LEROUEL,S., KABBAI,M.

, TAVENAS,
F. & BOUCHARD,
R. (1985). G&technique 35, No. 2, 159-180

Stress-strain-strain rate relation for the compressibility

of sensitive natural clays

S. LEROUEIL,* M. KABBAJ,* F. TAVENAS* and R. BOUCHARDt

Four types of oedometer tests (constant rate of strain nical problems and it is not surprising that
tests, controlled gradient tests, multiple-stage loading hundreds of papers have been written on this
tests and creep tests) were carried out on a variety of subject. Most of the studies are theoretical,
Champlain sea clays. Results obtained on five sites are analytical or numerical and more than 25 differ-
used to demonstrate that the rheological behaviour of
ent rheological models have been suggested for
these clays is controlled by a unique stress-strain-
strain rate relationship. This relationship can be simply
clays. Simply, four families of rheological mod-
described by two curves (upl-E, and (u,‘/ud)-E,). The els can be defined: if e is the void ratio, u,’ the
significance, the practical implications and the limita- vertical effective stress, t the time, & = aelat and
tions of the model are also discussed. 6,’ = au,‘/,% these families can be represented by
the following equations.
Quatre types d’essais oedom&riques (a vitesse con-
stante de d&formation, 21gradient contr616, g charge- R(u,‘,e)=O (1)
ment en plusieurs &tapes et avec fluage) ont BtB
effect& sur diverses argiles marines Champlain. LB
R(u,‘, e, t) = 0 (2)
rtsultats obtenus sur cinq sites sent utilis% pour R(u,‘, e, a,, d) = 0 (3)
dtmontrer que le comportement rheologique de ces
argiles est r&i par une relation unique contrainte- R(uv,‘, e,i)=O (4)
dtformation/vitesse de d&formation. Cette relation
peut &tre d&rite de fawn tr&s simple par deux courbes Equation (1) is representative of models in
(a,-l, et (a,‘/~,‘)-s,). L’article analyse aussi la signifi- which the effective stress-void ratio response of
cation, les implications pratiques et les limitations du
the soil is unique and independent of time or
modtle.
strain rate. This is the case for the classical
NOTATION Terzaghi theory of consolidation in which a
linear effective stress-void ratio relation is as-
compression index
coefficient of secondary compression with
sumed. The model suggested by Davis &
Raymond (1965) based on a linear void ratio-
respect to the void ratio
void ratio
logarithm of effective stress relation is also of
this type. Although equation (1) is widely used
afdat
in practice, it has been recognized ever since
oedometric modulus (du,‘/da,)
Buisman (1936) that the void ratio varies under
initial height of the specimen
a constant effective stress and thus that equation
time
pore pressure at the base of the specimen (1) is not sufficient to describe the rheological
back pressure behaviour of clays completely. Koppejan (1948),
pore pressure distribution parameter Bjerrum (1967) and Hansen (1969) have prop-
osed models represented by equation (2) in
vertical strain (AH/H,)
which the void ratio is a function of the effective
a&,/at
stress and time. A major difficulty, however, is
total vertical stress
preconsolidation pressure encountered with these models when an origin
vertical effective stress for time must be defined, particularly when the
applied load varies with time.
aa,‘/at
The models corresponding to equations (3)
and (4) overcome this difficulty since the be-
INTRODUCTION haviour of the material depends only on its
One-dimensional consolidation has always been present conditions and is not a function of previ-
considered one of the most important geotech- ous history. Taylor & Merchant (1940) were the
first to suggest a model of the type represented
by equation (3) in which the rate of change in
Discussion on this Paper closes on 1 October 1985.
For further details see inside back cover.
void ratio is a function of the effective stress, the
* Laval University, Quebec. void ratio and the rate of change in effective
t Techmat, Quebec. stress. This suggestion has been followed by

159
 160 LEROUEIL, KABBAJ, TAVENAS AND BOUCHARD

numerous researchers (Gibson & Lo, 1961; Wu, remoulded clay indicate a behaviour somewhere
Resendiz & Neukirchner, 1966; Poskitt & Bird- in between the predictions of theories A and B.
sall, 1971; Sekiguchi & Toriihara, 1976). Mod- Therefore it was decided to initiate a detailed
els corresponding to equation (4) show a unique research programme on consolidation. In the
relationship between the effective stress, the first stage, different types of oedometer tests
void ratio and the rate of change in void ratio; (CRS tests, controlled gradient tests (CGTS),
they can be represented in an e-u,’ diagram by stage loading tests and long-term creep tests)
P =constant lines called isotaches by Suklje were carried out on a variety of clays, and from
(1957, 1969a). Various equations have been these experimental data obtained on small speci-
suggested to define these isotaches. The first and mens a rheological model has been developed.
also the simplest corresponds to Taylor’s theory The objective of this Paper is to present these
B (1942) results. In the second stage of the research, the
i=au,‘+be+c
behaviour of clays observed in situ is compared
Other more complex equations have been with the behaviour observed in the laboratory;
suggested by Barden (1965), Suklje (1969b), results in terms of the preconsolidation pressure
Poorooshasb & Sivapatham (1969), Battelino have been published (Leroueil, Samson &
(1973) and Hawley & Borin (1973). Bozozuk, 1983; Leroueil, Tavenas, Samson &
The rheological models proposed were seldom Morin, 1983; Morin, Leroueil & Samson, 1983).
assessed experimentally or only on the basis of a A detailed analysis of the stress-strain-strain
few laboratory test results. Experimental rate behaviour of well-defined clay layers under
studies, however, have been performed on three test embankments will be the subject of
natural clays (Crawford, 1965; Jarrett, 1967; another paper.
Berre & Iversen, 1972; Sallfors, 1975; Larsson,
1981) and on resedimented clays (Smith &
Wahls, 1969; Aboshi, 1973; Burghignoli, 1979), CLAYS INVESTIGATED
but in each study only one type of test was used: The clays investigated were formed in the
constant rate of strain (CRS) tests by Crawford Champlain sea which occupied the St Lawrence
(1965), Smith & Wahls (1969) and Sallfors lowlands and the Ottawa valley from approxi-
(1975); constant rate of loading test by Jarrett mately 12 500 to 10 000 Before Present. Gen-
(1967) and Burghignoli (1979); total load ap- eral physical and mechanical properties of these
plied by steps by Berre & Iversen (1972) and marine clays have been presented by Leroueil,
Larsson (1981). It is thus difficult to obtain an Tavenas & Le Bihan (1983): the organic matter
overall view of the rheological behaviour of content is usually less than 1% and the sensi-
clays from these studies. As a result this abun- tivity greater than 15. According to Lebuis,
dant literature has modified neither the common Robert & Rissmann (1982), less than 50% of
practice based on the Terzaghi theory nor the the clay fraction is phyllosilicates and amorph-
way of thinking on clay behaviour. ous materials, the remainder being primary min-
About half a century after the work of Ter- erals, mainly quartz, plagioclase and potassium
zaghi (1923) and Buisman (1936) engineers are feldspar.
still undecided on whether there is a combina- Eleven sites were investigated by Terratech
tion of primary and secondary compressions Ltd, Montreal, and Laval University, Quebec,
during primary consolidation of thick layers of and 14 series of tests were carried out to study
clay. Suklje (1957), Barden (1965) and Bjerrum the preconsolidation pressure of the clays
(1967) have expressed the idea that there is a (Leroueil, Tavenas, Samson & Morin, 1983).
combination of both processes (theory B, from The oedometer test data have been reanalysed
Ladd, Foott, Ishihara, Schlosser & Poulos, in detail to extract the rheological behaviour of
1977), while Leonards (1977) and Ladd et al. these clays (Kabbaj, Leroueil & Tavenas, 1984).
(1977) think that secondary compression de- Typical results obtained for five representative
velops only after primary consolidation is com- sites are presented here. The main characteris-
pleted (Au-O), (theory A from Ladd et al., tics of the clays are shown in Table 1. The
1977). Here again, the experimental evidence is natural water content varies from 63% to 89%,
almost non-existent or not convincing: the con- the plasticity index ranges between 19 and 43,
solidation test results obtained by Berre & the liquidity index varies from 1.1 to 2.7 and the
Iversen (1972) on specimens of different heights preconsolidation pressure measured in conven-
which are often used to validate theory B were tional oedometer tests (Au,/u,= 0.5 and a re-
also used by Leonards (1977) to justify theory loading schedule of 24 h) ranges from 88 kPa to
A; the tests carried out by Aboshi (1973) on a 270 kPa.
 COMPRESSIBILllT OF NATURAL CLAYS 161

Table 1. Average geotechnkal properties of clays

Site Depth: Water content:* WL 1, Il. Clay (T&: $(cOn”):

m % fraction (fZi- (f&l kPa kPa
cone) vane)

Bat&an 7.3 79.6 (73.8-84.9) 43 21 2.7 81 125 25 6.5 88

Joliette 6.7 65.0 (63.1-66.5) 41 19 2.3 54 96 29 40 115
Louiseville 9.2 76.5 (73.8-79.4) 70 43 1.1 81 28 45 58 160
Mascouche 3.8 67.6 (64.1-70.9) 55 30 1.4 77 65 70 34 270
St CBsaire 6.8 84.8 (824-89.0) 70431.3 84 22 27 68 90

* Extreme values are given in parentheses.

