Vous êtes sur la page 1sur 47

STABILITY OP TAILINGS DAMS

cJ
by

N. David Carrier, 111


e

1.0 INTRODUCTION

Each year, the international mining industry processes


hundreds of millions of tonnes of earth and rock to extra
ct the
industrial, construction, and energy minerals that
are the
foundation of our modern technological civilization.
example, the phosphate industry in Florida annually excavaFor
tes
approximately 230 million cubic metres of overburden and
ore to
produce fertilizer; this volume of material is roughly
equivalent
to the original cut for the Panama Canal.

A large portion of ore is waste mineral material, commo


referred to as.tailings. nly
In some cases, such as copper, the
tailings constitute more than 99% of the original ore.
earlier times, and in a few places still today, the tailings In
were
either pushed over a nearby slope, or sluiced into a conven
river. ient
Because of the expanding quantity of tailings, the
increasing concern for the environment, and the growing
economic
need to conserve and recycle process water, most tailin
gs are
now deposited hydraulically in dams. [Actually, for many years,
the international dam engineering and construction commu
refused to recognize these structures as being dams. nity
starting
about two decades ago, it was realized that a number of
tailings
dams were among the largest structures ever built
consequently, deserved some serious attention.) , and,

Tailings dams are generally classified according


to the
method of construction: downstream, upstream, and cente
rline (see
Fig. 1). A downstream tailings dam is designed and constructed
basically the same as a conventional water-retentio
n dam, with
the exception that the former is built in stages over
many years
or decades, and the latter is usually built as
quickly as
possible. Consequently, a downstream tailings dam is
often
than a conventional dam because there is less tenden safer
cy for
cracking and more time to react to a problem.
Hence, the
stability of downstream tailings dams will not be consid
ered in
this paper.

The upstream construction method was developed by miners


a lower cost alternative to downstream construction. .as
There are
four basic zones in an upstream tailings dam:
starter dam,
shell, slimes, and internal drains (see Fig. 1).
The starter dam
is usually a conventional compacted embankment that
serves as a
retention dike when the initial tailings deposition
The shell is built on top of the starter dam and is begun.
is raised in the
upstream direction as the tailings dam grows. The
shell can be
constructed with compacted borrow materials but usually
consists
I) IlI
BCI, Lakeland, Florida, USA
)
/ j )f)( It
r
UI!

Un L) K1!H
CAL 0.: 00020040
0

0
U) z
C
C)
m
C,
m o2
-%
i
D

C) iD
0 cod:
%
0
Cl)

C) C
C)
0$ C)
CD
-. 0
C
13 (I,
C) (1) -o
CD C,
0
-1
0 m 0
(0 ci > 50
co (I)
CD
-% C,
0 z
U) 0
I z
m m
z I
-1 >
C) I
CD C)
0 0 00
D 50 -.4
0
> Cl)
Cl) z (I) Oo
C) -1 -,
>
C 50
0 coo
50 0
3 D
C) F: ci

UTIU?1A
0

rj H- (Cl rt
C) i 0
::ro H-CD r< hh
) Cl) Cl) CD Q
d-O
rt
Z
H- CD
(D 0
CD

o
CD
H CD
CD
o
,

Cl) CDJc-tU)

Cl)
o
frOClJ(nCl)
H
jJ
ji H- H-Cl

H- H-
o
Cljrt

., -Jwo l-
-
H CD-,
< CD
()0 0-.
CD CD
CD H

rt

CDCD(Dct
(D ::rcnCDl
o
H Li t fr1CD Li
I::; (ci
(J,c
( H

Cl)
CD,rt
CDLi
H-
Li Ct CD
Li U H- (I) Dl 0
t-.(fl
H
Ct CflDlrt
rtO CD Cl
-<0 H(D O-.Li
4

The slimes are the bulk of the waste mat


name implies, are generally fine sand to silt erial, and, as the
s and even clays.
The major stability concern with an upstream darn is the fact
the shell is raised over a portion of the that
shines.
The internal drains are similar to con
ventional dams. They
are constructed with graded sands and
gravels, perforated pipes,
geotextiles, etc., and are located
within the starter darn and
shell as required.

The centerline construction meth


od is a hybrid of the
downstream and upstream methods:
the shell is constructed both
upstream and downstream such that the
centerline of the dam rises
vertically. Because a portion of the shell is
slimes in centerline constructio raised over the
n, the evaluation of slope
stability is similar to that of ups
tream construction. Hence,
for the purposes of this paper,
centerline and upstream
construction will both be referred
to as simply upstream.
Geotechnical engineers generally
recognize that the shines
(and sometimes the shell) in an ups
tream tailings darn are loose
and contractive. That is, the soil will generate positiv
pressures during e pore
undrained shear (or exp erie nce volume
contraction during drained shear).
upstream tailings dam undergoes It is well -kno wn that an
undrained shear during an
earthquake. The resulting positive pore pressu
liquefaction of a portion of the tail res may cause
ings dam and thereby lead to
a flow slide with great loss of life
and property (c.f., Dobry
and Alvarez, 1967). Over the last two-and-a-half dec
seismic stability of upstream tail ades,
ings dams has been thoroughly
and intensively investigated. Although the details differ, the
consensus among geotechnical eng
ineers is that undrained shear
strength must be used in the ana
lysis of seismic stability. So,
this paper will not consider seis
mic stability, but will be
limited to static stability.

Why bother with just ordinary


static stability?
For two reasons:

First, in the analysis of static


slope stability, there is
not yet a consensus regarding the
proper shear strength to
use: Some geotechnical engineers
use drained strength, while
others use undrained strength.

Second, despite empirical guideli


nes regarding the rate of
dam raising, the length of beach,
time of desiccation, etc.,
static slope failures continue
to occur too often (see
Poulos, 1988 and Vick, 1991 for
recent compilations).

In summary, then, the subject of


stability of upstream tailings this paper is the static
dams.
5

2.0 DRAINED V5. UNDRAINED STRENGTH ANALYSIS

Raising an upstream tailings dam is a special case of staged


construction. Fortuitously, Ladd (1991) has recently published
his comprehensive and monumental 1986 Terzaghi Lecture regard
ing
stability evaluation during staged construction. He writes:
... considerable controversy and confusion exist
concerning what type of stability analyses should be
used for staged construction projects, both during the
design process and later to check stability during
actual construction. Stability evaluations to assess
the safety of existing (italics in original) structures
also face this technical issue. ...the real issue
concerns the assumed (or implied) drainage conditions
during potential failure.

To prove his point, Ladd lists some of the geotechnical


engineers who believe that a staged construction analy
sis should
be done with consolidated-drained shear strength, and
those
engineers who believe that it should be done with cons
olidated-
undrained strength. I have reproduced Ladds list of refer
ences
in Table 1, and added some recent references specifically
regarding upstream tailings dams. If you have ever felt
uncomfortable regarding which type of shear strength
to use in a
staged construction analysis, reviewing Table 1 should
be of some
comfort and consolation: you are in good company! The US Navy
has even been schizophrenic, recommending undrained
strength in
1971 and drained strength in 1982! And ICOLD, which should be
the final arbiter in these matters, takes no posi
tion at all,
only advising that Design methods and criteria should
conform
with current state-of-the-art knowledge (ICOLD, 1989).
Consequently, I recommend the following:

Rule 1: Geotechnical engineers should avoid designing


upstream tailings dams.

Because (a) Tailings dams are unique structures, in that they


are generally raised over a period of many years
or decades. During this extended timeframe, the
geotechnical engineer usually cannot exert
sufficient control on the project: execution of
his design is left to the mining company. Often,
he is at the mercy of the miners.

(b) Even if the miners follow the design very


carefully, the exact location of the slimes-shell
interface is unknown and unknowable.

(c) Even if the location of the slimesshell


interface is known with sufficient accuracy,
geotechnical engineers cannot agree among
themselves regarding the slope stability analysis
of an upstream tailings dam. If a failure
occurs, and a lawsuit ensues, there are plenty
of
other geotechnical experts to testify against the
designer.
C) a
TABLE 1

STABILITY ANALYSIS FOR STAGED CONSTRUCTION


AND/OR AN UPSTREAM TAILINGS DAM
(based on Ladd, 1991)

ained Strenath Parameters Undrpjne Strength pprameters


CD/ESAa UESAb CU/QRS/USAC
C ,c) c Cc..)

Bishop and Berrum (1960) Moore (1970) Lobdell (1959)


Lamba and Whitman (1969) Janbu (1973, 1979) Casagrande and Wilson (1960)
Parry (1972) Svano (1981) BrinchHansen (1962)
Schmertmann (1975) Finn et al. (1990) Barron (1964)
Janbu (1977) Lo et al. (1991) Terzaghi and Peck (1967)
Rivard and Lu (1978) US Army (1970, 1978)
Tavenas et al. (1978) US Navy (1971)
Tavenas and Leroueil (1980) Bromwell (1984)
Pilot et al. (1982) Poulos et al. (1985)
US Navy (1982) Poulos (1988)
Vick (1983) Carrier et al. (1989)
Murray and Symoris (1984) Schiffman and Carrier (1990)
Wroth and Houlsby (1985) Massey et al. (1990)
Stauffer and Obermeyer (1988) Ladd (1991)
Lo and Klohn (1990) This Paper
Abadjiev (1990)
Gelfand et al. (1990)
Brett (1990)

No Position: ICOLD (1982, 1989)

aCD = ConsolidatedDrained bUESA = Undrained Effective ConsolidatedUndrained


ESA Effective Stress Analysis Stress Analysis QRS Casagrandes ConsolidatedUndrained
(Ladd, 1991)
USA = Undrained Strength Analysis
(Ladd, 1991)
rt -QDrt
)
crc) (fl(D
c) 0 : CD -C 0
ft I-.
J_.c) Ift
Cl) Cl) Cl) U) I- 0 pJJ 001-p-1
(0 1))CD
rtrtbZ 0 ti c: ft ti 0 I-. W
I.-, tI o 0 0 CD tm 0 (1) CD ti,-
b 1fl -CD0)ti0)COti
(0 I-(fl(DC (:lCl) ttiC)CD0)titi0 (0 -c-t
OVle-CDQJ (DrtCD 0
t) ft a.rttiO1 01< ftCD-t CD CD l-- tTti (0(0 0 )--1 l--QU) c ED
cI
0 V
1< tic) ft- b(fl CD
tic-ta. Cl) CD CD. a.0)l--. 0)-CD
CD Q. I- CD rn 1 lyCD i--
.1 rttm CD oCn
(0 o ti
tn Q,. I.ti Nc) - ti (0(0 Hti Cl
lt,O Cr) CD 1)3<
CD ft:(,C) CD (Dc*CD0)<
- . CD Cl) 0 CDI-Qrt a.0)(D0) ,-,,_-, 1)) OO
1. QJ b
rj0 (Do rtr1-
ft
r-c-t ,<
0)
rr
o (0 Cl) 0
CD CD CD OCD tma.CD ftCD :1
CD 0
ft -- 1ltil -C- Cl) CD
o
fth4 og-O1<0 CDft CD 0.
,
C
1 D
ti ftc)
o ) CD titiCD cn
Cfl(fl CD c-t 0 It) CDCD1)j ft(D Cl.
CD
1 CD 0 ED ti C) 1 tm::C 0 )1

