Thaw Consolidation Properties of Fine-Grained Permafrost

Soils of the Mackenzie Valley, Canada

Simon Dumais, Ph.D., Aff.M.ASCE1; and Jean-Marie Konrad, Ph.D.2

Abstract: In this study, the thaw consolidation properties of fine-grained permafrost soils are investigated in terms of the characteristics of
the relationships between the effective stress (σ ′v ), the void ratio (e), and the hydraulic conductivity (kv). A total of 182 individual soil samples
are included in the analysis covering a wide range of hydromechanical properties. The data are obtained from geotechnical studies undertaken
Downloaded from ascelibrary.org by Alex Banos on 05/16/24. Copyright ASCE. For personal use only; all rights reserved.

for the Canadian Arctic Gas Pipeline project in the 1970s in the Mackenzie River Valley, Canada. The investigated characteristics are defined
in agreement with the definition formulated by previous thaw consolidation models for fine-grained soils. Based on the general interpretation
of the behavior of thawing soils, relationships are developed for the compression index of the thawed soil, the residual stress, the hydraulic
conductivity change index of the thawed soil, and the initial hydraulic conductivity of the thawed soil. All properties are influenced by the ice
content, which is characterized by the initial thawed void ratio for ice-rich soils and by the thawed void ratio for ice-poor soils. The liquid
limit, the clay content, and the median grain size of the fine fraction are used as predictive parameters in combination with the initial ice
content. The median grain size of the fine fraction yields the lowest error for the prediction of the characteristics of the σ ′v − e − kv relation-
ships. DOI: 10.1061/JCRGEI.CRENG-720. © 2023 American Society of Civil Engineers.

Introduction consolidation test results obtained mostly from the Canadian Arctic
Gas Pipeline project in the 1970s is used. The general behavior of
The input parameters for modeling thaw consolidation of perma- thawed fine-grained soils is analyzed to determine the factors con-
frost are the hydraulic conductivity of the thawed soil and the rela- trolling the thaw consolidation properties. Empirical relationships
tionship between volume change and applied stress (Morgenstern are then formulated. It is believed that the developed relationships
and Nixon 1971; Foriero and Ladanyi 1995; Dumais and Konrad may be used for sites located outside the Mackenzie River Valley in
2018). These parameters are characterized by the relationship be- the same way that some spatially specific thaw settlement relation-
tween the effective stress (σ ′v ), the void ratio (e), and the hydraulic ships (Luscher and Afifi 1973, Speer 1973; Johnson et al. 1984) are
conductivity (kv). Proper determination of the consolidation proper- commonly used in engineering practice.
ties of thawed soils is paramount for an accurate assessment of thaw
consolidation in engineering practice. Characteristics of the σ ′v –e–kv Relationships
Generally, intact permafrost samples are required to determine
the σ ′v − e − kv relationships experimentally. These samples need Recent developments in thaw consolidation modeling call for a
to remain frozen while they are taken from the field to the labora- more complex definition of the input properties (Qi et al. 2013;
tory. This poses a significant logistical challenge and can be very Zhang and Michalowski 2015; Dumais and Konrad 2018; Yu
costly due to the general remoteness of permafrost areas. Further- et al. 2020; Dayarathne et al. 2022; Xiao et al. 2022). A simple,
more, the experimental evaluation of complete σ ′v − e − kv relation- flexible, yet thorough, framework has recently been developed to
ships is time-consuming. The use of empirical relationships and define the thaw consolidation properties of fine-grained permafrost
data from the literature is common in permafrost engineering prac- soils (Dumais 2019; Dumais and Konrad 2022).
tice to limit the costs associated with the laboratory testing of intact Fig. 1 presents the framework that defines the characteristics of the
σ ′v − e − kv relationships of thawed soils. The volume change behav-
samples. However, the empirical relationships available for thaw
ior of ice-poor soils is described by a semilogarithmic linear σ ′v − e re-
consolidation are mainly limited to the evaluation of thaw settle-
lationship. For ice-poor soils, the slope of the log σ ′v − e relationship is
ment (Luscher and Afifi 1973; Speer 1973; Johnson et al. 1984).
given by the compression index of the thawed soil. For ice-rich soils,
The objective of this paper is to investigate the characteristics of
which contain melt water in excess of the amount that the soil skeleton
the σ ′v − e − kv relationships of fine-grained thawed soils from the
can absorb upon thawing, a bilinear σ ′v − e relationship is adopted.
Mackenzie River Valley, Canada. A compendium of thaw
First, the excess water is drained out without compressing the soil skel-
eton. Then, water is drained out of the soil after compressing the soil
1
Dépt. de génie des mines, de la métallurgie et des matériaux, 1065, av- skeleton in an ice-poor equivalent manner. The e − kv relationship is
enue de la Médecine, Pavillon Adrien-Pouliot Univ. Laval, Québec, QC, defined by way of a semilogarithmic linear curve with a slope defined
Canada G1V 0A6 (corresponding author). ORCID: https://orcid.org/0000 by the hydraulic conductivity change index of the thawed soil.
-0002-9839-1366. Email: [email protected]
2 The σ ′v –e relationship of ice-poor soils for σ ′v > σ ′0 is mathemat-
Dépt. de génie civil et de génie des eaux, 1065, avenue de la Médecine,
ically given by
Pavillon Adrien-Pouliot Univ. Laval, Québec, QC, Canada G1V 0A6.
′ 
Email: [email protected] σv (e)
(e − e*0 ) = −Cc* log (1)
Note. This manuscript was submitted on November 22, 2022; approved σ ′0
on May 14, 2023; published online on October 20, 2023. Discussion period
open until March 20, 2024; separate discussions must be submitted for in- The complete σ ′v –e relationship of ice-rich soils is given by
dividual papers. This paper is part of the Journal of Cold Regions Engi-
neering, © ASCE, ISSN 0887-381X. for e*i < e < e*0 σ ′v (e) = σ ′0 (2)

© ASCE 04023020-1 J. Cold Reg. Eng.

