Proceedings of the 16th International Conference on Soil Mechanics and Geotechnical Engineering

© 2005–2006 Millpress Science Publishers/IOS Press.

Published with Open Access under the Creative Commons BY-NC Licence by IOS Press.
doi:10.3233/978-1-61499-656-9-3579

Indirect estimation of segregation potential based on soil index properties

Évaluation indirecte de potentiel de ségrégation basée sur des propriétés d'index de sol

J.-M. Konrad
Dept. of Civil Engineering, Université Laval ,Québec, Canada

ABSTRACT
A new approach to estimate segregation potential values using the frost heave response of two reference soils is presented. The refer-
ence characteristics consist of a relationship between segregation potential at zero overburden pressure, specific surface area and aver-
age grain size of the fines fraction for two artificial soil mixtures in which the clay mineral is poorly crystallized kaolinite. The predic-
tion of segregation potential values using the reference frost heave characteristics approach is more robust and reliable than other
empirical approaches which do not specifically distinguish between clay and nonclay fines.
RÉSUMÉ
On présente une nouvelle approche pour estimer les valeurs du potentiel de ségrégation en utilisant la réponse au soulèvement dû au
gel de deux sols de référence. Les caractéristiques de la référence consiste en une relation entre le potentiel de ségrégation à une pres-
sion sus-jacente nulle, la surface spécifique et la grosseur moyenne de la fraction des particules fines pour deux mélanges de sols arti-
ficiels dans lesquels le minéral argileux est un kaolin faiblement cristallisé. La prédiction des valeurs du potentiel de ségrégation en
utilisant l’approche des caractéristiques de soulèvement dû au gel de la référence est moins erratique et plus fiable que les autres ap-
proches empiriques qui ne distinguent pas spécifiquement entre l’argile et les particules fines non argileuses.

1 SEGREGATION POTENTIAL OF FINE-GRAINED The analysis of frost heave data on several fine-grained
SOILS FROM SOIL INDEX PROPERTIES soils confirmed that the segregation potential of saturated
soils with no applied surcharge, SPo, was best related to the
From a phenomenological point of view, frost heave mechan- average size of the fines fraction (<75�m), d50(FF), its spe-
ics can be regarded as a problem of impeded drainage in a cific surface area, Ss, and the ratio of the material's water con-
layered medium to an ice-water interface that exists in the tent to its liquid limit, w/wL. The relevancy of this empirical
frozen soil at the segregation-freezing front. Konrad and relationship is justified by the fact that frost-susceptibility is
Morgenstern (1980, 1981, 1982) developed a simplified controlled by the water movements in capillary channels
model for water migration in freezing soils and showed that at which can be related, however limited this may be, to an av-
the onset of formation of the final ice lens in step-freezing erage grain size of the fines fraction and to the amount of ad-
tests, i.e. near thermal steady state, the pore-water velocity sorbed water, which is associated with the specific surface
entering into the unfrozen soil, vu, was proportional to the area. Also, given a soil, it is clear that fabric, hence grain ar-
temperature gradient in the frozen fringe, Grad Tf, provided rangement, influences the capillary channel geometry and
that the suction at the pore-freezing front was constant. therefore the frost-susceptibility. Furthermore, it was also es-
tablished that soil density needs to be considered in a relative
Pw − Pu � P −P � manner in order to cover a wide range of soils. It was thus
vu = K f = �� w u K f ��GradTf = SPo .GradTf (1) proposed to use the ratio of water content to liquid limit,
d � Ts � w/wL, as an indicator of relative soil packing. In general,
when a soil is sedimented and consolidated under its own
where Pw is suction at the ice lens, weight in a natural environment, w/wL is about 0.7±0.1.
Pu suction at the frost front, Structured clays, such as Champlain Sea clays, may display
Kf overall hydraulic conductivity of the frozen w/wL values much larger than 0.7, as high as 1.3, while de-
fringe structured clays present w/wL values close to 0.7.
Ts segregation-freezing temperature For w/wL of 0.7, the best-fit empirical relationship was
d thickness of frozen fringe given as:

The constant of proportionality, i.e. the slope of the linear re- SPoSs = [116 – 75 log d50(FF)] 103 mm4/(C.s.g) (2)
lationship between vu and Grad Tf, is defined as the segrega-
tion potential, SPo for zero applied overburden pressure. where d50(FF) is expressed in µm.
Konrad (1999) demonstrated that a successful approach for The frost heave response of quarry fines from several loca-
assessing the segregation potential from soil index properties tions in the Province of Quebec was studied in the laboratory
must consider at least the following key factors: using one-dimensional step-freezing tests with free access to
i) grain size distribution and fines content; water. Figure 1 presents this relationship for both fine-
ii) clay mineralogy; grained soils given by Konrad (1999) and the quarry fines.
iii) soil fabric; With one exception (basalt fines), the quarry fines have rep-
iv) overburden pressure resentative points (open circles) that are located below the
proposed relationship, which was already a lower bound

3579
value for w/wL values close to 0.7. By comparing the spe-
cific surface area of each quarry fines sample to the soils
given in Konrad (1999) (Figure 2a), it appears that all fine-
grained soils used by Konrad (1999) to define the proposed
empirical relationship had specific surface areas of the fines
fraction (d < 75 µm) larger than 30 m2/g. The quarry fines,
however, had specific surface areas less than 8 m2/g except
for the basalt fines with an Ss value of 12.6 m2/g.
These data suggest strongly that frost-susceptibility must
account for the presence of clay minerals in the soils. This is
not surprising considering that the particles of most of the
clay minerals are platy whereas nonclay minerals are com-
posed of bulky shaped particles predominantly rock frag-
ments (Mitchell 1993).
The shape of these small particles and their crystal struc-
ture influence, in turn, the unfrozen water content when the
soil is frozen. As discussed by Konrad (1999), unfrozen wa-
ter in a frozen soil can be partitioned into capillary water in
the pores far from the soil particle surface and adsorbed water
that can be considered as strongly influenced by the mineral
surfaces. Water mobility in capillary channels is greater than Fig 2b. Segregation Potential of a variety of soils.
that in the adsorbed water films, as the water molecules are
strongly oriented and structured (Hoekstra 1969). As dis- can be related to the amount of adsorbed water present in the
cussed above, frost heave is related to water flow through the frozen soil. Increasing adsorbed water content will be associ-
frozen fringe, thus through the network of capillary water ated with increasing values of specific surface. It was thus
channels. As different clay minerals have different values of postulated by Konrad (1999) that for a given value of d50(FF),
specific surface, the specific surface area of the fines fraction the amount of capillary water was inversely related to the
specific surface and that a consistent relationship between
SPoSs and d50(FF) should exist in fine-grained soils, which
was indeed supported by available data.
The specific surface may also indicate, at least in a qualita-
tive manner, the relative importance of clay minerals in the
fines fraction. Data on sand-silt and clay mineral mixtures by
Rieke et al. (1983) and on crushed granite- kaolinite clay mix-
tures (Konrad and Lemieux 2004) show that the specific sur-
face area of fines with 10% kaolinite clay and 90 % granitic
fines is 10.5 m2/g while it is 10 m2/g in sand–silt mixed with
25% of poorly crystallized kaolinite clay. In granitic fines
with kaolinite clay fractions larger than 50%, the specific sur-
face areas are larger than 20 m2/g. It is thus postulated that
fines in which nonclay minerals play a predominant role dis-
play generally a specific surface area smaller than about 10
m2/g. For these soils, there is obviously less adsorbed water
and the proposed relationship between segregation potential
at zero overburden and index properties established for fine-
grained soils in which fines are predominantly clay minerals
Fig. 1 Comparison of frost heave data from the quarry fines with
cannot adequately be used in soils with nonclay fines.
available data from various soils.
An empirical relationship between segregration potential
and index properties for soils with clay and nonclay fines us-
ing two reference soils.
A close examination of Rieke et al. (1983) frost heave data
on sand/silt and kaolinite mixtures indicates that both specific
surface area and average particle size increase with increasing
clay mineral content. Although there is thus no clear cut be-
tween nonclay and clay fines, it is proposed to define a lower
boundary for which the frost heave response is still dominated
by the clay minerals in the fines. The data from Rieke et al.
(1983) can be used to define this boundary since the fines
composition is well known. Two sand-silt-kaolinite clay mix-
tures: 80Sa-10Si-10K and 80Sa-4Si-16K were choosen as
reference soils. For these mixtures, the total fines content is
20% and it is thus appropriate to consider that the mixture’s
frost heave behaviour is dominated by the fines. Further-
more, it is also clear that with kaolinite clay fractions of re-
spectively 50 and 80%, both mixtures will have a frost- heave
response that is closer to soils with clay fines. The reference
line is thus taken as a linear relationship between specific sur-
face area and log d50(FF) for d50(FF) values larger than 1 �m.
Fig. 2a Specific surface area of a variety of soils.
For d50(FF) values smaller than 1�m, the reference specific

