Sie sind auf Seite 1von 9

Rotta, G. V., Consoli, N. C., Prietto, P. D. M., Coop, M. R. & Graham, J. (2003). Géotechnique 53, No.

5, 493–501

Isotropic yielding in an artificially cemented soil cured under stress


G . V. ROT TA ,  N. C . C O N S O L I ,  P. D. M . P R I E T TO, † M . R . C O O P { a n d J. G R A H A M }

The work simulates, in the laboratory, the formation of a Nos travaux simulent, en laboratoire, la formation d’un
cemented sedimentary deposit in which cement bonding dépôt sédimentaire cimenté dans lequel la cimentation se
occurs after burial and under geostatic stresses. Isotropic produit après enfouissement et sous certaines contraintes
compression tests were carried out on artificially ce- géostatiques. Des essais de compression isotrope ont été
mented specimens made with variable cement contents. pratiqués sur des spécimens artificiellement cimentés con-
After consolidating the samples to the uncemented nor- tenant diverses quantités de ciment. Après avoir consolidé
mal compression line, the specimens were allowed to cure les échantillons jusqu’à la ligne de compression normale
for 48 h before testing. The curing confining stresses non cimentée, nous avons laissé ces spécimens durcir
ranged from 50 to 2000 kPa, and were intended to pendant 48 h avant de pratiquer les essais. Les contraintes
represent soil elements at different depths in the fictitious confinantes dues au durcissement allaient de 50 à 2000
sedimentary deposit when the cementing occurred. The kPa et devaient représenter des éléments du sol à différ-
contribution of the cement bonds to soil compression and entes profondeurs dans le dépôt sédimentaire fictif pen-
the changes in the isotropic yield stress as a function of dant la cimentation. Nous avons analysé le rôle des
void ratio and cement content were analysed. The results agglomérats cimentés dans la compression du sol et dans
showed the importance of the void ratio during the les changements d’efforts de tension isotrope comme fonc-
formation of cement bonds and also of the degree of tion du taux de pores et de la teneur en ciment. Les
cementation for the compressive behaviour of the ce- résultats ont montré l’importance du taux de pores pen-
mented soil, and demonstrated that the variation in yield dant la formation des agglomérats cimentés et également
stress with void ratio and cement content is dependent on du degré de cimentation quant au comportement compres-
the curing stress and independent of the stress history. sif du sol cimenté ; ils ont démontré aussi que la variation
des efforts de tension en fonction du taux de pores et de la
KEYWORDS: compressibility; laboratory tests; stiffness; struc- teneur en ciment dépend de la contrainte de durcissement
ture of soils et ne dépend pas de l’historique de contrainte.

INTRODUCTION AND BACKGROUND Santamarina (2000) have seen a loss of stiffness as a result
For clayey soils, structure has been defined as the combina- of sampling that they believed resulted from the disturbance
tion of bonding and fabric (Burland, 1990; Leroueil & of interparticle contacts and breakage of cement bonds.
Vaughan, 1990). The effect of structure is to give the soil a Similarly, Coop & Willson (2003) attributed lower than
strength and stiffness that are greater and to allow the expected stiffnesses, observed in triaxial tests for oil reser-
natural soil to exist at a higher volumetric state than that of voir sandstones, to the breakage of the cement bonds as a
the same material in a reconstituted state. In sands, structure result of sampling as they were unloaded from in-situ states
has often been equated solely with interparticle bonds, at several kilometres depth. Fernandez & Santamarina
although Dusseault & Morgenstern (1979) and Cuccovillo & (2001) confirmed experimentally that a sand that had been
Coop (1999) have identified the influence that the fabric of cemented under pressure could have its interparticle bonding
geologically aged sands may have on the peak strength. The damaged by unloading.
effect of structure has been observed on a wide range of The other principal problem associated with the testing of
natural soils and weak rocks of both sedimentary and natural sands is that, depending on their geological origin,
residual origins, and also for artificially cemented soils, as there can be a high spatial variability, both in the degree of
might be created, for example, by soil improvement. cementation and in the nature of the particles. An alternative
Much of the literature investigating the effect of structure is therefore to use artificially cemented specimens made up
is based on laboratory testing of natural soil specimens through the addition to the soil of a cementitious agent, such
retrieved from the field (e.g. Burland, 1989, 1990; Leroueil as Portland cement, gypsum, or lime (e.g. Dupas & Pecker,
& Vaughan, 1990; Airey & Fahey, 1991; Smith et al., 1992; 1979; Clough et al., 1981; Coop & Atkinson, 1993;
Clayton et al., 1992; Airey, 1993; Petley et al., 1993; Cuccovillo & Coop, 1993; Huang & Airey, 1993, 1998; Zhu
Cuccovillo & Coop, 1997, 1999; Kavvadas et al., 1993; et al., 1995; Prietto, 1996; Consoli et al., 2000, 2001;
Lagioia & Nova, 1995; Consoli et al., 1998). This approach, Schnaid et al., 2001). This allows the simulation of natural
however, presents some difficulties resulting from the dis- cemented soils in the laboratory and the qualitative under-
turbance to the structure that can occur during the sampling standing of the behaviour of structured soils without exces-
process (Clayton et al., 1992). In natural sands, Stokoe & sive sample variability or any bias due to sample
disturbance.
Cementation may arise in natural sands through a variety
Manuscript received 27 March 2002; revised 13 January 2003. of processes. In some cases, the cement has been deposited
Discussion 1 December 2003. soon after deposition when the sand was at a shallow depth.
 Department of Civil Engineering, Federal University of Rio
This is typical, for example, of carbonate sands with calcium
Grande do Sul, Brazil.
† School of Engineering and Architecture, Catholic University of
carbonate deposited from supersaturated pore fluid to form
Pelotas, Brazil. calcarenites (Clough et al., 1981; Airey & Fahey, 1991;
{ Imperial College of Science, Technology and Medicine, Uni- Cuccovillo & Coop, 1993), so that Coop & Atkinson (1993)
versity of London, UK. formed their artificially bonded sands at zero confining
} University of Manitoba, Canada. stress. For these soils, burial is therefore subsequent to

