2
Physical and Thermal
Properties
Frozen soil is a four-component system consisting of soil parti-
cles, ice, water, and voids. The particles (mineral and/or organic
matter) come in various sizes and shapes with a thin film of
unfrozen water coating most mineral grains. The voids are filled
with ice, unfrozen water, and air. Ice may be distributed uni-
formly throughout the soil mass, or it may have accumulated in
the form of irregular or stratified ice inclusions. Larger ice
masses may form as a result of processes associated with ice
wedges and pingos. Frozen soil classification involves identifica-
tion of the soil phase, adding characteristics associated with the
frozen soil, and describing ice found in frozen ground.
The mechanical properties of frozen soil at a given tempera-
ture may vary from relatively brittle to plastic, depending on
the unfrozen water content. The presence of solutes in the pore
water will alter the ice content and the frozen soil behavior. Sea-
sonal temperature changes are responsible for the frost action
process, involving both heaving and loss in stability on thawing
for surface soil layers. The dependence of frozen soil behavior
on temperature requires computations that involve soil thermal
properties. These topics are presented here in Chapter 2.
2.1 Composition and Structure of Frozen
Ground
Soil Types
The term soil is used by civil engineers to describe the agglom-
eration of mineral and organic materials extending from the
ground surface down to solid rock. On the basis of their origin,
these materials can be divided into two groups: those that
resulted from chemical and physical rock weathering, and those
that are primarily of organic origin. In cold regions, weathering
is much slower and the soil blanket may be relatively thin. If the
products of rock weathering remain at the place of origin, they
constitute a residual soil. Transportation of these materials by
glacial action, wind, and/or water will alter the soil profile and
is responsible for the formation of various landforms. Bare
rock may remain in some areas except for pockets of soil cover.
20
Organic materials are formed primarily in situ, either by the
growth and subsequent decay of plants, such as peat mosses, or
by the accumulation of fragments of skeletons or shells of
organisms. Organic soils may refer to a transported soil consist-
ing of the products of rock weathering with a mixture of
decayed plant matter. Measurement of the organic content
requires separating mineral solids from the organic material.
The destruction of the organic solids by ignition (ASTM Desig-
nation D2974) is the method commonly used by engineers to
determine the organic content. When sufficient organic matter
is present to influence the soil properties, tests to measure these
engineering properties should be performed.
Soil conditions at a proposed construction site are usually
explored by means of test borings or test pits. A boring log con-
taining the name of each soil and the elevation of its bound-
aries is normally prepared. Sand and gravel properties are
described qualitatively by the terms loose, medium, and dense,
whereas those of clays are described by hard, stiff, medium, and
soft. In all cases, information that will permit identification and
classification of the frozen materials, their water and ice con-
tent, and frost susceptibility characteristics is required. Quanti-
tative information is normally obtained by means of laboratory
tests on relatively undisturbed samples or by suitable field tests
(Chapter 10). Because temperature greatly influences the prop-
erties and behavior of frozen solids, its measurement should be
included in any field investigation. Several soil types commonly
used by engineers in describing both frozen and unfrozen soils
are defined in the following paragraphs.
Bog soils are associated with tundra soils in an intricate geo-