0
l)J c-to
1 (DC tl CD
I-.. CD ft H
ti< rt
(J CD - C
H CD U-i CD z CD
-C tl o CD
h 0 Cl) CD Cl)
ft ftCD O .
cn -Cc) 1
(Oft C tma.-< t
0
4 Q
CD - a. lQ. %(D Cl) t
CDCDVIti CD 1) EDO) a.
0DJ rt I.. Ct C) a.CD :i -4
ED ti Wl CD
ED< ft
J
1)3 CD O.CD CD
Qkti l0Q) Cl) 1 1)
(D - .
CD -$l ftC)j CD1O COc) o-- rtti
tnp, C) lCtCDQCDI .
fr U) 0CDbCDO0)CD<
(D CD
lv )4 ti CDi lCD pi C) (I) tt C-)
C)
o (Dftr< IC Cr) 0
Lrt
C) . tm -1),
(n_. CDCD CD CD
(DO Cl
c o
1)) i- 0 og
cn 1))
(Dc) Coo
Ct CDftl-
I.. ft ft-ICft
CD tiCDtib-t U)
:Il-J tJ C).CD0)c) I.-. (Oft
(DCt OC)U) Q1< 0) EY
CD 0 flU) ti CD< ft
O.<rt 1 Q ti 0
C Dftft C) -i 0
1
CD 0CD l i-. I-,
CD -. 1 CD CD i- c ft t ft tCtc 1
C(t)ti
< I 0
C!) CU-ic) Eli )-- 0 CD 1)j ftfr ctti l CD -C CD
HCD CD(fl 0) tl tmi-z ti D31)l-Cfl (flU) (nti
CDO O)r1- ftCftftc) 0-Q
o oo i c-tcttn C CD CD -01)) CDU)
0 CD -) rt_. cl-ti0)O 0)-OP)(D CtCD
0(0 (0 - 1)) CDCD1))H, (DCD(D til-U)1<P)C
o tj h I ICDCDtm C1)tm0)(O I-
o i- C) . CD ,rt(DQ (0 I-I-rtQ) ct I 0) CD Ic)
CD Ct 1)30)
LQ c
t:1 1(D 0 rtrtCfl Q CD
Cfl 1i< CD
Ij
8

for normally consolidated soils, such as in an upstream tailings


dam (in increasing order of accuracy):

Good: c/a = 0.22 0.03 for homogeneous sedimentary


clays plotting above
Casagrandes A-line

= 0.25 0.05 for silts and organic clays


plotting below the A-line

where c, = in situ undrained shear strength

= shear stress on the failure plane at


failure

= (ff qf COS

qf = S 1
((a 03)
)
2
max / = peak undrained shear
strength of soil

= effective stress friction angle


corresponding to the peak, undrained shear
strength

= vertical effective consolidation stress

Better: c/a from CK


U
0 (consolidated-ariisotropic (KQ)
undrained) direct simple shear tests using
the SHANSEP reconsolidation technique
(stress history and normalized soil
engineering properties: Ladd and Foott,
1974)

Best: from CK
U compression, direct simple shear,
0
and extension tests using either the
Recompression technique (Bjerrum, 1973) or
SHANSEP, corrected for strain compatibility
(Koutsoftas and Ladd, 1985)

Ladd also discusses important soil behavior effects,


such as
sample disturbance, aging, strain rate, anisotropy,
rotation of
principal planes, etc. (c.f., Jamiolkowski et al., 1985),
as well
as correction for end effects in three-dimensional
slope
stability analyses (Azzouz et al., 1981, 1983).

In addition, Ladd concludes that the calculated facto


r of
safety for staged construction can be less than 1.5,
depending on
the degree of the site investigation and the field
monitoring;
but he does not recommend how much less.

Now, the USA methodology is primarily concerned with


soils
that generate positive pore pressures during
shear (i.e.,
contractive soils). What if the shearinduced pore press
ures are
approximately zero or negative? In that case, Ladd (and just
about everyone else) recommends that the consol
idateddrained
C)

(fl 0fl)U)ti10(flU) iio c c U)iUH


rt
(flU)
O0--(-Dct tiCD 0 U 1rtCD-
rtrt
t5C)O OCD I-.. C) :3 U) U) U)U)OQ) (t P3 C-lCD
CD U)C1 C) CD (1 rt U; rtrU)0V)ti --4 CDU)
3.:1 i-j. U) .0 UC- CD CD
C)- 1
-3 ci-I0e- ICo CU
(tU) 0 Dfl.(t) 0 -4 i-- U) 0 ,r10U
r(D tCDC) 00 U)
<OCDZCl CD -
CDQ< ) CDrt Cl ii--
- :3 ClU I,_-
CD (D-< &;, OCDci.
U) Ico UJ
- QCDH.U) 0
U)rt
U) I.
:1(t) CD rtU) U rt(D 0
t Cl CD MQ 0 0 rt ct H <
CD
.. (C>ttIH <.0
< U) 1 c: )-. ? U) CD CD 0i H).MCD
(I) U)rt CtU) U) -CD :3
:3 (t 0 UCt CDfr--
L0rI UCD .0
CD o:ii- 0 :3 - - I- CU CD CD CD
0.
< I<bC.flCD CD. 0 Q(DO UI-
U) U)U)r:3- C-I U)
CD <tTrt H)
n 4 0
(flfr..CU0
Cfl(.f -(C) 1
- C. I- C) H) U)
0 ortD CDU) 1
U) (U 0Q d:3 :30 c
t, a 0
gL 0
1 CD - rt fri CD rtr.Ofr- CD
( CD C1< Co i C/). < ..- UI 1
Q C) o.
o CD tU) C, N .-. Ui
CD U) CU U; CD 0
CD- 1 CD
0
OCD -) CD
fr-:1 ,-. UU) 4
U) .0
rt : U)ClU)HrtO (_t_
<CD U rt 0 :1U) .0
Q p.. CU CD H)Z0
CDCDCD 0U) U) o*ti-< :3
U) CD U) :1 (Ofr- CD0 OCfl:3frU). rt
Hl - U) H
0. CU CD
rt U) C--
CD C) ,- f?
(Ort U CDfr-tICfl C_C) :3
C- . CD ti -
r
0 t 0 (t CUCU) (D
rt .rt..- .CD
01 ti rt :3fr-- tfr
d
1 U)cL D .DUIUZ U
CD C-iU)ty
t Uti(U 1< ct U) Cl
t (OCt s2CD rt.U)tC :3 (OCU
U) -- 0)- C-I
CD(Ot.W(D rt0
U: C) Cl2.CDP-1Cl 00 (Do
:3 CD CD 01
rt :3Z 0-
c_ C) CD rt(U (3
U)U) U) - C-- (t
i ti(U CD Uct CD
CD fr- U)CD -. -.0
CD rtU) 0 -
z. ctU) ,-4 H)
CD U)
U) (U ,. t t :3 1,rt0rr 01
-.- U)CDU) CD CD tU 3- ti- HCD
00 Ct eCDC-.4O C-i
tt-i 0 tT< U rjrt CU
1 CDQ (t)CD
CD3. Z C
CD 0 CDU
U) jC-t(j)U) <(U
CD CD Ct 04
:3.-)
U) () rt(OQC-
0(t) CD cs 0
U)U)
t3 (DC-C_ CD (tiC-IC
rtt CD 0 (-trn .J I_C :3 t U) cz
U) C I-
CD ti 0 CD C-< 4
C_-
U)CDrtctU) 3. 4 0
CD CD .0 fr- - ctOI CD C--U)
(t)U)CU U) ci g
CD -. (U CD CC> tiCD 0 0 Y0 0
rt(jrt Ct(U tiCD (U :3 U CD .--. I-.(U I-. 0 U)
<0..CD 0 -Cl
0 (U 0 1rt UC-O o tci rtO c - Cl0 CD C--. CD
cn 1-.. Urt )ci0C-
CDQ. t3.< rtrtti rtCD < Q 0 I- U) 0 -4.
1 U) .1 CD CD Cl 0 CD (C) QU) rtCl - CDCDH)U)fr- CD0 0 tCt0 CD
C) UJ< 0U)U)iU) (DU)U)Lfr1 C)
tPi3. H (D :30 rtti C_-CD-.<CD C- CDCDti Ct
rtCD . 1 CD U ti I-.. U0 CU CD I.
<U) 1U1O(U-.CDU)0LJCD I.. Cl C U) 0 it1rt--U)U) U) U
CD 1 -1 rttr Cl CD - U) ) ((I CD
10
1

Cu

LP METASTABLE IF
w
Cr,

w
Cl-i

SHEAR STRAIN

Cu

Cr,
Cl-i
w
c/-i Cr

Lii STABLE IF
Cl, Td <Cr

SHEAR STRAIN

Td = driving shear stress

Cu = peak undrained shear strength (=qf cosp)

Cr = steady state, or residual, shear strength ( q COS 4r)

FIGURE 3: Undrained Steady State Shear Strength


(after Poulos, 1988)
____

11

However, there is not complete agreement on how to determine the


steady state strength. Poulos (1988) emphasizes that the
required tests must be carefully conducted and that only a few
geotechnical laboratories are properly equipped. Whereas, Seed
(1987) and Seed and Harder (1990) have back-calculated shear
strengths from the failures of sand embankments and have found
that the field strengths (1.7 to 36 kPa) were significantly less
than the corresponding laboratory values. Instead, Seed and his
colleagues propose a correlation between the standard penetration
test blowcount and the residual strength, Still others
Cr.
(Jefferies et al., 1990; McLeod et al., 1991; Lo et al., 1991)
have attempted to normalize Seeds shear strengths in terms of
either the effective stress or the overburden height. The
expression proposed. by McLeod et al. is:

Cr
= 0.022 (+0.00410.014) _cs
6
)
1
(N
0
vc
where (N
o
6
)
1 _ = standard penetration test blowcount in
an equivalent clean sand, standardized
for 60% energy ratio, normalized to an
effective overburden stress of 1 tsf
(96 kPa) (Seed and Harder, 1990; Vick,
1991)

= (N
6
)
1 0 +

where = correction for silt content (Seed,


1987)

Silt Content (%)

0 0
10 1
25 2
50 4
75 5
and (N
6
)
1 0 = CER N
1

where CER = correction factor, such that the


effective energy delivered to the drill
rods equals 60% of the theoretical
freefall energy

= 0.75 to 1.30 (Seed et al., 1985)

= 1.0 for a safety hammer used with two


wraps of the rope around the pulley
(most commonly used in the United
States)

= 0.75 for a donut hammer used with two


wraps of the rope around the pulley
(also used in the US)

and 1
N = SPT blowcount normalized to an
effective overburden stress of 1 tsf
12

= CNN

where N = SPT blowcourit (as measured in the


field)

= correction coefficient for the


effective overburden stress at the
depth where the SPT was performed

= 0.4 to > 1.6 (see Fig. 4 in Seed et


al., 1983 and Fig. 4 in Seed et al.,
1985)

Hence, ) 60-Cs
1
(N = CERCNN +

[If you do not understand this, it.is only because you


have been paying attention!)

