Está en la página 1de 4

International Journal of Rock Mechanics & Mining Sciences 40 (2003) 601604

Discussion
Discussion on the paper Ultimate bearing capacity at the tip of a
pile in rockPart 1: theory
William G. Pariseau*
Department of Mining Engineering, University of Utah, 315 WC Browning Building, Salt Lake City, UT 84112-0113, USA
Accepted 16 February 2003
1. Introduction
The authors of the subject paper (International
Journal of Rock Mechanics and Mining Sciences
2002;39(7):83346) obtain estimates of upper bounds
to bearing capacity of an end-loaded pile after a detailed
development and analysis of the problem within the
context of traditional plasticity theory and a two-
dimensional specialization to slipline theory. The
problem is essentially one of a at, cylindrical indenter
penetrating a rigid-plastic medium and is based on stress
equations of equilibrium and a yield condition. In case
of axial symmetry, rather than plane strain or plane
stress, some assumption about the intermediate princi-
pal stress is made. The resulting system then consists of
three equations in three unknowns and leads to upper
bounds of collapse loads after integration along the
sliplines that are also the stress characteristics of the
system. The authors use the popular HoekBrown (HB)
failure criterion. Because HB implies two distinct shear
strengths for a given mean normal stress, restriction of
HB is necessary to avoid physically meaningless
conclusions. While I suspect that most authors using
HB tacitly adopt the needed restriction, perhaps an open
discussion of HB and two shear strengths, associated
envelopes and a x-up with alternatives, is in order.
2. Two shear strengths from HB
Consider HB in one of its earlier forms [1]
s
1
s
3
mC
o
s
3
sC
2
o

1=2
; 1
where compression is positive and s
1
; s
3
; C
o
; s and m
are major principal stress, minor principal stress,
unconned compressive strength, material constant
and scale factor, respectively. The scale factor varies
between 0 and 1 and is an empirical adjustment for the
difference between intact rock (s 1) and rock mass
unconned compressive strengths. Rock mass uncon-
ned compressive strength is C
os
C
o
Os: The material
constant m may be expressed in terms of the unconned
compressive and tensile strength (T
o
) by performing
thought experiments and observing that the tensile stress
at failure is negative and numerically equal to tensile
strength. Thus,
m
C
o
T
o

T
o
C
o
r
1
r
; 2
where intact rock (s 1) is assumed for the discussion
here. Many rock types have a ratio of compressive to
tensile strength (r) between 10 and 20, so the constant m
is approximately equal to this ratio. If rock mass tensile
strength is scaled in the same way as unconned
compressive strength, then T
os
T
o
Os and m Osr
1=r which is larger than laboratory m:
Transformation of HB to a normal stressshear stress
representation involves the denitions,
s
m

s
1
s
3
2C
o
s
1
C
o
s
m
t
m
;
t
m

s
1
s
3
2C
o
s
3
C
o
s
m
t
m
;
3
of dimensionless mean normal stress and maximum
shear stress, respectively. Thus, after transformation to
these new variables, HB has the form
t
m

m
8

2

1
8

2
16ms
m

1
8

2
m
2
16s; 4
that is a parabola when plotted in a normal stressshear
stress plane. After division of Eq. (4) by m=8
2
; the
*Tel.: +1-801-581-5164; fax: +1-801-585-5410.
E-mail address: wgparis@mines.utah.edu (W.G. Pariseau).
1365-1609/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S1365-1609(03)00030-3
results are the primed dimensionless forms
t 1
2
2s
0
m
c; t
0
m

t
m
m=8
;
s
0
m

s
m
m=8
; c 1
16s
m
2
: 5a; c
Eq. (5a) is clearly a parabola when plotted in a normal
stressshear stress plane and is essentially that presented
by Serrano and Olalla. However, the bold line plots of
HB by Serrano and Olalla appear to be parabolas
incorrectly centered on the normal stress axis, e.g., their
plots in Figs. 3 and 4. Fig. 1 here shows a true plot of
Eq. (5a).
The lack of symmetry with respect to the normal
stress axis is a cause for concern because of the physical
implications. A consequence of the lack of symmetry is
two values of shear strength for each value of mean
normal stress, a physically improbable event. Solving
Eq. (5a) for the maximum (dimensionless) shear stress at
failure results in two numerically different shear
strengths. Thus,
t
0
m
172s
0
m
c
1=2
: 6
For example, in case of intact rock (s 1) and a
reasonable ratio of unconned compressive to tensile
strength of 10, c 1:16: The dimensionless strengths of
rock in pure shear s
m
0 implied by HB are then
0.077 and 2.077. These values are the intercepts of the
HB parabola in Fig. 1.
3. Envelopes and circles
A circle is also plotted in Fig. 1 (an HB circle) that
has a center at (0, 1) and a radius of c
1=2
1:077: This
circle is not a Mohrs circle representing a stress state at
failure. The equation for this circle in dimensionless
form consistent with the previous notation is
s
0
s
0
m