J. Cold Reg. Eng., 2024, 38(1): 04023020

Downloaded from ascelibrary.org by Alex Banos on 05/16/24. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Framework for thaw consolidation of fine-grained soils.

′  thawed soil kv0 *

is the hydraulic conductivity at the thawed void
σ v (e)
for σ ′v (e) ≥ σ ′0 (e − e*0 ) = −Cc* log ratio, that is, the hydraulic conductivity of the soil elements at
σ0′
the residual stress. The hydraulic conductivity change index of
The e–kv relationship is given by the thawed soil Ck* is the slope of the log kv–e curve.


kv (e)
(e − e*0 ) = Ck* log *
(3)
kv0
Description of Investigated Soils
The characteristics of the σ ′v − e − kv relationships are summa-
rized in Table 1. The void ratio of the soil in its frozen state is The study of the properties of thawed fine-grained soils is based on
the frozen void ratio ef. The void ratio of the soil thawed in un- the analysis of a compendium of data mostly from the Canadian
drained conditions is the initial thawed void ratio e*i . The volumet- Arctic Gas Pipeline project. The analysis is limited to nonorganic
ric changes between the frozen void ratio and the initial thawed fine-grained saturated soils. A total of 183 permafrost soil speci-
void ratio are attributable to phase change only such that they are mens are considered in the current investigation: 165 are from
related by the Mackenzie Valley in North-Western Canada (NESCL 1975,
ef 1976, 1977a, b, c, d) and 18 are from the site of the experimental
e*i = (4) warm-oil pipeline near Inuvik (Watson et al. 1972). The index
1.09
properties of the investigated soils are presented in Fig. 2.
The residual stress σ ′0 is defined as the effective stress that the The thaw consolidation and thaw settlement experimental meth-
soil skeleton can sustain when it is thawed in undrained conditions. ods, which are thoroughly described in the Canadian Arctic Gas
It is therefore the effective stress within the soil elements upon Pipeline project reports and by Watson et al. (1972), are briefly de-
thawing. The definition of the residual stress and its application scribed here. A modified oedometer in which the settlement and hy-
for thaw consolidation modeling are thoroughly covered in previ- draulic conductivity can be measured was used. Intact frozen core
ous studies (Morgenstern and Nixon 1971; Nixon and Morgenstern samples with a diameter slightly smaller than that of the cell were
1973; Dumais 2019; Dumais and Konrad 2018, 2019, 2022). The first thawed under a small applied load. The deformation was mea-
thawed void ratio e*0 is the void ratio of the soil elements at the re- sured after dissipating all excess pore-water pressure. The samples
sidual stress. For ice-poor soils, the thawed void ratio is equal to the were then subjected to incremental loading. The deformation was
initial thawed void ratio. For ice-rich soils, the initial thawed void measured for each loading step such that the void ratio could be de-
ratio is larger than the thawed void ratio due to the presence of ex- termined. The hydraulic conductivity was measured directly in the
cess melt water. The compression index of the thawed soil Cc* is the cell at selected loading stages for the consolidated samples using a
slope of the log σ ′v –e curve. The initial hydraulic conductivity of the falling head permeameter. In this study, the results of these tests are
analyzed in terms of the effective stress, the void ratio, and the hy-
Table 1. Characteristics of the σ ′v –e–kv relationships of thawed draulic conductivity.
fine-grained soils
Symbol Characteristic
ef Frozen void ratio Methodology
e*i Initial thawed void ratio
e*0 Thawed void ratio The raw laboratory data of each individual soil specimen were an-
Cc* Compression index of the thawed soil alyzed in order to establish the relationships between the effective
σ ′0 Residual stress stress, the void ratio, and the hydraulic conductivity. The analysis
Ck* Hydraulic conductivity change index of the thawed soil
* was performed as a function of the characteristics of the σ ′v − e − kv
kv0 Initial hydraulic conductivity at the thawed void ratio
relationships introduced previously.

© ASCE 04023020-2 J. Cold Reg. Eng.

J. Cold Reg. Eng., 2024, 38(1): 04023020

 (a) (b)
Downloaded from ascelibrary.org by Alex Banos on 05/16/24. Copyright ASCE. For personal use only; all rights reserved.

(c) (d)

(e) (f )

Fig. 2. Frequency histograms of some index properties of all investigated soils: (a) frozen void ratio; (b) median grain size of the fine fraction;
(c) liquid limit; (d) plastic limit; (e) sand content; and (f) clay content.

The σ ′v –e–kv relationships of selected ice-poor soils are pre- by extrapolating the linear relationship to the initial thawed void
sented in Fig. 3. The residual stress and the initial hydraulic con- ratio would be infinitely small and would hold no physical mean-
ductivity are obtained by extrapolating the experimental ing. For ice-rich soils, the compression index of the thawed soil
relationship to the thawed void ratio. The soils presented in and the hydraulic conductivity change index of the thawed soil
Fig. 3 can be considered ice-poor since the residual stress found can be determined within the linear portion of the relationships,
by extrapolation is in a range of realistic values. It is therefore as- as shown in Fig. 4.
sumed that the thawed void ratio is equal to the initial thawed void The soils were categorized as ice-poor and ice-rich soils in terms
ratio. of the behavior described previously such that 99 specimens are
The σ ′v –e–kv relationships of a selected ice-rich soils are pre- ice-poor, and 84 specimens are ice-rich. The relationship between
sented in Fig. 4. For ice-rich soils, the residual stress value obtained the effective stress and the void ratio was determined for 170 soil

Fig. 3. Properties of selected thawed ice-poor fine-grained soils (NESCL 1977a).