The value of Cr/GC for all of the cases considered by


McLeod et al. varies from 0.05 to 0.20.

With clay tailings, the following empirical expression for


remolded undrained shear strength can be used (Carrier and
Beckman, 1984):

=
(e-X)
(e) /
6.33
where = {0.00449(PI) [4.14 +

a = 0.0208(PI)[1.192

= 0.143

X = 0.00449{6.14(PL) (P1) [4.14 +

= 0.027(PL) 0.0133(PI)[1.192 +

j)

= -6.33

e void ratio

and where P1 = plasticity index (>10%)

PL = plastic limit (%)


ACT = activity = PIf(2 microns)

(Atterberg limits are run on the No. 200 sieve fraction:


see Section 4.0.)

For example, if PL = 27, P1 = 13, ACT = 0.3, and e


(hence, the liquidity index = 0.77), then
= 1
Cr/c1c = 0.02.
13

2.3 Ladd and Poulo3: Undrained Shear Strength (Cohesive and


Cohesionless Boils)

The Poulos design philosophy can be summarized as follows:


As a last resort, at least be sure that the steady state strength
is as great as the driving stress. I concur with this philosophy
with respect to upstream cohesionless tailings dams. But there
can be situations in which the cohesionless soil is not strongly
contractive and the Poulos criterion may lead to an embankment
design for which the conventional factor of safety is
significantly less than 1.5. Consequently, I recommend that both
a Ladd analysis (strain-compatible undrained shear strength with
F.S. -1.5) and a Poulos analysis (undrained steady state shear
strength with F.S. -1) be performed. Both criteria for factor of
safety must be met.

On the other hand, if the Poulos criterion is applied to


cohesive slimes, it can lead to an extremely conservative design:
the residual undrained shear strength can be ten times smaller
than the peak value = 0.02 vs. c/c7! = 0.2). Even taking
into account the difference in the recommended factors of safety,
the Poulos criterion would necessarily recplire a much wider shell
and/or a flatter downstream slope than is generally practical.
Of course, the construction of an upstream tailings dam on top of
cohesive slimes is difficult, but it is not impractical (c.f.,
Massey et al., 1990). And in fact, Poulos et al. (1985b)
evaluated the stability of slightly cohesive (P1 = 13) alumina
red mud, deposited on a shallow slope. The static slope
stability factor of safety, based on the undrained steady state
shear strength, was calculated by them to be just 0.26. However,
they concluded that the peak undrained shear strength could be
used instead (which gave a F.S. >1.5), on the grounds that the
strain-to-peak was large for the red mud (-3O)
; whereas, the
strain-to-peak for a typical contractive cohesionless sand is
only 1%. This reasoning appears to be in contradiction to the
Poulos (1988) quote above (.... it should usually be assumed that
sufficient progressive strain occurs to cause the peak strength
to be destroyed.)

Because the Poulos criterion has been primarily applied to


slopes composed of cohesionless soils, I recommend that it be
used with caution and judgment with respect to cohesive soils.

2.4 An Example of How Not to do a Static Stability Analysis


(Based on Drained Strength Parameters)

Shown in Fig. 4 is a simplified cross section through an


upstream tailings dam that failed during operation a few years
ago. More than two million cubic metres of shell and slimes
material flowed several kilometres downstream. The geometry of
the dam is shown just prior to the flow slide, which occurred
under static conditions. Fortunately, no one was injured or
killed, but the property and environmental damage totaled
millions of dollars.

The dam was designed by a very experienced and respected,


world-class engineering firm, and the design team had been
working closely with the mining company for several years.
However, the design team used an ESA (effective stress analysis)
CADO 0002008E
0

SLIMES
330

120 pcI

250
SHELL
30
122 pcI
_%______

.,.........

STARTER DAM
37
130 pcI

PERMEABLE FOUNDATION 0
37
130 pcI

0 100 200 300 400 500 800 700 800

HORIZONAL DISTANCE (ft)

FIGURE 4: Cross Section of an Upstream Tailings Dam That Failed Under


Static Conditions: Just prior to failure, the calculated slope
stability was greater than 1.7, based on an Effective Stress
Analysis (ESA)

fl.ATF121A
15

to evaluate the stability of the darn. In my opinion, the


designer made three fatal errors, as follows:

1. The design effective friction angle, , of the slim


es
was considered to be greater than the friction angl
e of
the shell; i.e., the slimes were thought to be stron
ger
than the shell. (Designers of other dams have used the
same for both materials. The assumption that the
slimes are stronger or equal to the shell is sedu
ctive
and treacherous, because it means that the desig
ner
does not have to worry about the location of
the
shines-shell interface: how convenient!] As a resul
t,
all of the critical circles in the slope stabi
lity
analyses were primarily in the shell material:
instability of the slimes was never really cons
idered.
2. The pore pressures used in the stability analyses
were
not the measured values from the many piezomete
rs
installed in the dam. As shown in Fig. 5, the tailin
gs
dam was founded on a highly permeable, coarse
sandy
material, which acted as an underdrain. Consequ
ently,
the measured pore pressures were less than hydrostat
ic.
If the measured pore pressures had been used
in the
stability analyses, it would have led to a steep
er
slope. From experience, the designer knew that this
steeper slope was not safe; so he chose to use grea
ter
pore pressures instead, based on the vertical dista
nce
below-the phreatic surface (or hydrostatic),
in order
to arrive at a flatter, more reasonable slop
e. When
the designer had to use manufactured pore
pressure
data, this should have been a clue that an ESA
analysis
was not appropriate. Ladd (1991) writes, Although
some engineers may conduct ESA with low c
values
and/or high u [pore pressure) values in an attem
pt to
guard against the possibility of an undrained
failure,
this approach should not be considered as
a reliable
replacement for a USA. Apparently, the designer
figured that the darn could be raised indefinit
ely as
long as the measured pore pressures were less
than the
design pore pressures (i.e., hydrostatic).
In reality,
the suction at the bottom of the tailings pond
helped
to keep the darn stable: this beneficial effect
decreased as the height of the dam increased
.
3. The designer believed/assumed that excess pore
pressures would dissipate rapidly; i.e., that
the shell
and shimes material were sufficiently perm
eable that
drained conditions would control behavior:
Here the
designers reasoning is a little murky:

(a) If he were thinking of just the excess pore


pressures caused by consolidation as the level
of
the slimes is raised, then the piezorneter data
clearly showed that these pore pressures were
not
fully dissipated. (Well, maybe not clearly: An
internationally recognized consultant who
was
called in after the failure, examined the
piezometer data and said, Obviously the shin
es
were more than fully consolidated: the pore
pressure was less than hydrostatic.
16

Pore Pressure Head (ft)


0 50 100 150 200
0
200

50
150
- 4-,
4-

100 c
-c
-4- .2
100 4-
a) 0
>
o -

w
150

50

200
0

FIGURE 5: Pore Pressure vs. Depth at Stci. 200


Within the Shell and Slimes of the
Tailings Dam Shown in Fig. 4
17

Regrettably, he had forgotten tha


t in the case of
vertical seepage through a slightl
y compressible,
fully consolidated soil, the por
e pressure would
be close to zero (c.f., Lambe and
Whitman, 1969;
Carrier et al., 1989). The fact that the pore
pressure was significantly gre
ater than zero
meant that the slimes were in fact still
consolidating.) Standard laboratory tests run
before the failure had shown tha
t the slimes were
non-plastic and this is perhaps
one reason that
the designer might have felt tha
t the slimes were
consolidated. Later, we determined that the
fines fraction of the slimes had a small
plasticity (plasticity index 10) and this
certainly retarded the rate of con
solidation (see
Section 4.0).

(b) But if the designer were thinki


ng of just the
shear-induced excess pore pre
ssures along a
hypothetical slip surface, would those pore
pressures dissipate? For the moment, let us
assume that the slinies and she
ll materials were
completely cohesionless and tha
t the rate of
raising was slow enough such tha
t there were no
consolidation pore pressures
. Would the shear
induced excess pore pressures dis
sipate?
I like to think of an upstream
tailings dam as
being similar to a giant tria
xial test. For a
drained, strain-controlled triaxial tes
time-to-failure, tfi is given t, the
by Bishop and Henkel
(1962) as:

tf 20 h
2 (for drainage at
=
both ends)
3 c,,

where h = height of triaxial sample


c, = coefficient of consolidation of
soil
Even a cohesionless sand has
a measurable c: an
upper limit value for a nor
mally consolidated,
contractive sand might be 1 cm
/s.
2
5000 cm (a modest tailings dam If h 50 ni
108 s ), then tf = 1.7 x
5 yr. So, if a tailings dam is goi
fail under drained condition ng to
s, one must assume
that the strain rate is extreme
ly slow. This may
have been what the designer
of the dam in Fig. 4
thought.

However, a tailings dam is mo


re akin to a stress
controlled triaxial test. Onc
e the peak deviator
stress is reached (the stre
ss condition assumed
in a slope stability analysis)
the strain rate to
the steady state, or residu
al, strength for
cohesionless, undrained contra
ctive sand is about
170%/s (c.f., Been and Jefferies, 1985).
consequently, if an upstrea
m tailings dam is

I
18

going to fail, it will occur very fast, and it


will be under undrained conditions.

Hence, whichever excess pore pressures the


designer was thinking of, neither would dissipate
rapidly.

Would the designer have been better off doing a UESA


(undrained effective stress analysis)? In principle, yes; in
practice, no.

In a recent paper advocating an UESA, Lo et al. (1991) state


that, In checking the factor of safety, it is important to
incorporate appropriate undrained excess pore pressure due to
field loading in the stability analysis. Actual construction
pore pressure should be regularly monitored to ensure that they
do not exceed those values assumed in the analysis. This, of
course, is simply the effective stress principle, one of the
basic tenets of soil mechanics.

The problem is that in actual practice, a UESA requires


these three things: a complete definition of the stress field
within the tailings dam; the value of Skemptons pore pressure /
parameter Af under those variable stress conditions; and the rate
of consolidation of the slimes beneath the shell. The first of
these requirements is certainly difficult; the second is more
difficult and maybe impossible; and the third is beyond our
capabilities at this time. (see Section 2.5).

The difficulty with the UESA is illustrated in Fig. 6.