2
t
0
1
2
t
0
m
1
2
; 7
where points on the circle have the primed but not
subscripted dimensionless normal and shear stress
values. A family of such HB circles would have centers
along the line of maximum shear stress equal to 1, as
may be inferred from Fig. 1.
Mohr circles must have centers along the normal
stress axis (maximum shear stress equal to zero). One
such circle is shown in Fig. 1 for a dimensionless mean
normal stress of unity. This circle represents failure for
the portion of HB in the upper (positive) half plane.
A reversal of the sign of the maximum shear stress
would require enlarging the circle to intersect the HB
parabola in the lower (negative) half plane. This
situation shows graphically the violation of conven-
tional wisdom that states the sign of the shear stress has
no importance to failure.
The HB parabola has an envelope of HB circles that
may be found from Eq. (5a) or Eqs. (6) and (7) in the
usual way (e.g., [2]). The selected two equations contain
two parameters (mean normal stress and maximum
shear stress). Use of Eq. (5a) or Eq. (6) in Eq. (7) to
eliminate the maximum shear stress results in
s
0
s
0
m

2
t
0
1
2
2s
0
m
c: 8
The partial derivative of Eq. (8) with respect to the
remaining parameter, the mean normal stress, shows
that
s
0
s
0
m
1: 9
Elimination of the parameter between Eqs. (8) and (9)
gives the equation of the envelope of HB circles. Thus,
t
0
1
2
2s
0
c 1: 10
The envelope of HB circles is also a parabola that has
the same form as HB. The procedure may be generalized
by using any function f s
0
m
that gives the radius of a
circle relative to the considered horizontal line of centers
in the normal stressshear stress plane. The general
result for the envelope of considered circles in normal-
ized dimensionless variables used previously is
t
0
1
2
f
2
1 f
0

2
; 11
where the prime on f means derivative with respect to its
argument. In Eq. (11), f is considered to be a function of
the normal stress s
0
: This equation shows that an
envelope exists only when jf
0
jo1: When an envelope
exists, the slopes of HB (Eq. (5a)) and the envelope of
Fig. 1. HoekBrown failure criterion plotted in a dimensionless mean
normal stress.
W.G. Pariseau / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 601604 602
HB circles (Eq. (10)) are related by
sinf tanc; 12
where f is the inclination of the HB envelope to the
normal stress axis and c is the inclination of HB to the
normal stress axis. This result is given by Hill [3] for
conventional Mohrs circles and failure criteria that are
symmetric with respect to the normal stress axes. The
qualication (Eq. (12)) indicates that an envelope exists
provided c j jop=4: Interestingly, in case of HB, this limit
occurs at the normal stress axis intercept t
0
m
0:
If consideration of HB and Mohr circles representing
failure is restricted to the upper half plane (positive
shear stress), then a partial envelope equation may be
found by a similar procedure. In this case,
s
0
s
0
m
1
1
2s
0
m
c
1=2
: 13
Solution for the mean normal stress s
0
m
in terms of the
shear stress s
0
requires solution of a cubic equation:
s
0
s
0
m
1
2
2s
0
m
c 1: 14
When the proper solution is obtained, back substitu-
tion into Eq. (8) eliminates s
m
and gives the envelope of
Mohr circles associated with HB in the upper half plane.
Although an algebraic solution is possible in principle,
the effort seems unwarranted for the simple reason that
a failure envelope is not needed in application to
problems of engineering interest. Indeed, a failure
criterion expressed in terms of principal stresses is much
more practical for computation because only the
principal stresses need to be calculated. A second
computation of failure plane orientations and associated
normal and shear stresses is avoided.
4. A x for HB and alternatives
The physically questionable implications of HB and
any other failure criterion expressed in terms of
principal stresses that turn out not to be symmetric
with respect to the normal stress axis after transforma-
tion, e.g., [4,5], can be avoided by restricting applic-
ability to positive shear stress values and then reecting
the restricted portion of the considered criterion about
the mean normal stress axis. The mean normal stress is
restricted to values greater than the normal stress axis
intercept, while reection is achieved by simply using the
absolute value of shear stress. In case of HB, this x
has the restriction
s
0
m
X
1 c
2
: 15
This restriction also corresponds to the range of HB
in the upper half plane over which an envelope exists. In
the plane of s
3
; s
1
; this restricts s
3
to values greater than
C
o
=mE T
o
:
Imposition of the absolute value condition gives
jt
0
m
j 1 2s
0
m
c
1=2
: 16
Inspection of Eq. (16) shows that a lesser value of
mean normal stress than one satisfying Eq. (15) leads to
an imaginary number in Eq. (16) and is physically
meaningless. A plot of the restricted HB given by
Eq. (16) is shown in Fig. 1.
Restricted HB has a sharp nose as does the well-
known MohrCoulomb (MC) criterion that has a
dimensionless form
t
0
m
as
0
m
b; a
C
o
=T
o
1
C
o
=T
o
1

