See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/237377865

Design and performance evaluation of vertical shafts: rational shaft design

method and verification of design method

Article  in  Canadian Geotechnical Journal · January 2011

DOI: 10.1139/t88-034

CITATIONS READS

14 4,116

2 authors, including:

P. K. Kaiser
Laurentian University
219 PUBLICATIONS   10,939 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Strength of Veined Brittle Rocks View project

Distributed Brillouin Sensing (DBS) for underground mining applications View project

All content following this page was uploaded by P. K. Kaiser on 09 July 2015.

The user has requested enhancement of the downloaded file.

 gg- \A-1

"

,
I
320

Design and performance evaluation of vertical shafts: rational shaft design method and
verification of design method
R. C. K. WONG l AND P. K. KAISER2
Department of Civil Engineering, University of Alberta, Edmonton, Alta., Canada T6G 2G7
Received April 1, 1987
Accepted Janwuy 20, 1988

Ground defonnations around axIsymmetric S1iaftS cannot be detennmed wlthtlie design approaclies currently availlible,
which are mostly based on plasticity methods. The convergence-confinement method (usually applied to tunnels), with
consideration of gravitational effects and the three-dimensional conditions near a shaft, is proposed as a tool to predict
fonnation pressure on a shaft and radial ground displacements. It is shown that the behaviour of a shaft is govemed by (1) the
mode of yield initiation dominated by the in situ stress state and the soil strength parameters and (2) the extent of the yield zone
that develops if wall displacements are allowed to occur during construction.
Closed-fonn solutions are presented to approximate the pressure-displacement relationship for cohesionless and cohesive
soils. Results from this approach compare well with those obtained by finite element analyses. The conventional design
methods that provide the minimum support pressures required to maintain stability are not conservative. These pressures are
generally less than those actually encountered if ground movements during construction are restricted with good ground
control.
Key words: shaft, design method, support, interaction, yielding, stress, displacement, earth pressure, arching.

Les defonnations du terrain autour de puits axisymetriques ne peuvent pas etre detenninees avec les approches de calcul
couramment disponibles qui pour la plupart sont basees sur des methodes de plasticite. La methode de convergence-
confinement, qui s'applique habituellement aux tunnels, avec la prise en compte des effets gravitationnels et des conditions tri-
dimensionnelles pres du puit, est proposee comme outil pour prewre la pression de fonnation sur Ie puit et les deplacements
radiaux du terrain. L'on demontre que Ie comportement du puit est regi par (1) Ie mode d'initiation de l'6coulement plastique
contr6le par l'etat des contraintes en place et les parametres de resistance du sol et (2) l'etendue de la zone de plasticite qui se
developpe lorsque des defonnations de la paroi peuvent se produire au cours de la construction.
L' on presente des solutions directes pour obtenir une approximation de la relation pression -deplacement pour les sols
coherents et pulverulents. Les resultats de cette approche concordent bien avec ceux obtenus au moyen des analyses en
elements finis. Les methodes de calcul conventionnelles qui fournissent les pressions de soutenement minimales requises pour
maintenir la stabilite ne sont pas securitaires. Ces pressions sont en general inferieures a celles qui agissent reellement lorsque
les mouvements en cours de Construction sont limites par un bon contr6le du terrain.
Mots eMs : puit, methode de calcul, soutenement, interaction, ecoulement plastique, contrainte, deplacement, pression des
terres, arc-boutement. .
[Traduit par la revue]

Can. Geotech. 1. 25, 320-337 (1988)

1. Introduction methods mentioned above. Most importantly, the actual pres-

The design of a shaft in soil consists of two main steps: sures are often greater than those predicted in this manner.
(1) design of the shaft lining to prevent instability of shaft wall No closed-fonn solutions are currently available for the
and (2) estimation of the soil movement associated with shaft determination of shaft wall displacements and movements of
construction. Although these two tasks are interrelated, they the surrounding ground. In practice, the displacements are
are usually handled separately. Except for simplicity, there are often controlled or limited by choice of a suitable factor of
no practical reasons why one should separate the detennination safety and excessive yielding is prevented by an appropriate
of lining pressure and ground deformation. construction sequence.
Most of the currently available design approaches are based The convergence-confinement method (CCM) , which
on soil plasticity considerations or plas~city equilibrium accounts for most relevant design parameters (e.g., in situ
methods (PEM) (e.g., Terzaghi 1943; Berezantzev 1958), stress field, ground and support properties, and construction
whereby the formation pressures on the shaft lining are deter- sequence), provides an analytical frainework to predict the for-
mined by satisfying equilibrium of yielded, plastic ground mation pressure and the soil deformation simultaneously. So
behind the support. Alternatively, limit equilibrium methods far, it has been applied successfully only to circular, horizontal
(LEM), assuming hypothetical rupture s~aces, are used for underground openings where the stress component parallel to
design (e.g., Prater 1977). The actually expected and observed the axis of the opening has little effect on the opening
pressures depend, however, on such factors as ground defor- behaviour and where the three-dimensional conditions near the
mation, in situ stress, and ground strength -deformation prop- tunnel face can be approximated.
erties. They differ substantially from those predicted by the Since the analysis of a shaft near the ground surface is a truly
three-dimensional problem, the CCM in its common fonn is
'Present address: Esso Resources Canada Ltd., Calgary Alta., not adequate to describe the behaviour of a shaft. The effect of
Canada 1'20 2B3. gravity or vertical forces must be considered. This can be
2Present address: Geomechanics Research Centre, Laurentian Uni- achieved by combination of plasticity or limit equilibrium
versity, Sudbury, Ont., Canada P3E 2C6. techniques with the CCM. In this manner, the ground conver-

PriDted in Canada I Imprim6 au Canada

 3C - disc 76p
WONG AND KAISER: 1 321

(b) ( C)
FIG. 1. Modes of yielding: (a) mode A, 0"\ - O"r = max; (b) mode B, O"y - O"r = max; (c) mode C, 0"\ - O"y = max.

gence curve for the shaft wall can be derived and the required actual conditions in the field. They provide either a lower limit
support pressure can be rationally assessed. The limits of of support pressure that is needed to prevent collapse or a
applicability of conventional methods for shaft design can then design pressure that should never be exceeded in reality. For
be evaluated properly and field measurements can be inter- comparison with field measurements, a performance model,
preted based on a rational shaft performance model. such as the one presented later, is required that predicts the
The scope of this paper is to establish typical shaft behav- actual conditions for a given construction sequence. Most of
ioural modes and to quantify the responses of a shaft to excava- the conventional methods are design and not performance
tion, by use of the convergence-confinement concept, in models.
terms of support pressure and radial wall displacement. First,
conventional design methods are briefly and critically 2.1 Terzaghi's method
reviewed, and the proposed convergence-confinement Terzaghi (1943) already recognized that the problem of shaft
method is introduced. Behavioural modes ranging from elastic design was not as simple as merely rotating a tunnel in a
response through yield initiation and propagation to ultimate weightless medium and provided an elegant, approximate sol-
collapse are discussed. In the second part, results of the pro- ution for considering the effect of gravity, if it dominates shaft
posed CCM and the conventional design methods are com- behaviour. He assumed that stress concentrations near a ver-
pared with results from finite element analyses. Throughout tical hole would cause yielding due to the stress difference
the paper, it is assumed that the magnitude of the horizontal between tangential and radial stresses in an annulus around a
stess is independent of its orientation (Uh = u~. shaft (mode A: due to Ut - U r ; see Fig. la). Even though the
The technique proposed in this paper has also been applied tangential stresses are generally greater than the vertical,
to several case histories reported by Lade et al. (1981), Miiller- equalization of vertical and tangential stresses near the shaft
Kirchenbauer et al. (1980), and Kaiser and Wong (1984), and was assumed. Based on this approximate stress distribution,
good agreement was observed. The results of this comparison the limit equilibrium state of a downward-sliding, cylindrical
will be presented in a later publication. element of yielded ground was assessed. Hence, the support
pressure calculated by Terzaghi (1943) is the minimum pres-
2. Review of previously proposed shaft design techniques sure required to prevent this mode of instability.
An example of this critical support pressure distribution is
Two groups of basic shaft design models can be identified: shown for a cohesionless soil in Fig. 2. It is only valid ifverti-
those considering gravity or the influence of the vertical princi- cal plasticity and mode A-type yielding is possible, that is, at
pal stress (e.g., Terzaghi 1943; Berezantzev 1958; Prater high K(,-values, as will be demonstrated later.
1977) and those treating a shaft as a two-dimensional, "hole- Terzaghi noticed that his solution neglected the effect of
in-a-plate" problem (e.g., Abel et al. 1979). As will be seen, principal stress rotation near the shaft due to nonzero shear
the latter approach is only valid under certain field stress con- stresses (McCreath 1980) and proposed the use of a reduced
ditions. friction angle cjJ*:
In all cases, support pressures are calculated that correspond
to one specific, often limiting, state of equilibrium and various cjJ* = cjJ - 5° for 30° < cjJ < 40°, c = 0
design criteria are evoked to ensure acceptable performance in for the Mohr-Coulomb failure criterion.
terms of ground deformation. A factor of safety is applied to
gain a sufficiently large margin against failure or to control 2.2 Berezantzev's method
deformations. Alternatively. stability and serviceability are Berezantzev (1958) proposed a method of earth pressure cal-
achieved by provision of sufficient support pressure to avoid culation for cylindrical retaining walls whereby the state of
onset of plasticity (allowable stress design). These approximate stress around the shaft is described by two differential equa-
methods of controlling deformations provide safety margins of tions of eqUilibrium. He used the Mohr-Coulomb failure cri-
unknown magnitude. Some conservative designers still support terion as a condition of plastic equilibrium and also made the
the concept of design for the virgin state of stress and are will- assumption of equal principal stresses (ut = Uy = U10 Ur = U3)
ing to accept high support cost to ensure an unnecessarily high to render the problem statically determinate. However, only
and unknown safety margin. one set of slip surfaces (mode B: due to Uy - ur) is assumed to
These design approaches are not intended to predict the occur inside the yielded soil (Fig. Ib).
322 CAN. GEOTECH. J. VOL. 25,1988