3580
surface area is set to a constant value of 25.95 m2/g as illus- pirical relationships are proposed for characterizing the frost
trated in Figure 2a. heave response of fine-grained soils:
The frost heave response of the reference soils is presented
in Figure 2b by the solid line which gives the values of SPo For Ss/Ssref < 1: (5)
as a function of average particles size of the fines fraction.
As for the reference specific surface area, the reference line SPo/SPoref = Ss/Ssref if w/wL = 0.7±0.1
for SPo is also taken as a linear relationship between SPo and SPo/SPoref = 0.08 + 1.42 Ss/Ssref if w/wL > 0.8
log d50(FF) for d50(FF) values larger than 1 �m. For d50(FF)
values smaller than 1 �m, the reference SPo is set to a con- and
stant value of 489 mm2/C.d.
In summary, the proposed reference characteristics to be For Ss/Ssref > 1: (6)
used for the analysis of frost heave response of various fine-
grained soils can be summarized by the following equations: SPo/SPoref = (Ss/Ssref)-0.85 if w/wL = 0.7±0.1
SPo/SPoref = 1.5 (Ss/Ssref )-0.55 if w/wL > 0.8
for d50(FF) < 1 µm: (3)
Ssref = 25.95 m2/g
SPoref = 489 mm2/C.d 2 CONCLUDING REMARKS
and
for d50(FF) > 1µm: (4) The frost heave response of quarry fines from several loca-
Ssref = 25.95 – 11.7 log d50(FF) tions in the Province of Quebec was studied in the laboratory
SPoref = 489 – 232 log d50(FF) using one-dimensional step-freezing tests with free access to
with d50(FF) expressed in µm. water. The frost heave response was quantitatively related to
the segregation potential parameter defined as the ratio of wa-
Available frost heave data, given in Konrad 2005, were revis- ter intake rate and temperature gradient in the frozen fringe
ited with respect to the reference frost heave characteristics close to thermal steady state. Comparison of the segregation
given above. The results of this analysis are summarized on potential values obtained from these laboratory tests with
Figure 3 which presents the normalized value of segregation available data on fine-grained soils revealed the importance to
potential, SPo/SPoref, versus the normalized value of specific include clay mineralogy and overburden effects in any predic-
surface area, Ss/Ssref. As discussed by Konrad (1999), it is tive empirical relationship, especially when fines are non-
important to consider the influence of soil density, i.e. of clays. A new approach based on reference frost heave char-
w/wL, on frost heave characteristic since it affects particularly acteristics was thus developed to improve the relationship
unfrozen water content, hydraulic conductivity of unfrozen proposed by Konrad (1999). The reference frost heave char-
soil and compressibility. The data presented in Figure 3 are acteristics consist of a relationship between segregation po-
thus separated into four classes: tential at zero overburden pressure, specific surface area and
i) soils with normalized specific surface areas smaller than average grain size of the fines fraction for two artificial soil
1 and with normalized water contents of about 0.7; mixtures in which the clay mineral is poorly crystallized kao-
ii) soils with normalized specific surface areas smaller than linite.
1 and with normalized water contents larger than 0.8; The study led to the following results:
iii) soils with normalized specific surface areas greater than • Consistent relationships were obtained between nor-
1 and with normalized water contents of about 0.7; malized segregation potential, normalized specific sur-
iv) soils with normalized specific surface areas larger than 1 face area and normalized water content for a variety of
and with normalized water contents larger than 0.8. fine-grained soils with clay and nonclay fines;
• When nonclay fines are predominant, i.e. for normal-
ized specific surface areas less than 1, normalized seg-
regation potential increases linearly with increasing
values of normalized specific surface area;
• When the fines consist predominantly of clay miner-
als, i.e. for normalized specific surface areas greater
than 1, normalized segregation potential decreases with
increasing values of normalized specific surface area;
• For fine-grained soils with a significant fraction of
nonclay fines, the influence of overburden pressure is
extremely important and can, at first approximation, be
related to the average particles size of the fines fraction.
• The proposed approach based on reference frost
heave characteristics has been successfully applied to
several well-graded glacial tills in which the fines
ranged from nonclays to clays (see Konrad, 2005).
• The prediction of segregation potential values using
the reference frost heave characteristics approach is
more robust and reliable than the empirical approach
proposed by Konrad (1999) which did not specifically
distinguish between clay and nonclay fines. Further-
Fig. 3. Normalized segregation potential vs. normalized specific sur- more, the use of normalized values of the key parame-
face area. ters eliminates unrealistic values when the average grain
size of the fines are larger than 30 µm.
Figure 3 establishes clearly that consistent relationships
between normalized segregation potential, normalized spe-
cific surface area and normalized water content are obtained
for all the data, even for nonclay fines. The following em-