493
494 ROTTA, CONSOLI, PRIETTO, COOP AND GRAHAM
cement bonds formation. In other cases, such as the quartzi- mm), and 7·5% clay (D , 0·002 mm). The Atterberg limits of
tic sandstone (Greensand) tested by Cuccovillo & Coop the portion passing a no. 40 sieve are liquid limit of 22 and
(1999), cementation again occurred at shallow depths, but in plastic limit of 15. The specific gravity of solids is 2·70, and
this case the cementitious agent was iron oxide deposited X-ray diffraction showed that the fine portion is predomi-
from water flowing through the soil, which occurred after it nantly kaolinite. The cement agent used was rapid-hardening
had undergone a strong overconsolidation through burial and Portland cement (initial setting time of 3·25 h), which allows
erosion. In contrast, many sands are bonded progressively homogeneous cementing in the specimens in a very short
with burial to great depths, and this is more typical of the period. The grain size curves for the soil and the Portland
silica and chlorite bonding of the quartzitic reservoir sand- cement used are shown in Fig. 1.
stones investigated by Coop & Willson (2003).
Fewer authors have investigated the effect of cementing a
sand while it is under a confining stress. An exception was Specimen preparation, curing and testing procedures
Zhu et al. (1995), although in this case the state of the sands The specimens (50 mm diameter 3 100 mm high) were
during the formation of the cement bonds was not on the prepared by initially mixing the relevant quantities of dry
normal compression line as in the tests carried out here. soil and powdered Portland cement; the water was then
Cementation was otherwise achieved by mixing the soil and added to the mixture and further mixing was performed until
cement prior to application of a curing confining stress. The it was homogeneous in appearance. Next, following the
most common cements used have been gypsum (e.g. Coop undercompaction method proposed by Ladd (1978), the
& Atkinson, 1993; Huang & Airey, 1998) and Portland soil–cement mixture was compacted in three layers into a
cement (e.g. Consoli et al., 2000; Schnaid et al., 2001). 50 mm diameter cylindrical split-mould, to a target void
Ismail (2000) has identified that gypsum provides a better ratio of 0·65, moisture content of 19%, and cement contents
model of a carbonate bonding than Portland cement, but that of 0 (uncemented), 1, 2 and 3% by weight of dry soil. The
the percentage of cement required to achieve a given bond artificially cemented specimens prepared as described are
strength is much higher than for natural cementing pro- characterised as weakly to moderately cemented soils de-
cesses. To simulate better the natural cementing process, pending on the classification criteria considered (e.g.
Ismail et al. (2000) therefore developed the CIPS process, Beckwith & Hansen, 1982; Rad & Clough, 1985;
which allows a calcite cement to be deposited between the Hardingham, 1994), and reflect the actual conditions of
soil particles while the soil is under a confining pressure. many naturally cemented soil deposits and cement-stabilised
The formation of cement bonds under a confining stress soil layers.
means that the soil will have different void ratios at different For any experimental work that requires artificially ce-
depths as a result of the current in-situ stresses and stress mented specimens to be prepared, reproducibility is crucial
history, and it is this effect that the research reported here to the consistency of the test results. In the present work,
was aimed at investigating. Eighteen isotropic compression the reproducibility of the specimens was evaluated in terms
tests were carried out to simulate, in the laboratory, the of both density and degree of cementation by examining the
formation of the cemented matrix in soil elements situated experimental variability of their physical characteristics and
at different depths in a fictitious natural soil deposit. The unconfined compressive strengths. For the whole set of 18
soil specimens were cemented under a variety of confining specimens prepared (see Table 1), the average values of the
stresses, in this case using Portland cement and isotropic moulding moisture content, the void ratio, and the degree of
stresses, as the aim was to establish a conceptual framework saturation were 0·194, 0·66 and 0·797, with coefficients of
for such soils rather than model a specific cementing pro- variation of 1·5%, 0·8% and 1·3% respectively. The uncon-
cess. The creation of cement bonds in this way models a soil fined compressive strengths, averaged from three indepen-
in which cementation occurred as a single event with a dent values, were 120, 180 and 271 kPa respectively for the
constant confining stress during the cementing process. This cement contents of 1%, 2% and 3%, with coefficients of
is distinct from the gradual evolution of cement bonds with variation that decreased with increasing strength from 8·3%
burial discussed above, which would be difficult to repro- for the lowest cement content to 4·4% for the highest.
duce artificially, but represents a first step in the conceptual The unconfined compression tests were performed follow-
modelling of such processes. The process used might model, ing the general procedure recommended by the ASTM
for example, the deposition of cement from a pore fluid standard (ASTM, 1991a). Before being tested at a strain rate
flowing through a sand stratum, or the cement stabilisation of 1·14%/min, the specimens were cured under zero confin-
of a soil layer as a method of soil improvement. The use of
a range of confining stresses during cementing and the 100
variety of cement contents investigates the interaction be-
tween the effects of the in-situ stress, void ratio, and
quantity of cement deposited on the subsequent behaviour. 80
The compressive response of the cemented soil was then
Percent finer by weight

evaluated, principally in terms of the variation of yield stress


as a function of the void ratio existing during cementing, the 60
stress history, and the cement content.