N
4>p= 30·
C = a
y = 20 kNIM 3

x/. "' - 1 - sin ~

.,

Berezantzev

Terzaghi

0.00 0.25 0.50 0.75 1.00

Ps I 8y
FIG. 2. Earth pressure distributions.

The earth pressure is detennined by Sokolovsky's numerical

"step-by-step" computation technique. An example of the I
resulting earth pressure distribution is shown in Fig. 2. It is
~I
almost- identical to the one predicted by Terzaghi. However,
Berezantzev's approach predicts sliding along a set of nearly ( b)
cone-shaped surfaces. With the assumption of equal major
FIG. 3. Arching near shaft: (a) horizontal section, 1-1; (b) vertical
principal stresses, mode A-type yielding along a set of vertical section through shaft.
planes is also possible to facilitate sliding along the conical
surfaces.
can be reasonably well predicted by Prater's method (A = Ko).
2.3 Prater's method
However, a further reduction of the radial pressure will cause
Prater (1977) approximated the sliding surface predicted by
arching, as illustrated by Fig. 3, and the vertical stress will
Berezantzev by a cone and conducted a limit equilibrium
drop while the tangential stress increases. Ultimately, vertical
analysis assuming a Mohr-Coulomb failure criterion for the
and tangential stresses equalize when the ground yields in both
sliding surface. At the vertical sides of a pie-shaped element,
modes, A and B. This ultimate condition is modelled by Bere-
he introduced a Wedging force T with a radial outward compo-
zantzev and approximated by Terzaghi. Prater's method does
nent and argued that a coefficient A for the determination of T
not predict the same minimum pressure required to maintain
should fall between the active earth pressure coefficient Ka and
stability. It is too conservative at shallow depth and too uncon-
the coefficient at rest Ko and that it should not be unity as
servative at greater depth.
assumed by Terzaghi or Berezantzev. The resulting earth pres-
sure distribution, shown in Fig. 2 for a cohesionless soil, indi- 2.4 "Hole-in-a-plate" approach
cates that this analysis leads to the unreasonable result of zero Many authon; have developed analytical procedures to cal-
support pressure requirement for stability of a deep shaft (e.g., culate the extent of yielding, the equilibrium support pressure,
H/a > 9 for cp = 30°, c = 0). and the related deformations for circular openings in a unifonn
This deficiency must be attributed to the fact that the magni- stress field and in perfectly plastic or strain-weakening ground
tude of the wedging force T is not limited. At depth, this force (e.g., Brown et al. 1983). These methods provide the ground
is actually controlled by mode A-type failure stresses and can- convergence curve, which can be combined with the support
not, as implicit in Prater's analysis, reach infinity at great confinement curve to predict an equilibrium state of ground-
depth. Prater's approach does not predict a critical (minimum) support interaction. Mode A-type failure is tacitly assumed for
support pressure, as Terzaghi, but the earth pressure at the ini- this approach.
tiation of mode B-type failure. Lower radial support pressures An early attempt to apply this concept for the interpretatiop
are sufficient to maintain stability if stress redistribution due to of field measurements was presented by Abel et al. (1979) in a
mode A-type yielding is permitted during shaft wall deforma- back analysis of a well-documented deep shaft. In their calcu-
tion. This is illustrated by Fig. lb. For Ko less than a critical lation of lining pressures, they neglected the rock displace-
value Ker (to be defmed later), radial stresses may decrease ments before liner installation but considered the effect of
while tangential stresses increase until fail~ occurs due to the elastic, shrinkage, and creep deformations of the liner. In a
deviatoric stress difference (uv - ur) in mode B. At this point, reevaluation of this case history, McCreath (1980) found that
A = ulUvo is somewhat greater than Ko and the earth pressure the CCM could be applied successfully to explain, without
 WONG AND KAISER: 1 323

consideration of gravity, the shaft perfonnance of this deep must be introduced by application of reasonable design criteria
shaft in yielding rock. Because ground defonnations were to ensure satisfactory shaft perfonnance.
restricted by the shaft support, the measured ground pressures
exceeded, as would be expected, the minimum pressures pre- 3.1 Horizontal arching
dicted by the methods proposed by Terzaghi and Berezantzev. 3.1.1 Support pressure
In summary, the existing methods are tied to specific condi- In this section, the support pressure is calculated by uncoupl-
tions and thus none of them permits prediction of the actual ing the horizontal from the vertical arching. Only equilibrium
shaft perfonnance. These methods do not indicate when gravity in the horizontal plane is considered.
effects are relevant or when the limits of applicability of the Prior to excavation, a soil element adjacent to the shaft wall
-------,-,
"=- ho-'-'l,-"e'---o
-i-= n-=-=a---'-p:..::l--'-'
at'---e-7'''=--
ap-'-p=--r---'---'
oac--'=h----='h-
av-'--e-'-=-b-ee
--'-'n'--'-'-r=-a
---'--.:C e -'-'Jc""h'--e~
dC::
. -=T='h""e"---
se d c-'
efi-=
l'_
-----tiHUbjeeted-to-in-situi!tresses-;---%e-exeavation-ofthe-shaft-may~--
ciencies are largely overcome by the proposed design method be simulated by progressively reducing the support pressure
described in the following. (or radial stress). For the axisymmetric case in elastic soil, the
stress distributions are given by [1] - [3] (Terzaghi 1943). For
notations see Appendix A.
3. New shaft design method-proposed shaft design and [1] Uvo = "'(h = Po
performance prediction method
The convergence-confinement method (CCM) provides a [2] Ur = KoPo - [KoPo - pd (a/r)2
conceptual framework that can be analytically fonnulated to
predict the interrelationship between stresses and displace- [3] Ut = KoPo + [KoPo - pd (air?
ments in the soil near an underground opening (e.g., Fenner
1939; Pacher 1964). The extent of the yield (or plastic) zone These equations are derived based on the following assump-
can also be estimated by this method under certain, well- tions: (i) uv , Un and Ut are principal stresses, (ii) shear stresses
defined conditions. The CCM is adopted and proposed in the along the shaft wall are negligible, and (iii) the bottom of the
following as a rational approach to predict shaft behaviour. In shaft is remote.
this manner most relevant factors (e.g., in situ stress, soil As Pi is reduced the stress differences increase and the
strength, and defonnation properties as well as many construc- strength of the soil may be exceeded. The onset of the plastic-
tion details) can be included in the analysis. ity and the mode of yield initiation depends on the value of Ko
and the strength parameters of the soil. For simplicity, the fol-
The behaviour of a shaft is affected and near the surface
lowing derivations are for cohesionless materials only. The
dominated by gravitational forces. It is a truly three-dimen-
fonnulations for cohesive soils are given in Appendix B.
sional problem and all three stress components (ut , uv , and ur)
For a purely frictional, elastic, perfectly plastic material with
must be considered. The mode of yielding, its initiation and
a linear Mohr-Coulomb failure criterion, the maximum stress
propagation, depends on the relative magnitude of these
ratio that may be sustained is
stresses and, hence, on the initial in situ state of stress or Ko. In
order to apply the two-dimensional' 'hole-in-a-plate" model to [4] Ut/U3 = N = tan2 (11"/4 + tP/2)
determine the relationship between support pressure and shaft
wall displacement, it is necessary to understand the Three modes of yield initiation can be calculated from
mechanisms of shaft behaviour, so that an adjustment can be [1]-[3], and the corresponding support pressures, Pi> are
made to the CCM. [5] Pi = 2KoPo/(N + 1) (mode A: Ut - ur)
The excavation of a shaft can be simulated by a stress relief
(.::lur at the shaft wall) causing the surrounding soil to defonn [6] Pi = Po/N (mode B: UV - U r)
both horizontally and vertically. Excessive stress relief will
induce yieldirig near the opening, causing pennanent plastic [7] Pi = (2Ko - N)po
defonnations. The magnitude of support pressure, wall dis-
placement, and extent of plastic zone are related. The stress The largest value of Pi will govern which mode of yield initi-
relief during shaft excavation causes stress redistribution near ation occurs. In tenns of Ko-values, failure is initiated if
the opening, and thus induces vertical and horizontal arching [8] (N + 1)/2N < Ko < (N + 1)/2 (mode A)
as illustrated by the arrows in the yielded zones in Fig. 3. This
figure illustrates the development of horizontal arching around [9] Ko < (N + 1)/2N (mode B)
a circular shaft subjected to an isotropic stress field. The radial
stress relief causes a tangential or horizontal hoop stress [10] Ko> (N + 1)/2 (mode C)
increase in a horizontal plane, i.e., in the tangential direction.
Vertical arching with convex downward stress trajectories For cohesionless soil with tP = 30° or N = 3, failure is initi-
(Handy 1985) arises when a plastic zone of limited extent tends ated if
to move vertically downward (Fig. 3b). It develops one-sided 0.67 < Ko < 2.0 (Fig. la) (mode A)
or between the shaft support and the surrounding unyielded
soil mass. Collapse is prevented if stable arches can be main- Ko < 0.67 (Fig. Ib) (mode B)
tained by sufficient support pressure.
In the following, the relationship between the support Ko > 2.0 (Fig. lc) (mode C)
pressure, Pi> the extent of the plastic zone, R, and the wall dis-
placement, ui> or the ground convergence curve (GeC) is Mode A is the mode commonly evaluated for tunnels in
derived considering horizontal and vertical arching. yielding ground. Mode C, although possible, has been
It must be stressed that the proposed model is a perfonnance neglected in the following analysis because it is seldom of
prediction and not a design model. Adequate safety margins practical significance in soft ground.
324 CAN. GEOTECH. J. VOL. 25, 1988