TEST PROGRAMME AND EQUIPMENT first necessary to analyse the eventual influence
Various types of oedometer tests have been of the rate of increase of the vertical effective
performed in this study stress &G = au,‘lat. To do so, it was decided to
carry out, on the clay from Batiscan, CRS tests
(a) Multiple-stage loading tests with a stress in which there is a constant E,= as,/& and a
increment ratio Aa,/a, = 0.5 and reloading continuous increase in uV’, and creep tests in
at the end of primary consolidation ((MSL),) which a,’ is constant and thus & =auV’/at is
or after 24 h ((MSL),) (in this test, drainage equal to zero after the end of pore pressure
was allowed only at the top of the specimen dissipation (Bouchard, 1982).
and the pore pressure was measured at the On the clay from St CXsaire, three CRS tests,
bottom). three CGTs and one (MSL), test were carried
(b) CRS tests (Smith & Wahls, 1969; Wissa, out (Samson, Leroueil, Morin & Le Bihan,
Christian, Davis & Heiberg, 1971) 1981). The results obtained on this clay, and
(c) CGTs in which the pore pressure difference those obtained on the clay from Batiscan, are
hub between the top and the bottom of the presented in detail in this Paper.
specimen was maintained constant (Lowe, On the three other clays selected (Joliette
Jonas & Obrician, 1969) (Samson et al., 1981); Louiseville and Mas-
(d) Creep tests in which the specimen was first couche (Leahy, 1980)), 3-6 CRS tests, one CGT
loaded step by step to an initial stress equi- and up to two multiple-stage loading tests were
valent to the in situ effective vertical stress; carried out. Only reduced data are presented.
then, it was loaded in one step to the final The 200 mm dia. Lava1 sampler (La Rochelle,
stress and maintained under this stress for Sarrailh, Tavenas, Roy & Leroueil, 1981) was
more than 70 days. used on the sites from Batiscan, Louiseville and
In CRS tests and CGTs, the average strain E,= Mascouche and the samples were stored in a
AH/H, or the void ratio e can be related to an humid room at 8 “C. All the tests on the clays
average effective stress calculated with the fol- from these sites were performed at Lava1 Uni-
lowing equation versity, Quebec. Except for the CGTs, all the
oedometer tests were performed without back
U”‘=a”-uuo-a(u~-U~) (6)
pressure on specimens 19.0 mm high and
where u, is the total vertical applied stress, u. is 50.8 mm in diameter. The CGTs were carried
the applied back pressure and ub is the pore out in a specially designed cell (Samson et
pressure measured at the base of the specimen. al., 1981) in which a back pressure of 100 kPa
Smith & Wahls (1969) and Wissa et aI. (1971) was applied at the top of the specimen. The
suggested a value of 0.67 for (Y. Although the specimens were 1.5 mm high and 55 mm in
value is in agreement for low values (below O-4) diameter. For all these tests, synthetic filter cloth
of dub/da,, Janbu, Tokheim & Senneset (1981) of low compressibility and high durability was
have indicated slightly higher values of (Y when installed at the top and the bottom of the
dub/du, becomes higher (a = 0.70 for du,/du, = specimen.
0.7; a! = 0.75 for dub/da,= 0.92). The (Y values On the St CBsaire and Joliette sites the clay
used in this study were those suggested by Janbu was sampled with a modified 70 mm dia.
et al. Geonor piston sampler with an area ratio of
Since the aim of the study was to determine 11% and samples were stored in a humid room
the rheological behaviour of natural clays, it was at 10 “C. The tests were performed by Terratech
 162 LJZROUEIL, KABBAJ, TAVENAS AND BOUCHARD

Effective stress CT,‘: kPa

Batiscan
15 - CRS tests: s-1
-“p,, = 143x10-5
--~“,=5~30x10-6
-.- &=2.13x10m6
-,..- &,=5.33x,0m7
20 -----iv5=,.0,x,0-7
~,,=,+Bx,0-8
H
<-
I
25 &- OB
Fig. 1. Typical CRS oedometer tests on Batisean day

Ltd, Montreal, with cells (Samson et al., 1.7 x ~O-‘S-~ and 4 x lo-’ s-l. Typical results
1981) in which a back pressure of 100 kPa was are presented in Fig. 1 where the origin of strain
applied at the top of the specimen and the pore is taken at u”’ = 65 kPa, a stress equal to the
pressure was measured at the base. The speci- vertical effective stress in the field at the depth
mens were also 15 mm high and 55 mm in of the specimens. This choice has only been
diameter. made to be able to compare CRS tests results
In all tests, the vertical displacements meas- with the creep tests in which the clay was first
ured were corrected for the filter papers and loaded to (T,~’ and settlement measurements
system compressibility. Because of the natural started from this state. Fig. 1 shows that, the
variability of the initial void ratio, even if it was higher the strain rate, the higher the pore pres-
small for the clays considered, all the test results sure increase at the base of the specimen and
were interpreted in terms of strain E, rather than that, for strain rates lower than 5 x 10m7s-‘, the
in terms of void ratios. excess pore pressure is so small that it cannot be
measured. It is also shown that, at a given strain,
RHEOLOGICAL BEHAVIOUR OF NATURAL the higher the strain rate, the higher the effec-
CLAYS tive stress, which is consistent with observations
by Crawford (1965), Sallfors (1975) and
Clay from Bat&an Leroueil, Samson & Bozozuk (1983). It should
Batiscan is situated on the north shore of the be noted, however, that the .& curve intersects
St Lawrence River, about 110 km west of the other curves significantly; among the 18
Quebec City. The clay tested was taken at a tests, three have shown such behaviour and they
depth of 7.25-7.46m. The water content is will be discussed later.
about 80%, the plasticity index 21 and the Nine creep oedometer tests were also carried
liquidity index 2.7. The preconsolidation pres- out on this clay. The final stresses varied be-
sure, as determined from a conventional tween 67 kPa and 151 kPa. The results, which
oedometer test, is 88 kPa (Table 1). are shown in Fig. 2, are consistent with earlier
Eighteen CRS tests were carried out with the observations (Leonards & Girault, 1961; Bjer-
strain rate varying from test to test between rum, 1967; Leroueil, 1977). For high stress in-
 COh4PREtSSlB~ OF NATURAL CLAYS 163

Time: min

20

25 t

Fig. 2. Creep oedometer tests on Bat&an clay

creases, the Ev-lOg t curves have a typical S The curves shown in Fig. 3 can thus be consi-
shape; however, for small stress increases, the dered as Ed-& relationships under constant
curves have a continuously increasing slope with effective stress when t-, < 1O-6 s-l.
time. The same results are also shown in Fig. 3 It is worth noting the shape of these E,--E,
in an &,--log E, diagram. curves, particularly those obtained for stresses of
At the beginning of the test, there were excess 90 kPa, 98 kPa and 109 kPa. There is first a
pore pressures and the effective stresses were rapid decrease in the strain rate at low strains
smaller than the applied stresses. Unfortunately, followed by an approximately constant strain
since pore pressures were not measured in these rate with the strain increasing between 1% and
tests, the effective stresses were unknown. How- 3% (a strain corresponding to the passing of ad
ever, from the shape of the curves shown in Figs in CRS tests, Fig. 1). At higher strains, the strain
2 and 3 and from experience with other clays, it rate decreases progressively.
is thought that the excess pore pressures are From the creep test results, it is possible to
small and the effective stress is approximately define effective stress-strain relations for strain
equal to the applied stress when 6, G 10m6ssl. rates 6, of 10m6s-l, ~O-‘S-~, lo-* ski and even

Strain rate Q,:s-’

IO_‘0 10-8 1o-7
b”_”
67’