© ASCE 04023020-3 J. Cold Reg. Eng.

J. Cold Reg. Eng., 2024, 38(1): 04023020

Downloaded from ascelibrary.org by Alex Banos on 05/16/24. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Properties of thawed ice-rich fine-grained soils (Sample 3A-1-9; NESCL 1976).

specimens, while the relationship between the hydraulic conductiv- swelling behavior of the clay-silt mixture determines the volume
ity and the void ratio was determined for 88 soil specimens. The change behavior upon thawing. Swelling is mainly effective within
investigated soils cover a wide range of mechanical and physical the micropores. Because the micropore distribution is controlled by
properties such that the general applicability of the analysis can the clay-size particles and the clay minerals, the volume change be-
be assumed. havior upon thawing is expected to be a function of the clay content
in the soil, that is, a function of the soil composition.
The hydraulic conductivity of thawed fine-grained soils can also
be analyzed in terms of the mechanisms acting on the soil skeleton
General Interpretation of the Properties
upon freezing and thawing. Generally, it is the finer clay particles
of Thawing Soils
that control the hydraulic conductivity. In unfrozen soils, the clay-
The hydraulic conductivity and the volume change properties of size particles form an organized structure. The suction forces cause
fine-grained soils upon thawing are controlled by the soil composi- a structural rearrangement of the clay-size particles upon freezing.
tion and by the soil fabric. Fine-grained soils can be conceptualized This causes the clay packets to collapse into a denser and more dis-
as mixtures of silt and clay particles. The soil composition and the persed structure. Upon thawing, the collapsed clay packets occupy
soil fabric of thawed fine-grained soils can be discussed in terms of less void space than in the unfrozen soil. The hydraulic conductiv-
the behavior of the silt and clay phases (Chamberlain and Gow ity of thawed soils is therefore higher than the hydraulic conductiv-
1979; Konrad 1999). Fine-grained soils are characterized by their ity of the soil prior to freezing. Regarding the soil composition,
macropore and micropore distribution. The macropore space is soils with higher clay content have a lower hydraulic conductivity
mainly controlled by the coarser silt particles that control the pack- than soils composed predominantly of coarser silt particles.
ing. The micropore distribution within the soil skeleton is con- As previously mentioned, the behavior of the soil upon thawing
trolled by the clay-size particles and the clay minerals. is controlled by the fabric of the soil. The fabric of the soil is de-
An appreciation of the mechanisms acting on the soil skeleton fined as the arrangements of the particles and the association be-
upon freezing is essential to explain the behavior of fine-grained tween the particles. For the same soil composition, the particles
soil upon thawing. Closed-system freezing with no access to exter- of fine-grained soils may be arranged in different ways, which af-
nal water is first considered. As the temperature decreases, the ice fects the volume change behavior and the hydraulic conductivity.
first forms in the macropores. Unfrozen water is then found in the For thawed fine-grained soils, the fabric of the soil can be discussed
micropores in the form of free water and adsorbed water films in terms of the initial ice content, and primarily with regard to the
around the clay minerals. The temperature continues to decrease, difference in behavior between ice-poor and ice-rich soils. The
leading to the formation of additional ice in the pores, which even- main difference between ice-poor and ice-rich soils is the formation
tually starts to penetrate in the micropores. The unfrozen free cap- of segregated ice in the form of ice lenses in ice-rich soils. Two ide-
illary water and the adsorbed water then comes into contact with ice alized clay-slit mixtures are now considered: one is ice-poor, and
such that it is in a state of suction according to the laws of thermo- one is ice-rich. It is assumed that the suction forces acting in the mi-
dynamics. This causes a reduction in the amount of capillary water cropores are similar in closed-system and open-system freezing. In
and of adsorbed water. The clay-size particles and clay minerals are other words, the arrangements of the particles in the micropores
then subjected to suction forces, which induce a structural rear- after freezing are comparable irrespective of the presence or ab-
rangement in the pores leading to overconsolidation. The suction sence of segregated ice. Yet, there is a difference in the fabric of
forces are mainly effective within the micropores. thawed ice-poor and ice-rich fine-grained soils, which can be attrib-
Upon thawing, the soil particles do not move back to their orig- uted to the presence of excess ice upon thawing.
inal position. However, the overconsolidated soil elements within Upon thawing of an ice-poor soil, the overconsolidated soil el-
the micropores can absorb the free melt water. This leads to a swell- ements absorb the free melt water found in the macropores. When
ing of the overconsolidated clay elements. For the sake of simplic- thawed under an applied load higher than the residual stress, some
ity, the state path followed upon thaw consolidation is defined by of the melt water is simultaneously drained out due to the compres-
the swelling curve of the soil upon thawing. In other words, the sion of the soil skeleton. The space in the macropores is therefore

© ASCE 04023020-4 J. Cold Reg. Eng.