40
EXPERIMENTAL PROGRAMME
Materials
20
The soil samples used in the present work, derived from Soil
weathered sandstone, were obtained from the region of Porto
Portland cement
Alegre, in southern Brazil. The soil is classified as non-plastic
0
silty sand (SM) according to the Unified Soil Clas- 0.0001 0.001 0.01 0.1 1 10
sification System. The grain size distribution is 27·8% Grain size: mm
medium sand (0·2 mm , D , 0·6 mm), 33·4% fine sand
(0·06 mm , D , 0·2 mm), 31·3% silt (0·002 mm , D , 0·06 Fig. 1. Grain size distribution
ISOTROPIC YIELDING IN AN ARTIFICIALLY CEMENTED SOIL 495
Table 1. Summary of the isotropic compression tests
Notation Cement content: Curing/initial isotropic Curing/initial Incremental yield Initial bulk
% stress: kPa void ratio stress: kPa modulus: MPa
ISO(0)0-3000 0 – 0·65 – 6·0
ISO(1)100-6000 1 98 0·62 80 25·9
ISO(1)250-6000 246 0·57 159 45·7
ISO(1)500-6000 499 0·53 216 57·0
ISO(1)1000-6000 998 0·47 385 79·8
ISO(2)100-6000 2 104 0·62 190 36·4
ISO(2)250-6000 254 0·57 267 68·0
ISO(2)500-6000 494 0·54 380 147·0
ISO(3)100-6000 3 99 0·62 260 48·4
ISO(3)250-6000 245 0·57 315 167·0
ISO(3)500-6000 492 0·54 498 267·2
ISO(3)600-6000 596 0·51 587 393·9
ISO(3)700-6000 733 0·49 572 395·0
ISO(3)1000-6000 974 0·48 679 440·2
ISO(3)2000-6000 1960 0·43 989 563·9
ISO(3)500-6000 484 0·54 476 237·0
ISO(3)50-6000 48 0·54 498 –
ISO(3)50-6000 48 0·49 662 –
ISO(3)500–6000 was consolidated up to 500 kPa, cured, and unloaded to 50 kPa before isotropic loading.
ISO(3)50–6000 was consolidated up to 500 kPa, unloaded to 50 kPa, and then cured before isotropic loading.
ISO(3)50–6000 was consolidated up to 1000 kPa, unloaded to 50 kPa, and then cured before isotropic loading.

ing stress for a 48 h period and soaked in water during the triaxial cell using a set of linear variable differential trans-
last hour to saturate them. This procedure resulted in an formers (LVDTs) with a 25 mm linear range and 0·001%
average of 19·4% moisture content and 79·7% degree of resolution, using a 12-bit data logger and 10 V power supply.
saturation. To assess the effect of non-saturation on the These internal devices were mounted on the top cap and
unconfined compression response, suction measurements directly on the specimen following the arrangement de-
were made for a number of specimens prior to soaking. scribed by Lingnau et al. (1995) and Blatz & Graham
Values of only 10–15 kPa and 22 kPa were obtained, respec- (2000). Corrections were made to the axial strain measure-
tively, by using the paper filter technique (ASTM, 1991b) ments to take into account both porous stone and paper filter
and the miniature suction probe, as described by Ridley & compliance errors. The curing stresses were applied only
Burland (1993). Considering that these values were expected after the LVDTs had been mounted and set at their electrical
to have reduced significantly due to soaking, any suction zero, so all the volumetric strains that occurred during
effects were considered to have only a marginal influence on curing were monitored and the void ratio at the end of the
soil strength. curing stage could be calculated.
For the isotropic compression tests, the specimens were
moulded with the same moisture content and degree of
saturation as for the unconfined tests, and with a void ratio TEST RESULTS AND ANALYSIS
of 0·66. They were placed in the triaxial cell and submitted, Figures 2–4 present the results obtained from the isotropic
prior to the onset of the cementitious bonds, to confining compression tests carried out on specimens cured under
stresses ranging from 50 to 2000 kPa, before being cured for different confining stresses and for cement contents of 1%,
48 h at a constant confining stress. The time it took to 2% and 3% respectively. A summary of the test results is
prepare (mix and compact) and set up the specimen in the given in Table 1. The identification of the tests follows the
triaxial cell was always less than 1 h, which is much shorter general nomenclature ISO(x)y-z, where x is the cement
than the initial setting time of the Portland cement used. To content, y is the curing confining stress, and z is the maxi-
increase saturation and reduce suction effects, de-aired water mum isotropic stress reached in the test.
was percolated through the specimens during the curing The curve ISO(0)0-3000, presented in all graphs, was
period, under a maximum hydraulic head equivalent to a obtained from a specimen moulded without the addition of
15 kPa pressure difference between the ends of the speci- cement and represents the intrinsic soil response in its
mens. Although full saturation was not guaranteed, suction destructured (uncemented) state. All the other curves were
effects, which had already been evaluated as negligible for obtained from specimens moulded with cement addition and
the unconfined compression tests, were expected to reduce cured at different confining stresses, prior to cement bonds
even more as the saturation increased during the initial formation. The first point plotted on each curve, for the
consolidation of the specimens. cemented specimens, represents the void ratio and the iso-
Isotropic compression was then carried out by applying tropic stress at the end of the curing procedure and the
consecutive, but not necessarily equal, increments of stress. beginning of isotropic compression. The subsequent points
A new increment was applied only after the remaining represent the end of each stress increment. It can be readily
excess pore pressure was smaller than 5% of the stress noticed that the first point of each curve lies approximately
increment. At this point, further increases in volumetric on the isotropic compression curve obtained for the unce-
strain were not significant. The pore pressure was monitored mented specimen. This shows that, when submitted to the
by an electrical transducer connected to the bottom of the curing confining stress, the specimens with cement addition
specimen while drainage was permitted only through an behaved similarly to the uncemented reference specimen,
outlet line connected to the top cap. Since the specimens regardless of the cement content. This supports the supposi-
were not fully saturated, the volumetric strains were calcu- tion that the cementitious bonds were formed only after the
lated from axial and radial strains measured inside the initial consolidation of the specimens, and indicates that any
496 ROTTA, CONSOLI, PRIETTO, COOP AND GRAHAM

0.65 0.65
Primary yield point
Primary yield point
ISO(0)0-3000
0.60 ISO(0)0-3000 0.60
ISO(2)100-6000
ISO(1)100-6000
ISO(2)250-6000
ISO(1)250-6000
0.55 0.55 ISO(2)500-6000

Void ratio
Void ratio

ISO(1)500-6000
ISO(1)1000-6000
0.50 0.50

0.45 0.45

0.40 0.40

0 10 100 1000 10000 0 10 100 1000 10 000


p′: kPa p′: kPa
(a) (a)