5-

g lo-
..
w’
.E
E 15-
x
Bat&an
creep tests
20 -

25-

Fig. 3. Strain-strain rate variation for creep tests on BaGscan clay

 164 LEROUEIL, KABBAJ, TAVENAS AND BOUCHARD

Effective stress CT,‘:kPa 1 and 4, it is evident that the strain rate is a very
important factor influencing the rheological be-
haviour of Batiscan clay. It is important to know
which of equations (3) and (4), or in terms of
strains the equations

No,‘, E,, d”‘, .&) = 0 (7)

E,,6”) = 0
R'(a,', (f3
is representative of the clay behaviour.
In the CRS tests & is constant and the effec-
tive stress is continuously increasing, while in
the creep tests ui is constant (for t-, G 10m6s-l)
.E and thus 6”’ = 0. Therefore, any effect of uV’
E
c should be clearly identified from the test results.
uJ 15
: To analyse this point, the effective stresses ob-
Batiscan tained in the various tests at strains of 5%) 10%)
creep tests 15% and 20% are plotted against the strain rate
Strain rate 8,: s in Fig. 5. At strains of 5%, 10% and 15%,
20- 0 lom6 well-defined &-a,’ relations, independent of the
A 10m7 test used, are found, indicating that the rate of
0 1om8 increase in effective stress d-,’ has no influence
0 10-S on the rheological behaviour of clays and that
25- there is a unique o;‘--E,+, relationship for this
clay.
Fig. 4. Constant 8, curves deduced from creep tests
At a strain of 20%, the scatter is important:
OIIBat&an clay
for example, at a strain rate of 1.07 X ~O-‘S-~,
one CRS test gives an effective stress that is
10m9s-l. This is done in Fig. 4 and it can be seen 12 kPa in excess of the average curve while
that these curves are very similar in shape and another gives an effective stress that is 14 kPa
position to the ones observed in CRS tests (Fig. below this curve. For CRS tests with strain rates
1). of 1.7 X lO-‘s~’ and 3.6~ lo-‘s-’ the effective
From the two sets of curves presented in Figs stresses are significantly higher than the effective

Batiscan clay

I I

10-7 10-B
Strain rate 6,: s-’
Fig. 5. Effective stress-strain rate variation at various strains for Bat&an clay
 COhPRJ33SIBUlTY OF NATURAL CLAYS 165

stresses deduced from creep tests. It is difficult Effective stress 0,‘: kPa
to say whether the scatter (or differences) ob- O0 50 100 150 200 250
I
served at 20% strain is due to fundamental
rheological clay behaviour or to the natural var-
iability of the specimens. This latter possibility is
real, as small variations in grain size distribution, 5-
water content and particle arrangement have a
significant influence on the stress necessary to
compress the soil at high strains. However, for
g IO-
strains up to 1.5% and strain rates usually en- ..
:
countered in the laboratory, it can be considered
.G
that there is a unique stress-strain-strain rate !I
relationship, independent of the rate of effective fX 15-
stress change.
The existence of such a (T,‘--E,--E, relationship
has been confirmed by two special CRS tests in - SPI test
20 -
which the strain rates were changed at various -- SP2 test
strains. The test SPl (Fig. 6) started with a strain i,, = 2.70 x 1Om6s-’
rate E,, = 2.7 x 1O-6 s-r up to a strain of 3.7%, d,,=1.o5x1o-‘s~’
25 - 1,,=1.34x10-5s-’
then was switched to E,* = 1.05 X ~O-‘S-~ up to
a strain of 7.2% and back again. The test SP2
began with a strain rate & = 1.05 x lo-‘s-’ up
to E,= 4.5%, then was switched to FV1 up to 3OL
E,= 8.8% and back again. A strain rate I& =
1.3~ 10e5 s-r was also used for the SP2 test at Fig. 6. Special CRS oedometer tests on Batiscan clay
strains between 17.7% and 21.8%. The results
in Fig. 6 clearly show that there is a unique a narrow range, while four are out of this range
stress-strain-strain rate relationship for this and will be discussed later. For the creep tests,
clay, at least for strains less than 16%. At strains the strain rate varied continuously during the
larger than 23%, the two & curves are differ- test, but at a given strain there was a definite
ent, probably because of small differences in the strain rate and it was possible to normalize the
specimens. However, the results prove that effective stress (equal to the applied stress when
there is still a strain rate effect at strains as high E, < 10m6s-r) with respect to the preconsolida-
as 23%. Graham, Crooks & Bell (1983) also tion pressure associated with the strain rate at
observed such strain rate effects at large strains that strain (Fig. 7). For example, if we consider
on a clay from Belfast. the creep test in which a, = 121 kPa is applied,
From the stress-strain curves presented in after 2000 min the strain is equal to 15.5% (Fig.
Figs 1 and 4, it can be observed that the strain 2, point A), the strain rate of the specimen at
on passing the preconsolidation pressure up’ is that moment is 2 x ~O-‘S-~ (Fig. 3, point A) and
about 1% for all the tests. A a,‘+, relationship the preconsolidation pressure a,’ corresponding
can thus be determined in the same manner as to this strain rate is 94 kPa (Fig. 7, point A). It is
that by which a,‘-& curves were established at then possible to plot u,‘/u,‘(E,) (121/94= 1.29)
other strains (Fig. 5). Both series of a,‘-& against E,= 15.5% (Fig. 9, point A). All the
points deduced from CRS (Fig. 1) and creep creep tests were interpreted in this way and the
tests (Fig. 4) and plotted in Fig. 7 are on the results are plotted in Fig. 9.
same average curve. As for the CRS tests (Fig. 8), with the excep-
The o,,‘-E, and up’-& curves presented in Figs tion of the two more rapid tests carried out with
5 and 7 have roughly the same shape and it a strain rate of 4.0~ 10-5s-‘, they all fall in a
appeared that the complete u,‘--E,-& relation- very narrow range up to strains of 15%. At
ship for the clay could be normalized. This was larger strains, the range obtained with 14 tests
verified in the following way: for CRS tests, the enlarges and, moreover, the two slowest CRS
vertical effective stress u,’ was normalized with tests (6, = 3-6 x lOmas-r and 1.7 X 1O-8 S-I) go
respect to the preconsolidation pressure up’(&) out of this range. There are two possible reasons
associated with the corresponding strain rate on for the behaviour in these two tests: the first is
the average curve (Fig. 7). The a,‘/~,‘(&) ratios associated with the natural variability of the spec-
were plotted against F, for the 18 tests (Fig. 8). imens; the second is associated with a thixotropic
Fourteen normalized test curves for strain rates hardening of the clay skeleton at very small
between 1.4 x 10m5s-r and 1.07 X lo-’ s-’ fall in strain rates of the type observed by Leonards &
166 LEROUEIL, KABBAJ, TAVENAS AND BOUCHARD

Strain rate i,: s-’

10-e 2 5 lo-’ 2 5 10-S 2 5 10-s 2 5 10-a
170 I , I 1 I I I 1 1 1 I

Batiscan clay
150- 0 CRS tests
0 creep tests
(from Fig. 4)

130-

Fig. 7. Variation in preco~~~olidation pressure with strain rate for Bat&can clay

Altschaeffl (1964). No definite answer can be from a different viscous behaviour of the clay at
given and this point will need more considera- very high strain rates. Such normalized stress-
tion in future comparisons between CT,‘-E,-k, strain curves with a different shape have been
relationships in the laboratory and in situ. The observed on other clays.
two more rapid CRS tests give normalized Considering now the creep tests (Fig. 9), all
curves which are markedly different from the the normalized data fall within the range ob-
others; in particular, they are less steep. During tained from the 14CRS tests; even at the slow
these tests, the pore pressures measured at the strain rate of 10-8s-’ (circled points), the data
bottom of the specimens were very high with points are well in the middle of the range. Thus,
Aub/AuVvalues as high as 0.89 and the method
of interpretation is probably doubtful under Normalized effective stress ~,‘/a,‘(~,)
these conditions. Another explanation arises 0.5 1.0 1.5 2.0
O0 I I I
Normal&d effective stress o,‘/a,‘(i,)

0.5 1.0 1.5 2.0 2.5

O0 I I I

5-

Limits for 14 tests

. 0

l<
ae
..
3
.;
L
lo
t
Limits for 14 CRS
(Fig. 8)