J. Cold Reg. Eng., 2024, 38(1): 04023020

 reduced due to the concurrent drainage of water into the overcon-
solidated soil elements and outside the soil skeleton. In other
words, the macrostructure of ice-poor soils mostly collapses upon
thawing.
Upon thawing of an ice-rich soil, there are different operating
seepage mechanisms. As the ice-rich soil thaws, the melt water
in excess is drained out of the soil skeleton without compressing
the soil skeleton. Unidimensional vertical consolidation is assumed
such that the excess melt water seeps through the macropores of the
thawing soil to reach the surface of the soil. Meanwhile, the over-
consolidated soil elements are thawing and absorbing free melt
water. In contrast to ice-poor soils, the water absorbed by the over-
consolidated soil elements from the macropores is replenished by a
continuous seepage of excess melt water. The macrostructure is
sustained by the pressure head within the macropores. As all the ex- Fig. 5. Relationship between the median grain size of the fine fraction
Downloaded from ascelibrary.org by Alex Banos on 05/16/24. Copyright ASCE. For personal use only; all rights reserved.

cess water is drained out of the bulk soil, a consolidation of the soil and the clay content of investigated soils.
occurs in such a way that a part of the macrostructure collapses.
Still, the fabric of thawed fine-grained soils can maintain more
macropore space due to the seepage of excess water upon thawing. Furthermore, the void ratio is used as the state variable in thaw con-
solidation modeling (Dumais and Konrad 2018, 2022). Like unfro-
zen soils, the use of the thawed void ratio may be redundant with
Parameters Controlling the Properties of Thawed the use of the liquid limit, the clay content, or the median grain
Fine-Grained Soils size of the fine fraction for ice-poor soils. However, using the initial
thawed void ratio in addition to the liquid limit, the clay content, or
The previous discussion highlighted the importance of the soil the median grain size of the fine fraction may be valuable for ice-
composition and of the soil fabric in governing the properties of rich soils to characterize the impact of the excess ice.
thawed fine-grained soils. We will now explore the index properties
that are indicative of the composition and fabric of thawed fine-
grained soils to support the development of empirical relationships Difference between Ice-Poor and Ice-Rich Soils
for the characteristics of the σ ′v − e − kv relationships. The difference between ice-poor and ice-rich soils regarding their
First, the liquid limit can be used in continuity with several stud- thaw consolidation behavior is thoroughly discussed in other pub-
ies, which have successfully developed empirical relationships for lications (Dumais 2019; Dumais and Konrad 2022). The main dis-
the compressibility of fine-grained soils from the liquid limit tinction is that ice-rich soils contain excess ice (van Everdingen
(Azzouz et al. 1976). 2005). This means that there is melt water in excess of the amount
Second, the soil composition of fine-grained soils can be char- that the soil skeleton can sustain upon thawing. This entails that
acterized by the clay content. As discussed in the previous section, thaw consolidation of ice-rich soils is a combination of drainage
the behavior of thawed clay–silt mixtures is largely controlled by of the excess water and of expulsion of pore water, while thaw con-
the content of clay-size particles, which are most affected by solidation of ice-poor soils is attributable only to the expulsion of
freezing. pore water.
Third, the median grain size of the fine fraction, d50 ff , can also The empirical relationships proposed in the next section are
be used to characterize the soil composition of fine-grained soils. based on the distinct behaviors of ice-poor and ice-rich soils. Be-
Clay particles may differ in size depending on their origin and min- cause the objective is to establish the characteristics of the
eralogy. The size of clay particles is known to impact the consoli- σ ′v − e − kv relationships, the thaw consolidation behavior as de-
dation properties of fine-grained soils. The median grain size of the scribed in section “Characteristics of the σ ′v –e–kv Relationships”
fine fraction is used to give an indication of the mineralogy of the cannot be used as an input. Instead, it is essential to be able to iden-
clay phase. The median grain size of the fine fraction is therefore tify the presence of excess ice. Conveniently, it is not necessary to
believed to be a more indicative parameter of the soil composition determine with precision the excess ice content. As a first estimate,
than the clay content. The median grain size of the fine fraction has the presence of visible ice in frozen soil samples generally indicates
been used successfully to develop empirical relationships for the the presence of excess ice (Castagner et al. 2023). Alternatively, the
frost susceptibility, a property highly correlated with the hydraulic presence of supernatant water when a frozen soil sample is allowed
conductivity of the soil (Konrad 1999). Fig. 5 presents the correla- to thaw and settle freely in a container is indicative of excess ice
tion between the clay content and the median grain size of the fine (Kokelj and Burn 2003).
fraction for the investigated soils. As expected, the median grain
size of the fine fraction decreases as the clay content increases.
However, there is some variation, which indicates that the investi-
gated soils may contain different sizes of clay particles. Empirical Relationships for the Properties of Thawed
Finally, the initial void ratio is a parameter often used to develop Fine-Grained Soils
empirical relationships for the compressibility of fine-grained soils
(Azzouz et al. 1976). In unfrozen soil, the initial void ratio is used The compendium of literature data is now used to develop empir-
as a proxy to characterize the impact of the soil fabric and the soil ical relationships to determine the characteristics of the
composition on the compressibility. The initial thawed void ratio σ ′v − e − kv relationships. The liquid limit, the clay content, the me-
can therefore be used as an index property to develop empirical re- dian grain size of the fine fraction, and the initial thawed void ratio
lationships. It is convenient to use the initial thawed void ratio be- are used as index properties. The relationships presented in this sec-
cause it is directly related to the frozen void ratio by Eq. (4). tion can be used to determine the parameters of Eqs. (1)–(3).

© ASCE 04023020-5 J. Cold Reg. Eng.