Stress Distribution Stress State

around Opening at Shaft Wall

(1) T

tI
I

The boundary between the two remaining modes of yield ini- and the stresses within the plastic zone (r < Rtr) are
tiation is then given by a critical Ko-value: Kcr = (N + 1)/2N.
[13] (lr = Pi(r/a~-1
For each of these two modes, the relationship of support pres-
sure and extent of plastic zone can be derived separately. [14] (It = Npi(r/a~-1
3.1.1a Initiation of yielding ofmodeAatKo > K.:r (Fig. 4a) As the tangential stress decreases during yielding it will
~ventually become equal to the vertical stress. At this stage Pi
Figure 4b shows the sequential stages of stress redistribution
IS equal to Kapo. Substitution of this value into [12] provides
for mode A as the internal support pressure is reduced.
During the first stage (Fig. 4b (1», the vertical stress the extent Rtr (for (It = (Iv > (lr and (lvo).
remains constant, the radial stress decreases, while the tangen- Further reduction of the internal pressure (stage 3) will cause
both the tangential and the vertical stress to decrease and arch-
tial stess increases according to [1] -[3]. Yield initiation of
ing will now occur in horizontal and vertical directions. Near
mode A occurs if the condition
the opening, where (It = (Iv < Po, both modes of yielding (A
[11] (It!(lr =N or Pi = 2KoPo/(N + 1) and B) are evident. The extent of this zone is denoted by Rvr •
Between Rvr and Rtr> yielding is still governed by mode A only
is reached. The vertical stress at yield initiation always acts as and Rtr can be determined from [12]. The extent of the plastic
an intermediate stress. zone Rvr (modes A and B) is found by using the condition (It =
Further reduction of the fictitious internal support pressure
(Iv = Po (i.e., [14] and [1]):
causes the propagation of a plastic zone, and the tangential
stress decreases to satisfy the failure criteria (Fig. 4b (2». The [15] Rvr = a(po/pYI(N-l)
extent of the plastic zone is (Ladanyi 1974)
After propagation of yielding, the plastic zone contains a
[12] Rtr = a[2Kopo/(N + l)pd l/(N-l) region where mode A exists alone (Fig. Sa) and a region, close
 WONG AND KAISER: 1 325

The plastic zone Rtr is as before ([12]):

Mod.s A. B Mod.s A. B [22] Rtr = a[2Kopo/(N + l)pd ll(N-l)
For the range Rtr < r < Rvro the radial stress distribution is
governed by the stress equilibrium with the tangential stress
( [2]):
[23] Ur = KoPo - (KoPo - u;)(Rtr/r)2
where-u' -= 2[(, from
(a) (b) [20] and [22]).
FIG. 5. Extent and types of yield zones (vertical sections through Substituting r = Rtr and u; into [23] provides the radial stress
shaft): (a) Ko > Ker; (b) Ko < K er · distribution in the outer plastic zone (Rtr < r < Rvr)' Conti-
nuity of radial stresses at the elastic zone boundary (ur = KaPo)
locates the extent of the plastic zone, Rvr> as

(Rv~V = [Ko(~)(Pi~;:ol»)]2/(N+l)
to the shaft, where modes A and B occur simultaneously.
3.1.1 b Initiation of yielding by mode B at Ko < Kcr (Fig. 6a) [24]
Figure 6b shows the stress distributions in a horizontal sec- a) (Ko - Ka)
tion through the shaft at a particular depth in sequential stages
of reducing the internal support pressure. Again, after propagation of yielding, the plastic zone con-
At stage 1, yielding initiates at the wall in mode B. The tains a region where only one mode (mode B) exists (Fig. 5b)
stress distributions are given by [1] - [3] and the internal pres- and a region, close to the shaft, where modes A and B occur
sure is Pi = uv/N = Kapo at yield initiation. simultaneously.
As Pi is further reduced, Ut still increases due to wedging and Equations [1] - [24] are valid for conditions where horizon-
U v decreases due to vertical arching during downward displace- tal arching dominates. The influence of gravity was, so far,
ment of the soil near the wall (stage 2 in Fig. 6b). Yielding neglected and will, in section 3.2, be derived separately.
propagates outward from the wall. The radial and tangential 3.1.2 Wall displacement (uj)
stress distributions are given by [2] and [3] until the stresses at Once the relationship between the support pressure and the
the wall satisfy the failure criteria extent of the plastic zone has been established, the correspond-
[16] ut!ur =N ing wall displacement induced by stress relief can be deter-
mined. Some restrictive assumptions must be made to obtain a
The extent of the plastic zone due to the mode B yielding at closed-form solution for the ground convergence curve (Pi -
this stage is determined by equating radial stresses at the Ui relationship). For horizontal arching, it is, as before,
elastic-plastic interface for stress continuity. At this bound- assumed that solutions for plane strain condition constitute a
ary, the radial stress in the elastic region is given by [2], and in reasonable approximation, and that the radial wall conver-
the plastic zone, by gence is due to changes in Ut and Ur only.
[17] ur = uv!N and Uv = Po For the initial, elastic response, the wall displacements may
be calculated using Lame's equation:
Equating [2] and [17] yields
[25] Ui = a[(Kopo - Pi) (1 + /I)]/E
[18] (Rv~2 _ KoPo - Pi
After yielding is initiated, the wall displacement depends on
\ a) (Ko - KJ Po the adopted constitutive model of the soil. Brown et al. (1983)
At the end of stage 2, the tangential stress becomes equal to presented a summary of currently available GeC formulations
the vertical stress and the support pressure is given by [11], or for different material models and for the idealistic case of a
circular opening in an isotropic stress field under plane strain
[19] Pi = 2KoPo/(N + 1) conditions. For simplicity, the displacements may be approxi-
Mode A is initiated at the wall. mated by application of the solution proposed by Ladanyi
Further relief of Pi> stage 3, causes a simultaneous decrease (1974):
of U v and U t (satisfying the failure criteria given by [16] and [26] Ui = a{1 - [(1 - eav)/(1 + Av)]1/2}
[17]). At this stage, the maximum extent of the plastic zone
(Rvr in Fig. 6b) is still governed by mode B-type yielding, but where eav and Av are plastic dilation and deformation param-
the radial stress distribution will differ from that in [2] because eters defined by Ladanyi (1974).
of the need to satisfy [16]. Hence, in order to determine the This implies that the entire yield zone is treated as if mode A
radius of the plastic zone at this stage, the radial stress distribu- would exist throughout. The extent of the plastic zone, needed
tion needs to be calculated. for [26], is assumed to be given by the maximum extent
Figure 6b, stage 3, shows that inside the plastic zone (r < described by [12] and [24], respectively.
Rtr), where yielding in modes A and B occurs (ut = Uv > ur),
3.2 Vertical arching
the radial and tangential stress distributions are (Ladanyi 1974)
3.2.1 Support pressure
[20] Ur = Pi(rla)N-l In the previous section, the support pressure was calculated
[21] U = Npi(rla)N-l to satisfy equilibrium in a horizontal plane only and vertical
t
equilibrium was neglected. It has been shown that different
if the influence of the vertical stress reduction is neglected. types of yielding are induced if sufficient deformations due to
326 CAN. GEOTECH. J. VOL. 25,1988