2x10-5s creep tests

x 15

II 67 1121
v 78 0133
20- ‘80 x139
Batiscan
CRS tests 088 *I51
20-
*lo9
0 i,=4~0x10-5s-’
0 Point for which L, = 1O-Bsm’ \
0 ~“=4.0X10-5S-’
0 i,=3.6~1O-~s-’
25- 25-
n c,=1~7x10~8s ’
Fig. 8. Normal&d effective stress-strain
relationship Fig. 9. Normahzed effective stress-strain relationship
deduced from CRS oedometer tests on Batisam clay deduced from creep oedometer tests on Batiscan clay
 COMPFtESSJB~ OF NATURAL CLAYS 167

it can be concluded that for all the creep and effect is very clear: at a given strain, the higher
CRS tests, with the exception of those carried the strain rate, the higher is the effective stress.
out at very high strain rates and in which the Three CGTs were also carried out with con-
pore pressure is high (and possibly with the stant pore pressure differences Au, between the
exception of those carried out at slow strain top and the bottom of the specimen of 3.6 kPa,
rates when at large strains), for Batiscan clay 15 kPa and 35 kPa. The corresponding stress-
there is a unique stress-strain-strain rate rela- strain curves are shown in Fig. 11(a). The higher
tionship which can be normalized and rep- the pore pressure difference, the higher is the
resented by only two curves: a,‘-& and effective stress at a given strain. This is consis-
(+“‘/uD’(+&,. tent with the results obtained from CRS tests
since the higher the pore pressure difference the
Clay from St Chaire higher is the strain rate (Fig. 11(b)). Also shown
The site of St CCsaire is located in the on this figure is the fact that the strain rate
Champlain sea basin, about 50 km south east of reaches a maximum at a strain corresponding
Montreal. The clay specimens used for this study approximately to the passage of the preconsoli-
were taken at a depth of 6.55-7.04 m. They are dation pressure (3-5% for this clay sampled
characterized by an average water content of with a 70 mm dia. piston sampler) and then
85%, a liquid limit of 70, a conventional pre- decreases.
consolidation pressure of 90 kPa and a sensitiv- One multiple-stage loading test ((MSL),) with
ity of 22 (Table 1). a stress increment ratio AuJcr, = 0.5 and reload-
Three CRS tests were carried out at strain ing at the end of primary consolidation (more
rates of 2.2 x 10s5 s-l, 4.5 x 10e6 s-’ and 6.67 X precisely when Au,,= 1 kPa) was also carried
lo-‘s-i. The results are shown in Fig. 10. At a out. The average strain and the pore pressure
strain rate of 6.67 x ~O-‘S-~, the excess pore measured at the base of the specimen for the
pressure is negligible. However, at a strain rate 65-99kPa, 99-149kPa and 149-226kPaload-
of 2.2 x 10e5 s-r, the excess pore pressure is ing steps are plotted as functions of the
high, with du,/doV as high as 0.5. In the nor- logarithm of time in Fig. 12. The end-of-loading
mally consolidated clay range, the strain rate points are plotted in a u,I--E, diagram (Fig. 13)

Effective stress 0,‘: kPa

5-

10 -
s
..
w’
.E
E
x
15 -
St Cbsaire
CRS tests

-- ~,=4.50x10-~s-’
6 = 2.20 x 10-S s-1
20- - ”

Fig. 10. CRS oedometer tests on St CCaire clay (after Samson et al., 1981)
 168 LEROUEIL, KABBAJ, TAVENAS AND BOUCHARD

Effective stress 0,: kPa Strain rate 8,:s

oo- 50 100 150 200 1 = 2 5 10-S 2 5 10-a

‘;’ ’
v, \ ‘.
5- \ ‘. \ \
1 !
I ’I
I
s.. 10- II ’I
:
.g
L

Fl 15-
:
\\ ’1,
St C&aim \ ’ St Cbaire
\ \ 1 I
CGTs \ CGTs
\\ l’ I
zo- - Au, = 3.6 kPa \\
I - Au, = 3.6 kPa
‘\\ : -- Au,= 15 kPa
-- Au,=lSkPa I
-- Au,=35kPa Y : -- Au,=35kPa
25-
(a) b)

Fig. 11. Controlled gradient oedometer teats on St Cksaire clay (after Samson et al.,
1981)

Time t: min

Effective stress 0,‘: kPa

50 100 150 200 250
.E 16 \ ,
’
eE .*.
\
v) 20 .-.-. \
-.--._
‘..
24 St Cbsaire . .. . 5-
(MSL), test
\:
\:
(a)
1:
lo- \:
I:
s !i
..
: \.
c 15- \:
0,=99kPa
‘1
L
-- o, = 146 kPa
‘. z
-.- o, = 226 kPa
\.
20 - St G%aire
‘.
\ (MSL), test
\
-u”‘=o,-u,-o66b
. . . ..g”‘=(T.-_ub
25 -

30-

Fig. 13. Effective str-train curves deduced from

Fig. 12. Multiple-stage loading oedometer test on St multiple-stage loading oedometer tests on the St
Cbsaire clay (after Samson el al., 19%1) C&ire clay
 COMPFESSlBlLlTY OF NATURAL CLAYS 169

and the classical oedometric curve (broken line) (b) Even if there is an eventual small error in
is drawn. Also shown in this figure with full the calculated average effective stress, the
lines are the stress-strain curves obtained from stress-strain curves shown in Fig. 13 are
the data presented in Fig. 12 and a calculation representative of the real behaviour. They
of the average effective stress by using equation are very different from the classical
(6) with an cx value of 0.66. This last calculation oedometric curve (broken line) showing, dur-
is not rigorous just after loading, when the ex- ing the first phases of the loading, a rapid
cess pore pressure is high, but it gives a higher increase in the effective stress with small
limit for u,‘. An extreme lower limit is given by strains followed by a progressive increase in
considering the effective stress at the base of the the effective stress with a significant accumu-
specimen: a,’ = a,---~~ (dotted line in Fig. 13). lation of strain. Berre & Iversen (1972) ob-
The difference between these two curves is rela- served similar behaviour in a Norwegian
tively small and the error in the average effec- clay.
tive stress calculated by using equation (6) is
As for the clay from Batiscan, it is possible to
limited to a few kilopascals. The results shown
verify for the clay from St CCsaire whether there
in Figs 12 and 13 call for some remarks.
is a unique u~‘--E,-& relationship and whether
(a) The loading from 65 kPa to 99 kPa is typical this relationship can be normalized. To do so, it
of the behaviour of eastern Canadian over- is first necessary to define the preconsolidation
consolidated clays. The duration of the pressure-strain rate relation. For the CRS tests
primary consolidation is relatively short (Fig. (Fig. lo), the preconsolidation pressures ob-
12(b)) and the &,-log t curve has a slope tained at strains of 3-5% are plotted against the
that increases continuously with time. The imposed strain rates in Fig. 14. In the CGTs, the
Aub-log t curve for loading from 99 kPa to strain rate is varying continuously (Fig. 11(b)).
149 kPa is typical of a clay passing its pre- It is possible, however, to associate the precon-
consolidation pressure with a well-defined solidation pressure with the strain rate of the
step between 10 and 60 min, as previously specimen when passing this preconsolidation
established by Leroueil, Le Bihan & pressure. This is done in Fig. 14 and it can be
Tavenas (1980). The accumulated strain concluded that the up’-& relationship is unique
(about 17%) during this step is very impor- for CGTs and CRS tests.
tant. Finally, the behaviour observed during For all the tests, at a given moment, the clay
the loading from 149 kPa to 226 kPa is typi- specimen has an average strain E,, an average
cal of a normally consolidated clay. strain rate 6, and an average effective stress a,‘.

Strain rate d,:sm’

St @sake
0 CFIS tests
A CGTs

8OL
Fig. 14. Variation in preconsolidation pressure with strain rate in odometer tests
on the St C&mire clay (after Samson et al., 1981)
170 LEROUEIL, KAJ3BA.I.TAVENAS AND BOUCHARD

Normal&d effective stress o,‘/a.‘t&) given soil, any stress-strain-strain rate relation-
ship for the soil may easily be reconstructed as
shown in Fig. 16. Equations (9) and (10) may
also be used directly in numerical methods to
analyse consolidation problems for that soil.
Equations (9) and (10) may be combined to
obtain the general rheological equation

g 10
&=f-’
( ‘)
- (+”
Ed4
(11)