graphic pattern. They occur in swales, on wide terraces, and on
the flat topography associated with upland positions. Bog soils
are generally formed from mixed sedge-sphagnum, and they
are commonly 0.50 to 1.20 m thick—and sometimes as much
as 9 m thick in draws. These soils are mildly to strongly acidic.
Ground ice includes ice in pores, cavities, voids, or other
openings in soil or rock, including massive ice. It may be sea-
sonal or perennial, varies greatly in amount from place to place,
depending on the type of material in which it is found, avail-
 PHYSICAL AND THERMAL PROPERTIES
ability of moisture, and rate of freezing. Massive ice describes
large masses (at least 10–100 cm), including ice wedges, pingo
ice, and ice lenses.
Ice wedge is a massive, generally wedge-shaped body with its
apex pointing downward, composed of foliated or layered, ver-
tically oriented, commonly white ice, from about 100 mm to 3
m or more wide at the top, tapering to the apex at a depth of 1
to 10 m or more. Some ice wedges extend downward to a depth
of 25 m and may have shapes dissimilar from wedges.
Muck is a dark brown to black soil, composed of a mixture
of well-decomposed organic material and fine-grained soil.
Muck has a higher proportion of silt than peat. Muskeg (or peat
land) is any terrain covered by a layer of peat.
Palsa is a round or elongated mound, with a maximum
height of about 10 m, composed of a peat layer overlying min-
eral soil. It has a perennially frozen core that extends from
within the covering peat layer downward into or toward the
underlying mineral soil.
Permafrost, or perennially frozen ground, is defined as soil or
rock having temperatures below 0 °C during at least two con-
secutive winters and the intervening summer. Moisture in the
form of water and ground ice may or may not be present.
Thickness may vary from a few centimeters to more than 100
m. Materials in a perennially frozen condition include all soils
satisfying the thermal condition.
Soil wedge is generally a wedge-shaped, downward-tapering
body of soil, different in structure (and possible texture) from
the surrounding soil. The wedge may be produced by a
repeated frost cracking and infilling with soil where no ice
wedge was ever present.
FIGURE 2-1 Mass–volume relationships for frozen and unfrozen soil.
Source: T. H. W. Baker 1991.
21
Talik is a layer or body of unfrozen ground within the per-
mafrost, “closed” when entirely surrounded and “open” when
only partially surrounded by frozen ground (see Fig. 1-8).
Till is an unstratified glacial deposit of boulders, gravel,
sand, silt, and clay. It covers the rock surface in areas that were
previously glaciated.
Varved clay consists of alternating layers of darker silty clay
and gray inorganic silt. The thickness of layers rarely exceeds 12
mm, but thicker varves may be encountered. The soil materials
were transported into freshwater lakes by meltwater from gla-
ciers. Undesirable properties of both silts and soft clays are
associated with varved clays.
Phase Relationships
Frozen soils are a four-phase system consisting of solid parti-
cles, ice, unfrozen water, and gas or air. The disappearance of
ice on thawing gives the three-phase system associated with
unfrozen soils. Soil particles (solids) are small grains of differ-
ent minerals and/or fragments of organic matter. The voids
contain ice, unfrozen water, and air in various proportions. It is
convenient to represent the different phases schematically in
terms of their volumes and masses, as is done in Fig. 2-1.
Engineering practice usually involves the measurement of
the sample volume V, the total mass M, the mass of dry solids
Ms, and the mass of water Mw = Mi + Muw. The remaining val-
ues are calculated using the appropriate mass-volume relation-
ship. Volumetric ratios can be determined directly using quan-
tities from the phase diagram (Fig. 2-1). The void ratio, e, is
defined, as
22 FROZEN GROUND ENGINEERING

