 5. Large inward movements at the
a = 0.5. base of the fully excavated shaft were recorded which corre-
sponded to large pressures at this depth. Before the fluid removal,
3.2. Temporary stabilization using fluid pressure expansion of the shaft lining was observed due to larger pressures
exerted by the liquid inside the shaft than the outside soil. Similar
In this technique, the soil to be excavated is replaced by a flex- observations were made by Kusakabe et al. (1985) in a series of
ible rubber bag filled with liquid or gas. The liquid level, or gas centrifuge tests conducted to investigate the influence of axisym-
pressure, is reduced in stages to simulate the shaft excavation pro- metric excavation on buried pipes. Fig. 5 further shows that the
cess. This technique is generally used in centrifuge testing due to measured pressures are higher than the calculated using Ber-
the restrictions in modeling excavation during the test. ezantzev method.
Lade et al. (1981) conducted a series of centrifuge tests to inves- Konig et al. (1991) carried out a series of centrifuge tests to
tigate the lateral earth pressure against shafts in sand. A cylindrical study the effects of the shaft face advance on a pre-installed lining.
tub of 850 mm diameter and 695 mm deep was used as the test The model shaft consisted of two sections: an upper section made
container in which dry fine Leighton Buzzard sand (c = 15.35– of rigid tube to simulate the installed lining, and a lower section
15.5 kN/m3, / = 38.3°) was placed by pluvial deposition. The lining made of rubber membrane to model the unsupported area of the
was formed using a 0.35 mm thick Melinex sheet. The soil inside excavation. At the initial condition, the membrane was pressurized
the shaft was replaced by two different liquids: ZnCl2-solution with air to equilibrate the pressure exerted by the soil. To simulate
with density similar to that of the soil and paraffin oil with a den- the shaft face advance, the air pressure was incrementally reduced.
sity of 7.65 kN/m3. The excavation process was modelled by The lateral movement of the rubber membrane was monitored
removing the liquid in four stages and the liquid level was moni- using LVDT’s embedded in the sand; the stresses in the shaft lining
tored. The readings of eight strain gauge sets installed along the were monitored using strain gauges installed at different distances
lining were recorded and used to calculate the lateral earth pres- from the end of the lining. Results indicated that for dry sand, only
sure. Earth pressure cells and LVDT’s were used to monitor the a small support pressure was needed to maintain stability. How-

Fig. 6. Semi-cylindrical model shaft and earth pressure distribution for smooth and rough walls (adapted from Fujii et al. (1994)).
194 T. Tobar, M.A. Meguid / Tunnelling and Underground Space Technology 25 (2010) 188–197

ever, there was a significant load transfer to the lining closest to Researchers have adopted simplified models to simulate the radial
the excavation face due to arching and stress redistribution in displacement of the lining (Fujii et al., 1994; Imamura et al., 1999),
the soil. or took advantage of the radial symmetry to model only a portion
of the problem (Herten and Pulsfort, 1999; Chun and Shin, 2006).
3.3. Mechanically adjustable lining Tobar and Meguid (2009) developed a mechanical system that al-
lowed for the modeling of both the full shaft geometry as well as
In this technique, a mechanical system is used to move a rigid the radial displacement of the lining.
shaft lining in order to simulate the soil displacement that may oc- Using a mechanically adjustable shaft model, Fujii et al. (1994)
cur during the excavation process. Using this technique, it is possi- conducted centrifuge tests to study the effects of wall friction and
ble to impose a homogeneous radial displacement along the entire soil displacements on the earth pressure distribution around rigid
shaft height at a controlled rate. However, the mechanism required shafts. The lining was made of an aluminum cylinder of 60 mm in
to model the inward movement of the shaft lining is challenging. diameter split vertically into two semi-cylinders; one-half was

0
S = 0 mm
1 S = 0.5 mm

Normalized depth, h/a

2
Motor
Load 100 cm
cells 3

5
0 2 4 6 8 10
100 cm Earth pressure (kN/m2)
a) Test setup b) Measured earth pressure
Fig. 7. Quarter-cylinder model shaft and the earth pressure distribution (adapted from Herten and Pulsfort (1999)).