First, assume that an element of soil within the tailings dam is
fully consolidated under K 0 conditions, as shown in Fig. 6a: This
might correspond to the slimes in the middle of the pond after
the consolidation pore pressures have dissipated. Now, assume
that an increment of shear stress due to a hypothetical slope
failure is imposed, thereby generating excess pore pressure. The
resulting effective stress path would terminate at the strength
envelope (a, a), depending on the value of Skemptons pore
pressure parameter, Af, as shown. The maximum available
undrained shear strength for a UESA is given by:

UESA = u UESA COS

C COS + 1c
0
l+K
sin
, r +
1K
0
(2Af -1)
= 2 l+K 0
cos,
1 + (2Af l)sin

And this would be the same undrained shear strength for a


USA at the same initial effective stress (a = a), because
the effective stress path is the same for both cases. And if Af
= constant (independent of a) and c = 0, the usual case for
normally consolidated soils, then the shear strength normalizes:
-S

CA O.: 0002010C n C:

Sr.,

b
Ic4
b

I.,

b = ta
(
1 sin4p)
IcJ
b

A
ci) = 0.5 LC
a)
Af 0
K
=
I.
S USA u UESA
C) (all Af)
1

a)
-c
Cl)
E
E \
x
o
\ cr= cr
\

c7+U =

Average Effective Normal Stress, p =

2 2

FIGURE 6a: Conditions; Fully Consolidated (u=O)

F1ATR1 1 A
CA 0002011C

1 (sin
tan )

UESA
Af)

,c=
1
o. Jc

Average Effective Norma Stress,


2 2

FIGURE 6b: 0 Conditions; Consolidating (u>O)


K

FM T8 121 A
C - NO.: 0002012C C)

b
1
b tan (sinp)

U USA
0 su UESA
0 K
0
a; SUUSA= SUUESA
4.., 0
K
(I) (Af = 1)
L.
a
a)
-c
(1)

E \
E \
><
a \
;= o.,c

a+cr =
Average Effective Normal Stress, P = 2 2

FIGURE 6c: K Conditions; Consolidating or Fully Consolidated;


No Rotation of Principal Planes
C NO.: 000201.3C
0

t4)

I-)

b
1
tan (slnp)
b

0
I)
K
(I)
C) UESA
L S 0
K
-4--, USA
Cl)

0
C)
C/)

E
D
S

=
Average Effective Normal Stress, P =
2 2

FIGURE 6d: K Conditions; Consolidating or Fully Consolidated;


Rotation of Principal Planes

P.4 TR 121 A
Ci JO.: 0002014C
C 0

b = (sincP)
1
ta

0
(I,
(1) UESA
L.

C,)
\
a) Aq
-c
(1
E N

E
x
ci
N
N

=
Average Effective Normal Stress, P =
2 2

FIGURE 6e: General Conditions; Consolidating;


Rotation of Principal Planes

flA TR 121 A
4
2L

u UESA sin q, [Af(1 )


0
K + Ko] - Cu UESA
Cos
1 + (2A 1)sin p

So far, no problem.

Now consider an element of soil that is undergoing K 0


, 1c = vc and the excess pore
consolidation (Fig. 6b). Given K
0
pressure u, then cj = a, can be calculated and c UESA can be
determined as above, and is equal to C for all values of Af.
Still no problem.

Next consider an element of soil that is either undergoing


consolidation or is fully consolidated under K < K 0 conditions
(Fig. 6c). In order to determine c one would have to know
(measure) A = f(K). For a USA, one just uses the test results
for K
0 conditions. Hence, as shown in Fig. 6c:

Cu TJ > Cu UESA if Af < 1


Cu us = C VESA A = 1
C < Cu UESA if Af > 1

For normally consolidated soils, Af 1, and, thus, Cu USA


c UESA and the USA is conservative.

Next consider an element of soil that is either undergoing


consolidation or is fully consolidated under K conditions, but
the principal planes have been rotated through an angle, 0, such
that a < (Fig. 6d). Now, in order to determine C UESA
one would have to know Af = f(K,O). The USA again ignores that
complication, but because < Cu USA < tJESA even for Af
= 1, as shown in Fig. 6d.

Finally, consider an element of soil that has undergone an


increment of shear (e.g., due to a raising of the shell) and is
now consolidating. Furthermore, the principal planes have been
rotated. This is the general case for the slimes beneath the
shell. In order to determine c UESA during construction, not to
mention prior, one must know Af = f(K,O,tq, etc.). Even if this
were possible, it is a daunting bookkeeping task.

Thus, in principle, a UESA is rigorously correct; but in


practice, unwieldy and perhaps impossible to execute. However,
for normally consolidated soils, which is the usual case for
upstream tailings dams, Af 1 and consequently, C usi < C UESA
and thus, a USA is conservative.

2.5 Limitations to Undrained Strength Analysis

General

Estimating the undrained shear strength prior to


construction requires that the pore pressure distribution vs.
time be accurately predicted: This is beyond present
geotechnical capabilities. Although considerable progress has
been made in the analysis of onedimensional, large strain, non
linear consolidation (cf., Gibson et al., 1967; Somogyi, 1979,
1980; Somogyi et al., 1981; Carrier et al., 1983; Yonq and
Townsend, 1984; Montgomery and Leach, 1984; Moudgil and
Somasundaran, 1985; Mesri and Choi, 1985; Schiffman and Carrier,
1990; Murphy and Williams, 1990; Feldkamp and Belhoinine, 1990), so
far, very little work has been done in two- and three-dimensional
consolidation (c.f., Somogyi et al., 1984; Gibson et al., 1990).

Hence, whereas accurate predictions of pore pressure can be


made for the slimes occupying the majority of the tailings pond,
reliable predictions cannot yet be made for the slimes beneath
the shell (c.f., Ladd, 1991). As a result, the present
consolidation models can only be used for guidance prior to and
during construction of the darn. The geotechnical engineer must
be prepared to iterate the design as field measurements of pore
pressure become avajiable.

Ladd: Undrained Strength Analysis

As described in Section 2.1, Ladd recommends sophisticated


K
-
0 direct simple shear tests to determine the undrained shear
strength for projects involving staged construction. However,
for the special case of upstream tailings dams, the laboratory
tests must be run on reconstituted samples (see Section
3.2).
Hence, in my opinion, rigidly following Ladds recommendatio
ns
may lead to a false sense of security. I would rather use
simpler tests, even CIU triaxial tests, in combination with
a
healthy dose of common sense.

Poulos: Undrained Steady State Shear Strength

The Poulos methodology also suffers from the presen


t
inability to recover undisturbed samples of loose tailing
s: Re
constituted samples must be tested.

Furthermore, in order to make use of the data from standard


penetration tests, the measured blowcounts must undergo
three
stages of empirical correction. The resulting shear strength
ratio (c/a) can then be too high by a factor of nearly
three.
In my opinion, the present SPT correlations are too approx
imate
for the stability analysis of an upstream tailings dam.

In addition, there is no agreed upon methodology to predict


SPT blowcounts prior to deposition of the sand. Hence, the
present SPT correlations cannot even be used for guidan
ce for the
initial dam design.

So, even if the correct type of shear strength is


established
(i.e., consolidatedundrained), the appropriate value
of the
shear strength is still very difficult to determine.
As Poulos
(1988) stated,

Computation of the driving shear stresses is the major


focus of all conventional stability analysis techniques
that have been published. But current techniques for
selection of strength introduce a far greater chance
for major errors than do the stress computations.

Consequently, I recommend the following:


26

Rule 3: Geotechnical engineers should avoid designing


structures in which the shear strength is difficult
to determine, such as an upstream tailings darn.

3.0 UNDRAINED ANALYSIS DURING CONSTRUCTION

As discussed above, present geotechnical capability does not


permit an accurate prediction of undrained behavior of an
upstream tailings dam prior to construction; hence Rules 1 and 3.
But this does not mean that upstream tailings dams cannot be
constructed and operated safely. However, it does mean that
analysis, and possibly redesign, must be performed throughout
the useful life of the structure. This is simply the
observational method espoused by Terzaghi, reiterated
specifically for upstream tailings dams.

It is very comiion for a geotechnical engineer to


periodically check the stability of a tailings dam by means of a
drained analysis. However, as discussed in Section 2.4, a
drained analysis can be unconservative and can have catastrophic
consequences. Undrained analyses must also be made during
construction and these are our recommendations:

3.1 Laboratory Testing

Representative tailings samples (shines and shell materials)


must be obtained, prior to and during construction. These
samples do not have to be undisturbed, which is generally next
to
impossible anyhow. Instead, the samples can be reconstitut
ed in
the laboratory at various water contents; the pore water should
be the same as the mill recycle water. Of course, if long-te
rm
chemical reactions can occur in the tailings pond
(c.f.,
Troncoso, 1990a; Gelfand et al., 1990), re-constituting
samples
may be difficult. Hence, it is very important to monitor pH,
Eh,
conductivity, dissolved solids, cation concentration, etc.

The re-constituted samples are then consolidated to variou


s
stress levels and corresponding void ratios, and the
undrained
shear strength measured. Important secondary factors include
-
0
K
consolidation; direct simple shear, rotation of principal
planes,
etc.; but the primary consideration is to somehow measu
re the
undrained shear strength, and even standard triaxial shear
tests
can be used (see Section 2.5).

Additional characterization tests should also be run, as


discussed in Section 4.0.

3.2 Field Measurements

At a minimum, an array of piezometers must be installe


d in
the shell and in the slimes. The type of piezometer to be used
in the slimes depends on the expected drainage conditi
ons at the
bottom of the pond: (1) If the bottom is drained, then the pore
pressures will be less than hydrostatic and a simple casagr
ande-
NO
type piezometer can be used. (2) If the bottom is sealed,
the pore pressures will probably then
be greater than hydrostatic
(depending on the rate of con
solidation) and it could be
extremely awkward to read the piez
ometers while floating in a
boat on the tailings pond: The
water level in the piezometer
could easily be over the head of
the field person. Instead, an
electric, vibrating wire, or pneum
atic piezometer should be used.
Whichever piezometer type is
selected, it must have the
capability of being extended upw
ards as the tailings pond rises.
And, given the period of service
required, it will be necessary
to re-calibrate the piezorneters
periodically.
Once the pore pressures have
possible to calculate the effect been measured, it is then
ive stresses within the shell and
slimes. Whence, the distribution
of undrained shear strength can
be predicted based on the suite
of laboratory tests discussed in
Section 3.1. Finally, the factor of safety
using a total stress (undrained can be calculated
) analysis. In general, the
undrained shear strength will incr
ease with depth; in order to
utilize a standard limiting equilib
rium slope stability computer
program (circular or wedge-shaped
failure surface), it will be
necessary to input multiple soi
l layers, each with a zero
friction angle and a different valu
e of cohesion corresponding to
the undrained shear strength
at that depth. If the undrained
shear strength is anisotropic, then
modify the input shear strength it wil l als o be necessary to
for each trial failure surface,
a very time-consuming process.
Alternatively, if the source cod
for the slope stability computer e
program can be modified, then it
is possible to write a custom
ized program that automatically
handles all of the shear strengt
h bookkeeping. This was done for
the isotropic case by Bromwell
(1984; see Fig. 10 in Ladd, 1991)
and for the anisotropic case by
Azzouz et al. (1981).
As a supplement and check on thi
s minimal procedure, we also
recommend that samples be reco
vered from the tailings pon
Slimes samples are required d.
in and below the active pond
and also from beneath the she area,
ll. The latter require drilling
through the shell from the dow
nstream face. Thus, appropriate
access to the downstream face
(berms, benches, etc.) should
incorporated into the design be
of the dam from the beginning.
Samples from the active pond
area can be obtained manually
(Carrier and Keshian, 1979) and/or with standard
drilling/sampling equipment.
Manual sampling methods should
used to refusal (often, 10 be
to 15 m in depth), and mechan
methods below that level. ical
Where the pond water is deep
sampling from a boat or bar enough,
ge is relatively straightfo
Where there is no water and the rward.
beach is soft, sampling is a
more difficult and expensive. lot
Plywood boards laid on the bea
will usually suffice for manual ch
access and sampling. Mechanize
equipment may require earthen d
pads or other forms of surf
stabilization; often, portable dri ace
lling tripods, although labo
intensive, are preferable. r-

Whichever method of sampling


usually cannot be obtained. is used, undisturbed samples
Fortunately, however, the wat
content of saturated samples can er
be preserved. The measured voi
ratios can then be compared wit d
h the predicted void ratios,
latter being based upon the cal the
culated effective stresses
the laboratory measuremen plus
ts of consolidation. If necessary,
adjustments can be made to the predicted undraineci snear
strength, always erring on the conservative side.