m
8
.
;
b
1
C
o
=T
o
1

m
8
.
: 17a; c
The MC criterion (17a) is plotted in Fig. 1 using the
same numerical values in the plot of HB. In this regard,
when s
0
m
0:5; the actual mean normal stress roughly
equals the one-half unconned compressive strength. In
fact, s
m
m=8C
o
9:9=16C
o
0:62C
o
and the ma-
jor principal stress is about equal to the unconned
compressive strength.
Thus, over a range of mean normal stress from zero to
C
o
, there is little to choose between problematic HB and
the well-known MC criteria. The difference at s
0
m
1 is
less than 3%. This range is very practical because
unsupported rock excavations are unconned; wall
failure is essentially at unconned compressive strength.
Even when conned by support or reinforcement, the
conning pressure introduced is usually small compared
with rock strength. Comparisons of yield zones about
circular tunnels obtained from MC and HB by Soanos
and Halakatevakis [6] show differences of about 10% or
less.
Over an extended range of mean normal stress or
conning pressure, experience with test data leads one to
expect a decrease in slope of the failure criterion plot, a
decrease in the angle of internal friction, f: A better t than
MC may then be possibly obtained by a nonlinear form
jt
m
j
n
as
m
b; 18
where the dimensionless maximum shear stress and mean
normal stress are dened in Eq. (3), a and b are
dimensionless material constants, and n is an exponent
[7]. Solution for the maximum shear stress gives
t
m
7as
m
b
1=n
; 19
where a positive root is implied. The dimensionless
constants are given by
a
1
2

n

1
2r

n
1
2

1
2r
; b
1
r

1
2

n

1
2r

n
1
1
2r
; 20
where the ratio r C
o
=T
o
:
W.G. Pariseau / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 601604 603
When n 1; Eqs. (18) and (19) reduce to MC. When
n 2; a parabola is obtained that is symmetric with
respect to the normal stress axis. Super-parabolas are
obtained when n > 2; while sub-parabolas are obtained
for 1ono2: Sub-parabolas have pointed noses. As the
exponent n becomes very large, a tension cut-off appears
and the slope tends to zero.
The dimensionless nature of the plots in Fig. 2 results
in all graphs passing through the point (0.5, 0.5), that is,
a=2 b
1=n
0:5: Restricted HB is also shown in
Fig. 2. The plots suggest that a suitable choice of
exponent n for a sub-parabola, n-type criterion, would
result in a close match to HB. The advantage of
an n-type criterion is automatic symmetry with respect
to the mean normal stress axis; no restriction is
necessary.
The sharp nose of MC, restricted HB and the n-type
sub-parabolas raises an issue regarding a physically
suitable shape of failure criteria plots in the region of
tensile stress. This is an old issue, but one that seems to
be overlooked in recent times. Various suggestions for
rounding the nose of sharp-nosed failure criteria have
been made (e.g., [8]), but since there are few test data in
this important region, the choice seems to be one of
personal taste and mathematical convenience. In any
case, rounding produces slopes greater than 1 and thus a
region where an envelope does not exist. For these
reasons, failure envelopes are incidental features to
mathematically simple and physically appropriate cri-
teria. In case of anisotropy, and in three dimensions, the
situation becomes more complex and therefore demand-
ing of greater care in selecting candidate functions for
failure criteria.
References
[1] Hoek E, Brown ET. Underground excavations in rock. London:
The Institution of Mining and Metallurgy, 1980. p. 140.
[2] Ford LR. Differential equations. New York: McGraw-Hill, 1955.
p. 2313.
[3] Hill R. The mathematical theory of plasticity. Oxford: Clarendon
Press, 1950. p. 295.
[4] Price NJ. Fault and joint development in brittle and semi-brittle
rock. Oxford: Pergamon, 1966. p. 26.
[5] Bieniawski ZT. Estimating the strength of rock masses. J S Afr Inst
Min Metall 1974;74:31220.
[6] Soanos AI, Halakatevakis N. Equivalent tunneling Mohr
Coulomb strength parameters for given HoekBrown ones. Int J
Rock Mech Min Sci 2002;39(1):1317.
[7] Pariseau WG. Post-yield mechanics of rock and soil. Mineral
Industries, vol. 36(8). The Pennsylvania State University,
University Park, 1967. p. 16.
[8] Nadai A. Theory of ow and fracture of solids, vol. II. New York:
McGraw-Hill, 1963. p. 47688.
Fig. 2. N-type failure criteria plot. MC: n 1; parabola; n 2; super-
parabola; n 10; sub-parabola; n 15; hb=restricted HB. C
o
=un-
conned compressive strength.
W.G. Pariseau / International Journal of Rock Mechanics & Mining Sciences 40 (2003) 601604 604

También podría gustarte