Fig. 8. Semi-cylindrical model shaft and the measured earth pressure using a shape aspect ratio, H/a, of 4.286 (adapted from Chun and Shin (2006)).
 T. Tobar, M.A. Meguid / Tunnelling and Underground Space Technology 25 (2010) 188–197 195

instrumented with small stress transducers and horizontally corresponds to 0.2% of the wall height, H (1.6% of the shaft radius).
moved using a motor to simulate the radial displacement of the The maximum extent of the yield zone was found to be approxi-
shaft lining. Details of the apparatus are shown in Fig. 6a. The mod- mately 0.7 times the shaft diameter.
el shaft was placed into a rectangular soil container and Toyoura Herten and Pulsfort (1999) took advantage of the radial symme-
dry sand was rained around it up to 200 mm in height, H. Four tests try of the problem and modeled only one quadrant of the shaft. The
were conducted for different densities and wall friction conditions. test setup consisted of one quarter of a cylindrical shaft with 0.4 m
The measured earth pressure versus normalized depth for dense in diameter and 1 m long. The model shaft was placed along one
sand (/ = 42°, c = 14.7 kN/m3) and different wall friction is shown corner of a rectangular box of 1  1 m in plan and 1.2 m in height.
in Fig. 6b along with the earth pressure calculated from Ber- To minimize the wall friction, the walls were lubricated using Tef-
ezantzev method. The experimental results show good agreement lon film and oil. The test container was filled using pluvial deposi-
with the theoretical solution of Berezantzev (1958). Little change tion with dry fine sand of / = 41° in dense state (36% porosity). The
in the measured earth pressure was reported at displacements shaft lining was horizontally moved using a motor to simulate the
greater than 1% of the wall height, H (6.6% of the shaft radius), radial displacement of the shaft. Details of the test setup and the
and the wall friction was found to have a negligible effect on the results of one of the four tests conducted are shown in Fig. 7. Little
measured earth pressure distribution. change in the measured lateral earth pressure occurred for wall
Imamura et al. (1999) developed a model shaft similar to that displacements greater than 0.05% of the wall height (0.25% of the
used by Fujii et al. (1994). However, the instrumented semi-cylin- shaft radius).
der was horizontally translated using an external mechanism at- Chun and Shin (2006) conducted model tests to study the ef-
tached to a motor. Air-dried Toyoura sand with / = 42° and fects of wall displacement and shaft size on the earth pressure dis-
c = 15.2 kN/m3 was used during the four centrifuge tests conducted tribution using a mechanically adjustable semi-circular shaft. The
to study the development of the active earth pressure around lining was made from an acrylic semi-cylinder that was cut longi-
shafts and the extent of the yield zone. They concluded that the tudinally into three equal segments, i.e. each span an angle of 60°,
earth pressure decreases with increasing wall displacement until to accommodate the changes in diameter during testing. Transver-
it coincides with Berezantzev’s solution at a wall displacement that sally the shaft was divided into five equal segments; some of them

Fig. 9. Details of the full axisymmetric shaft apparatus and the measured results (Tobar and Meguid, 2009).
196 T. Tobar, M.A. Meguid / Tunnelling and Underground Space Technology 25 (2010) 188–197

were used as sensitive areas for load cells installed behind the lin- radially; it consisted mainly of steel hinges that connected the seg-
ing. Fig. 8a shows a schematic of the model. ment holders to central nuts. These nuts pass through a central
The soil container used was a rectangular box, 0.7 m wide, 1 m threaded rod extended along the shaft axis. As the axial rod was ro-
long and 0.75 m deep filled with dry sand (/ = 41.6°; c = 16.4 kN/ tated, the nuts moved vertically, pulling the segment holders radi-
m3; Dr = 81%). Three different shaft radii, a, equal to 0.175, 0.15 ally inwards and consequently the shaft lining was uniformly
and 0.115 m, and a constant depth, H = 0.75 m, were tested. The re- translated.
ported earth pressure versus depth at various wall displacements The model shaft (0.15 m in diameter and 1 m long) was placed
for a smooth shaft and aspect ratio, H/a, equal to 4.286 are pre- into a circular container of 1.22 m diameter and 1.07 m depth. The
sented in Fig. 8b. The results indicate that earth pressure decreased container was filled with coarse dry sand (/ = 41°; c = 14.7 kN/m3)
with increasing wall movement and became minimum when the using pluvial deposition. The axisymmetric active earth pressure
wall movement reached 0.6–1.8% of the wall height. In Fig. 8 the fully developed when the wall displacements, S, ranged between
earth pressure calculated from Berezantzev and Terzaghi methods 0.2% and 0.3% of the wall height, H. It was concluded that for
are shown for comparison. It appears from this comparison that S P 0.1% H, the measured pressures fell into the range predicted
the measured earth pressures are higher than that predicted from by Cheng and Hu (2005); and that at S P 0.3% H, the measured
Berezantzev; Terzaghi‘s distribution falls between the measured pressures closely followed the pressure distributions calculated
earth pressure at S equal to 0.43 and 1.87 mm (0.06% and 0.25% using Terzaghi (1943) and Berezantzev (1958) methods.
of the wall height, H). Chun and Shin (2006) found that soil failure
extended a distance of approximately one shaft radius from the 3.4. Discussion of experimental investigations
outer perimeter of the lining.
Tobar and Meguid (2009) conducted a series of tests under nor- Table 1 shows a summary of the required wall displacement for
mal gravity to investigate the changes in lateral earth pressure due establishing active conditions. To simplify the design and opera-
to radial displacement of the shaft lining. The developed apparatus tion of the shaft models, simplified mechanisms were used to re-
allowed for the modeling of both the full geometry of the shaft and duce the shaft diameter uniformly. It is evident that no
the radial displacement of the lining. It was built using six curved agreement has been reached among researchers as to the required
lining segments held vertically using segment holders (Fig. 9a). A wall movement to reach active conditions. The displacement ran-
simple mechanism was developed to translate the lining segments ged from 0.05% to 1.8% of the shaft height as shown in Table 1. This