Obviously, it would be preferable to measure the in situ


undrained shear strength directly, instead of predicting it
indirectly. Unfortunately, the options are few:

First, undisturbed sampling of these very soft materials, as


previously stated, is next to impossible, and probably not
worth the expense to even try.

Second, measuring the undrained shear strength with a cone


penetrometer involves subtracting one large number from
another:

where = stress measured at tip of cone


penetrometer
= total vertical stress in the soil
= cone bearing capacity factor

Often, when this approach has been attempted with soft


slimes, the resulting S has been found to be negative.
Conseqi.iently, there is no basis for confidence, even when
the calculation yields a positive value. Furthermore, the
value of N varies by at least a factor of two, depending on
whose bearing capacity theory is used.

Other field methods, such as the flat plate dilatometer, the


pressuremeter, etc., suffer from the same limitation as the
cone penetrometer. To be able to use any of these methods,
either the overburden total stress must be accurately known
(preferably measured), or its influence must be eliminated.
One possibility is to drill a large cased hole with drilling
mud, such that the diameter of the boring is significantly
greater than that of the cone penetrometer. A total stress
transducer near the tip of the cone penetrometer could then
be used to measure the pressure of the drilling mud, or
pseudo-overburden. Alternatively, it may be necessary to
pressurize the drilling mud in order to prevent bottom heave
of the shines. With an accurate measurement of the imposed
total stress, it might be possible to obtain a reliable
estimate of S. This approach is probably worth
investigating further.

Third, a vane shear test works well when the shines are
primarily clayey. But many shinies contain a significant
amount of sand and in these cases, it is not altogether
clear what is being measured with a vane shear. Perhaps the
solution for a particular tailings material would be to run
a series of vane shear tests in prepared beds of material
with a known undrained shear strength, in order to develop
correlation factors applicable to the field measurements.
Fourth, geophysical methods have the potential to provide
global estimates of shear strength, rather than
values at individual points. Spectral-Analysisof -Surf ace
Waves (SASW) appears to be particularly attractive (Stokoe
et al., 1988, 1989). In the SASW method, both the source
and the receivers are placed directly on the soil surface;
i.e., without borings. Through sophisticated signal
processing and analysis, a detailed profile of shear wave
velocity vs. depth is produced. In the case of a tailings
dam, the shear wave velocity could be measured on re
constituted slimes samples (e.g., with resonant column
tests); and then the field measurements could, in principle,
be correlated with the void ratio and the undrained shear
strength. Evaluation of tailings dam stability by means of
the SASW method certainly deserves more investigation.

If a geotechnical engineer accepts a commission to design an


upstream tailings dam, the scope must include analysis and design
during construction. Because the construction period may span
years or even decades, there must be a strong working
relationship between the owner and the engineer. Obviously, this
level of commitment may be difficult to attain. However, in my
opinion, only if this condition is satisfied should a
geotechnical engineer even consider making an exception to Rules
1 and 3. Hence,

Rule 4: The only exception to Rules 1 and 3 is when (a) The


geotechnical engineer follows Rule 2; and (b) The
geotechriical engineer has a close working
relationship with the client throughout the design-
life of the tailings dam.

4.0 HOW TO KNOW WHEN AN tflDR7INED STRENGTH ANALYSIS IS


REQUIRED

As stated earlier, in general, both a drained and an


undrained strength slope stability analysis should always
be
performed: Whichever yields the lower factor of safety contro
ls
the behavior, and, hence, the design of the tailings dam.
However, there are some cases where experience shows that
one or
the other ccnditiori will clearly control: For example, when
dealing with phosphatic clays, which are highly plastic, we
never
bother with a drained analysis; undrained shear strength
always
controls design.

In order to better predict whether a given tailings material


will behave as drained or undrained, we have modified some of
the
basic definitions of soil mechanics.

Geotechnical engineers are used to thinking of soils as


being either sand-like in behavior (i.e., drained), clay-li
ke
(undrained), or siltlike (somewhere in between). In accordance
with the Unified Soil Classification System (ASTM D
2487;
originally devised by Casagrande, 1947, 1948), if more than
50%
by weight of a soil is coarser than a No. 200 sieve (74 micron
s),
5U

then the soil is sand; and, hence, sand-like in behavior. If


xnore than 50% is finer than a No. 200 sieve, then it is either a
silt or a clay. In order to distinguish between silt and clay,
Atterberg limits (ASTM D 4318) are then run on the minus No. 40
sieve (0.42 mm) fraction. Once the major soil type is determined
(sand, silt, or clay), appropriate adjectives are often added to
the description, such as clayey sand. But even with the
modifiers, the behavior is presumed to be that of the major soil
type.

Unfortunately, the USCS is two-dimensional: It does not


recognize that the water content of the soil can make a sand
behave as if it were a clay; i.e., undrained instead of drained
(Carrier, 1988). For example, a mixture of phosphatic clay,
sand, and water can behave undrained if there is as little as 15%
by dry weight of phosphatic clay. According to the USCS, this
soil would clearly be a sand (85% coarser than a No. 200 sieve),
regardless of the water content. And yet, this soil can behave
as if it were a pure clay if the water content is sufficiently
high such that the sand particles are not in contact with each
other.

Because of these limitations with the USCS, we have devised


the following protocol when dealing with a slurried material:

First, refer to Fig. 7a. This is a ternary diagram


developed by Scott and Cymerman (1984) which is very useful for
displaying the sand-like and clay-like regimes of a given soil or
tailings material. As shown in Fig. 7a, the three extreme points
on the diagram represent sand, fines (silt and clay are lumped
together), and water. The various parameters shown in Fig. 7a
are defined as follows:

The water content, w, along the outside of the left axis, is


defined conventionally:

w
w= xlOO%

where W = weight of water


= dry weight of fines (-No. 200 sieve)
W = dry weight of sand (+No. 200 sieve)

The solids content, S, along the outside of the right axis


is a term used by the mining industry:

= WF + Ws &JSt
S x 100%
WW+WF+WS U

The solids content and the water content are inversely related as
follows:

w = ( 1) x ].0O
S
31

WATER

900, 10
20
20
20
\10
233 30
11
5 9.
.* 150 40

.1 /,. q:/ \
C
9,
50
100
Cf
67 2 60
eT/$/ eF

1i,,1 / 1
CONSTANT TOTAL VOID RATIO
\
1
70

(eT)
25 80

/ \ \\ 90

0
0 10
5
/ /
20 30 40 50
i
60
0.5\

70
0.2 0.1
\\
80
\\
90
0
100
100
SAND Fines in Dry Solids, F (%) FiNES
(S1t/ay)

FIGURE 7a: Ternary Diagram


I
(after Scott and Cyrnerman, 1984)
32

WATER

aD 0
Segregation Limit

900
Qu1ck clay

400
Liquid Limit of Clay 200%
233

.1 150

c
o10
()
ci
67 60
Plastic Limit
43 of cloy = 60%
344+4

90

0 100
0 30 40 50 60 70 80 90
SAND Fines in Dry Solids, F (z) FiNES
(st/ccy)
essox 0.635
es 0.92
max
DR = 0%
Probably Controctive

N
0
N
r Definitely Controctive
(ClayLike) Probably Dilative

0 Very Probably Definitely Dilative


Contractive (Sand Like)

FIGURE 7b: Example with Phosphatic Clay and Sand


33

1 -.

WATER \O

900 10

400 20

6 \ ,* 130L1C
233 30 \ 9
0

. 150 40.
p
100 503,

25 4
%/
H IHH
4 80

SATURATED - --
011
/ \ 100
0 10 20 30 40 50 60 70 80 90 100
SAND Fines In Dry Solids, F (%) FiNES
(SH/cy)

Segregating

[:j:jjjj Nonpumpabie
Pumpabie: Non sereatTng

FIGURE 7c: Example with Oil Sand Sludge


(Scott and Cymerman, 1984)
3L

Fines in dry solids, F, along the outside of the bottom axis


is:

= WF
F

The sand-tofines ratio, SFR, along the inside of the bottom


axis is:
w
SFR=
Wy

The fines in dry solids and the sandto-fines ratio are inversely
related as follows:
100%
1 + SFR

The total void ratio, eT, along the inside of the left axis,
is equivalent to the conventional void ratio:

vv
eT =
VF +

where V = volume of voids (water when saturated)


VF = volume of fine particles
V = volume of sand particles

For a saturated material, the water content and the total void
ratio are directly related:

eT
w = x 100%
G

where G = specific gravity of solid particles

(In Fig. 7a, it has been assumed that G5 = GF = 2.7. For some
tailings materials, such as oil sand sludge, Gs and GF are not
equal and adjustments must be made to the scales of the ternary
diagram.)

Constant values of total void ratio plot horizontally across the


ternary diagram.

The fines void ratio, eF, along the inside right axis, is
the same as a conventional void ratio, except that the sand
particles are ignored:
35

Vv
e
F

The fines void ratio is directly related to the total void ratio
as follows:
eF = eT(SFR + 1)

Thus, at an SFR = 0 (no sand), e = eT.

Constant values of fines void ratio plot diagonally from the


sand apex to the corresponding scale along the inside right
axis.

The sand void ratio, e, defines the void ratio of the sand
skeleton and is given by:
- Vv+VF
es

The sand void ratio is directly related to the total void ratio
as follows:
eT(SFR + 1) + 1
e =
SFR

Values of constant e plot diagonally across the lower lefthand


corner of the ternary diagram, as shown. If SFR = (no fines),
= eT.

The fines void ratio is directly related to the sand void ratio
as follows:
eF = SFR
5
e 1

Then, for a particular slurried material:

1. Measure the particle size distribution (ASTM


D 422), including a hydrometer analysis.
Determine the -No. 200 sieve fraction and
calculate the SFR as shown above. Determine
the 2 micron fraction of the No. 200
fraction (i.e., not of the whole sample,
just of the fines).