Equation (11) may be experimentally defined

0’
St CIsaire clay in different ways. The method described earlier,
.c
E which led to the establishment of equations (9)
x and (10) and the corresponding figures 7, 9, 14
15 d Au,=3,6kPa 0
and 15 for the Batiscan and St Cesaire clays,
derives equation (11) with reference to the pre-
consolidation pressures a,‘. This approach ac-
CRS tests counts for the fact that a,’ is a useful parameter,
20- o &=2.16~10~~s- representative of the entire geological history of
G &=4.5x16~6s-’
natural clays. However, this approach requires
that the strain at a,’ is approximately constant
for the material considered. Equation (11) could
25- also be determined by referring to any selected
strain E,“, i.e. by defining the two relationships
Fig. 15. Normalfzed effective stress-strain relationship
for the St Cksaire day
o,‘(&o) = f)(&) (12)
o”‘/o”‘(E”O)= g’(c,) (13)
It is thus possible to normalize a,’ with respect
to the preconsolidation pressure corresponding or by determining the stress-strain relation at a
to the strain rate on the average curve drawn in given strain rate .&, and generalizing in the
Fig. 14. The results are shown in Fig. 15. Except forms
for the CRS test carried out at a high strain rate
of 2.2 x 1o-5 s-l, all the results are in a very a”‘(&) = g”(EJ (14)
narrow range around a single line, proving again u;/u,I(E”cJ = P(;-“) (15)
that the clay possesses, in the range of strain
rates usually encountered in the laboratory, a From a practical point of view the experimen-
unique stress-strain-strain rate relationship. tal approach selected in this Paper, which led to
The existence of a unique stress-strain-strain equations (9) and (10) appears to be the most
rate relationship for natural clays supports the suitable for natural clays.
statements of Leroueil, Samson & Bozozuk It should be kept in mind that this rheological
(1983) and Ieroueil, Tavenas, Samson & Morin model has a limitation at high strain rates, the
(1983) that the preconsolidation pressure ob- stresses calculated with equation (6) being
tained in a multiple-stage loading test can be higher than expected. This was observed on the
associated with the strain rate (or the domain of clays from Batiscan and St CCsaire and on many
strain rates) measured at the end of the two or other Champlain sea clays for strain rates grea-
three loadings just after the preconsolidation ter than 1.5 x ~O-‘S-~ and dub/da, greater than
pressure has been passed. 0.45. The reasons are not well understood but
could be due to the method of interpretation,
RHEOLOGICAL MODEL FOR NATURAL and in particular to unrepresentative average
CLAYS values of uV’ and e,, and/or to a change in the
The experimental data presented here indi- rheological behaviour of the clays at high strain
cate that the rheological behaviour of natural rates.
clays may be completely described by the two
GENERALIZATION OF STRESS-STRAIN-
equations
STRAIN RATE RELATIONSHIP
up’ = f(&) (9)
Other eastern Canada clays
U”‘IUP = g(.%) (10)
On clays from other sites of the Champlain
Once these two relationships are known for a sea basin, similar test series were performed by
 COh4PRESSlBILITY OF NATURAL CLAYS 171

Fig.16. Suggested rheological model for natmd clays

Terratech Ltd, Montreal (Samson et al., 1981), solidation pressure-strain rate relation is shown
and the geotechnical section of Lava1 University, in Fig. 17(a) while the normalized stress-strain
Quebec (Leahy, 1980). The data have been relation ((a,‘/~,‘)-e,) is shown in Fig. 17(b). If
analysed and the results have been presented by the variation in the preconsolidation pressure
Kabbaj et al. (1984). The behaviour was essen- with strain rate is similar to the variations ob-
tially the same as previously described and only served for the clays from Batiscan and St
typical results are shown here. Cesaire, the normalized stress-strain curve is
The clay from the Joliette site was sampled much flatter, the stress necessary to reach a strain
and tested using the same technique as for the St of 20% being 2.7 times the preconsolidation
CCsaire clay. The main properties of this clay pressure, compared with 1.5 and 1.35 times for
are presented in Table 1. Its conventional pre- the clays from Batiscan and St Cesaire.
consolidation pressure is 115 kPa. Three CRS The clays from the Louiseville and Mascouche
tests (E, = 2.2 x 10m5s-l, E, = 4.5 x lop6 s-l and sites were sampled and tested using the same
6, = 6.7 x 10e7 s-l), one CGT (Au, = 10 kPa) and technique as for the Batiscan clay. Their main
two multiple-stage loading tests (one with re- properties are presented in Table 1 and their
loading after 24 h (MSL),,, and one with reload- conventional preconsolidation pressures are re-
ing after the end of primary consolidation spectively 160 kPa and 270 kPa. On both clays,
(MSL),) were carried out. The preconsolidation series of CRS tests with strain rates varying
pressures were associated to strain rates, as de- between 5.2 x lo-’ s-l and 1.8~ 10e5 s-l and
scribed earlier for the CGTs and the CRS tests; one CGT with a pore pressure difference hub =
for the multiple-stage loading tests, the strain 15 kPa were carried out. The preconsolidation
rate or range of strain rate was that obtained at pressure-strain rate relations are shown in Figs
the end of the two or three loadings just after 18(a) and 19(a) while the normalized stress-
a,’ had been passed, as suggested by Leroueil, strain relations are shown in Figs 18(b) and
Samson & Bozozuk (1983) and Leroueil, 19(b).
Tavenas, Samson & Morin (1983). The precon- The three series of test results presented in
172 LEROUEIL, KABBAJ, TAVENAS AND BOUCHARD

Strain rate 0,: s-’

10~8 2 5 10-7 2 5 10-e 2 5 10-s 2 5 10-d

rnlW ’ I I 1 1 1 I ! 1
!?? Joliette clay
‘i 170-
0 CRS tests
F
A CGT
z
a 150 - m(MSL), test
p!
P @MSL), test

.! 1=-
;i
rr
a
g 110-

Normalized effective stress o,‘/o,‘(P,)

O0 o-5 1.0 1.5 2.0 2.5 3.0 3.5
I I I f

5-

?I JO-
:

OAu, = 10 kPa

(MSL), test A
(MSL), test 0
25-
(b)

Fig.17. ~~'4, and (cT.,'/u~')-E,

relations
for Joliette
cfay(after
Samson etal.,
1981)

Figs 17-19 show very little scatter and confirm In contrast, the (od/crP’)-~V curves vary mar-
the existence for Champlain sea clays of a uni- kedly from clay to clay. Figs 15 and 17(b) indi-
que u,‘--E,--E, relationship which can be expres- cate that to reach a strain of 20% at a constant
sed by equation (11). strain rate an effective stress of 1.35 times the
Considering Figs 7, 14, 17(a), 18(a) and 19(a), preconsolidation pressure is necessary for the
it can be seen that the variation in the precon- clay from St CCsaire, while a stress of 2.7~~’ is
solidation pressure with strain rate is very simi- needed for the clay from Joliette.
lar from clay to clay. From these results and Series of special oedometer tests were also
others obtained on Champlain sea clays, performed by other researchers on eastern
Leroueil, Tavenas, Samson & Morin (1983) Canada clays: Crawford (1965) and Vaid,
showed that the uP’-E, curves can be normalized Robertson & Campanella (1979) performed
with respect to the preconsolidation pressure CRS tests, and Jarrett (1967) carried out
obtained at a reference strain rate. The results oedometer tests in which the rate of loading was
are presented in Fig. 20 where the reference constant. In each case, the published data show
strain rate is 4 x 10m6s-‘; they indicate that the a significant strain rate effect but do not allow a
change in AoP’/uP’ with strain rate is about the detailed analysis in terms of stress-strain-strain
same for all Champlain sea clays. rate. However, as a first approximation, the
 COMPRESSIBILITY OF NATURAL CLAYS 173

Strain rate 8,: 5’

10-S 2 5 10-7 2 5 10-S 2 5 10-s 2 5 10-a
2401 , I 1 I I I I I I I ,
Louiseville clay
0 CRS tests
A CGTs
!3Conventionel test

Normalizedeffective stress ~,‘/a,‘(&)

0 Au=lSkPa

W
Fig. 18.(~~‘4, and (uv’/crp’)-~, relations for L.ouisedle clay

stress-strain curves obtained in CRS tests by about 3.5% and the strain rate. The shape is
Crawford (1965) and Vaid et al. (1979) can be very similar to the shapes observed for Champ-
normalized in the manner described here. lain sea clays. Fig. 21(b) shows points of the
stress-strain curves obtained in CRS tests nor-
Clays from outside eastern Canada malized with respect to the preconsolidation
Series of test results obtained on clays from pressure obtained at the corresponding strain
outside eastern Canada have also been pub- rate (Fig. 21(a)). The scatter of the data is very
lished and can be used to verify the general small and the rheological behaviour of the
validity of the model established here. B&ckebol clay can be described by the two
Natural clays. SZlfors (1975) carried out a curves in Fig. 21 or by equation (ll), following
series of six CRS tests on specimens of intact the principle established for Champlain clays.
clay taken at a depth of 7 m at the site of On clay specimens from the same site of
BHckebol, Sweden. The strain rates used varied B&kebol, but taken at a depth of 8 m, Larsson
between 1.66 x lo-’ s-l and 5 x 10m7s-l. Fig. (1981) carried out eight multiple-stage loading
21(a) shows the relationship between the pre- oedometer tests with load increment ratios
consolidation pressure obtained at a strain of (Aa,/cJ of almost unity and defined such that
174 LEFtOUJZIL, KABBAJ, TAVENAS AND BOUCHARD