Stress Distribution Stress State

around Opening at Shaft Wall

",'
:'.,,'.
...
~

~ .. ', .
Slip lines

(2)
' ..:
,,
.,- .-- ... _-
..
'.
__ Elastic-plastic
interface
....
:.
~ I- Rvr

( 3) Rvr
t, - - -<Tv
--
t
(a)
,/
I

",
,/

., O'r O't=O'v < O'vo

, ,," O'r
1" ~Rtr

(b)
FIG. 6. Mode B yielding, Ko < Kcr •

radial stress relief are pennitted. Three modes, A, B, and C, created if radial stress relief is pennitted. Hence, two sets of
have been identified for yield initiation and propagation (Figs. slip surfaces exist in the inner zone near the shaft wall and only
1 and 5). For each mode, the plastic flow occurs along slip sur- one set in the outer zone. Sliding along any set of slip surfaces
faces whose ultimate shear strengths have been reached. The could occur due to gravity. However, unstable blocks tending
direction and shape of these failure surfaces differ for each to slide inward toward the shaft opening must be bounded by
mode, as indicated in Fig. 1. Since the shear resistance along two sets of slip surfaces for a kinematically possible failure
these surfaces has been fully mobilized due to the far-field mode. This condition is satisfied when modes A and B occur
stresses, the soil mass tends to slide along these surfaces simultaneously. In the outer zone, for mode A-type yielding,
toward the shaft under its own weight. Consequently, to pre- only vertical sliding along spiral-shaped surfaces is possible
vent instability, a support pressure, Pg' must be applied to the without overcoming additional resistances. No radial sliding
shaft wall in the area where gravity dominates. Horizontal component exists. For the case of mode B-type yielding (in the
arching leads to lower pressures in this region and Pg represents outer zone), the inclined slip surfaces are formed initially but
the total support pressure required to maintain stability. This blocks are prevented from sliding inward because of wedging
phenomenon is referred to in this paper as "gravity effect." action and because the ultimate strength has not yet been
Vertical arching may, however, develop if sufficient vertical reached for tangential loading. The gravitational support pres-
movement is pennitted. From the orientation of the slip sur- sure, Pg' required to prevent instability therefore arises only
faces shown in Fig. 1, it is evident that gravity effects are more from the inner zone with two sets of slip surfaces. Neverthe-
dominant in modes A and B than in mode C. In mode C, slid- less, because of the close proximity of the state of stress in the
ing occurs in the tangential direction along spiral-shaped sur- outer zone to the failure state, the gravitational support pres-
faces with no component in the radial direction. sure is calculated in this paper based on the maximum extent
As illustrated earlier (Fig. 5), two zones of yielding contain- for both zones.
ing one or both of two types of yield modes (A and (or) B) are This support pressure, Pg' can be determined by application
 WONG AND KAISER: 1 327

of one of two classical approaches: (1) limit state equilibrium and used as an indicator of the width of the surface settlement
(LEM) and (2) plastic equilibrium (PEM). Prater (1977) trough.
adopted the fIrst approach to calculate the support pressure
required to prevent initiation of sliding along conical failure 3.3 Summary of design procedure based on convergence-
surfaces without considering the influence of horizontal arch- confinement method
ing. The influence of yield propagation, the effect of the extent It is suggested that the behaviour of a shaft can be described
of the yield zone, and the related stress redistribution were not and treated by independently considering horizontal and verti-
included in this approach. Consequently, the total support cal arching near the shaft. These two arching actions can be
pressure could not be correctly determined. quantifIed by use of the CCM with inclusion of gravity effects.
-----=---...I""'h=e--=s=ec=o=n'""dr-ca=p=p=ro=a~cth--tfi~a=s--=c=o'=mm=o=nI='=y-tbee=n~u=sea~t=o~d2-=e=te=rmt~n~e~~ln-this-fashion, it is possible to derive a relatioDshi~
the lateral pressure exerted by a flowing mass, e.g., for hori- the support pressure and the radial displacement. The CCM for
zontal stress determination in silos (Kendal 1980). Handy shaft design consists of the following steps (refer to Fig. 8):
(1985) applied this approach after some adjustment for soil- 1. Identify the mode of yielding near the shaft wall. It depends
wall interaction to estimate the lateral pressures on retaining on the initial in situ stress (Ko); it governs the extent of the
walls, and demonstrated satisfactory agreement with results plastic zone and the shape of convergence curve.
from model tests. The conditions near a shaft resemble those 2. Calculate the ground convergence curves for a specifIc ele-
near a retaining wall. Hence, the plastic equilibrium approach vation hi and determine the extent of the plastic zone, using the
is adopted in the following to calculate the total support pres- appropriate two-dimensional model (Fig. Sa). Neglect gravity
sure, Pg. This pressure is determined by assuming that slip sur- effects.
faces were created by horizontal arching. It represents the total 3. Select a specifIc wall displacement, us, based on a servicea-
support pressure required to maintain stability. bility design criteria and establish pressure versus depth and
Figure 7 shows the forces acting on a horizontal soil element plastic zone versus depth relationships from results of step 2
within the yield zone. Summation of the vertical stresses on the (Figs. 8b and 8c; full line).
element, for simplicity assumed to be uniform, leads to 4. With the confIguration of the plastic zone around the shaft,
determine the gravitational support pressure distribution, Pg'
[27] ['Y -
dO"y = 211"0"y f.~slLsR + KWlLwa\ldh with depth (Fig. 8c; dashed line).
a)J
A \ sm 5. To the two pressure distributions due to horizontal (full line)
and vertical (dashed line) arching form an envelope of required
where A = sectional area of the yield zone (= 1I"(R - a)2); support pressure (Fig. 8c; shaded) for the specifIed wall dis-
a = angle to the inclined yield surface (a = 90° for mode A; placement us'
a = 45° + cPI2 for mode B); Ks = stress ratio at wall-soil 6. Adjust the pressure envelope at the bottom of the shaft for
interface (= K a); ILs = frictional coefficient at soil-soil inter- the reduced pressure caused by face effects (Panet and Guenot
face; and ILw = frictional coefficient at soil-wall interface. 1982).
Integration of [27] provides the vertical stress, O"y, distribu- 7. Multiply pressure envelope by an appropriate load factor to
tion with depth, and the total horizontal pressure required to arrive at a design envelope. The resulting pressure should be
prevent instability is thus given by the plastic equilibrium con- checked as not to exceed pressure at rest.
dition It follows from the above calculation steps that the resulting
[28] Pg = KaO"y design pressure distribution envelope depends on an assess-
ment of acceptable or achievable ground movements.
An example is presented in the following section.
Equations [27] and [28] show that the support pressure, Pg'
increased with growing extent of yield zone, R. For a given 4. Verification of design method-comparison of proposed
size of yield zone, this pressure induced by mode A-type yield- solution with numerical simulation
ing is higher than that by mode B because of differences in the The numerical examples generated by application of the
orientation of the shear plane (Le., a). Hence, if two zones of fInite element method (FEM) are intended to compare results
different yielding mode occur, Pg is governed by the outer yield with the proposed method. They are not designed to simulate
zone radius for mode A yield initiation whereas the radius of any particular case history but are aimed at illustrating
the inner yield zone dominates design for mode B yield initia- mechanisms of typical shaft behaviour. For example, the mag-
tion because this zone has a steeper boundary (larger a) and nitudes of Ko were chosen to represent two extreme conditions
thus a greater gravity effect. (Ko < Ker and Ko > Ker) and create two different modes of
Equations [27] and [28] are applicable only for cohesionless yielding.
soils. The gravity effect in cohesive soils differs because the
shear strength of cohesive soils is independent of confIning 4.1 Finite element analyses
pressure. This case is presented in Appendix C. The fInite element program SAFE, developed at the Univer-
sity of Alberta by Chan (1985), was used for this comparison.
3.2.2 Displacements near the shaft collar The soil near the shaft was discretized for an axisymmetric
A solution for displacements considering gravity is not avail- shaft by pie-shaped elements. The confIguration of the mesh,
able and it is recommended, as a fIrst approximation, to consisting of 8-node quadrilateral isoparametric elements, is
neglect gravity for the prediction of wall displacements and 'to shown in Fig. 9. Zero displacement boundary conditions were
determine the convergence using the approach given in Section assumed at three boundaries (AB, BC, and CD) for the initial
3.1.2. stress determination by "switch on gravity." The boundary
The extent of the yield zone near the ground surface is not AB, representing the wall of a 2 m diameter shaft, was then
defIned because of zero confIning stresses at this level. How- allowed to move inward to simulate shaft construction. The
ever, by extrapolation from depth, its extent can be estimated specifIed displacement profIle is shown in Fig. 9. It enforces
328 CAN. GEOTECH. l . VOL. 25, 1988