J. Cold Reg. Eng., 2024, 38(1): 04023020

Downloaded from ascelibrary.org by Alex Banos on 05/16/24. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

Fig. 6. Compression index of (a) ice-poor; and (b) ice-rich thawed fine-grained soils.

Compression Index of the Thawed Soil relationships. In Fig. 6(b), the relationships between the initial
thawed void ratio and the compression index of the thawed soil
Fig. 6 presents the relationships between the initial void ratio and
are presented as a function of the median grain size of the fine frac-
the compression index of the thawed soil. In Fig. 6(a), a relation-
tion. The data points are presented in terms of a range of median
ship is presented for ice-poor soils for which the initial void ratio
grain size of the fine fraction value from which relationships for
is equal to the thawed void ratio. For the analyzed data set, the
the average value of the range are developed. The relationship
use of a predictive parameter in addition to the thawed void ratio for ice-poor soils is reproduced in Fig. 6(b). Once again, the com-
does not lead to any significant increase in the precision of the em- pression index of thawed soils increases with the initial thawed
pirical relationship. A unique relationship in terms of the thawed void ratio. Also, Fig. 6(b) shows that the compression index of
void ratio is thus adopted for ice-poor soils. In unfrozen fine- thawed soils increases as the size of the soil particles decreases.
grained soils, the regression analysis based on the combination of For the sake of illustration, the data point of an individual range
the initial void ratio and an additional independent variable also of value is presented in Fig. 7. In Fig. 7, ice-poor and ice-rich val-
presents insignificant increases in accuracy over the regression ues alike are presented for soils with a median grain size of the fine
analysis based on the initial void ratio alone (Azzouz et al. fraction varying from 4 × 10−3 to 2 × 10−3 mm. The variability of
1976). In these cases, the initial void ratio can be used as a proxy the data with the predictive relationship for ice-rich soils appears
for characterizing both the soil fabric and the soil composition. to be significant. However, it should be mentioned that the illus-
As expected, Fig. 6(a) shows that the compression index of thawed trated predictive relationship is for the average median grain size
ice-poor soils increases with the thawed void ratio. of the fine fraction value, while the data point spreads over a
For ice-rich soils, the effect of the excess ice on the macrostruc- range of value. The accuracy of the predictive relationships is
ture has been discussed previously. It is therefore warranted to use best appreciated in terms of the predictive errors presented in
an additional predictive parameter to establish empirical Table 2 rather than visually from the graphical representation of
the relationships.
Table 2 presents the predictive relationships for the compression
index of the thawed soil for ice-poor and ice-rich soils in terms of
different index properties. The best-fit relationships for ice-poor
soils and for the median grain size of the fine fraction presented
in Table 2 correspond to the graphical relationships presented in
Fig. 6. Relationships developed in terms of the liquid limit and
the clay content are presented in addition to the relationship devel-
oped in terms of the median grain size of the fine fraction. For each
relationship, the mean absolute error (MAE) is indicated to show
the accuracy of the relationship. Given the general variability of
geotechnical properties, the mean absolute error values obtained
for all relationships for the compression index of the thawed soil
are highly satisfactory. For ice-rich soils, the median grain size of
the fine fraction appears to be the best predictive parameter because
the mean absolute error of the relationship is the lowest.

Residual Stress
Fig. 7. Compression index of thawed fine-grained soils with a median
Fig. 8 presents the relationships between the thawed void ratio of
grain size of the fine fraction between 4 × 10−3 and 2 × 10−3 mm.
ice-poor soils and the residual stress in terms of the median grain

© ASCE 04023020-6 J. Cold Reg. Eng.

J. Cold Reg. Eng., 2024, 38(1): 04023020

 Table 2. Relationships for the compression index of the thawed soil
Index parameter e*0 = e*i MAE e*0 < e*i MAE
wL (%) Cc* = 0.74 log e*0 + 0.22 0.050 Cc* = (0.0081wL − 0.019) log e*i + (0.0033wL + 0.037) 0.061
clay% (%) Cc* = (0.0051clay% − 0.18) log e*i + (0.0015clay% + 0.096) 0.056
d50 ff (mm) Cc* = (−0.11 log d50ff + 0.080) log e*i + (−0.097 log d50ff − 0.082) 0.041

stress at a constant thawed void ratio increases as the size of the

soil particles decreases.
Table 3 presents the predictive relationships for the residual
stress in terms of different index properties. Once again, the median
grain size of the fine fraction appears to be the best predictive pa-
rameter because the mean absolute error of the relationship is the
lowest. The mean absolute error value of a little bit higher than
half an order of magnitude is highly satisfactory.
Downloaded from ascelibrary.org by Alex Banos on 05/16/24. Copyright ASCE. For personal use only; all rights reserved.

Hydraulic Conductivity Change Index of the Thawed Soil

Fig. 9 presents the relationships between the initial void ratio and
the hydraulic conductivity change index of the thawed soil. The
same approach as described for the compression index of the
thawed soil is adopted here. In Fig. 9(a), a relationship is presented
for ice-poor soils. Once again, a unique relationship in terms of the
thawed void ratio is adopted for ice-poor soils. Fig. 9(a) shows that
Fig. 8. Residual stress of thawed fine-grained soils. the hydraulic conductivity change index of thawed ice-poor soils
increases with the thawed void ratio.
For ice-rich soils, Fig. 9(b) presents the relationships between
size of the fine fraction. While it is possible to determine the com- the initial thawed void ratio and the hydraulic conductivity change
pression index of thawed ice-poor fine-grained soils from the index of the thawed soil as a function of the median grain size of the
thawed void ratio alone, the residual stress varies significantly as fine fraction. The relationship for ice-poor soils is reproduced in
a function of the soil composition. As expected, the residual stress Fig. 9(b). Once again, the hydraulic conductivity change index of
increases as the thawed void ratio decreases. Also, the residual thawed soils increases with the initial thawed void ratio. Also,