Strain rate d,: s-’

10-B 2 5 lo-’ 2 5 10-e 2 5 10-S 2 5 IO-4
440‘ 1 I , , I I I I 1 1-n
Mascouche clay
g
” 400 0 cFEtests
-
b A CGT
E 0 Conventional test
z
m36c-
c
P
s
‘g 320 -
0

(a)
Normalized effective stress a,‘/a,‘(B,)
0.5 1.0 1.5 2.0
CI?-

5,-

Mascouche clay
CGT
0 Au,=lSkPa
c!?S tests
0 d, = 1.76x 10-5s-’
n 1,=6.0~10-~s-’
20
0 &=3.5xlo~~s-’
A &=2~1~1O~~s~’
A Q,=1.05x10-~s-’
* i, = 5.2~ IO-‘s-’
25
Fig. 19. ~~‘4, and (u,‘/u~‘)-E, relations for Mascooche thy

1.3- 0 Batiscan V Mascouche ,+90/o

t Joliette + St CBsaire ;
1’2 - 0 Louiseville
7 l.l For other sites and symbols, see Leroueil, *
Q” Tavenas, Samson & Morin (1983)

$ l.O-
::
2 0.9-

>
b 0.8-

0.7 - / - -

0.6 - I I I I
10-8 10-7 10-a 10-S 104
Strain rate 8,: se’
Fig. 20.Nomwlized precondidation pressurwdrain rate relationship for
Champlain sea days (from Jaxooeil, Tavenas, Samson & Morin, 1983)
 COh4PRESSlBlLlTY OF NATURAL CLAYS 175

Strain rate iv: s-’

10-e 2 5 10-5 2 5 10-s 2 5 10-S 2 5 104

1101 1 T I I 1 , 1 1 T I I 1

B;ickebl (7 m)
CRS tests

(a)

Normalized effective stress o,‘/o,‘(&:,)

Bckebol (7 m)

7
A C7-34, 8,=2.0~10-~s-’

1
20
0 c74, ~,=3Glx10~s-'
0 c7-5, ~,=7.5xlo~=s-’
A c7-6, b, = 1.7 x lo-Es-’
25
b)
Fig. 21. and (a,‘/~,‘)-E,
up’-& relations for Biickebol clay at a
depth of 7 m (after SCillfors, 1975)

the applied stresses were different for the eight clay from Belfast, Northern Ireland, when
specimens. These tests gave almost continuous Graham et al. (1983) performed a special CRS
stress-strain curves. In these tests, the end of test in which the strain rate was changed at
primary consolidation has been obtained at various strains. The same researchers reported
strain rates higher than 10-6s-1 and from the strain rate effects on the preconsolidation pres-
test results Larsson was able to draw curves sure of various other clays from Belfast, North-
corresponding to strain rates of 10-6s-1, ern Ireland, Drammen, Norway, and Winnipeg,
~O-‘S-~ and 10-8s-1. From these curves, it was Canada. Whereas Graham et al. suggested a
possible to estimate the preconsolidation pres- linear relationship between the preconsolidation
sures which are plotted against the strain rate in pressure and the logarithm of strain rate, most
Fig. 22(a). The normalized effective stress-strain of the published data (the average curve ob-
curves, drawn in Fig. 22(b), superimpose, con- tained from 14 Champlain sea clays (Fig. 20),
firming the validity of the proposed model. clay from BHckebol (Figs 21 and 22), clay from
A strain rate effect was also observed on a Drammen (Graham et al., 1983) and the clay
176 LJZROUEIL, KABBAJ, TAVENAS AND BOUCHARD

Strain rate 8,: s-’ able strain rate effects. This can be considered as
a particular case of the proposed model (equa-
tion (11)) in which the variation in the precon-
solidation pressure (or the effective stress at a
given strain or void ratio) with strain rate is
negligible.
Finally, when considering remoulded clays,
the behaviour represented by equation (11) may
be not general since, as shown by Leonards &
Altschaeffl (1964), there could be a strengthen-
P
F Normalized effective stress o,‘/o,‘k,) ing of the bonds between particles.

DISCUSSION
In the previous sections, a rheological model
for natural clays has been established. In this
section, the physical meaning of this model and
its practical implications will be discussed.

Significance of the proposed model

As previously shown, the rheological be-
haviour of sensitive natural clays is completely
described by two curves ((+p)-E, and (~:/a,‘)-
E,). The (u~‘/u~‘)--E, curve has an important
physical meaning since it represents the reaction
of the structure of the clay and thus its mineral-
ogy and geological history.
35 1 The al)‘-& curve is representative of the capa-
b) bility of the clay skeleton to creep. For Champ-
Fig. 22. up’-iv and (u~‘/u,,‘)--E, relations for Bickebol lain sea clays, Leroueil, Tavenas, Samson &
clay at a depth of 8 m (after Larson, 1981) Morin (1983) showed that the a,‘-& curves had
similar shapes and could be normalized as
shown in Fig. 20. For other clays with different
from St Jean-Vianney (Vaid et al., 1979)) indi- mineralogies, other trO’-&, curves can be ex-
cate a decrease in the strain rate effect with pected. However, a compilation made by Mesri
decreasing strain rate. & Godlewski (1977) shows that the C,,/C, ratio,
Remoulded clays. Burghigffoli (1979) pre- in which C,, = Ae/A log t and C, = Ae/A log a,’
sented results of a series of constant rate of are obtained from oedometer tests, is about the
loading oedometer tests on a remoulded clay same (O-04) for most non-organic natural clays.
from Fiumicino, Italy. The preconsolidation Such a constant C,,/C, ratio implies (Mesri &
pressures were not well defined in these tests Choi, 1979) that the variation in up’ per
and the approach used for the natural clays, with logarithm cycle of time or strain rate is about
normalization with respect to up’, could not be the same for these materials.
used as such. A detailed analysis of the results For clays such as the Champlain sea clays, for
(Kabbaj et al., 1984) showed, however, that the which the variation in the preconsolidation pres-
rheological behaviour of the remoulded sure with strain rate is known (Fig. 20), one
Fiumicino clay can be described by two curves stress-strain curve obtained from a CRS
corresponding to equations (14) and (15) and is oedometer test is sufficient to determine the ov’-
thus consistent with the model proposed here. F,--E, relationship of the clay considered com-
Smith & Wahls (1969) carried out series of pletely.
CRS oedometer tests on kaolinite and calcium In the model proposed, the E,= constant
montmorillonite. The pore pressures measured curves can be normalized with respect to the
in tests on montmorillonite were as high as 88% preconsolidation pressure that they define. This
of the applied total stress and, as indicated by means that, for the two curves E, = E,, and & =
Smith and Wahls, in such conditions the in- & showing the preconsolidation pressures uPl’
terpretation of the tests becomes questionable; and up*‘, at a given strain the effective stresses
these results have been disregarded. The results are such that
obtained on kaolinite are very different from %I U”2’
_=-
other results since they do not present notice-
(16)
UPI’ up*’
 COMPREssIBlLlTY OF NATURAL CLAYS

s.
:
.5
c
63
- &=10-6s-’
- 8, = 10-5 s-1
-- 6, = 10-e s-’
-- &=,O-Ps-’
-.- 1, = 10-7 s-’
-.- 6, ZE lo- s-’

___ Schematic CGl ---- Schematic

multiple-stage
loading test

Effective stress cr”‘: kPa Effective stress a,‘: kPa

Fig. 23. schematic state path followed daring a con- Fig. 24. Schematic state path followed dming a
trolled gradient oedometer test multiple-stage loading oedometer test