Table 1
Comparison of the required wall displacements for active condition.

Prototype Model Required wall movement (S) to reach active condition Soil
Semi-cylinder (non-segmented)  Fujii et al. (1994) Dense sand
Sa S P 1%Hb
or
S P 6.6% ac
 Imamura et al. (1999)
S = 0.2% H
or
S = 1.6% a

Quarter-cylinder (non-segmented)  Herten and Pulsfort (1999) Dense sand

S = 0.05% H
or
S = 0.25% a

Semi-cylinder (segmented)  Chun and Shin (2006) Dense sand

0.6% H < S < 1.8% H
or
0.15% a < S < 0.4% a

Full cylinder (segmented)  Tobar and Meguid (2009) Loose sand

S P 0.2% H
or
S P 2.5% a

a
Radial wall displacement.
b
Wall height.
c
Shaft radius.
 T. Tobar, M.A. Meguid / Tunnelling and Underground Space Technology 25 (2010) 188–197 197

Table 2
Advantages and disadvantages of selected shaft modeling technique.

Method Advantages Disadvantages

Shaft sinking  Suitable for modeling shafts constructed using the  Causes soil disturbance.
sinking technique.  High shear stresses can develop along the shaft.
 Difficult to assess the effects of the shear stresses along the wall on the lateral
earth pressure.
Pressurized bags  Can be used to simulate shaft excavation under 1 g and  Applicable for flexible shaft linings.
in a centrifuge.
Liquid bag  Simplifies modeling initial stress state in centrifuge.  The liquid inside the bag may exert more pressure in the centrifuge than the
 Can simulate the excavation advance process. soil outside.
 Large inward deformation may occur at the base of the excavation.
Air bag  Flexibility to readjust air pressure during testing.  The pressure imposed along the model shaft is based on average theoretical
 Suitable for modeling small sections of the excavation. value.
 Does not simulate the excavation advance.
Mechanically  Easy to model the translation displacement of the shaft  Limited to rigid lining models.
adjustable lining wall.  Involves oversimplification of the geometry or the radial displacement of the
 Can be used under 1 g or in a centrifuge. soil around the shaft.
 Facilitates the installation of pressure cells behind the  Does not simulate the excavation advance.
lining.