Table 3. Relationships for the residual stress

Index parameter e*0 = e*i MAE
wL (%) σ ′0 = exp [(e*0 −0.014wL − 0.42)/(−0.0014wL − 0.012)] 100.897
clay% (%) σ ′0 = exp [(e*0 −0.0049clay% − 0.82)/(−0.00063clay% − 0.060)] 100.750
d50 ff (mm) σ ′0 = exp [(e*0 +0.25 log d50 ff − 0.45)/(0.020 log d50 ff − 0.060) ] 100.671

(a) (b)

Fig. 9. Hydraulic conductivity change index of (a) ice-poor; and (b) ice-rich thawed fine-grained soils.

© ASCE 04023020-7 J. Cold Reg. Eng.

J. Cold Reg. Eng., 2024, 38(1): 04023020

 Table 4. Relationships for the hydraulic conductivity change index of thawed soils
Index parameter e*0 = e*i MAE e*0 < e*i MAE
wL (%) Ck* = 0.30 log e*0 + 0.12 0.035 Ck* = (0.0035wL − 0.018) log e*i + (0.0019wL + 0.021) 0.044
clay% (%) Ck* = (0.0018clay% − 0.099) log e*i + (0.00070clay% + 0.067) 0.033
d50 ff (mm) Ck* = (−0.074 log d50ff − 0.014) log e*i + (−0.028 log d50ff + 0.024) 0.031

soil. Once the soil does not contain any more excess melt
water, the behavior is ice-poor equivalent (Dumais and Konrad
2022). This means that once the soil is at the thawed void ratio,
the effective stress within the soil is equal to the residual stress,
and the compression of the soil skeleton is governed by a semilo-
garithmic linear relationship between the void ratio and the effec-
tive stress. The importance of the transition from the ice-rich
state to the ice-poor equivalent state should be emphasized in
Downloaded from ascelibrary.org by Alex Banos on 05/16/24. Copyright ASCE. For personal use only; all rights reserved.

terms of the link between the initial thawed void ratio and the
thawed void ratio.
As discussed previously, the excess ice contained in ice-rich
soils contributes to the preservation of macropore space in the
thawed soil. The macropore space is expected to influence the com-
pression behavior of the ice-poor equivalent soil. In other words,
there should exist a link between the initial void ratio of an ice-rich
soil, its thawed void ratio, and its compression index. The relation-
ship between the initial void ratio of ice-rich soils and the compres-
Fig. 10. Initial hydraulic conductivity of thawed fine-grained soils. sion index is shown in Fig. 6. As the initial void ratio increases, the
compression index increases due to the increasing macropore space
in the thawed soil.
Fig. 9(b) shows that the hydraulic conductivity change index of
Fig. 11 presents a schematic representation of a method pro-
thawed soils increases as the size of the soil particles decreases.
posed to determine the ice-poor equivalent thawed void ratio for
Table 4 presents the predictive relationships for the hydraulic
ice-rich soils. The thawed void ratio of the ice-poor equivalent
conductivity change index of thawed soils in terms of different
soil can be found by using the relationship between the compres-
index properties. Once again, the mean absolute error values are
within an acceptable range. For the hydraulic conductivity change sion index of the thawed soil and the thawed void ratio presented
index of thawed soils, the clay content appears to be marginally in Fig. 6(a). The compression index of the thawed ice-rich soil is
more precise than the median grain size of the fine fraction. first determined from the initial thawed void ratio. Then, the thawed
void ratio of the ice-poor soil with the same compression index
value is determined. The value hence determined can be used as
Initial Hydraulic Conductivity of the Thawed Soil the thawed void ratio of the ice-rich soil.
Fig. 10 presents the relationships between the thawed void ratio of
ice-poor soils and the initial hydraulic conductivity in terms of the Use of Empirical Relationships in Practice
median grain size of the fine fraction. As expected, the initial hy-
draulic conductivity increases twice: when the thawed void ratio in- The use of empirical in practice may lead to cost savings associated
creases and when the size of the particles increases. with geotechnical investigations. However, the uncertainty associ-
Table 5 presents the predictive relationships for the initial hy- ated with such relationships may also increase the cost of the infra-
draulic conductivity in terms of different index properties. The structure as the design needs to consider a higher factor of safety. A
mean absolute error values of each of the three considered param- thorough geotechnical investigation is paramount and cannot be re-
eters are slightly under one order of magnitude. However, the rela- placed by empirical relationships. It should be noted that the pro-
tionships can be considered accurate given the general variability of posed relationships are based on the knowledge of the frozen
the hydraulic conductivity of fine-grained soils. void ratio, which typically requires an intact frozen sample. The ad-
ditional costs associated with laboratory testing of these samples
might not be too significant. Empirical relationships are best used
Discussion in combination with actual testing results from the site. They
may be used to complete unavailable information such as the hy-
draulic conductivity or to develop site-specific relationships by
Note on Ice-Rich Soils
combining test results with the trend suggested in this study. In
As shown in Fig. 1, thaw consolidation of an ice-rich soil is first this regard, the relationships proposed in this study can be used
characterized by the drainage of the excess water out of the bulk for sites outside the Mackenzie Valley.