This equation can also be written as curves obtained from CRS tests (Fig. 10) and
CGTs (Fig. 11(a)) on the clay from St Cesaire.
log a,r’ -log a,*’ = log c+’ -log a@ (17)
In multiple-stage loading tests, the strain rate
which indicates that in an a,-log u,’ diagram the is directly related to the excess pore pressure. At
distance log a”,‘-log uVz’ between two E, = the beginning of a loading period, pore pres-
constant curves is a constant. sures and thus strain rates are high and the state
path lies on corresponding E, curves, exhibiting
Clay response during testing relatively high effective stresses. With time, the
In an E”--CT.,’ diagram such as Fig. 23, the strain rate decreases and the stress-strain curve
model is depicted by a family of curves, each intersects the &=constant curves as shown in
corresponding to a particular strain rate. At a Fig. 13 and schematically represented in Fig. 24.
given strain, the higher the strain rate, the The behaviour is entirely controlled by two in-
higher is the effective stress. During one test, the dependent laws which are Darcy’s law and the
stress-strain followed in the diagram will vary u~‘--E,--E, law of the clay skeleton. Between
depending on the type of test and the corres- points D and E, the strain rate is controlled by
ponding evolution of strain rates. excess pore pressures and Darcy’s law: this is
In a CRS test, the soil follows an & = constant the primary consolidation period. If the test is
cuNe. If the strain rate changes from E,, to .&, extended to strains in excess of E,~, when the
as for the clay from Batiscan (Fig. 6), the effec- effective stress is essentially constant, the strain
tive stress jumps from the i,= E,r curve to the rate is then controlled by the viscosity of the
E, = E,z curve. clay skeleton: this is the secondary consolidation
In a CGT, the pore pressure difference be- period. During secondary consolidation, there
tween the top and the bottom of the specimen are movements of porewater, and thus, accord-
is constant and it was observed that the strain ing to Darcy’s law, excess pore pressures. How-
rate decreased while the strain increased in the ever, for small specimens these excess pore pres-
normally consolidated range. Considering a sures are very small and almost impossible to
hypothetical CGT, owing to this decrease in measure (for a specimen 2 cm high with a per-
strain rate between points B and C (Fig. 23), the meability of 10-9m/s, the excess pore pressure
effective stress-strain curve is steeper than the would be of the order of 0.1 kPa at a strain rate
curves deduced from CRS tests. This is evi- of 5 x lo-* s-l).
denced by comparing the effective stress-strain From the previous comments on the tests, the
178 LEROUEIL, KABBAJ, TAVENAS AND BOUCHARD

values of the oedometric modulus E,,= clay rebounds owing to unloading or relaxes
do,‘/de, or other strain parameters such as m,, under a constant strain. In relaxation tests, the
a, and C, depend on the type of test carried out strain is constant, the strain rate is thus constant
to measure them. and equal to zero, and in such conditions the
uniqueness of an effective stress-strain-strain
rate relationship implies a constant effective
Rheological model
stress. However, it is well known that it is not
To eliminate the effect of initial void ratio the case and that the effective stress decreases
differences on the experimental results, the with time. When analysing this, it must be re-
model was established by referring to strains. membered that the model proposed was de-
However, the proposed (+d-E,--F, relation is veloped considering total strains, while the clay
equivalent to a a,‘-e-i? form, i.e. to a rheologi- behaviour should probably be decomposed into
cal model represented by equation (4). In par- elastic and plastic strain components. In the
ticular, the E, = constant curves are equivalent to normally consolidated range, the elastic strains
the isotaches (i’ = constant) suggested by Suklje
are small compared with the plastic strains. The
(1957). Moreover, the tests carried out in the
model, which was developed mainly from obser-
present study (on small specimens and in a vations in the normally consolidated range, is
certain domain of strain rate) do not indicate a thus believed to represent essentially plastic
significant influence of the stress rate civ’ on the strains. In the overconsolidated state, either for
rheological behaviour of the clays. Consequently stresses lower than u,,’ or during relaxation, the
rheological equations of the type of equation (3) elastic strain component could become relatively
are not relevant for natural clays. more important and the model could be rep-
In contrast with most previous studies on the resentative of only part of the behaviour. There
rheological behaviour of clays, no mathematical is a need for detailed research on the oedomet-
equation is suggested. The reason is that the ric clay behaviour in the overconsolidated range
model is completely defined by the two experi- and during relaxation.
mental curves corresponding to equations (9)
and (10) which can be programmed in a compu-
Use and limitations
ter, without using specific equations. However,
any proposed equation should be of the general The rheological model proposed described by
form the two curves a,’ = f(.&) and u</upl = g(e,) can
be combined with a strain (or void ratio)-
a”’ permeability relation (E,-k or e-k), such as
&=f-’
( )
-
g(G)
(11) those described by Tavenas, Jean, Leblond &
Leroueil (1983), to resolve problems of consoli-
The model presented here is mainly based on dation. It should be kept in mind, however, that
data obtained in the normally consolidated the model has been established on small speci-
range; the behaviour in the overconsolidated mens and for strain rates usually encountered in
range is not as clear. The CRS tests carried out the laboratory, and thus, for the moment, it
by Sallfors (1975) show separated 6, = constant must be used only under such conditions to
curves, the strain moduli E,,=du”‘/da, de- interpret and understand clay behaviour in the
creasing with the strain rate, but this is not so laboratory. It will probably be the basis for a
evident from the CRS tests performed in the reanalysis of the current practice for estimating
present study, the scatter in the results in the settlements and settlement rates. However, be-
overconsolidated range being more significant. fore using this model for field applications where
However, creep tests in the same range of stress the strain rates are much lower and the clay
show increases in strain with decreasing strain layers much thicker, it is necessary to verify its
rates (Tavenas, Leroueil, La Rochelle & Roy, validity under these conditions and in particular
1978, Fig. 3), indicating separated &, = constant to determine how the a,‘--&, curve extrapolates
curves. Without clear evidence of the real be- at low strain rates. A research programme on
haviour but accepting, to a first approximation, a this problem is in progress at Lava1 University,
normalization of the uV’-E,--E, behaviour (Figs Quebec.
15, 17(b), 18(b), 19(b) and 21(b)), the model
shown in Figs 16 and 23 with separated E,= CONCLUSION
constant curves is suggested. Various types of oedometer tests (CRS tests,
The model proposed has been established on CGTs, multiple-stage loading tests and creep
the basis of tests in which the strains were tests) were carried out on a variety of Champ-
always increasing and it must be used only under lain sea clays. From the test results a simple
such conditions; it should not be used when the rheological clay model has been established. The
 COMPFESSlBILlTY OF NATURAL CLAYS 179