can be attributed to the difference in the testing conditions, shaft par divers Savants, et lûs dans ses Assemblées, Paris, vol. 7 (volume for 1773
published in 1776), pp. 343–382.
geometry, and wall movement technique used in each study. It is
Fujii, T., Hagiwara, T., Ueno, K., Taguchi, A., 1994. Experiment and analysis of earth
therefore recommended that large-scale experiments be con- pressure on an axisymmetric shaft in sand. In: Proceedings of the 1994
ducted using full shaft geometry to account for the gravity effects International Conference on Centrifuge, Singapore, pp. 791–796.
and confirm these conclusions. Table 2 presents the advantages Herten, M., Pulsfort, M., 1999. Determination of spatial earth pressure on circular
shaft constructions. Granul. Matter 2 (1), 1–7.
and disadvantages of the experimental techniques discussed in Imamura, S., Nomoto, T., Fujii, T., Hagiwara, T., 1999. Earth pressures acting on a
the previous section. deep shaft and the movements of adjacent ground in sand. In: Kusakabe, O.,
Fujita, K., Miyazaki, Y. (Eds.), Proceedings of the International Symposium on
Geotechnical Aspects of Underground Construction in Soft Ground. Balkema,
4. Conclusions Rotterdam, Tokyo, Japan, pp. 647–652.
Konig, D., Guettler, U., Jessberger, H.L., 1991. Stress redistributions during tunnel
A comparative study of the theoretical and experimental meth- and shaft constructions. In: Proceedings of the International Conference
Centrifuge 1991, Boulder, Colorado, pp. 129–135.
ods used to determine the earth pressure on cylindrical shafts has Kusakabe, O., Tsutomu, K., Akira, O., Nobuo, T., Nobuaki, N., 1985. Centrifuge model
been presented. For shallow shafts (H 6 2a), the theoretical solu- test on the influence of axisymmetric excavation on buried pipes. In:
tions provide approximate estimates of the active earth pressure Proceedings of the 3rd International Conference Ground Movements and
Structures. Pentech Press, London, England, pp. 113–128.
distribution. As the shaft depth exceeds its diameter, the solutions
Lade, P.V., Jessberger, H.L., Makowski, E., Jordan, P., 1981. Modeling of deep shafts in
become more sensitive to the ratio between the vertical and hori- centrifuge test. In: Proceedings of the International Conference on Soil
zontal arching and therefore only a range of earth pressure values Mechanics and Foundation Engineering, Stockholm, Sweden. vol. 1, pp. 683–
691.
can be calculated. No agreement has been reached among
Liu, F.Q., Wang, J.H., 2008. A generalized slip line solution to the active
researchers as to the magnitude of wall movement required to earth pressure on circular retaining walls. Comput. Geotech. 35 (2), 155–
establish active conditions around the shaft and further investiga- 164.
tions are therefore needed. Liu, F.Q., Wang, J.H., Zhang, L.L., 2009. Axi-symmetric active earth pressure obtained
by the slip line method with a general tangential stress coefficient. Comput.
Geotech. 36 (1–2), 352–358.
Acknowledgements Prater, E.G., 1977. Examination of some theories of earth pressure on shaft linings.
Can. Geotech. J. 14 (1), 91–106.
This research is supported by the Natural Sciences and Engi- Rankine, W.J.M., 1857. On the stability of loose earth. Philos. Trans. Roy. Soc. London
147, 9–27.
neering Research Council of Canada (NSERC) under Grant No. Rowe, P.W., 1969. Progressive failure and strength of sand mass. In: Seventh
311971-06. International Conference on Soil Mechanics and Foundation Engineering, vol. 1,
pp. 341–349.
Sherif, M.A., Ishibashi, I., Lee, C.D., 1982. Earth pressures against rigid retaining
References
walls. Am. Soc. Civ. Eng., J. Geotech. Eng. Div. 108 (GT5), 679–695.
Sherif, M.A., Fang, Y.S., Sherif, R., 1984. Ka and Ko behind rotating and non-yielding
Berezantzev, V.G., 1958. Earth pressure on the cylindrical retaining walls. In: walls. J. Geotech. Eng. 110 (1), 41–56.
Conference on Earth Pressure Problems, Brussels, pp. 21–27. Terzaghi, K., 1920. Old earth-pressure theories and new test results. Eng. News-
Bros, B., 1972. The influence of model retaining wall displacements on active and Record 85 (13), 632–637.
passive earth pressure in sand. In: Proceedings of the 5th European Conference Terzaghi, K., 1934. Large retaining-wall tests. Eng. News-Record 112 (5), 23.
on Soil Mechanics and Foundation Engineering, Madrid, vol. 1, pp. 241–249. Terzaghi, K., 1943. Theoretical Soil Mechanics. John Wiley & Sons, New York.
Cheng, Y.M., Hu, Y.Y., 2005. Active earth pressure on circular shaft lining obtained Terzaghi, K., 1953. Anchored bulkheads. Am. Soc. Civ. Eng. Proc., ASCE 79, 39.
by simplified slip line solution with general tangential stress coefficient. Tobar, T., Meguid, M.A., 2009. Distribution of earth pressure on vertical shafts.
Chinese J. Geotech. Eng. 27 (1), 110–115. In: 62nd Canadian Geotechnical Conference, Halifax, September 2009, CD,
Cheng, Y.M., Hu, Y.Y., Wei, W.B., 2007. General axisymmetric active earth pressure 6pp.
by method of characteristics – theory and numerical formulation. Int. J. Walz, B., 1973. Left bracket apparatus for measuring the three-dimensional active
Geomech. 7 (1), 1–15. soil pressure on a round model caisson right bracket. Baumasch. Bautech. 20 (9),
Chun, B., Shin, Y., 2006. Active earth pressure acting on the cylindrical retaining wall 339–344 (in German).
of a shaft. South Korea Ground Environ. Eng. J. 7 (4), 15–24. Westergaard, H.M., 1941. Plastic state of stress around deep well. Civ. Eng. (London)
Coulomb, C.A., 1776. Essai sur une application des regles des maximis et minimis a 36 (421), 527–528.
quelques problems de statique relatifs a l’architecture. Memoires de Yu, H.S., 2006. Plasticity and Geotechnics. Springer. p. 326.
Mathématique et de Physique, Présentés a l’Académie Royale des sciences,