Table 5. Relationships for the initial hydraulic conductivity of thawed soils

Index parameter e*0 = e*i MAE
* −8
wL (%) kv0 = 1.3 × 10 exp[−0.25wL + (0.08wL + 12.85)e*0 ] 100.977
*
clay% (%) kv0 = 1.1 × 10−12 exp[−0.077clay% + (−0.050clay% + 20.5)e*0 ] 100.906
*
d50 ff (mm) kv0 = 2.2 × 10−10 exp[3.1 log d50ff + (2.02 log d50ff + 23.6)e*0 ] 100.932

© ASCE 04023020-8 J. Cold Reg. Eng.

J. Cold Reg. Eng., 2024, 38(1): 04023020

 circumstances. First, the coefficient of consolidation is relatively
constant when the changes in void ratio are small. This is rarely
the case for thawing fine-grained soils. Second, the coefficient of
consolidation of the soil is constant for a certain ratio between
the compression index of the thawed soil and the hydraulic conduc-
tivity change index of the thawed soil. This ratio defined by Tave-
nas et al. (1983) is shown in Fig. 12. Fig. 12 also shows the ratio as
a function of the soil samples included in this study. For the small-
strain consolidation theory to be valid, the points should be on the
dashed line. It is clear in Fig. 12 that most soils considered in the
current study do not meet the criteria and that the large-strain con-
solidation theory should be used for modeling thawing permafrost
soils. It should be acknowledged that correction factors were pro-
posed for the Nixon and Morgenstern (1973) theory to consider
some large-strain effects in thaw consolidation, although only par-
Fig. 11. Determination of the ice-poor equivalent thawed void ratio for
Downloaded from ascelibrary.org by Alex Banos on 05/16/24. Copyright ASCE. For personal use only; all rights reserved.

tially (Dumais and Konrad 2018).

ice-rich soils from the compression index of the thawed soil.

The accuracy of the proposed relationships is highly satisfac- Conclusion

tory. A part of the uncertainty arises from the general variability
of geotechnical properties and of laboratory test results for frozen Relationships for the characteristics of the σ ′v − e − kv relationships
soils. This is especially true for the hydraulic conductivity. It is nor- of thawed fine-grained soils have been developed for soils from the
mal to have a larger error for the hydraulic conductivity relation- Mackenzie Valley. For ice-poor soils, the compression index of the
ships. The proposed relationships were developed based on thawed soil and the hydraulic conductivity change index of the
physically meaning parameters and not simply on statistical thawed soil are functions of the thawed void ratio. For ice-rich
correlation. soils, these two characteristics are functions of the initial thawed
void ratio and of index properties, which characterize the soil com-
position and the soil fabric. The liquid limit, the clay content, and
Large-Strain Consolidation of Thawing Soils
the median grain size of the fine fraction are used as predictive pa-
The validity of using the small-strain consolidation theory for mod- rameters, with the median grain size of the fine fraction yielding the
eling thawing soils can be evaluated from the large amount of data greatest precision. The thawed void ratio of ice-rich soils can be ob-
analyzed in this study. Therzaghi’s theory was used in the develop- tained from the ice-poor equivalent soil with the same compression
ment of the Morgenstern and Nixon (1971) thaw consolidation the- index. The residual stress and the initial hydraulic conductivity are
ory. The conditions for the validity of small strain consolidation functions of the thawed void ratio and of index properties. The pro-
theory are presented in Fig. 12 in comparison with the values of posed empirical relationships are defined in agreement with the def-
the data from the Mackenzie Valley. The small-strain consolidation inition of the σ ′v − e − kv relationships of thawed fine-grained soils
theory assumes that the coefficient of consolidation of the soil is formulated by Dumais and Konrad (2022). Hence, they can be used
constant during consolidation. This is valid under two to define the input parameters required for the numerical modeling
of thaw consolidation with the model formulated by Dumais and
Konrad (2018).

Data Availability Statement

All data that support the findings of this study are available from the
corresponding author upon reasonable request.

References

Azzouz, A. S., R. J. Krizek, and R. B. Corotis. 1976. “Regression analysis

of soil compressibility.” Soils Found. 16 (2): 19–29. https://doi.org/10
.3208/sandf1972.16.2_19.
Castagner, A., A. Brenning, S. Gruber, and S. V. Kokelj. 2023. “Vertical
distribution of excess ice in icy sediments and its statistical estimation
from geotechnical data (Tuktoyaktuk Coastlands and Anderson Plain,
Northwest Territories).” Arctic Science. e-First https://doi-org.acces
.bibl.ulaval.ca/10.1139/as-2021-0041.
Chamberlain, E. J., and A. J. Gow. 1979. “Effect of freezing and thawing
on the permeability and structure of soils.” Eng. Geol. 13 (1–4): 73–92.
https://doi.org/10.1016/0013-7952(79)90022-X.
Fig. 12. Validity of the small-strain consolidation theory for investi- Dayarathne, R., B. Hawlader, R. Phillips, and D. Robert. 2022. “One- and
two-dimensional finite element modelling of thaw consolidation.” Can.
gated soils.
Geotech. J. 59 (7): 1130–1145. https://doi.org/10.1139/cgj-2021-0120.

© ASCE 04023020-9 J. Cold Reg. Eng.