TABLE 1. Input data for finite element analyses of shaft

Soil property
No. of tJ> qu E 'Y
analysis Soil model (%) (kPa) (MPa) (kN/m3) II Ko Ker K.
SMI Elastic, perfectly plastic, associated flow rule, 33 0 38 20 0.29 0.409 0.647 0.295
. Mohr-Coulomb (constant with depth)
SM2 As above 33 0 38 20 0.40 0.980 0.647 0.295
=(oonstant-witJj:Qepth)=
CMI Elastic, perfectly plastic, von Mises 0 60 40 20 0.495 0.980 0.625 -0.864
(constant with depth)

a constant wall displacement (ui) along the. shaft except near

the base.
Three analyses were performed to investigate the shaft
I
behaviour in purely cohesionless and cohesive soils. For cohe-
sionless soil, an elastic, perfectly plastic model and the Mohr-
Coulomb yield criteria with associated flow rule were used and
a set of typical soil parameters was selected (Table I). Two
Ko-values, 0.409 and 0.980, were chosen to create two typical
modes of yielding.
d:b
Because of the assumed associated flow rule for the cohe-
sionless soil, deformations due to dilation and, hence, arching
effects will be exaggerated as compared to what might be
expected in reality. For cohesive soil, the unconfined compres-

W
sion strength of the ground was assumed to be constant with \(\----'--
depth and the elastic, perfectly plastic model with the von
Mises yield criteria with no volume change was employed.
A minimal surcharge (equivalent to 3 or 4 m of cohesionless R
or cohesive soil, respectively) was applied at the ground sur- I
face, to overcome the problem of yielding at zero confinement FIG. 7. Effect of gravity during vertical arching.
near the ground surface. Some numerical convergence prob-
lems were experienced. This is commonly encountered in
FEM analysis involving nonlinearity, plasticity near the limit 3.8 m) for case SM2 is plotted in Fig. 13 for comparison with
equilibrium state (De Borst and Vermeer 1984; Griffiths and stresses calculated by the CCM. This particular section is
Koutsabeloulis 1985). The incremental displacement step after remote from the ground surface and the shaft bottom. Stress
formation of a plastic zone was selected at ufa = 0.01 %, distributions at two displacement stages are plotted: first, when
resulting in 4-10 iterations for reasonable convergence toler- yielding at the wall is initiated (at ufa = 0.26%) and, second,
ances. after much yielding has occurrred. The results in Fig. 13 indi-
cate that yield initiation occurred in mode A, as expected. The
4.2 Results from finite element analysis
stress distributions predicted by the CCM agree well with
4.2.1 Stress distribution
those from the FEM.
The radial, tangential, and vertical stress distributions
around the shaft after the final displacement step are plotted in 4.2.2 Radial displacement and plastic zone radius devel-
Figs. 10-12 for three soil or stress conditions. Yielding has opment
occurred in all cases. In SMI (Fig. 10), yielding was induced Figures 14-16 present results of the radial support pres-
by mode B, and the tangential stress is always the intermediate sure - wall displacement and support pressure - plastic zone
stress during propagation. In contrast, in cases SM2 and CMI, radius relatioJ),ships obtained from the FE analyses (designated
mode A governs the yield initiation and the vertical stress is by crosses) and by the CCM (full lines). The numbers on the
always the intermediate stress. However, the stress distribution GeC plot (Pi - Ui) correspond to the different stress stages
patterns are similar for the three cases, except near the lower described in Figs. 4 and 6. In zone I, the ground responds
boundary. The radial stress decreases toward the shaft wall elastically; line E separates elastic from plastic behaviour.
because of stress relief and causes stress redistribution as hori- Beyond line E yielding is induced for SMI (Fig. 14) by mode
zontal arching develops. The tangential stresses initially B, and this results in a nonlinear response curve. Further stress
increase in the elastic and plastic regions but decrease during relief causes a reduction of the tangential stress near the wall
yield propagation close to the shaft. until it becomes equal to (Iv. This state is represented by line T.
The drastic reduction of vertical stresses near the shaft wall For cases SM2 and CMI, the mode of yielding differs, i.e.,
indicates that arching in vertical planes near the elastic - plastic initiation by mode A. Lines E and T in case eMI are parallel
interface develops in all cases (the shaft wall friction is zero). and vertical because yielding in purely cohesive soil is induced
As a consequence, a slight increase of the vertical stress is at a constant stress difference equal to the compressive strength
observed at the interface. of the soil.
The stress distributions at one horizontal section (depth h = The Pi - R plots demonstrate outward propagation of the
 WONG AND KAISER: 1 329

Face effect

Radius of Wall Displacement suppon Pressure p

Plastic Zone
(a) (b) (c)

FIG. 8. Shaft design approach based on convergence-confinement method with gravity effect: (a) GeC at various depths without gravity
effect; (b) extent of plastic zone and pressure distribution at Us without gravity effect; (c) pressure distribution from CCM with gravity effect.

plastic zone due to support pressure reduction. The results pre- ~

dicted by the CCM agree reasonably well with those of the FE
analyses. I A
~-.~--.---------.------------,
D

4.2.3 Extent of plastic zone with depth

The extent of the plastic zone in a vertical section at two dis-
C
I

placement levels obtained by both methods, FEM and CCM,

is presented in Fig. 17. Because of the coarseness of ·the FE
mesh, the extent varies in a steplike fashion for the FE analysis
but the results of the two methods agree well, considering that
yielding is only indicated at the integration points of each ele-
ment. For the enforced wall displacement profile (Fig. 9), the
radius of the plastic zone decreases with depth for cohesionless
soils and forms a cone, whereas it is constant for a shaft in
cohesive soils, forming a cylinder (above a short cone). In
cohesionless soils, at the sam~ displacement, a larger plastic 4 6

zone develops for case SM1 at lower field stress (Ko = 0.41). DISTANCE

This implies that the pressure due to gravity in case SM.1 (Ko < FIG. 9. Finite element mesh for shaft analyses.
Ker) is more dominant than in case SM2 (Ko > Ker)' However,
the support pressure at a fixed deformation level is greater in
case SM2 because of higher in situ stress (see Fig. 18). ilIUund the shaft were not dominant in these cases because rela-
tively small displacements (ufa = 0.3-0.56%) were imposed.
4.2.4 Pressure distribution at wall
Large movements could not be simulated because of conver-
From the acc's (Figs. 14-16), the support pressures
gence problems. Nevertheless, the CCM provides an excellent
required to maintain equilibrium at a given displacement are
tool to predict the ground pressure at specified, large displace-
plotted in Fig. 18 for three displacements levels together with
ment levels.
results from the FE analysis. Both methods give consistent
results except at the bottom boundary. These figures clearly 4.3 Gravity effect and vertical arching
show that the support pressure is a function of wall displace- Because of difficulties in simulating behaviour at large dis-
ment and in situ stress. Pressures predicted by the CCM for placements by FE analysis, several analytical examples have
ufa = 0.8%, that is, 8 mm wall movement, are shown in been generated to illustrate the role of gravity for shaft design.
Fig. 18. The support pressure determined from the limit equili- Results determined by application of [27], [28]. and [C.1] for
brium method proposed by Berezantzev (1958) is included for cohesionless and cohesive soils are plotted in Figs. 19 and 20.
comparison. Large displacements would have to be allowed in Dimensions of the shaft and the shape as well as the extent of
order to obtain these minimum pressures. the assumed plastic zone are also shown.
The support pressure for a constant wall displacement Figure 19 indicates that for cohesionless soil, the support
increases steadily with depth for all three cases, even at hfa > pressure due to gravity effect increases as the yield zone
4, where the methods by Terzaghi (1943) and Berezantzev expands. The distribution of the support pressures depends
(1958) predict constant minimum support pressures. The rapid heavily on the assumed configuration of the plastic zone. It
increase at a depth of more than 6 m is an artifact of the reduces to zero at depth for cone-shaped plastic zones. For
enforced displacement distribution near the shaft bottom. cylindrical yield zones, as assumed by Terzaghi (1943), a con-
Gravity effects and vertical arching within the plastic zone stant pressure would be approached. Near the surface, the
330 CAN. GEOTECH. J. VOL. 25,1988

o~----------------------.~,-~------------------.~~--------------------.

J:
Ii: 4
ILl
C

234 5 6 7 2 3 4 5 6 7 2 3 456 7
DISTANCE

FIG. 10. Stress contours at ufa = 0.3% (SMl): (a) (JfO (b) (Jt. (c) (Jv.

O · ,-,,~~----------------,

80 '.r-. 80
1..-

o 120
J:
I-
a.
ILl ,
C r--
160 :..--- 160

~
o ~ 190
(c)
(0 ) (b)

23456 7 2 3 4 5 6 7 2 3 4 5 6 7
DISTANCE

FIG. 11. Stress contours at ufa = 0.5% (SM2): (a) (Jr. (b) (Jt> (c) (Jv.