2. Measure the Atterberg limits on the -No. 200


fraction, not on the No. 40 fraction. This
is important, because often the fines
control the behavior of the slurry.
fActually, we would prefer to make this
t-h, hrnndarv (2
* microns), but we are unaware of a practical
method to do this. If such a method were to
be developed, then we would also re-define
the sand-tofines ratio accordingly.] Also,
the soil should not be oven-dried prior to
running the Atterberg limits; and
supernatant pond water should be used to
adjust the water content. Calculate the
activity based on the plasticity index of
the -No. 200 fraction, divided by the 2
micron fraction determined in Step 1.

3. Measure the minimum void ratio (ASTh D 4253)


and the maximum void ratio (ASTM D 4254) of
the i-No. 200 fraction. The maximum void
ratio defines the approximate void ratio, or
density, at which the sand particles just
come into contact.

For example, consider a uniform sand mixed with phosphatic


clay, such that:

+No. 200 80%


No. 200 = 20%
2 micron = 50% of -No. 200 fraction
SFR = 4.0

Liquid limit of clay = 200% (eL = 5.40)= e oJ LL


Plastic limit of clay 60% (e = l.62) oJ PL
Activity = (20060)/50 = 2.8 (No. 200 fraction)

esmin = 0.35 (uniform spheres,


hexagonal close-
packing)
e max = 0.92 (simple, facecentered
cubic packing)

These data are plotted in Fig. 7b. In general, if the sand


void ratio, e, of any mix is greater than e max and the fines
5
void ratio, eF, is greater than e, then the soil will behave
undrained. These conditions are necessary and sufficient to
assure contractive, claylike behavior. In this example, if the
water content of the mix were 50% (as shown in Fig. 7b), then e 5
> e max 1.94 vs. 0.92; and eF > eL: 6.75 vs. 5.40. Hence, the
mix would be clay-like. At the specified sand-tofines ratio of
4, the mix would definitely behave like a clay if the water
content were as low as 40%, and as high as 113%. (Above the
latter water content, the sand particles would not remain
suspended in the phosphatic clay and, hence, the mix would be
unstable.)

If the sandtofines ratio of this soil mix were altered,


such that SFR < 1, it would be classified as a clay in accordance
with the USCS. And, if it were a natural deposit, with SFR 0
and the water content greater than the liquid limit (eT
= e? >
eL), a geotechnical engineer would be alert to the possibility
that the soil might even be a quick clay. The point of this
example is that if SFR > 1, and e > e and eF > eL, then the
soil will still behave like a clay evenmax
though the USCS says it
is a sand.
The e max arid eL lines do not define the boundary
undrained and drained behavior; just that if they between
are both
exceeded, undrained behavior is assured. In addition, a
sand can
behave like a clay (i.e., undrained), even if it cont
ains no
fines. This can occur if the sand void ratio (in this case,
and SFR = e, = ) exceeds the critical void ratio, e
= eT ;
as
first defined by Casagrande (1936). As is now wellknown, when
a sand experiences large shear strains under drained
conditions,
it approaches a constant volume, corresponding to
its
regardless of the initial void ratio prior to shear.
If the
initial void ratio is less than the sand is said to be
dilative because it expands to reach
e; and if the initial void
ratio is greater than it is contractive. If the same sand
is sheared under undrained cOnditions, it will generate
negative
pore pressure (i.e., sand-like behavior) if its void
ratio is
less than e; and positive pore pressure (clay-like behavi
or) if
its void ratio is greater than A slurrydeposited sand can
easily attain a metastable void ratio that is less than
5
e but
greater than A small triggering event can then leadmaxto an
undrained flow failure.

Casagrande gave the following example of a cohe


sionless
material failing in undrained shear under static cond
itions:
The lack of stability of very finegrained,
saturated materials must also be considered
in
the placing of spoil banks of chemical wastes,
which often consist of extremely fine powders.
Excepting for a dry surface crust, the voids of
such materials are filled with water. The large
capillary pressures render it seemingly very
stable, capable of standing on high vertical
banks. Yet disturbances of any sort, or the
formation of shrinkage or tension cracks with
subsequent infiltration of surface water,
resulting in deformations of the mass, will
result in local liquefaction of the material
which may spread quickly over large areas and
lead to a slide. An example of such a slide,
shown in Fig. 5 [see Fig. 8], had disastrous
consequences. In the background of this picture
one can see still standing the undisturbed banks
of the deposit, 100 feet 1 high.
Later in the same paper, Casagrande concluded that

The [void ratio) in the loose state of many


cohesionless soils, particularly medium and fine,
uniform sands, is considerably above their
critical [void ratio). Such materials in their
loose state tend to reduce their volume if
exposed to continuous deformation. If the voids
are filled with water and the water cannot escape
as quickly as the deformation is produced, then
a
temporary transfer of load on to the water takes
place, arid the resulting reduction in friction
impairs the stability of the mass, which can
lead, in extreme cases, to a flow slide.
0 0

- Ft Ft (i V 0) (Dt5 CD U) WI) Cl) :E 0) 0) 0 CD


(CD -tQ 0) (DO 0)<FtCDO tI () kDHHcI) C) X H
(nco 0 CD U) H< 1 CD 0 CD U)- Ft 0
U) FtQ) z a t- 00) 0I-t 0)HtCD
FtU)0ci 0
0 H H.0-. -0. -1
1 <0) OW CD0-. --- tfl CD
0 -1 OCD < ,Ft U)
- 0)
CD
CDHl)
0
H :5CD CD cni < CD Co
<Ct- H- O t Ft
0 CX)Ut00 CD
H H- 00 U) Ft 0 -.CDF tH
ft p-CD -
cr0(0rtU) Ft 00o CtH
s1U)Xd 1) Ft Ft -xj
H , H Cl) 0) H
ILCD (D a
hft .- tCD FtFtO
FtCD 0) <HU -1 o - -
:5 -.
OCDH
H-O )
HO CD CD :5HH FtO
0 o*:
rt -< Ft()
CDH. XHH.Ct H
0 CD <(D CD - <QH
0)
H- Ut
H H
o a
l0)1lct H _ <OCDFtQ U) Ftp- OFt:5 Ft
CD H- o
:5 H CD ci Ftc Cl)
ClQ:5 0)ttt CoU)N(1Q - H.H
CD
0) CD 0)
CD<c-tCD Lfl :5 H
on CD 0
- ct 0
H t)
:5 0 CD -j U) Ft
0) 0
CD 0 Ft
H CD :i H Ft
II)
,JLD0) t HCDOFt 0 Ft
a0- CDFt (1HCD 0 (D
0
c on O
OR H. TJ) (D(D 0
o:j H,
(0 HflU) Cl
Oul Cl)t5
(DCD CD oCD 0).: (fl(D 0
t-Q H << gI U)FtO Ft Cl)
W0
0 < ctH.FtH(1 FtO H- H
H C0i 0
0CD
CD CD U) ft
H- 0
CD 0) 0
HH-CD U) Co f-DO -
Ft ID CD ct- H 1 Ct-U) CD
(DQWO co<J ftft
Ft HH CD 0)
U)
H
0
.
f-LD 0
H - H a H
0) 0 H<< (fl0)
Ft H-C 0
H Ft
U) 0CD CD U) Ft FtQ
cn- H.0) H tUtaCD
0)0) 1 0 0 -< H- o U)< H-
H Cl) Cl hn 11 00 Itt a..
(0 - H- H- (0
v cu HCD.
CD 1fl
U
0 )
U)
(10 a
:5H- -CD - (flH.
H-::I Cl) -go
(DO CDI) CD
. If-DHD
CD -g
tn
(1(00) CD 0) Ct Cl) H, 0 CDCDCo H:5 ClClt
0
-ID (DH :5U)fl& 0
H H--O
Ft H
OFt.f-C)(D(1
ftCD (0 1
t5t.Ft CD-t (Cl Ft
Ft f-n H.0HFtHH. (0 0)
Cn
1 -
CD 11) CD
i rt(1 H-
Ui
ctrt CD o 0 HU) o (10)
0) Ft
U) CD CD0. Ct 0) Z(DH
H CD H- -0 0
tn
0)
QLQO 0)H 0 Ft Ft
Ft
(D<ClJCD (00) :5. 0) H 0HH. HO (J
H FtU) ft 0) UD0) (/) U) .DHFt f-D1 H
tftFt H a0CD0)CD Cot Hf-DH JHCD U (10
FtO CD FtFtO 0 (DFt (0P1JU1Ft (flH- CDCDUtrt(0 CD-O
CD rt(1CD CD M1 -. 0) (1- 0)-CD 0---- rtCt--<I _.CD
undrained shear strength is usually of more concern, I recommend
that undrained tests be used. Thus, not only is the behavior oi
the tailings material characterized for the purposes of the
ternary diagram, but valuable shear strength data are also
obtained, as discussed in Section 3.1.

However, if the slurry in a tailings dam is normally


consolidated (which is the usual case; the only exception being
sub-aerial deposition), then additional guidelines are as
follows:

1. If e5 > e and eF > e, then the slurry is very


probably contractive. Even at the plastic limit,
corresponding to an effective confining stress of
approximately 1000 kPa, a normally consolidated clay
still generates positive pore pressure during undrained
shear.

2. If e > e and eL > eF > e, then the slurry


is probably contractive.

3. If e
5 < e and eF > eL, then the slurry is
probably dilative.

4. And if e5 < e and eL > eF > e, then the


slurry is definitely dilative.

Shown in Fig. 7c are the results of a detailed laboratory


testing program for one particular mine tailings material: oil
sand sludge mixed with sand (Scott and Cynierman, 1984). Various
boundaries have been defined, including: Sand-Like; Clay-Like;
Solid; Nonpumpable; Puinpable and Nonsegregating; and Segregating.
Thus, a ternary diagram can be used to summarize the behavior of
a tailings or dredged material. By comparing the pattern of one
ternary diagram with others, important insights regarding soil
behavior can also be obtained.

When in doubt, both a drained and an undrained slope


stability analysis must be performed (Rule 2).

5.0 CONCLUSIONS

Rule 1: Geotechnical engineers should avoid designing upstream


tailings dams.

But if you insist, then use a USA (undrained strength


analysis) to evaluate the slope stability, based on consolidated
undrained shear tests, as described by Ladd (1991) Just to be .

sure,

Rule 2: In general, geotechnical engineers should do both an


ESA (effective stress analysis: c, c) and a USA (ce);
whichever yields the lower factor of safety controls
the design of the embankment.

However, in the case of upstream tailings dams, where it is


usually impossible to obtain undisturbed samples of the loose
slimes material, I do not believe that it is necessary to run
CK direct simple shear tests, as Ladd recommends.
U
0 I believe
that CIU triaxial tests on reconstituted samples are
satisfactory, when used with care.