J. Cold Reg. Eng., 2024, 38(1): 04023020

 Dumais, S. 2019. “Modélisation de la consolidation au dégel à grandes NESCL (Northern Engineering Services Company Limited). 1977b. Thaw
déformations.” Ph.D. thesis, Dept. of Civil Engineering, Univ. Laval. settlement studies, proposed Arctic gas pipeline route. Calgary, AB,
Dumais, S., and J.-M. Konrad. 2018. “One-dimensional large-strain thaw Canada: NESCL.
consolidation using nonlinear effective stress–void ratio–hydraulic con- NESCL (Northern Engineering Services Company Limited). 1977c.
ductivity relationships.” Can. Geotech. J. 55 (3): 414–426. https://doi Thaw-settlement test data Fort Simpson to Zama Lake Area. Calgary,
.org/10.1139/cgj-2017-0221. AB, Canada: NESCL.
Dumais, S., and J.-M. Konrad. 2019. “Large-strain nonlinear thaw consol- NESCL (Northern Engineering Services Company Limited). 1977d.
idation analysis of the inuvik warm-oil experimental pipeline buried in Laboratory testing for thaw slope stability: Continuous permafrost
permafrost.” J. Cold Reg. Eng. 33 (1): 04018014. https://doi.org/10 zone. Calgary, AB, Canada: NESCL.
.1061/(ASCE)CR.1943-5495.0000179. Nixon, J. F., and N. R. Morgenstern. 1973. “Practical extensions to a theory
Dumais, S., and J.-M. Konrad. 2022. “Framework for thaw consolidation of of consolidation for thawing soils.” In Proc., Permafrost: North
fine-grained soils.” Can. Geotech. J. American Contribution to the 2nd Int. Conf., 369–377. Washington,
Foriero, A., and B. Ladanyi. 1995. “FEM assessment of large-strain thaw DC: National Academy of Sciences.
consolidation.” J. Geotech. Eng. 121 (2): 126–138. https://doi.org/10 Qi, J., X. Yao, and F. Yu. 2013. “Consolidation of thawing permafrost con-
.1061/(ASCE)0733-9410(1995)121:2(126). sidering phase change.” KSCE J. Civ. Eng. 17 (6): 1293–1301. https://
Johnson, T. C. E., E. C. McRoberts, and J. F. Nixon. 1984. “Design impli- doi.org/10.1007/s12205-013-0240-1.
cations of subgrade thawing.” In Frost action and Its control: A state of Speer, T. L. 1973. “Effects of ground-ice variability and resulting thaw set-
Downloaded from ascelibrary.org by Alex Banos on 05/16/24. Copyright ASCE. For personal use only; all rights reserved.

the practice report, 45–103. New York: ASCE. tlements on buried warm-oil pipelines.” In Proc., Permafrost: North
Kokelj, S. V., and C. R. Burn. 2003. “Ground ice and soluble cations in American Contribution to the 2nd Int. Conf., 746–752. Washington,
near-surface permafrost, Inuvik, Northwest Territories, Canada.” DC: National Academy of Sciences.
Permafr. Periglac. 14 (3): 275–289. https://doi.org/10.1002/ppp.458. Tavenas, F., P. Jean, P. Leblond, and S. Leroueil. 1983. “The permeability
Konrad, J.-M. 1999. “Frost susceptibility related to soil index properties.” of natural soft clays. Part II: Permeability characteristics.” Can.
Can. Geotech. J. 36 (3): 15. Geotech. J. 20 (4): 645–660. https://doi.org/10.1139/t83-073.
Luscher, U., and S. S. Afifi. 1973. “Thaw consolidation of Alaskan silts and Van Everdingen, R. O. 2005. Multi-language glossary of permafrost and
granular soils.” In Proc., Permafrost: North American Contribution to related ground-ice terms. Calgary, AB, Canada: International
the 2nd Int. Conf., 325–334. Washington, DC: National Academy of Permafrost Association.
Sciences. Watson, G. H., R. K. Rowley, and T. L. Speer. 1972. Inuvik frozen core
Morgenstern, N. R., and J. F. Nixon. 1971. “One-dimensional consolida- testing. Inuvik, AL: Mackenzie Valley Pipeline Research Limited.
tion of thawing soils.” Can. Geotech. J. 8 (4): 558–565. https://doi Xiao, W., E. Liu, X. Yin, G. Zhang, C. Zhang, and Q. Yu. 2022.
.org/10.1139/t71-057. “Numerical simulation of coupled liquid water, stress and heat for fro-
NESCL (Northern Engineering Services Company Limited). 1975. zen soil in the thawing process.” Eng. Comput. 39 (4): 1492–1510.
Geotechnical data report proposed Arctic Gas pipeline cross delta al- https://doi.org/10.1108/EC-03-2021-0138.
ternative route channel depth anomaly drilling and sampling program. Yu, F., P. Guo, Y. Lai, and D. Stolle. 2020. “Frost heave and thaw consol-
Calgary, AB, Canada: NESCL. idation modelling. Part 2: One-dimensional thermohydromechanical
NESCL (Northern Engineering Services Company Limited). 1976. (THM) framework.” Can. Geotech. J. 57 (10): 1595–1610. https://doi
Thaw-settlement test data: Fort Simpson Area. Calgary, AB, Canada: .org/10.1139/cgj-2019-0306.
NESCL. Zhang, Y., and R. L. Michalowski. 2015. “Thermal-hydro-mechanical
NESCL (Northern Engineering Services Company Limited). 1977a. analysis of frost heave and thaw settlement.” J. Geotech. Geoenviron.
Laboratory testing for thaw slope stability: Discontinuous permafrost Eng. 141 (7): 04015027. https://doi.org/10.1061/(ASCE)GT.1943
zone. Calgary, AB, Canada: NESCL. -5606.0001305.

© ASCE 04023020-10 J. Cold Reg. Eng.

J. Cold Reg. Eng., 2024, 38(1): 04023020