0.65 0.65

Primary yield point Primary yield point


ISO(0)0-3000 0.60 ISO(0)0-3000
0.60
ISO(1)100-6000 ISO(2)100-6000
ISO(1)250-6000 ISO(2)250-6000
0.55 0.55
Void ratio
ISO(2)500-6000
Void ratio

ISO(1)500-6000
ISO(1)1000-6000
0.50 0.50

0.45 0.45

0.40 0.40

0 1000 2000 3000 4000 5000 6000 0 1000 2000 3000 4000 5000 6000
p′: kPa p′: kPa
(b) (b)

Fig. 2. Isotropic compression response for the 1% cement Fig. 3. Isotropic compression response for the 2% cement
content specimens: (a) log scale; (b) linear scale content specimens: (a) log scale; (b) linear scale

influence of the grading alteration due to cement addition, that, for the cement content range investigated, the bulk
up to the 3% maximum used here, and creep effects, if they moduli tend to a unique low value as the curing void ratio
occurred at all, had negligible effects on the compression increases. This clearly demonstrates the coupled effect of
data at this stage of the test. This also illustrates one density and cementation on the mechanical behaviour of
advantage of using Portland cement over gypsum cement as bonded soils (e.g. Huang & Airey, 1998).
the cementitious agent in the present work. Coop & Atkin-
son (1993) found that the quantities of gypsum that had to
be added to achieve realistic strengths were so great that the
change of fines content had a significant impact on the Determination of yield stresses in isotropic compression
intrinsic behaviour. In contrast to some examples reported in the literature
The isotropic compression curves obtained for the ce- (e.g. Leroueil & Vaughan, 1990; Cuccovillo & Coop, 1997),
mented specimens are similar in shape to those reported there is no clearly defined yield in the curves presented in
in the literature for structured and naturally/artificially Figs 2–4. Difficulties in determining the yield point have
cemented soils (e.g. Leroueil & Vaughan, 1990). The speci- also been reported by other authors, for example Barksdale
mens are initially much stiffer than the soil in its des- & Blight (1997) for residual soils, Kavvadas et al. (1993)
tructured state, then becoming gradually softer as the for the Corinth marl, and Cecconi et al. (1998) for soft
isotropic stress increases. rocks. This has often been explained as a result of a gradual
From the isotropic compression curves presented in Figs onset of the breakage of the cement bonds. Also, the grading
2–4, it can be observed that, for the specimens prepared of the soil (e.g. Martins et al., 2001), a silty sand in the
with the same cement content, the stress increment neces- present work, and the micro-features of the cementitious
sary to cause a given reduction in void ratio increases as the agent (e.g. Ismail et al., 2002) can significantly add difficul-
curing stress increases. This is corroborated by the initial ties in identifying the yield point for cemented soils.
bulk moduli reported in Table 1 and plotted in Fig. 5, which Here the primary yield—that is, the point at which break-
also shows an interactive effect of the curing void ratio, at age of the cement bonds commences—was considered to be
the moment of formation of the cement bonds, and the when, in the linear scale, the stress–strain curve deviates
cement content on the pre-yield volumetric stiffness of the from the initial linear behaviour (Fig. 6). This definition is
cemented soil. From Fig. 5 it can be seen that the increase consistent with that used in other work (e.g. Cuccovillo &
in the initial bulk modulus caused by an increase in cement Coop, 1997). The arrows plotted in Figs 2–4 indicate the
content is much more pronounced at lower void ratios, and primary yield stresses of the cemented materials.
ISOTROPIC YIELDING IN AN ARTIFICIALLY CEMENTED SOIL 497
0.50
0.65
Primary yield point
Primary yield point
ISO(3)1000-6000
0.60 ISO(0)0-3000 0.48
ISO(3)100-6000
ISO(3)250-6000 a
0.55
Void ratio

ISO(3)500-6000 0.46

Void ratio
ISO(3)600-6000
0.50 ISO(3)700-6000
ISO(3)1000-6000 0.44
ISO(3)2000-6000
0.45

0.42
0.40

0 10 100 1000 10000 0.40


p′: kPa 0 1000 2000 3000 4000 5000 6000
(a) p′: kPa

Fig. 6. Determination of the primary yield stress in isotropic


0.65 Primary yield point compression
ISO(0)0-3000
0.60 ISO(3)100-6000
ISO(3)250-6000
Yield locus
ISO(3)500-6000
0.55
Void ratio

ISO(3)600-6000
ISO(3)700-6000
ISO(3)1000-6000
0.50
ISO(3)2000-6000
Void ratio

0.45
Post-yield
compression line
0.40 (PYCL)

0 1000 2000 3000 4000 5000 6000 7000


p′: kPa Intrinsic
(b) compression line
(ICL)
Fig. 4. Isotropic compression response for the 3% cement
ln p′
content specimens: (a) log scale; (b) linear scale
Fig. 7. Patterns of behaviour observed in the isotropic compres-
sion tests
0.70

feature had also been observed by Cuccovillo & Coop


1% cement content
(1999) for natural calcarenites of variable initial void ratio.
2% cement content The zone between the post-yield compression line and the
0.60 3% cement content ICL defines the structure permitted space (Leroueil &
Curing void ratio

Vaughan, 1990).
Figure 8 summarises the locations of the post-yield com-
pression line (PYCL) for each cement content, showing how

0.50
0.65 1% cement content
Yield loci 2% cement content
0.60 3% cement content

0.40
0 200 400 600 0.55 1% 2% 3%
Void ratio

Post-yield
Bulk modulus: MPa compression lines
(PYCL)
Fig. 5. Variation of initial bulk modulus with curing void ratio 0.50