Specifically regarding cohesionless tailings, I recommend


that both a Ladd analysis (strain-compatible undrained shear
strength with F.S. 1.5) and a Poulos analysis (undrained steady
state shear strength with F.S. -1) be performed. Both criteria
for factor of safety must be met. Regarding cohesive tailings,
the Poulos analysis may be impractically conservative and
judg-ment must be exercised.

Measuring the undrained shear strength of slimes, whether


peak or steady state, whether in the laboratory or in the field,
is very difficult. Hence:

Rule 3: Geotechnical engineers should avoid designing


structures in which the shear strength is difficult to
determine, such as an upstream tailings darn. This is
a corollary to Rule 1.

Rule 4: The only exception to Rules 1 and 3 is when (a) The


geotechnical engineer follows Rule 2; and (b) The
geotechnical engineer has a close working relationship
with the client throughout the design-life of the
tailings dam.

Finally, a conceptual framework has been described in which


a given soil is broken down into its three components: sand,
fines (silt and clay), and water. Calculations are then made to
determine if the in situ soil consists of independent sand
particles suspended in a matrix of fines and water; or instead,
consists of fines and water trapped within the interstices of a
sand skeleton. If the former, the soil is contractive (i.e.,
undrained, clay-like behavior) under almost all conditions; if
the latter, the soil is generally dilative (i.e., drained, sand
like behavior) if e5 <

ACKNOWLEDGMEWrB

I greatly appreciate valuable discussions with L.P. Moore


(BCI), J.A. Beriswill (formerly with BCI, now with Golder
Associates, Ltd.), L.G. Bromwell (formerly with BCI, now with
Consolidated Minerals, Inc.), C.C. Ladd (MIT), S.J. Poulos and G.
Castro (GEl Consultants, Inc.), K.H. Stokoe (University of
Texas), and S.G. Vick (private consultant). I also appreciate
K.
Been (Golder Associates, Ltd.) sending me a copy of his paper
prior to publication, and discussing his conclusions with me.
L4j

REFERENCES

Abadjiev, C. B. (1990) Improvement of a Tailings Darn Stability


After Nine Years of Operation, International Symposiumpj
Safety and Rehabilitation of Tailings Dams, ICOLD, Sydney,
Vol. 1, pp. 146155.

ASTM D 422 (1990) ParticleSize Analysis of Soils, Standard


Test Method.

ASTM D 2487 (1990) Classification of Soils for Engineering


Purposes, Standard Test Method.

ASTM D 4253 (1990) MaximUm Index Density of Soils Using a


Vibratory Table, Standard Test Methods.

ASTM D 4254 (1990) Minimum Index Density of Soils and


Calculation of Relative Density, Standard Test Methods.

ASTM D 4318 (1990) Liquid Limit, Plastic Limit, and Plasticity


Index of Soils, Standard Test Method.

AZZOUZ, A. S., Baligh, M. M., and Ladd, C. C. (1981) Three-


Dimensional Stability Analysis of Four Embankment Failures,
Proceedings, 10th International Conference on Soil Mechanics
and Foundation Engineering, Stockholm, Vol. 3, pp. 343-346.

Azzouz, A. S., Baligh, M. N., and Ladd, C. C. (1983) Corrected


Field Vane Strength for Embankment Design, Technical Note,
Journal of Geotechriical Engineering, ASCE Vol. 109, Nay, pp.
730734.

Barron, R. A. (1964) Discussion of Stability Coefficients for


Earth Slopes, by A. W. Bishop and N. R. Morgensterri,
Geotechnique, Vol. 14, December, pp. 360-361.

Been, K. and Jefferies, M.G. (1985) A state parameter for


sands, Geotechnique, Vol. 35, June, pp. 99-112.

Been, K., Jefferies, M.G. and Hachey, J. (1991) The critical


state of sands, Geotechnique, in press.

Bishop, A. W. and Bjerrum, L. (1960) The Relevance of the


Triaxial Test to the Solution of Stability Problems,
Proceedings, Research Conference on Shear Strength of
Cohesive Soils, ASCE, Boulder, pp. 437-501.

Bishop, A. W. and Henkel, D. J. (1962) The Measurement of Soil


Properties in the Triaxial Test, Edward Arnold, Ltd.,
London.

Bjerrum, L. (1973) Problems of Soil Mechanics and Construction


on Soft Clays: StateOfTheArt Report, Proceedings, 8th
International Conference on Soil Mechanics and Foundation
Engineering, Moscow, Vol. 3, pp. 111-159.

Brett, D. N. (1990) Duncan Colliery Tailings Disposal Filter


Dams, International Symposium on Safety and Rehabilitation
of Tailings Dams, IcOLD, Sydney, Vol. 1, pp. 296305.
BrinchHansen, J. (1962) Relationship Between Stability Anai.es
With Total and Effective Stress, SolsSoils, No. i, po. 28
41.

Bromwell, L. G. (1984) Stability Investigation of Tailings Dam


1, Tyrone Mine, Tyrone, New Mexico, Bromwell & Carrier,
Inc. under contract to Phelps Dodge Corp., submitted to New
Mexico State Dam Engineer.

Carrier, W. D., III (1988) Report on Session 2 Material


Properties, Hydraulic Fill Structures, ed. by D. 3. A. van
Zyl and S. G. Vick, ASCE Geotechnical Special Publication
No. 21, Fort Collins, pp. 203226.

Carrier, W. D., Illand Keshian, B., Jr. (1979) Measurement and


Prediction of Consolidation of Dredged Material,
Proceedings, 12th Annual Dredging Seminar, Texas A & N
University, College Station.

Carrier, W. D., III and Beckman, J. F. (1984) Correlations


between index tests and the properties of remoulded clays,
Geotechnictue, Vol. 34, June, pp. 211-228.

Carrier, W. D., III, Bromwell, L. G., and Somogyi, F. (1983)


Design Capacity of Slurried Mineral Waste Ponds, Journal
of Geotechnical Engineering, ASCE, May, pp. 699716.

Carrier, W. D., III, de Mello, L. G., and Moh, ZC (1989)


Environmental Impact in Geotechnical Engineering: Special
Lecture, Proceeding, 12th International Conference on Soil
Mechanics arid Foundation Engineering, Rio de Janeiro, in
press.

Casagrande, A. (1936) Characteristics of Cohesionless Soils


Affecting the Stability of Slopes and Earth Fills,
Contributions to Soil Mechanics. 1925-1940, Boston Society
of Civil Engineers, 1940, pp. 257276. originally published
in the Journal of the Boston Society of Civil Engineers,
January.

Casagrande, A. (1947) Classification and Identification of


Soils, Proceedings, ASCE, Vol. 73, pp. 783810.

Casagrande, A. (1948) Classification and Identification of


Soils, Transactions, ASCE, Vol. 113, pp. 901991.

Casagrande, A. (1976) Liquefaction and Cyclic Deforiation of


Sands-A Critical Review, Harvard Soil Mechanics Series No.
, Harvard University, Cambridge.

Casagrande, A. and Wilson, S. D. (1960) Testing Equipment,


Techniques and Errors: Moderators Report, Session 2,
Proceedings, Research Conference on Shear Strength of
Cohesive Soils, ASCE, Boulder, pp. 1123-1130.

Castro, G. (1969) Liquefaction of Sands, Harvard Soil Mechanics


Series No. 81, Harvard University, Cambridge, January, pp.
1112.
t

Chen, H. W. and van Zyl, D. J. A. (1988) Shear Strength and


VolumeChange Behavior of Copper Tailings Under Saturated
Conditions, Hydraulic Fill Structures, ed. by D.J.A. van
Zyl and S.G. Vick, ASCE Geotechnical Special Publication No.
21, Fort Collins, pp. 430451.

Dobry, R. and Alvarez, L. (1967) SeismiC Failures of Chilean


Tailings Dams, Journal of the Soil Mechanics and
poundations Division, ASCE, Vol. 93, November, pp. 237-260.

Feldkamp, J. R. and Beihonune, G. 11. (1990) Large-Strain


Electrokinetic Consolidation: theory and experiment in one
dimension, ceotechniqu, Vol. 40, December, pp. 557568.

Finn, W. D. L., Yogendrakumar, M., and Ledbetter, R. H. (1990)


Seismic Response Analysis of Tailings Dams, International
Symposium on Safety and Rehabilitation of Tailings Dams,
ICOLD, Sydney, Vol. 1, pp. 7-33.

Gelfand, R. E., Larina, E. A., and Panteleev, V. G. (1990)


Prediction of Ash Dumps Characteristics With Regard for
Their Hardening With Time, International Symposium on
Safety and Rehabilitation of Tailings Dams, ICOLD, Sydney,
Vol. 1, pp. 194203.

Gibson, R. E., England, G.L. and Hussey, M. J. L. (1967) The


theory of onedimensional consolidation of saturated clays,
Geotechnigue, Vol. 17, September, pp. 261-273.

Gibson, R.E., Gobert, A., and Schiffman, R. L. (1990) On Cryers


problem with large displacement and variable permeability,
Technical Note, Geotechnique, Vol. 40, December,
pp. 627-
631.

ICOLD (1982) Manual on Tailings Dams and Dumps, Bulletin 45.

ICOLD (1989) Tailings Dam Safety: Guidelines, Bulletin 74.

Jamiolkowski, M., Ladd, C. C., Gerinaine, J. T., and Lancellotta,


R. (1985) New Developments in Field and Laboratory Testing
of Soils: Theme Lecture 2, Proceedings, 11th International
Conference on Soil Mechanics and Foundation Engineering, San
Francisco, Vol. 1, pp. 57-153.

Jaribu, N. (1973) Slope Stability Computations, Embankment-Dam


Engineering, ed. by R. C. Hirschfeld and S. Poulos, John
Wiley & Sons, Inc., New York, pp. 47-86.

Janbu, N. (1977) Slopes and Excavations in Normally and Lightly


Overconsolidated Clays: State-Of-The-Art Report,
Proceedings, 9th International Conference on Soil Mechanics
and Foundation Engineering, Tokyo, Vol. 2,
pp. 549-566.
Janbu, N. (1979) Design Analysis for Gravity Platform
Foundations, proceedings, 2nd International Conference on
the Behaviour of Offshore Structures, London, Vol. 1,
pp.
407426.
r

Jefferies, M. G., Been, K., and Hachey, J. E. (1990) Influence


of Scale on the Constitutive Behaviour of Sand, 43rd
Canadian Geotechnical Conference, Quebec City.

Koutsoftas, D. C. and Ladd, C. C. (1985) Design Strengths for an


Offshore Clay, Journal of Geotechnical Engineering, ASCE,
Vol. 111, March, pp. 337355.

Kuerbis, R., Negussey, 0., and Vaid, 1. P. (1988) Effect of


Gradation and Fines Content on the Undrained Response of
Sand, Hydraulic Fill Structures, ed. by D.J.A. van Zyl and
S.C. Vick, ASCE Geotechnical Special Publication No. 21,
Fort Collins, pp. 330345.

Ladd, C. C. (1991) Stability Evaluation During Staged


Construction, The TwentySecond Karl Terzaghi Lecture,
Boston, 1986, Journal of Geotechnical Engineering, ASCE,
Vol. 117, April, pp. 537615.