J:
I-
a. 4
ILl
C

(o) (b) (c)

8
2 3 4 5 6 7 2 3 4 5 6 7 2 3 4 5 6 7

DISTANCE

FIG. 12. Stress contours at ufa = 0.56% (eMl): (a) (Jr. (b) (Jt> (c) (Jv.
 WONG AND KAISER: 1 331
240

220 Cohesionless, Ko=0.98 (SM2)

200

180

160

7 140 _~ C!) -.. ~ •• ~ -=-:- .-:::- -=- __

~ .......
•
•••_ - _ .. ~.:..!=" :: .:. :~.."": =.:::-4
. l!l:::- -- T
rn 120 : "Iii •. ~ - + FEM:
rn
. ~+/
~ 100
: I!I Radial Stress at uta _ 0.26%
, .m I!I/ + C!I Tangential Stress at uta _ 0.26%
16 + lJ. Vertical Stress at uta - 0.26%

: t/
40 I
CCM:
+ Radial Stress at uta _ 0.5%
x
$
Tangential Stress at uta _ 0.5%
Vertical Stress at uta _ 0.5%

Radial Stress at uta - 0.26%

- .... -Tangential Stress at uta _ 0.26%

Radial Stress at uta - 0.5%

20 - - -Tangential Stress at uta - 0.5%

o
1 .0 2.0 9.0 4.0 5.0 6.0 7.0 8.0 9.0
DISTANCE FROM WALL (m)
FIG. 13. Stress distribution at h = 3.4 m (SM2).

pressure is close to the active earth pressure. two depths: for a section close to the surface (shallow) and for
In cohesive soil (Fig. 20), the support pressure to prevent a level at greater depth (deep). This diagram illustrates two
instability due to gravity also increases for larger yield zones. features of practical significance: (1) horizontal arching pre..
However, a distinct difference in vertical arching action exists vails for small wall displacements, that is, when good ground
between cohesionless and cohesive soils. For cohesive soils, control is exercised, and at greater depths; and (2) vertical
support pressure applied along the upper portion of the shaft arching and, hence, gravity effects dominate only when
does not enhance the stability because the shear strength of the "excessive" yielding is permitted and only at shallow depth.
cohesive soil is independent of the confining pressure. The It follows, then, that previously proposed design methods
support pressure to inhibit collapse must be applied where a are seldom in agreement with modem construction procedures
collapse mechanism is possible, generally near the bottom and design requirements enforcing good ground control to
(Britto and Kusakabe 1983). minimize surface settlement and ground disturbance.
Combining Figs. 18 and 19 provides the expected pressure
5.2 Limitations of the proposed approach
distribution for case SMI (Fig. 21). Near the collar, to about
5 m depth, gravity dominates. Below this level, the pressure is Many assumptions had to be made to arrive at closed..form
solutions for the CCM. It is imperative to investigate for prac..
displacement controlled (shown for u/a ~ 0.5%). Because this
pressure represents the actually expected pressure, a load..fac .. tical applications how these assumptions and approximations
affect the accuracy of the result. For the simulation of horizon-
tored pressure should be used for design of the shaft lining.
The load factor should, however, be selected so that the result.. tal arching, the two-dimensional plane strain "hole-in-plate"
ing design pressure does not exceed the stress at rest. model was employed. The neglect of shear stresses between
horizontal layers, whose effects have been investigated in
detail by Terzaghi (1943), could lead to an underestimation of
5. Discussion the extent of the yield zone, and thus a slightly unconservative
5.1 Horizontal versus vertical arching support pressure. This was already recognized by Terzaghi,
Stress relief due to the shaft excavation causes horizontal and who recommended use of a reduced friction angle for compen-
vertical arching near a shaft. It is of practical importance to sation (see Section 2.1). It must be realized that the assumed
understand under what conditions horizontal or vertical arch.. two-dimensional plane strain condition does not prevail at
ing dominates shaft behaviour and becomes relevant for shallow depth where the gravity effect dominates. However,
design. The CCM provides an effective tool to depict the role honzontal arching normally initiates yielding near the surface
of each type of arching quantitatively. Figure 22 presents a and, hence, correctly reflects the true shaft behaviour before
schematic GeC with the support pressure normalized to initial gravity dominates. At greater depth, the model is a reasonable
state of stress (KoPo). This reduces the GeC for different ele.. approximation for all displacement levels and distributions.
vations to one unique curve, if horizontal arching is considered For the calculation of the support pressure to resist gravity,
only and gravity is neglected. The normalized support pressure the vertical stress distribution is assumed to be uniform at each
due to gravity, pgfKoPo, during vertical arching is sketched for depth. This assumption is contradicted by the results of the
 332 CAN. GEOTECH. 1. VOL. 25, 1988

120 120 240 ............- - . - - . - -............--.--.....--,

110
SM 1: 110
SM 1: m220 _ 220
'i 100
-CCM
X X FEM (b)
Ii
11.
100
-CCM
X x FEM (b)
11. 200
~
11. 200
~
..II:
~ 90 CD 180 CD 180
- 90
f
~ 80
CD
~ 80 ~ 160 ~ 160
III III o ~
III
CII 70
III
CD 70
f 140 f 140
11. 11.
~ ~
-.
~
Il.
60

50
-.
--------~=-~~~----_+4_----+_--9_~~~======~------~Il.~~~-~--~~~~----~~--~~~--------.~-
Il.
Il.
~ 120

~
~ 120

~
~ ~ UI 80 UI 80
UI 40 UI
~ ~ 60
ia 30 ~ ~ ~ ~
G G
~ ~ II: 40 II: 40
G G
II: 20 II:
a 20 20
10 a 10

o 1 1.00 1.50 2.00

2 3 4 5 6 1.00 1.50 2.00 Radius of
Radius of Radial Displacement (mm)
Radial Displacement (mm) Plastic Zone (m)
Plastic Zone (m)
FIG. 15. Radial support pressure - displacement relationship and
FIG. 14. Radial support pressure displacement relationship extent of plastic zone (SM2, cohesionless, Ko = 0.98) at (a) h =
and extent of plastic zone (SMl, cohesionless, Ko = 0.41) at (a) h = 1.8 m, (b) h = 3.8 m, (c) h = 5.8 m, (d) h = 7.8 m.
1.8 m, (b) h = 3.8 m, (c) h = 5.8 m, and (d) h = 7.8 m.

r-.-...,..-,-...........,.........,

.
240

finite element analyses, which demonstrate that the vertical _220 220

stress increases rapidly away from the shaft wall. Thus, the • 200
~ ~ 200
assumption of a constant average (Iv results in a higher (conser- ! 180 ! 180
vative) support pressure Pg due to gravity effects ([27] and ~ ~
.160
: 160
[28]). The existence of shear stresses causes a rotation of the
E. 140
:
D: 140
yield plane (i.e., a reduction of the angle ex in [27]), and
neglect of this effect leads to a conservative estimate of Pg. 1: 120 1: 120
Ig. 100 IIl.l00
The effect of pore pressures is not considered in the pro- ~
UI
posed design approach, but it can be implemented without dif- 80 UI 80
ii
ficulties. Volume changes during pore pressure dissipation ~
•
60 'iii 60
can, however, cause significant deformations and related pres- II: 40 II: 40
a
sure increases. These time-dependent processes may govern
20 20
shaft. design and must be assessed separately.
5.3 Effect of variation of E and 'lu with depth on shape of 012945678 1.00 1.50 2.00 2.50
Radlu. of
plastic zone PI.atlc Zone (m)
For the previously presented analyses, Young's modulus, E,
FIG. 16. Radial support pressure - displacement relationship and
and shear strength, qu, were assumed to be independent of
extent of plastic zone (eM 1, cohesive, Ko = 0.98) at (a) h = 1.8 m,
depth. This assumption may not be valid in practice. Equations (b) h = 3.8 m, (c) h = 5.8 m, (d) h = 7.8 m.
governing the extent of the plastic zone depend on these
parameters. Consequently, shape and extent of the plastic zone
must be affected by this simplification. This effect has been in cohesionless soils. Hence, the method proposed by Bere-
studied by Wong (1986) for cohesionless and cohesive soils. zantzev (1958) seems appropriate for the determination of the
minimum support pressure requirement but for conditions of
5.3.1 Cohesionless soils
constant soil stiffness only.
For a constant modulus, the extent of the yield zone can be
determined as a function of depth using [12] and [18] and [24] 5.3.2 Cohesive soils
for modes A and B, respectively. For mode A and a constant Similar reasonings as for cohesionless soils are applicable to
imposed shaft wall displacement, yielding occurs on vertical cohesive soils. Equations governing the shape of the yield
(spiral) surfaces and its extent decreases rapidly with the depth. zones for modes A and B are given in Appendix B ([B.12] and
For mode B, yielding caused by the vertical-radial stress dif- [B. 18] -[B.24] respectively). Figure 23 illustrates possible
ference occurs along inclined (conical) surfaces following the configurations of the plastic zones for different cases. The
Rankine's slip lines and its configuration looks like the trun- plastic zone may be cone-shaped or cylindrical depending on
cated cone assumed by Prater (1977). the strength and deformation property variation with depth.
For soils with a linearly increasing modulus with depth, the The extent of the plastic zone generally diminishes rapidly at
profile of the yield zone will look as shown in Fig. 23. With a the bottom of the shaft because of the face effect.
constant wall displacement imposed, the magnitude of stress Britto and Kusakabe (1982, 1983, 1984) investigated theo-
relief with depth is larger and the extent of the yield zone is retically and experimentally the mechanisms of the collapse of
greater at depth. unsupported axisymmetric excavations in soft clays. Their
Figure 23 shows that cone-shaped yield zones may develop findings agree well with the shapes shown in Fig. 23.
 WONG AND KAISER: 1 333

,..cohesionless, Ko=O.41 Cohesionless, Ko=O.98 f.cohesive, Ko=O.98

fI I!) I
0.5 fI
b
i dJ
I
I!!