General patterns of behaviour 0.45


Figure 7 shows a schematic diagram illustrating the Intrinsic
patterns of behaviour observed in Figs 2–4, and defines the compression line
0.40
terms that will be used to describe the behaviour. After (ICL)
primary yield, the paths of the specimens of a variety of 100 1000 10 000
different initial void ratios all follow a post-yield compres- p′: kPa
sion line that is unique for each degree of cementation and
which converges with the intrinsic compression line (ICL) of Fig. 8. Summary of primary yield points, yield loci, and post-
the uncemented soil as the isotropic stress increases. This yield compression lines
498 ROTTA, CONSOLI, PRIETTO, COOP AND GRAHAM
it both expands and steepens as the cement content in- Fitting equation: ecure ⫽ A ln(∆p′y) ⫹ B
0.65
creases. By extrapolating the PYCLs and the ICL shown in Cement content A B R2
Fig. 8, it can be seen that convergence would occur even- 1% ⫺0.1026 1.0832 0.995
tually, but not before an isotropic stress as high as 30 MPa 0.60 2% ⫺0.1297 1.2988 0.994
had been reached, regardless of the cement content. This 3% ⫺0.1376 1.3771 0.959
slow convergence may seem surprising, given the relatively

Curing void ratio


modest cement contents. As discussed above, such a small 0.55 R S T
amount of cement fines would have a negligible effect on
the location of the intrinsic compression line, but the gradual
0.50
convergence indicates a much slower destructuration than R′ S′ T′
has been seen, for example, for some naturally cemented
sands (e.g. Lagioia & Nova, 1995; Cuccovillo & Coop, 0.45
1999). However, Coop & Atkinson (1993) also found a slow
convergence for an artificially cemented sand. The slow
convergence may also be a function of the base material 0.40
0 200 400 600 800 1000
used, which is gap graded (e.g. Thevanayagam & Mohan,
Incremental yield stress: kPa
2000). Martins et al. (2001) have shown how, for these soils,
initial differences in void ratio are not readily removed by Fig. 9. Variation of incremental yield stress with curing void
compression, even to high stresses. For the uncemented ratio
samples they tested, the initial differences in void ratio were
created by varying degrees of compaction. Here the initial void ratio of 0·54 on the normal compression line (˜ p9y
difference is essentially created by the delay in compression ¼ 0). By increasing the cement content at the same curing
brought about by the cement bonding. void ratio, the increment in the yield stress in isotropic
Also shown in Fig. 8 are the primary yield loci, which, compression is shifted to the right up to the corresponding
for this soil, lie between the ICL and PYCL and again cement content (point S for 1% and T for 3% cement
converge with the ICL as the isotropic stress increases. For content). In Fig. 10, the variation in the primary yield stress
the natural calcarenite tested by Cuccovillo & Coop (1999) ( p9y ¼ p9cure þ ˜ p9y ) is plotted against cement content, for
the yield locus practically coincided with the PYCL, perhaps four different curing void ratios. It can again be observed
because it had been cemented at a shallow depth prior to that, for specimens cured at the same density, the primary
burial and not under a confining stress as was done here. yield stresses increase with the increase in cement content.
Since the specimens were cemented at states on the ICL, the
(b) For specimens with the same cement content, the
yield locus necessarily lies to the right of the ICL, but for
primary yield stress increases with reducing void ratio
some soils this is not the case. For example, the Greensand
during curing.
tested by Cuccovillo & Coop had its yield point in compres-
sion below the ICL as it had been cemented after a strong For example, the specimen cured at the void ratio of 0·54
geological overconsolidation. and cement content of 3% (point T) will have an incremental
In Fig.8 the yield loci and the PYCL are represented yield stress in isotropic compression lower than the value
simply as straight lines. However, they cannot be straight obtained for the specimen cured with the same cement
over an extended pressure range, as clearly the lines cannot content but at the void ratio of 0·48 (point T9). Again, this
cross the ICL, and it might be expected that—no matter can also be observed in Fig. 10, where the primary yield
what the stress is when the cement cures—there should stresses increase with the curing void ratio for the specimens
always be some positive effect on the yield stress. The shape cured at the same cement content. This may arise from an
of the yield locus is examined in greater detail in the increase in density resulting in an increase in the number of
following section. contact points between the soil particles where the cement
can form a bond.
(c) The relative contribution of cementation to the primary
Factors determining the size of the primary yield surface
yield stress in isotropic compression decreases with
The increases in primary yield stresses in isotropic com-
reducing void ratio during curing.
pression for the artificially cemented soil are presented in
Fig. 9 as a function of the curing void ratio and cement
content. For this soil and cement, and for the method of 10000
specimen preparation used, the points plotted can be satis-
factorily fitted by logarithmic functions such as: Curing void ratio