Ladd, C. C. and Foott, R. (1974) New Design Procedure for


Stability of Soft Clays, Journal of the Geotechnical
Engineering Division, ASCE, Vol. 100, July, pp. 763786.

Lambe, T. W. and Whitman, R. V. (1969) Soil Mechanics, John Wiley


& Sons, Inc., New York.

Lo, R. C. and Klohn, E. J. (1990) Seismic Stability of Tailings


Dams, International Symposium on Safety and Rehabilitation
of Tailings Dams, ICOLD, Sydney, Vol. 1, pp. 90105.

Lo, R. C., Klohn, E. J., and Finn, W. 0. L. (1988) Stability of


Hydraulic Sandfill Tailings Dams, Hydraulic Fill
Structures, ed. by D.J.A. van Zyl and S.G. Vick, ASCE
Geotechnical Special Publication No. 21, Fort Collins, pp.
549572.

Lo, R. C., Klohn, E. J., and Finn, W. 0. L. (1991) Shear


Strength of Cohesionless Materials Under Seismic Loadings,
Proceedings, 9th Panamerican Conference on Soil Mechanics
and Foundation Engineering, Via del Mar, Vol. III, pp.
10471062.

Lobdell, H. L. (1959) Rate of Constructing Embankments on Soft


Foundation Soils, Journal of the Soil Mechanics and
Foundations Division, ASCE, Vol. 5, May, pp. 61-76.

Marcuson, W. F., III, Ballard, R. F., Jr., and Ledbetter, R. H.


(1979) Liquefaction Failure of Tailings Dams Resulting from
the Near Izu Oshima Earthquake, 14 and 15 January 1978,
Proceedings, 6th Panamerican Conference on Soil Mechanics
and Foundation Engineering, Lima, Vol. 2, Pp. 69-80.

Massey, A. J., Truscott, E. G., and Callum, J. C. (1990) Raising


of Existing Red Mud Disposal Ponds by the Upstream and
Downstream Methods, International Symposium on Safety and
Rehabilitation of Tailings Dams, ICOLD, Sydney, Vol. 1, pp.
224234.
L5

McLeod, H. N., Chambers, R. W., and Davies, M. P. (1991) Seismic


Design of Hydraulic Fill Tailings Structures, Proyeedin,
9th Panamericari Conference on Soil Mechanics and Foundation
Engineering, Via del Mar, Vol. III, PP. 10631081.

Mesri, G. and Choi, Y.K. (1985) Settlement Analysis of


Einbanknents on Soft Clays, Journal of Geotechnical
Engineering, ASCE, Vol. 111, April, pp. 441464.

Montgomery, R. L. and Leach, J. W., Editors (1984) Dredging and


Dredged Material Disposal, Proceedings of Dredging 84,
ASCE, Clearwater Beach, 1115 pages.

Moore, P. J. (1970) The Factor of Safety Against Undrained


Failure of a Slope, Soils and Foundations, Vol. 10,
September, pp. 81-91.

Moudgil, B. M. and Somasundaran, P., Editors (1985) Flocculation.


Sedimentation and Consolidation, Proceedings of an
Engineering Foundation Conference, Sea Island, 633 pages.

Murphy, S. and Williams, P. (1990) The Application of Finite


Strain Consolidation Theory to the Design of Tailings
Disposal Systems, International Symposium on Safety and
Rehabilitation of Tailings Dams, ICOLD, Sydney, Vol. 1,
pp.
27 02 81.

Murray, R. and Symons, I. F. (1984) Settlement and Stability of


Ejnbank.inents on Soft Subsoils, Ground Movements and Their
Effects on Structures, Surrey University Press,
pp. 321-352.
Parry, R. H. G. (1972) Stability Analysis for Low Embankinents on
Soft Clays, Proceedings, Roscoe Memorial Symposium on
Stress-Strain Behaviour of Soils, Cambridge University, pp.
643668.

Pilot, G., Trak, B., and LaRochelle, P. (1982) Effective Stress


Analysis of the Stability of Embankinents on Soft Soils,
Canadian Geotechnical Journal, Vol. 19, November,
pp. 433-
450.

Poulos, S. J. (1988) Strength for Static and Dynamic Stability


Analysis, Hydraulic Fill Structures, ed. by D. J. A. van
Zyl and S. G. Vick, ASCE Geotechnical Special Publication
No. 21, Fort Collins, pp. 452474.

Poulos, S. J., Castro, G., and France, J. W. (l985a)


Liquefaction Evaluation Procedure, Journal of Geotechnical
Engineering, ASCE, Vol. 111, June, pp. 772-795.

Poulos, S. J., Robirisky, E. I., and Keller, T. 0. (1985b)


Liquefaction Resistance of Thickened Tailings, Journal of
Geotechnical Engineering, ASCE, Vol. 111, December,
pp.
13801394.

Rivard, P. J. and Lu, 1. (1978) Shear Strength of Soft Fissured


Clays, Canadian Geotechnical Journal, Vol. 15, August,
pp.
382390.

Schiffman, R. L. and Carrier, W. D., III (1990) Large
l Strain
Consolidation Used in the Design of Tailings Impoundments,
jernatipna1 Symposium on 5afety and Rehabilitation o
Tailings Dams, ICOLD, Sydney, Vol. 1, pp. 156-174.

Schmertmann, J. II. (1975) Discussion on Measurement of In Situ


Shear Strength, Proceedig, Specialty Conference on In
Situ Measurement of Soil Properties, ASCE, Vol. 2,
pp. 175-
179.

Scott, J. D. and Cymerman, G. J. (1984) Prediction of Viable


Tailings Disposal Methods, Sedimentation Consolidation
Models, Prediction and Validation, ed. by R. N. long and F.
C. Townsend, Proceedings of a Symposium sponsored by ASCE
Geotechnical Engineering Division, San Francisco,
pp. 522-
544.

Seed, H. B. (1987) Design Problems in Soil Liquefaction,


Journal of Geotechnical Engineering, ASCE, Vol. 113, Augus
t,
pp. 827845.

Seed, H. B. (1979) Considerations in the Earthquake-Resistant


Design of Earth Dams, Geotechnique, Vol. 29, pp. 215263.

Seed, H. B., Lee, K. L., Idriss, I. N., and Makdisi,


F. I. (1975)
The Slides in the San Fernando Dams During the Earthquake
of February 9, 1971, Journal of the Geotechnical
Engineering Division, ASCE, Vol. 101, July, pp. 651688.

Seed, H. B., Idriss, I. N., and Arango, I. (1983)


Evaluation of
Liquefaction Potential Using Field Performance Data,
Journal of Geotechnical Engineering, ASCE, Vol. 109, March
,
pp. 458482.

Seed, H. B., Tokimatsu, K., Harder, L. F., and Chung,


R. M.
(1985) Influence of SPT Procedures in Soil Liquef
action
Resistance Evaluations, Journal of Geotechnical
Engineering, ASCE, Vol. 111, December, 1425-1 445.
pp.
Seed, R. B. and Harder, L. F., Jr. (1990) SPT-Based
Analysis of
Cyclic Pore Pressure Generation and Undrained
Residual
Strength, Proceedings, H. Bolton Seed Memorial Symposium,
Berkeley, Bitech Publishers, Ltd., Vancouver, Vol.
2, pp.
35137 6.

Somogyi, F. (1979) Analysis and Prediction of


Phosphatic Clay
Consolidation: Implementation Package, Report by Brornwell
Engineering to Florida Phosphate Council, Lakeland.

Somogyi, F. (1980) Large Strain Consolidation of


FineGrained
Slurries, presented at Canadian Society for Civil
Engineering 1980 Annual Conference, Winnipeg.

Sornogyi, F., Keshian, B., Jr., Bromwell, L. G.,


and Carrier, W.
D., III (1981) Consolidation Behavior of Impounded
Slurries, presented at ASCE National Convention, New
York.
-1/

somogyi, F., Carrier, W. D., III


, Lawver,J. E., and Beckman,
J.
-

F. (1984) Waste Phosphatic Clay


Disposal in Mine Cuts,
Sedimentation Consolidation Models, Prediction and
Validatiop, ASCE, San Francisco ed.
, by R. N. Yong and F. C.
Townsend, pp. 545-564.

Stauffer, P. A. and Obermeyer, J.


R. (1988) Pore Water Pressu
Conditions in Tailings Dams, Hyd re
raulic Fill Structures, ed.
by D. J. A. van Zyl and S.
G. Vick, ASCE Geotechnical
Special Publication No. 21, Fo
rt Collins, pp. 924939.
Stokoe, K. H., II, Nazarian, S.,
Rix, G. J., SarichezSalinero,
I., Shen, JC, and Mok YJ (19
88) In Situ Seismic Testing
of Hard-To-Sample Soils by Sur
face Wave Method, Recent
Advances in Ground Motion Evaluatio
Conference on Earthquake Enginee n, ASCE Specialty
ring and Soil Dynamics II,
Park City.

Stokoe, K. H., II, Rix, G. J.,


and Nazarian, S. (1989) In
seismic testing with surface situ
waves, Twelfth International
Conference on Soil Mechanics and
Foundation Engineering, Rio
de Janeiro, Vol. 2,
pp. 331334.
Svano, G. (1981) Undrained Eff
ective Stress Analyses, the
presented to the sis
Norwegian Institute of Technology,
Trondheim, in partial fulfill
ment of the requirements for
the degree of Doctor of Enginee
ring.
Tavenas, F. and Leroueil, S.
(1980) The Behaviour of Emban
on Clay Foundations, Canadia kinents
n Geotechnical Journal, Vol
17, May, pp. 236260. .

Tavenas, F., Blanchet, R., Gar


neau, R., and Leroueil, S.
The stability of stage-con (1978)
structed embankments on sof
clays, Canadian Geotechnica t
l Journal, Vol. 15, May
283305. , pp.

Terzaghi, K. and Peck, R.B.,


(1967) Soil Mechanics in Eng
Practice, John Wiley & Son ineering
s, Inc., New York.
Troncoso, J. H. (1985) Criti
cal State of Tailing Silty
Earthquake Loading, Sands for
Proceedings, II International
Conference, Soil Dynamics
and Earthquake Engineering,
QE II.
Troncoso, J. H. (1990a)
Failure Risks of Abandoned
Dams, International Sympos Tailings
ium on Safety and Rehabilit
of Tailings Dams, ICOLD, ation
Sydney, Vol. 1,
pp. 34-47.
Troncoso, J. H. (1990b) Seism
ic Responses of Tailings Dam
With Cohesionless Soils s Built
to Different Types of Gro
Motions, International und
Symposium on Safety
Rehabilitation of Tailings and
Dams, ICOLD, Sydney, Vol.
8289. 1, pp.

US Army (1970) Stability of


Earth and Rock Fill Dams,
Manual EM 11102-1902, Cor
ps of Engineers.
US Army (1978) Design and
Construction of Levees, Eng
Manual EM 111021913, Cor ineer
ps of Engineers.

Vous aimerez peut-être aussi