I!!
I
I
I!)

I!)
I
I!) I I!)/ II
I!! I!) I!! C!I I!!' C!I
I I

:/
I
2.0 II!
II!)
~ I!! I C!I
I
- - - - - - -- :I!! /I!) iii I!! C!I I
-E _'I!! I C!I
~ J I!! I I!) I

-•
I I
• r . 9.5
oI!J
~ ~ I!)/ I!!. C!I
I
a. I!! II!) ~ ~ I!! I I!)
I
Q 1m II!) r I!! I I!)

I
5.0 I
I
I!)
r I!! I C!l1

r I
I!! I!)
I!)

I!) I!!' I!) I

6.5
I, FEM:
C cuI a • 0.06 %
o 0 u I a • 0.30 %
C FEM:
C cuI a - 0.26 %
I
II!! ~
/'

~
o 0 u I a - 0.50 %

V
/
II!)
CCM:
_.- u I a • 0.06 %
- - u I a • 0.30 % ~
CCM:
- - - u I a - 0.26 %
- - u I a - O.SO %
I
•

fie .,/.,/
I!!
.,/
~
1o·£:
c CuI
uI
a - 0.20 %
a _ 0.56 %
I
I [e
6.0
1.6 2.2 2.6 1.0 1.6 2.2 2.6 1.0 2.2 2.6

Radial Distance from Shaft Wall (m)

FIG. 17. Comparison of extent of plastic zone (SMl, SM2, CMl).

\ - f' f~ i '\
0.4
!\ \ 5M1; Ko = 0.41 \ \ 5M2; K =0.98 \
CM1; Ko =0.98
0.8

1.2
I\ -~
\- ~
I
FEM:
D D u I a - 0.26 0/0 ,
i~
~ K.
FEM:
D D u I a - 0.26 0/0
o 0 u I a - O.SO 0/0
-
-\
\
\ FEM:
D D u I a _ 0.26 0/0
o 0 u I a - O.SO 0/0
o 0 u I a - O.SO 0/0
1.6 I \\- CCM: 1\\ CCM: \ CCM:
- __ Uo/a_O.260/0 - - - u I a - 0.26 0/0 - - - u I a - 0.26 0/0
i ~ - - u I a - 0.50 0/0 -\ - - u I a - O.SO 0/0
2.0

2.4 \ \-
I
- - u I a - O.SO 0/0
- - - u I a - 0.80 0/0 ~ ~
I I
- - - u I a - 0.80 0/0
-\ \
2.8 I- ~
I
~ -\\
3.2
I
I-
I
i . \ I.
I

.
I
i 3 •6 \ I \ I

\- \ i
'i' \
"
r' \ O

'" 4.4
'f'
I
: ~
I
I
I.
I
- \\ I.
I

4 .8 I- \ I. \ I.
~ I I
\, _\ I \

\
5 .2 • I
\: \
I \

\
I I
\
5.6
-
--~\ \ \. \:,\
~ .1
\ -\
6.0
.\ \ I

6.4
\ -"\
\ "\

."\
I

.\ \ - \ .' ' '

- \
\

\
I.
I

,
6.8

7.2
'-,-
"" '
),. I
'-
'-
...........
" I!I"' "--, ,
" "',,\
'
7.6 'Berezantzev \ ,-~, Berezantzev '- ~ ",
~
8.0
(with surcharge)

20 40 60 80
• -
100
(with surcharge)

40 80 120
....
'.::,

300
160 200 50 100 150 200 250
PRESSURE (kPa) PRESSURE (kPa) PRESSURE (kPal

FIG. 18. Comparison of radial pressure distributions (SMl, SM2, CMl).

334 CAN. GEOTECH. 1. VOL. 25, 1988

Cohesion less Soil:

a ~ 1 m
la~
0
'Y = 20 kN/rrr
<I> = 30°
I
/ Ks= Ka= Kw~ 0.33
I I80'ho,/ 2

E
H
.c 4
Extent of
c..
Q)

Yield Zone
c

o 5 10 15 20 25 30 35 40 45 50

Radial Support Pressure P 9 (kPa)

FIG. 19. Gravity effect due to vertical arching, cohesionless soils.

-ia I R

I H = 10 m
I
I
I a = 1m
c u = 40 kPa
\
\
I R(m) (kPa) \
p=LF-p;, LF=1·3

H
3 17
\
5 183 \
\
E
\
- 4 \
::c \
FIG. 20. Gravity effect due to vertical arching, cohesive soils. t-
eL \ I
LlJ
Cl
5
\ ,I
6. Conclusions 'I
The behaviour of shafts in cohesionless and cohesive soils,
6
/':\
including the transition from elastic response to yielding and to \
collapse, has been explored in detail. It depends on the mode 7
of yield initiation and propagation and the influence of gravity .
Stability of a shaft is enhanced by two types of arching, hori-
8 ____ ____ __ ______
zontal and vertical, that develop around a shaft opening if suf- ~ ~ ~~ ~ L-~~

o 20 40 60 80 100
ficient wall displacements are permitted during construction. It
was found that in cohesionless soils horizontal arching prevails RADIAL SUPPORT PRESSURE (kPa l.
at small wall displacements and at great depth, whereas verti- FIG. 21. Design pressure envelope for ufa = 0.8% (SM1).
cal arching is only mobilized after large displacements and
near the ground surface. For cohesive soils, when a distinct
collapse mechanism is established, the support pressure required support pressures with respect to displacement or
required to prevent the instability arises primarily from gravity serviceability design criteria. It differs from conventional
effects but yielding is caused and controlled by horizontal approaches in which the limit state of equilibrium is considered
arching. and, hence, reasonable safety margins must be selected care-
It is suggested that horizontal and vertical arching can be fully. The new design method has been verified in this paper
approximately quantified by application of the convergence- by comparison with results from finite element simulations.
confinement method with consideration of gravity effects. This The method provides reliable estimates of support pressure,
method provides a performance prediction model that predicts wall displacement, and extent of plastic zone. It constitutes a
 WONG AND KAISER: 1 335

PI
Horizontal Arching Vertical Arching
Ko Po

__ ~or~~ __ ©_:__f-7 ~
--L
, Ko Po

-----
Shallow Shaft\..
-~~-Pg
•

Shaft Wall Displacement

FIG. 22. Nonnalized ground convergence CUIVe, horizontal or vertical arching (cohesionless soils).

COHESI ONLE SS COHESI VE

MODE A (Ko>Kcr l MODE B (Ko< Kcrl MODE A (Ko> Kcr> MODE 8 (Ko< ~r
I T
E constant with depth E, qu constant with depth

y JT j7 lJ
E increasing with depth E, qu increasing with depth

IT IT IT lJ
E constant and qu increasing with depth

J7 y
•
FIG. 23. Shapes of plastic zones for cohesionless and cohesive soils.

rational design method as well as an approach for performance Acknowledgments

prediction and assessment of field measurements. Its validity The financial support provided by the Natural Sciences
was also verified by comparison with measurements from case and Engineering Research Council of Canada is gratefully
histories. This will be published in a future paper. acknowledged. The authors also wish to acknowledge D. R.
The proposed design method provides support pressure dis- McCreath for the preliminary work and for the stimulating
tribution with depth for specified wall displacements. Hence, it thoughts presented in his M.Eng. thesis.
was possible to evaluate the limits of applicability of conven-
tional design methods. These methods generally provide the ABEL, J. F., DOWIS, E., and RICHARDS, P. 1979. Concrete shaft
minimum support pressures required to prevent collapse. lining design. 20th U.S. Symposium on Rock Mechanics, Austin,
These values can, however, be reached only if the wall dis- TX, pp. 627 -633.
placements are large and, for most practical purposes, exces- BEREZANTZEV, V. G. 1958. Earth pressure on the c~lindrical retain-
sive and unacceptable. ing walls. Conference on Earth Pressure Problems, Brussels, Vol.
336 CAN. GEOTECH. J. VOL. 25, 1988

II, pp. 21-24. stress difference (mode A)