ecure ¼ Aln ˜ p9y þ B (1) 0.47
Primary yield stress: kPa

1000 0.54
where ecure is the curing void ratio (obtained at the end of
curing stage), ˜ p9y is the incremental yield stress, defined as 0.57
the difference between the primary yield stress in isotropic 0.62
compression, p9y , and the isotropic curing stress, p9cure , and A
and B are fitting constants.
100
From the analysis of Fig. 9, three basic conclusions are
possible:
(a) For specimens cured at the same void ratio, the primary
yield stress in isotropic compression increases with
increasing cement content. 10
0 1 2 3 4
This behaviour can be observed from Fig. 9 through the Cement content: %
straight line RST. The point R represents the value of the
incremental yield stress for an uncemented specimen with Fig. 10. Variation of primary yield stress with cement content
ISOTROPIC YIELDING IN AN ARTIFICIALLY CEMENTED SOIL 499
Observing two specimens prepared with the same cement cement was tested in isotropic compression. The specimen
content but cured at different void ratios (points S and S9 in was consolidated up to 500 kPa, cured for 48 h, and then
Fig. 9), the stresses resisted by interparticle friction can be unloaded to 50 kPa (ISO(3)500-6000 ). In this way, it was
obtained directly from the uncemented isotropic compression possible to obtain a specimen with a void ratio comparable
curve given in Figs 2–4, and the stresses resisted by to that obtained for the normally consolidated specimen
interparticle bonding by the straight lines RS and R9S9 in ISO(3)500-6000, but with a different OCR. The results are
Fig. 9, for 1% cement content. It can be observed that the reported in Table 1 and presented in Fig. 11, along with the
ratio between the stresses resisted by bonding and by friction isotropic compression curve for the reference-uncemented
decreases with reducing curing void ratio, demonstrating that soil.
the contribution of cementation is less effective in denser It can be readily noticed that, for one unloading/reloading
specimens. A better visualisation is provided in Fig. 10. cycle, the behaviour in isotropic compression is independent
Taking the example of the specimen at a void ratio of 0·57, of the unloading applied to the specimen after curing, even
the increase in cement content up to 3% caused an increase with an overconsolidation ratio as high as 10. This clearly
of about 106% in the primary yield stress (from 276 kPa to demonstrates that both the curing void ratio and the degree
560 kPa). However, for the specimens cured at a void ratio of cementation control the stress–strain behaviour of the
of 0·47, the increase in cement content increased the primary cemented soil in isotropic compression, and not the stress
yield stress from 1139 kPa to 1653 kPa, representing an history after bonds formation. In this case, unloading clearly
increase of only 46%. Cuccovillo & Coop (1999) made had no effect on the cement bonds. This is in contrast to the
similar observations for the natural calcarenite they tested, findings of Fernandez & Santamarina (2001), who found that
for which the effect of cementation is proportionately much sands cemented under a confining stress yielded more easily
greater at higher void ratios than at lower ones, so that the in swelling than in compression. Here the soil was unloaded
post-yield compression line and yield locus converged with by 450 kPa, which is similar to the incremental yield stress
the intrinsic normal compression line, as indicated in Fig.7. in compression of 476 kPa, and yet there was no apparent
Figure 10 also indicates that an increase in density, for adverse effect.
example by an increase in compaction energy, can be more
effective in increasing the primary yield stress, and hence
soil stiffness, than an increase in cement content. This might Influence of stress history prior to curing
be of paramount importance to engineering practice in To evaluate the influence of stress history prior to curing,
relation to the use of artificially cemented soils in earth- two additional specimens using 3% cement content (see
works. Table 1) were tested in isotropic compression. The first
The changes in the incremental yield stress in isotropic specimen (ISO(3)50-6000 ) was consolidated up to
compression as a function of the curing void ratio and 500 kPa, unloaded to 50 kPa, and then cured for 48 h before
cement content can be conveniently expressed by a simple being isotropically compressed up to 6000 kPa. The second
mathematical expression derived from the equations pre- specimen (ISO(3)50-6000 ) was consolidated up to
sented in Fig. 9. From a simple linear regression analysis of 1000 kPa, unloaded to 50 kPa, and then cured. In this way, it
the values presented in the table embedded in Fig. 9, the was possible to obtain two specimens with void ratios
coefficients A and B can be expressed as linear functions of comparable to those obtained for the normally consolidated
cement content (CC): specimens ISO(3)500-6000 and ISO(3)1000-6000 respec-
tively, but cured under different confining stresses and with
A ¼ 0:0175 3 CC  0:0883, R2 ¼ 0:909 (2)
different overconsolidation ratios.
The results are presented in Figs 12 and 13, along with
B ¼ 0:1470 3 CC þ 0:9591, R2 ¼ 0:932 (3)
the isotropic compression curve for the reference-uncemen-
Introducing the expressions for A and B in equation (1), the ted soil. Different slopes in the unloading and reloading
value of the incremental yield stress (˜ p9y in kPa) in lines of tests ISO(3)50-6000 and ISO(3)50-6000 can
isotropic compression is then given by be readily seen, reflecting the different states of the speci-
  men at each stage of the test: uncemented during unloading
0:147 3 CC þ 0:9591  ecure
˜ p9y ¼ exp (4) prior to curing, and cemented during reloading. Using the
0:0175 3 CC þ 0:0883 procedure previously described, similar incremental yield
where ecure is the curing void ratio and CC is the percentage
cement content.
The primary yield stress in isotropic compression
( p9y ¼ p9cure þ ˜ p9y ) obtained from Fig. 9 or using equation 0.65
(4) provides the intersection between the primary yield sur- ISO(0)0-3000
face and the isotropic axis (p9-axis) in p9:q9 stress space. So, 0.60 ISO(3)500-6000
decreasing the curing void ratio and/or increasing the cement ISO(3)500-6000*
content causes an increase in the primary yield isotropic Curing point
0.55
Void ratio

stress and, therefore, the expansion of the primary yield Primary yield
surface. So, in a normally consolidated deposit that has been point
cemented while under a geostatic stress through the precipi- 0.50
tation of a cementitious agent, it might be expected that there
will be an expansion of the primary yield surface with depth,
0.45
the magnitude of which depends on the variation of both the
curing density and the amount of cementitious agent.
0.40

1 10 100 1000 10 000


Influence of stress history after curing p′: kPa
To assess the effect of stress history after curing on the
primary yield stress, an additional specimen containing 3% Fig. 11. Effect of stress history after curing (OCR 10)
500 ROTTA, CONSOLI, PRIETTO, COOP AND GRAHAM
the increase of isotropic stress supported before primary
0.65
yielding will be the same for both specimens, as the incre-
ISO(0)0-3000 ment of loading beyond the curing stress will be carried
0.60 ISO(3)500-6000 entirely by the soil bonds, which are much stiffer than the
ISO(3)500-6000** soil skeleton.
Curing point
0.55
Void ratio

Primary yield
point SUMMARY AND CONCLUDING REMARKS
0.50 Eighteen isotropic compression tests were carried out on
artificially cemented specimens cured under different confin-
0.45
ing stresses and prepared with different cement contents.
The specimens were first consolidated to various confining
stresses along the normal compression line of the uncemen-
0.40 ted soil, simulating the formation of soil elements at differ-
ent depths in a fictitious sedimentary deposit. After
1 10 100 1000 10000 consolidation, the specimens were then allowed to cure and
p′: kPa cement. Isotropic compression was then applied incremen-
tally after curing. The test results allowed the evaluation of
Fig. 12. Effect of stress history prior to curing (OCR 10) the contribution of cementitious bonds to the volumetric
behaviour of the cemented soil as well as the variation in
primary and incremental yield stresses in isotropic compres-
sion as a function of curing void ratio. The conclusions can
be summarised as follows:
0.65
(a) For the artificially cemented specimens studied, the
ISO(0)0-3000 primary yield stress in isotropic compression is a
0.60 ISO(3)500-6000 function of the curing void ratio and cement content; it
ISO(3)500-6000*** is also dependent on the curing stresses, but is
0.55 Curing point independent of any increase in the OCR after curing.
Void ratio