3581
ACKNOWLEDGEMENTS Konrad, J.-M. 1999. Frost susceptibility related to soil index proper-
ties. Canadian Geotechnical Journal, 36: 403-417.
This research work was funded through the NSERC industrial Konrad, J.-M. 2005. Estimation of the segregation potential of fine-
chair (Chaire de Recherche sur l’Exploitation des Infrastruc- grained soils using the frost heave response of two reference soils
. Canadian Geotechnical Journal. In Press.
tures soumises au Gel). The authors wish to acknowledge the Konrad, J.-M. and N. Lemieux. 2004. Influence of fines on frost
contributions of F.D. Gilbert, I. Leclerc, A. Locat and E. heave characteristics of a well-graded base course material. Ca-
Voyer in the laboratory. Fruitful discussions with Dr. J. Coté nadian Geotechnical Journal. In Press
were appreciated. Konrad, J.-M., and N.R. Morgenstern, 1980. A mechanistic theory
of ice lens formation in fine-grained soils. Canadian Geotechni-
cal Journal, 17: 473-486.
REFERENCES Konrad, J.-M. and N.R. Morgenstern, 1981 The segregation potential
of a freezing soil. Can. Geot. J. , 18: 482-491.
Brandl, H. 2000. Freezing-thawing behaviour of soils and unbound Konrad, J.-M. and N.R. Morgenstern, 1982. Effects of applied pres-
road layers. First Central Asian geotechnical symposium, As- sure on freezing soils. Canadian Geotechnical Journal, Vol. 19:
tana/Kazakhstan. A.A. Balkema, Rotterdam 494-505.
Hoekstra, P., 1969. Water movement and freezing pressures. Soil Mitchell, J.K. 1993. Fundamentals of soil behaviour. Second ed.
Science Society of America Proceedings, Vol. 33, pp. 512-518. John Wiley, New York, 437 pp.
Konrad, J.-M. 1994. Frost heave in soils: Concepts and engineering. Rieke, R., Vinson, T.S., Mageau, D.W. 1983. The role of specific
16th. Canadian Geotechnical Colloquium. Canadian Geotechni- surface area and related index properties in the frost heave sus-
cal Journal, 31: 223-245. ceptibiliy of soils. In Proceedings of the 4th. International Con-
ference on Permafrost, Fairbanks, Alaska, pp. 1066-1071
.

3582