BRITTO, A. M., and KUSAKABE, O. 1982. Stability of axisymmetric radius of plastic zone induced by vertical-radial
excavations. Geotechnique, 32: 261-270. stress difference (mode B)
---1983. Stability of axisymmetric excavations in clays. ASCE r radial distance
Journal of Geotechnical Engineering, 109: 666-681.
radial displacement of shaft wall (at r = a)
---1984. On the stability of supported excavations. Canadian
Geotechnical Journal, 21: 338-348. ar radial stress
BROWN, E. T., BRAY, J. W., LADANYI, B., and HOEK, E. 1983. av vertical stress
Ground response curves for rock tunnels. ASCE Journal of Geo- at tangential stress
technical Engineering, 109: 12= 23. __al> a3 _ _ majoLand minor principal stress _ _ _ __
CHAN, D. 1985. Finite element analysis of strain-softening materials. T shear stress
Ph.D. thesis, Department of Civil Engineering, University of ah minor horizontal stress
Alberta, Edmonton, Alta. aH major horizontal stress
DE BoRST, R., and VERMEER, P. A. 1984. Possibility and limitations Pi = Ps internal or support pressure (radial stress) at wall
of finite element for limit analysis. Geotechnique, 34: 199-210. Po vertical initial in situ stress
FENNER, R. 1939. Untersuchungen zur Erlcenntnis des Gebirgs- stress due to gravity effect
druckes. Glueckauf, 74: 32-33. Pg
GRIFFITHS, D. V., and KOUTSABELOUUS, N. 1985. Discussion on "( unit weight of soil (= p g)
'Possibility and limitations of finite element for limit analysis.' p density of soil
Geotechnique, 35: 90-94. 11 Poisson's ratio
HANDY, R. L. 1985. The arch in soil arching. ASCE Journal ofGeo- E Young's modulus
technical Engineering, 111: 302-319. e av average plastic dilation (positive for volume
KAISER, P. K., and WONG, R. C. K. 1984. Interpretation of measure- decrease)
ment on tunnel in clay shale and shaft in Edmonton Till. Report Av parameter for calculating plastic deformation
submitted to Department of Water and Sanitation, City of Edmon- cp angle of internal friction of soil
ton, and Alberta Oil Sands Technology and Research Authority. N tan2 (45 0 + cp/2)
KENDAL, N. O. 1980. The practical design of silo and bunkers with unconfined compression strength of cohesive soil
reference to the Codes of Practice. International Conference on qu
Design of Silos for Strength and Flow, University of Lancaster, eu shear strength of cohesive soil
United Kingdom, Session 6. JLw wall friction coefficient
LADANYI, B. 1974. Use of the long-term strength concept in the JLs soil friction coefficient
determination of ground pressure on tunnel linings. Advances in K = A ratio of horizontal to vertical stress
Rock Mechanics, Thinl Congress of the International Society for Ks K at soil-soil boundary
Rock Mechanics, Vol. 2.B, pp. 1150-1156. Kw K at soil-wall boundary
LADE, P. V., JESSBERGER, H. L., MAKOWSKI, E., and JORDEN, P. ~ K at rest
1981. Modelling of deep shaft in centrifuge tests. Proceedings, Ka K in active state (= tan2 (45 0 - cp/2»
10th International Conference on Soil Mechanics and Foundation Kc critical K-value distinguishing mode A from mode B
Engineering, Vol. 1, pp. 683-691. r
MCCREATH, D. R. 1980. Analysis of formation pressure on tunnel
and shaft linings. M.Eng. report, Department of Civil Engineering, Appendix B. Mechanism of shaft behaviour in cohesive soil
University of Alberta, Edmonton, Alta. This appendix contains the equations governing shaft behav-
MULLER-K!RCHENBAUER, H., WALZ, B., and KLAPPERICH, H. 1980. iour in cohesive soils. Equations are designated in the same
Experimentelle und theoretische Untersuchungen zum Erddruck-
sequence as those for cohesionless soil. Thus, explanations
problem auf radial symmetrische Senkldisten und Schiichte. Grund-
bauninstitut der Technischen Universitiit, Berlin, Heft 7. given in the main text for cohesionless soils are generally appli-
PACHER, F. 1964. Measurements of deformations in a test gallery as a cable to this appendix. The notation is given in Appendix A.
means of investigating the behaviour of the rock mass and specify- The stresses in an elastic thick-walled hollow cylinder are
ing lining requirements. Felsmechanik und Ingenieurgeologie
(Rock Mechanics and Engineering Geology), Supplement I:
[B.l] a v = "(h = Po
149-161. [B.2] a r = ~Po - [~Po - Pi] (alr)2
PANET, M., and GUENOT, A. 1982. Analysis of convergence behind
the face of a tunnel. Tunnelling '82, The Institution of Mining and [B.3] at = ~Po + [~Po - Pi] (alr)2
Metallurgy, pp. 197-204.
PRATER, E. G. 1977. An examination of some theories of earth pres- For cohesive material the strength is assumed to be constant:
sure on shaft linings. Canadian Geotechnical Journal, 14: 91-106. [B.4]
TERZAGHI, K. 1943. Theoretical soil mechanics. John Wiley and
Sons, New Yorlc, NY. Hence, the support pressures for the three possible modes of
WONG, R. C. K. 1986. Design and performance evaluation of tunnels yield initiation are
and shafts. Ph.D. thesis, Department of Civil Engineering, Univer-
sity of Alberta, Edmonton, Alta. [B.5] Pi = KoPo - qu!2 mode A
[B.6] Pi = Po - qu modeB
Appendix A. List of symbols
[B.7] Pi = (2~ - l)po - qu modeC
a radius of shaft
H depth of shaft The largest value of Pi will govern the mode of yield initiation.
h depth measured from ground surface This can be expressed in terms of required ~-values:
A area of cross section
R radius of plastic zone
[B.8] [l - qu/2po] < ~ < [qu/2po + 1] mode A
Rtr radius of plastic zone induced by tangential-radial [B.9] ~ < [1 - qu/2po] mode B
 WONG AND KAISER: 1 337

modeC R
As for cohesionless soils, mode C has been neglected in the
following analysis. A critical K-value, Kcr, is used to distin-
Ja~
guish mode A from mode B.

ur) at Ko > Ka-

I

~
Mode A (ut -

~W
The yielding initiates at the condition of
____~ntJJL_~-~=2~p~-2pi=q'~----------------------~--_1

~
A further decrease in the support pressure causes a propagation
of a plastic zone to
H
fs
•
[B. 12] Rtr = a{exp [(KQpo - Pi)/qu - 112])
Stresses within the plastic zone Rtr are
Pg _J~~_
[B. 13] Ur = Pi+ qu [In (rIa)]
[B. 14] U t = Pi + qu [l + In (rIa)]

Additional stress relief causes yielding in mode B (together

with mode A and the extent of this zone, Rvro is FIG. C.l. Gravity effect and vertical arching in cohesive soil.
[B. 15] Rvr = a exp [(Pi - Po)/qJ
[B.23] ur = KQpo - (KQpo - u;)(Rtr/r)2
Mode B (uv - ur) at Ko < Kcr
Yielding initiates at where u; = 2KQpo - qu at r = R tr •
Substituting r = Rtr and u; into [B.23] yields the radial stress
[B.16] Pi = Uv - qu
distribution in the plastic zone (Rtr < r < Rvr)' Continuity of
As Pi is further reduced, the vertical stress decreases as a radial stresses at the elastic zone boundary (ur = Kauv) locates
result of the vertical arching and the tangential stress will the extent of the plastic zone, Rvr , as
increase until it becomes equal to the vertical stress at the wall.
At this state, the radial stress at the elastic -plastic boundary is
given by
[B. 24] (RvrV = (KQpo -
a)
~
qJ exp [2 KQP;u- Pi -
(1-KQ)Po+qu
+)]
[B. 17] Ur = Uv - qu and Uv = Po
The wall displacements are calculated based on the model
Equating this stress with that in the elastic zone given by [B.2] proposed by Ladanyi (1974). For elastic ground response, the
provides the radius of the plastic zone: . wall displacement is
[B. IS] (Rv~2 _ KQpo - Pi [B.25] Ui = a[(KQpo - Pi)(1 + JI}]/E
a J (KQ - l)po + qu
If yielding occurs, the displacements are
and the support pressure is
[B.26] Ui = a{l - [11(1 + Av)]1I2}
[B. 19] Pi = KQpo - qul2
Further relief in support pressure causes propagation of the Appendix C. Gravity effect and vertical arching in cohesive
plastic zone. The mode of yielding is still mode B. But near the soils
shaft wall, there exists a zone where the tangential stress is The shear resistance of cohesive materials is independent of
equal to the vertical stress. Within this zone, the radial and tan- confining pressure. Hence, applying horizontal pressure on the
gential stresses, governed by the failure criteria, are given as shaft wall does not enhance the strength of the soil in the
(Ladanyi 1974) plastic zone. Hence, it is necessary to identify possible col-
[B.20] Ur = Pi+ qu [In (rIa)] lapse mechanisms and to inhibit these by application of exter-
nal forces.
• [B.2l] U t = Pi + qu [l + In (rIa)] Figure C.l shows one possible collapse mechanism. The
and the extent of this zone is cyllildrical plastic zone will cave in (slide vertically) and exert
a force, W', on the conical soil mass underneath. A horizontal
[B.22] Rtr = a {exp [(KQpo - Pi)/qu - 112]) pressure must be applied to inhibit failure of this soil cone. The
required total force, Pg, is given by considering the force
The radial stress distribution in the zone of mode B yielding
equilibrium along the inclined failure surface of the conical
must be known for the determination of the extent of the plastic
soil mass:
zone. Between Rtr < r < Rvr (shown in Fig. 6b), conditions of
equilibrium provide the radial stress: [C.l] (W' + We) sin a - Picos a) - Fe = 0
View publication stats