Primary yield (b) The relative contribution of cementation to the soil


point behaviour in isotropic compression reduces with de-
0.50
creasing curing void ratio.
(c) A decrease in the void ratio existing during formation
0.45 of the cement bonds and/or an increase in the degree of
cementation causes an increase in the primary yield
stress in isotropic compression and, consequently, the
0.40
expansion of the primary yield surface of the cemented
1 10 100 1000 10000
soil.
(d ) At similar void ratios, the curing confining stress
p′: kPa
influences the primary yield stress, but does not affect
Fig. 13. Effect of stress history prior to curing (OCR 20)
the incremental yield stress.
(e) According to the simulation carried out in this study, it
is expected that there will be an expansion in the
primary yield surface with increasing depth in a
stresses were obtained for the specimens cured at similar
sedimentary deposit that is cemented uniformly at all
void ratios, but at different overconsolidation ratios (see
depths in a single event.
Table 1). At a given void ratio, the curing confining stress
( p9cure ) therefore influences the primary yield stress ( p9y ), but
not the incremental yield stress (˜ p9y ), as defined in equation
(1), which is kept practically unchanged. These tests confirm ACKNOWLEDGEMENTS
that not only the amount of cement controls the size and The authors wish to express their gratitude to CNPq
location of the yield envelope, but also the stress acting (National Council of Scientific and Technological Research)
when the cement bond is created. (Projects 520610/95–4 and 479804/01–0), the British Coun-
However, even if the yield stress is affected by the curing cil, and CAPES (Coordination of Training of High Educa-
confining pressure, the location of the post-yield compres- tion Graduate) (Project CAPES-British Council 088/99) for
sion line seems to remain unchanged. It may be that while the financial support to the research group.
primary yield represents the onset of bond degradation, the
strains in reloading are not sufficient to completely destroy
the bonds, the influence of which is only gradually reduced NOTATION
as the post-yield compression line converges with the intrin- A, B constant coefficients
sic compression line at large strains. D grain diameter (size)
The independence of the incremental yield stresses, at a e void ratio
given void ratio, from the curing confining stresses can be ecure curing void ratio
R2 coefficient of determination
explained by analysing the contribution of bonding to soil
p9 mean isotropic stress
behaviour during isotropic compression. When two identical p9cure mean isotropic stress at curing
soil specimens (that is, with similar cement contents and p9y primary yield stress in isotropic compression
void ratios) are cured under different confining stresses, the ˜ p9y incremental yield stress in isotropic compression
bonds are actually formed under no stress, as the curing CC cement content
stress, which is applied before bonding formation, is entirely ICL intrinsic compression line
supported by the uncemented soil skeleton. Consequently, PYCL post-yield compression line
ISOTROPIC YIELDING IN AN ARTIFICIALLY CEMENTED SOIL 501
REFERENCES Hardingham, A. D. (1994). Development of an engineering descrip-
Airey, D. W. (1993). Triaxial testing of naturally cemented carbo- tion of cemented soils and calcrete duricrusts. Proc. 1st Int.
nate soil. J. Geotech. Engng, ASCE 119, No. 9, 1379–1398. Symp. on Engng Characteristics of Arid Soils, Rotterdam,
Airey, D. W. & Fahey, M. (1991). Cyclic response of calcareous 87–90.
soil from the North-West Shelf of Australia. Géotechnique 41, Huang, J. T. & Airey, D. W. (1993). Effects of cement and dens-
No. 1, 101–121. ity on an artificially cemented sand. Proc. 1st Int. Symp.
ASTM (1991a). Standard test method for unconfined compressive on Geotech. Engng of Hard Soils–Soft Rocks, 1, Athens,
strength of cohesive soil, ASTM D2166–91. Annual Book of 553–560.
ASTM Standards. Philadelphia: American Society for Testing Huang, J. T. & Airey, D. W. (1998). Properties of artificially
and Materials. cemented carbonate sand. J. Geotech. Geoenviron. Engng, ASCE
ASTM (1991b). Standard test method for measurement of soil 124, No. 6, 492–499.
potential (suction) using paper filter, ASTM D5298–91. Annual Ismail, M. A. (2000). Strength and deformation behaviour of
Book of ASTM Standards. Philadelphia: American Society for calcite-cemented calcareous soil. PhD thesis, University of
Testing and Materials. Western Australia.
Barksdale, R. D. & Blight, G. E. (1997). Compressibility and Ismail, M. A., Joer, H. A. & Randolph, M. F. (2000). Sample
settlement of residual soils. In Mechanics of residual soils, preparation technique for artificially cemented sands. Geotech.
(G. E. Blight ed.) pp. 95–154. Rotterdam: A. A. Balkema. Test. J., ASTM 23, No. 1, 141–157.
Beckwith, G. H. & Hansen, L. A. (1982). Calcareous soils of the Ismail, M. A., Joer, H. A., Randolph, M. F. & Sim, W. H. (2002).
south-western United States. In Geotechnical properties, behav- Effect of cement type on shear behaviour of cemented calcareous