clay behaviour under one-dimensional compres- structural viscosity of soft clays by means of con-
sion is controlled by a unique stress-strain- tinuous consolidation tests. Proc. 7th Eur. Conf.
strain rate relationship which can be described Soil Mech. Fdn Engng, Brighton 2, 23-28.
Crawford, C. B. (1965). The resistance of soil struc-
by only two curves (a,‘-& and (cT,‘/u~‘)- E,).
ture to consolidation. Can. Geotech. J. 2, No. 2,
This model is a pariicular case of the isotache
90-97.
model proposed by Suklje (1957). Davis, E. H. & Raymond, G. P. (1965). A non-linear
From published data on secondary consolida- theory of consolidation. Gtotechnique 15, No. 2,
tion and strain rate effects, this model may be 161-173.
representative of a: large variety of natural clays. Gibson, R. E. & Lo, K. Y. (1961). A theory of
However, the model has been established on consolidation for soils exhibiting secondary compres-
specimens of small thickness in which the strain sion. Report No. 41, Norwegian Geotechnical In-
was continuously increasing, and, for the mo- stitute.
Graham, J., Crooks, J. H. A. & Bell, A. L. (1983).
ment, it must be used only in these conditions.
Time effects on the stress-strain behaviour of
natural soft clays. Gtotechnique 33, No. 3, 327-
ACKNOWLEDGEMENTS 340.
Hansen, B. (1969). A mathematical model for creep
The work carried out at Lava1 University was
phenomena in clay. Proc. 7th Int. Conf. Soil Mech.
supported by grants from the ministries of Fdn Engng, Mexico City, Specialty session 12, pp.
Transport and Education of Quebec and from the 12-18.
National Sciences and Engineering Research Hawley, J. G. & Borin, D. L. (1973). A unified theory
Council of Canada. The contribution of D. for the consolidation of clays. Proc. 8th Int. Conf.
Leahy and P. Leblond who performed some of Soil Mech. Fdn Engng, Moscow 1, No. 3, 107-119.
the laboratory tests is gratefully acknowledged. Janbu, N., Tokheim, 0. & Senneset, K. (1981). Con-
Part of the research was performed by Ter- solidation tests with continuous loading. Proc. 10th
ratech as an unsolicited proposal sponsored by Int. Conf. Soil Mech. Fdn Engng, Stockholm 1,
645-654.
the Division of Building Research of the Na-
Jarrett, P. M. (1967). Time-dependent consolidation
tional Research Council of Canada and was
of a sensitive clay. Bull. Am. Sot. Test. Mater 7,
financed by the Division of Building Research of No. 7, 300-304.
the National Research Council and Supply and Kabbaj, M., Leroueil, S. & Tavenas, F. (1984).
Services Canada under contract lSX79-00026. Dtueloppement d’une loi rhtologique repr&entatiue
The Authors thank J. P. Le Bihan, P. Morin and du comportement en laboratoire des argiles naturef-
L. Samson for performing the tests and for les. Internal Report, Department of Civil En-
constructive discussions. gineering, Lava1 University, Quebec.
Koppejan, A. W. (1948). A formula combining the
Terzaghi load compression relationship and the
Buisman secular time effect. Proc. 2nd Int. Conf.
REFERENCES Soil Mech. Fdn Engne, Rotterdam 3,32-37.
Aboshi, H. (1973). An experimental investigation on Ladd, C. C., Foott, R., Ishihara, K., Schlosser, F. &
the similitude in the consolidation of a soft clay, Poulos. H. J. (1977). Stress-deformation and
including the secondary creep settlement. Proc. 8th strength characte;isti& Proc. 9th Int. Conf. Soil
Int. Conf. Soil Mech. Fdn Engng, Moscow 4, No. 3, Mech. Fdn Engng, Tokyo 2, State of the art report,
88. pp. 421-494.
Barden, L. (1965). Consolidation of clay with non- La Rochelle, P., Sarrailh, J., Tavenas, F., Roy, M. &
linear viscosity. Gtotechnique 15, No. 4, 345-362. Leroueil, S. (1981). Causes of sampling distur-
Battelino, D. (1973). Oedometer testing of viscous bance and design of a new sampler for sensitive
soils. Proc. 8th Int. Conf. Soil Mech. Fdn Engng soils. Can. Geotech. J. 18, No. 1, 52-66.
Moscow 1, No. 1, 25-30. Larsson, R. (1981). Drained behaviour of Swedish
Berre, T. & Iversen, K. (1972). Oedometer tests with clays. Report No. 12, Swedish Geotechnical Insti-
different specimen heights on a clay exhibiting tute.
large secondary compression. Ge’otechnique 22, Leahy, D. (1980). Contribution h I’ttude du comporte-
No. 1, 53-70. ment oedomktique des argiles. MSc thesis, Depart-
Bjerrum, L. (1967). Engineering geology of Norwegian ment of Civil Engineering, Lava1 University,
normally consolidated marine clays as related to the Quebec.
settlements of buildings. Ge’otechnique 17, No. Lebuis, J., Robert, J. M. & Rissmann, P. (1982).
2,83-118. Regional mapping of landslide hazard. Int. Symp.
Bouchard, R. (1982). Eflet de la uitesse sur le com- Landslides, Linkbping, pp. 205-262.
portement oedomdtrique de I’argile de B&scan. Leonards, G. A. (1977). Proc. 9th Int. Conf. Soil Mech.
MSc thesis, Laval University, Quebec. Fdn Engng, Tokyo 3, Panel discussion, pp. 384-
Buisman, A. S. (1936). Results of long duration settle- 386.
ment tests. Proc. 1st Int. Conf. Soil Mech. Fdn Leonards, G. A. & Altschaeffl, A. G. (1964). Com-
Engng, Cambridge 1,103-107. pressibility of clay. J. Soil Mech. Fdns Div. Am.
Burghignoli, A. (1979). An experimental study of the Sot. Civ. Engrs 90, SM5, 133-155.
 180 LEROUElL, KABBAI, TAVENAS AND BOUCHARD

Leonards, G. A. & Girault, P. (1961). A study of (1981). Pressions de p&consolidation des argiles
one-dimensional consolidation test. Proc. 5th Int. sensibles. DSS Contract lSX79-00026. Division of
Conf. Soil Mech. Fdn Engng, Paris 1, 213-218. Building Research of the National Research Coun-
Leroueil, S. (1977). Quelques considerations sur le cil of Canada.
comportement des argiles sensibles. PhD thesis, De- Sekiguchi, H. & Toriihara, M. (1976). Theory of one-
partment of Civil Engineering, Lava1 University, dimensional consolidation of clays with cbnsidera-
Quebec. tion of their rheoloaical orooerties. Soils Fdns 16.
Leroueil, S., Le Bihan, J. P. & Tavenas, F. (1980). An No. 1, 27-44. - _ _
approach for the determination of the preconsoli- Smith, R. E. & Wahls, H. E. (1969). Consolidation
dation pressure in sensitive clays. Can. Geotech. J. under constant rates of strain. J. Soil Mech. Fdns
17, No. 3, 446-4.53. Div. Am. Sot. Ciu. Engrs 95, SM2, 519-539.
Leroueil, S., Samson, L. & Bozozuk, M. (1983). Suklje, L. (1957). The analysis of the consolidation
Laboratory and field determination of preconsoli- process of the isotache method. Proc. 4th Int. Conf.
dation pressure at Gloucester. Can. Geotech. J. 20, Soil Mech. Fdn Engng, London 1, 200-206.
No. 3, 477-490. Suklie, L. (1969a). Rheoloaical asoects of soil
Leroueil, S., Tavenas, F. & Le Bihan, J. P. (1983). mechanics. London: Wiley-Interscience.
Proprittes caracteristiques des argiles de Test du Suklje, L. (1969b). Consolidation of viscous soils sub-
Canada. Can. Geotech. J. 20, No. 4, 681-705. jected to continuously increasing uniform load. In
Leroueil, S., Tavenas, F., Samson, L. & Morin, P. New advances in soil mechanics, Vol. 1, pp. 199-
(1983). Preconsolidation pressure of Champlain 235. Praha.
clays: part II-laboratory determination. ‘Can. Tavenas, F., Jean, P., Leblond, P. & Leroueil, S.
Geotech. J. 20, No. 4. 803-816. (1983). The permeability of natural soft clays: part
Lowe, J. III, Jonas, E. & Obrician, V. (1969). Con- II, permeability characteristics. Can. Geotech. J.
trolled gradient consolidation test. J. Soil 20, No. 4, 645-660.
Mech. Fdns Div. Am. Sot. Ciu. Engrs. 95, SMl, Tavenas, F., Leroueil, S., La Rochelle, P. & Roy, M.
77-97. (1978). Creep behaviour of an undisturbed lightly
Mesri, G. & Choi, Y. K. (1979). Strain rate behaviour overconsolidated clay. Can. Geotech. J. 15, No. 3,
of the Saint-Jean-Vianney clay: discussion. Can. 402-423.
Geotech. J. 16, No. 4, 831-834. Taylor, D. W. (1942). Research on consolidation of
Mesri, G. & Godlewski, P. M. (1977). Time and stress clays. Series 82, Massachusetts Institute of Tech-
compressibility interrelationship. J. Geotech. Engng nology.
Div. Am. Sot. Ciu. Engrs 103, GT5, 417-430. Taylor, D. W. & Merchant, W. (1940). A theory of
Morin, P., Leroueil, S. & Samson, L. (1983). Precon- clay consolidation accounting for secondary com-
solidation pressure of Champlain clays: part I, in pression. .I. Math. Phys. 19, 167-185.
situ determination. Can. Geotech. J. 20, No. 4, Terzaghi, K. (1923). Die Berechnung der Durchlassig-
782-802. keitsziffer des Tones aus dem Verlauf der hyd-
Poorooshasb, H. B. & Sivapatham, T. (1969). Consoli- rodynamischen Spannungserscheinungen. Sber.
dation of sensitive clays exhibiting strong structural Wien. Akad. Wiss. 132, Nos. 3 & 4.
breakdown. Proc. 7th Int. Conf. Soil Mech. Fdn Vaid, Y. P., Robertson, P. K. & Campanella, R. G.
Engng, Mexico City, Specialty session 12, pp. 27- (1979). Strain rate behaviour of Saint-Jean-
37. Vianney clay. Can. Geotech. .I. 16, No. 1, 34-42.
Poskitt, T. J. & Birdsall, R. 0. (1971). A theoretical Wissa, A. E. Z., Christian, J. T., Davis, E. H. &
and experimental investigation of mildly non-linear Heiberg, S. (1971). Consolidation at constant rate
consolidation behaviour in saturated soil. Can. of strain. J. Soil Mech. Fdns Div. Am. Sot. Ciu.
Geotech. J. 8, No. 2, 182-216. Engrs 97, SMIO, 1393-1413.
Siillfors, G. (1975). Preconsolidation pressure of soft Wu, T. H., Resendiz, D. & Neukirchner, R. J. (1966).
high plastic clays. PhD thesis, Chalmers University Analysis of consolidation by rate process theory. I.
of Technology, Gothenburg. Soil Mech. Fdns Diu. Am. Sot. Ciu. Engrs 92,
Samson, L., Leroueil, S., Morin, P. & Le Bihan, J. P. SM6, 229-248.