iour and performance of calcareous soils, (K. R. Demars & soil. J. Geotech. Geoenviron. Engng, ASCE 128, No. 6, 520–529.
R. C. Chaney eds) vol. 1, pp. 16–35. Philadelphia: ASTM. Kavvadas, M. J., Anagnostopoulos, A. G. & Kalteziotis, N. (1993).
Blatz, J. & Graham, J. (2000). A system for controlled suction in A framework for the mechanical behaviour of cemented Corinth
triaxial test. Géotechnique 50, No. 4, 465–469. marl. Proc. 1st Int. Symp. on Geotech. Engng of Hard Soils–
Burland, J. B. (1989). Small is beautiful: the stiffness of soils at Soft Rocks, 1, Athens, 577–583.
small strains. Can. Geotech. J. 26, No. 4, 499–516. Ladd, R. S. (1978). Preparing test specimens using undercompac-
Burland, J. B. (1990). On the compressibility and shear strength of tion. Geotech. Test. J., ASTM 1, No. 1, 16–23.
natural clays. Géotechnique 40, No. 3, 329–378. Lagioia, R. & Nova, R. (1995). An experimental and theoretical
Cecconi, M., Viggiani, G. & Rampello, S. (1998). An experimental study of the behaviour of a calcarenite in triaxial compression.
investigation of the mechanical behaviour of a pyroclastic soft Géotechnique 45, No. 4, 633–648.
rock. Proc. 2nd Int. Symp. on Geotech. Engng of Hard Soils– Leroueil, S. & Vaughan, P. R. (1990). The general and congruent
Soft Rocks, Naples 1, 473–482. effects of structure in natural soils and weak rocks. Géotechni-
Clayton, C. R. I., Hight, D. W. & Hopper, R. J. (1992). Progressive que 40, No. 3, 467–488.
destructuring of Bothkennar clay: implications for sampling and Lingnau, B. E., Graham, J. & Tanaka, N. (1995). Isothermal
reconsolidation procedures. Géotechnique 42, No. 2, 219–239. modelling of sand-bentonite mixtures at elevated temperatures.
Clough, G. W., Sitar, N., Bachus, R. C. & Rad, N. S. (1981). Can. Geotech. J. 32, No. 1, 78–88.
Cemented sands under static loading. J. Geotech. Engng, ASCE Martins, F. B., Bressani, L. A., Coop, M. R. & Bica, A. V. D.
107, No. 6, 799–817. (2001). Some aspects of the compressibility behaviour of a
Consoli, N. C., Schnaid, F. & Milititsky, J. (1998). Interpretation of clayey sand. Can. Geotech. J. 38, No. 6, 1177–1186.
plate load tests on residual soil site. J. Geotech. Geoenviron. Petley, D., Jones, M., Fan, C. & Stafford, C. (1993). Deformation
Engng, ASCE 124, No. 9, 857–867. and fabric changes in weak fine-grained rocks during high
Consoli, N. C., Rotta, G. V. & Prietto, P. D. M. (2000). The pressure consolidation and shear. Proc 1st Int Symp on Geotech
influence of curing under stress on the triaxial response of Engng of Hard Soils–Soft Rocks, Athens, 1, 737–743.
cemented soils. Géotechnique 50, No. 1, 99–105. Prietto, P. D. M. (1996). Study of the mechanical behaviour of an
Consoli, N. C., Prietto, P. D. M., Carraro, J. A. H. & Heineck, artificially cemented soil. MSc thesis, Federal University of Rio
K. S. (2001). Behavior of compacted soil–fly ash–carbide Grande do Sul, Brazil (in Portuguese).
lime–fly ash mixtures. J. Geotech. Geoenviron. Engng, ASCE Rad, N. S. & Clough, G. W. (1985). Static behaviour of variably
127, 774–782. cemented beach sands. In: Strength testing of marine soils:
Coop, M. R. & Atkinson, J. H. (1993). The mechanics of cemented laboratory and in-situ measurements, (R. C. Chaney and K. R.
carbonate sands. Géotechnique 43, No. 1, 53–67. Demars eds.) pp. 306–317. Philadelphia: ASTM.
Coop, M. R. & Willson, S. M. (2003). On the behavior of Ridley, A. M. & Burland, J. B. (1993). A new instrument for
hydrocarbon reservoir sands and sandstones. J. Geotech. Geoen- the measurement of soil moisture suction. Géotechnique 43,
viron. Engng, ASCE (accepted for publication). No. 2, 321–324.
Cuccovillo, T. & Coop, M. R. (1993). The influence of bond Schnaid, F., Prietto, P. D. M. & Consoli, N. C. (2001). Characteriza-
strength on the mechanics of carbonate soft rocks. Proc. 1st Int. tion of cemented sand in triaxial compression. J. Geotech.
Symp. on Geotech. Engng of Hard Soils–Soft Rocks, Athens 1, Geoenviron. Engng, ASCE 127, No. 10, 857–868.
447–455. Smith, P. R., Jardine, R. J. & Hight, D. W. (1992). The yielding of
Cuccovillo, T. & Coop, M. R. (1997). Yielding and pre-failure de- Bothkennar clay. Géotechnique 42, No. 2, 257–274.
formation of structured sands. Géotechnique 47, No. 3, 481–508. Stokoe, K. H. & Santamarina, J. C. (2000). Seismic-wave-based
Cuccovillo, T. & Coop, M. R. (1999). On the mechanics of testing in geotechnical engineering. In GeoEng 2000, pp. 1490–
structured sands. Géotechnique 49, No. 6, 741–760. 1536. Pennsylvania, Technomic Publishing Co., Melbourne,
Dupas, J. M. & Pecker, A. (1979). Static and dynamic properties of Australia.
sand-cement. J. Geotech. Engng, ASCE 105, No. 3, 419–436. Thevanayagam, S. & Mohan, S. (2000). Intergranular state variables
Dusseault, M. B. & Morgenstern, N. R. (1979). Locked sands. and stress–strain behaviour of silty sands. Géotechnique 50,
J. Engng Geol. 12, No. 1, 117–131. No. 1, 1–23.
Fernandez, A. L. & Santamarina, J. C. (2001). Effect of cementa- Zhu, F., Clark, J. I. & Paulin, M. J. (1995). Factors affecting at-rest
tion on the small-strain parameters of sand. Can. Geotech. J. 38, lateral stress in artificially cemented sands. Can. Geotech. J. 32,
No. 1, 191–199. No. 1, 195–203.

Das könnte Ihnen auch gefallen