Vv rs
e= = -1 (2.1-1)
Vs rd

where Vv is the volume of voids and Vs is the volume of solids.

The solids density rs and dry density rd may also be used to
compute the void ratio (frozen or unfrozen). Typical void ratios
for several unfrozen soils are given in Table 2-1 for both the
loose and dense states. The porosity n is defined as

Vv
n= (100) (2.1-2)
V
where Vv is the volume of voids and V is the total soil sample
volume. The porosity is normally expressed as a percentage.
Typical values of n are also given in Table 2-1. Porosity may be
computed using the void ratio or solid and dry densities:

e r
n= =1- d (2.1-3)
1+e rs

Now consider the mass side of the phase diagram in Fig. 2-1.
The amount of water present in a given soil volume is described
FIGURE 2-2 Density variation for water and ice.
on the basis of the dry mass of soil.
Define the total water content w by the ratio

Mw MS
w= (100) (2.1-4) rs = (2.1-7)
MS Vs
where Mw is the mass of water (frozen and unfrozen) and MS is
the mass of dry soil solids. It is convenient to represent the total Mw
rw = (2.1-8)
water content w as Vw

w = wu + w i (2.1-5)
where wu is the unfrozen water content and wi is the ice con-
tent. The natural water content for unfrozen inorganic soils is
usually less than 100%. For some organic soils, it can range up
to 500% or higher. Determination of w in the laboratory
involves measuring the total soil mass M followed by drying to
a constant mass equal to MS . By difference, obtain the mass of
water Mw = (M – MS) and compute w using Eq. (2.1-4).
Density is the ratio that relates the volume side of the phase
diagram (Fig. 2-1) to the mass side. Bulk density (frozen or
unfrozen) is defined as mass per unit volume:
M MS + M w (2.1-6)
r= = = rd (1 + w )
V V
where rd is the soil dry density. Dry density serves as a basis for
describing the degree of soil compaction. The bulk density will
depend on the air, water, and ice content of the soil as well as
the density of the solid particles. Density (kg/m3) multiplied by
gravity (g = 9.807 m/s2) gives the soil unit weight (kN/m3).
Typical wet unit weights for the unfrozen soils listed in Table 2-
1 range from 11.2 kN/m3 for a colloidal clay to 24.5 kN/m3 for a
dense silty sand and gravel till. The colloidal clay in the loose
state will contain relatively larger amounts of water. For the
solid particles and water, densities are defined as
Typical values of rs for most soil solids range from 2,500 to
2,800 kg/m3 (2.5 to 2.8 Mg/m3). It is common to use rs equal to
2.65 Mg/m3 for sand and 2.7 Mg/m3 for clay. The density of
water varies slightly with temperature. At 4 °C, water has a den-
sity close to 1,000 kg/m3 (1 Mg/m3). This value is sufficiently
accurate for most engineering work. Ice has a density close to
916.8 kg/m3 at 0 °C, with slightly lower values being more typi-
cal, depending on the air content. Ice density will increase a
small amount with colder temperatures. The freezing of water
gives the variation of densities illustrated in Fig. 2-2. The spe-
cific gravity (Gs) of soil materials (referred to in footnote a in
Table 2-1) is defined as the ratio of solids density rs to the den-
sity of water rw:
rs M
Gs = = S (2.1-9)
rw Vs rw
The value Gs = 2.6 is typical for a mixture of mineral solids and
organic fragments. Organic solids may have Gs values in the
range 1.1 to 2.5, with 1.4 being typical. Other densities com-
mon to soil engineering include the saturated density rsat and
the submerged or buoyant density r¢.
MS + M w
rsat = (Va = 0, Sr = 100%) (2.1-10)
Vt
 PHYSICAL AND THERMAL PROPERTIES 23

TABLE 2-1 Typical Values of Soil Index Propertiesa

Particle size and gradation Voidsb Unit weightc (kN/m3)
Approx. size Approx. Submerged
range (mm) range Void ratio Porosity (%) Dry weight Wet weight weight
Approx. uniform 100%
D10 coeffici- emax emin nmax nmin Min. Mod. Max. Min. Max. Min. Max.
Dmax Dmin (mm) ent, Cu loose ecr dense loose dense loose AASHO dense loose dense loose dense

Granular materials

Uniform materials
Equal spheres — — — 1.0 0.92 — 0.35 47.6 26 — — — — — — —
(theoretical values)
Standard Ottawa 0.84 0.59 0.67 1.1 0.80 0.75 0.50 44 33 14.5 — 17.3 14.6 20.6 9.0 10.8
sand
Clean, uniform — — — 1.2–2.0 1.0 0.80 0.40 50 29 13.0 18.1 18.5 13.2 21.4 8.2 11.5
sand (fine or
medium)
Uniform, inorganic 0.05 0.005 0.012 1.2–2.0 1.1 — 0.40 52 29 12.6 — 18.5 12.7 21.4 8.0 11.5
silt

Well-graded materials
Silty sand 2.0 0.005 0.02 5–10 0.90 — 0.30 47 23 13.7 19.2 20.0 13.8 22.3 8.5 12.4
Clean, fine to 2.0 0.05 0.09 4–6 0.95 0.70 0.20 49 17 13.4 20.7 21.7 13.5 23.3 8.3 13.5
coarse sand
Micaceous sand — — — — 1.2 — 0.40 55 29 11.9 — 18.9 12.1 21.7 7.5 11.9
Silty sand and 100 0.005 0.02 15–300 0.85 — 0.14 46 12 14.0 — 22.9d 14.1 24.4d 8.8 14.5
gravel

Mixed soils
Sandy or silty clay 2.0 0.001 0.003 10–30 1.8 — 0.25 64 20 9.4 20.4 21.2 15.7 23.1 6.0 13.4
Skip-graded silty 250 0.001 — — 1.0 — 0.20 50 17 13.2 — 22.0 18.1 23.7 8.3 14.0
clay with stones or
rock fragments
Well-graded gravel, 250 0.001 0.002 25– 0.70 — 0.13 41 11 15.7 22.0 23.3e 19.6 24.5e 9.7 14.8
sand, silt, and clay 1,000
mixture

Clay soils
Clay (30–50% clay 0.05 0.5m 0.001 — 2.4 — 0.50 71 33 7.9 16.5 17.6 14.8 20.9 4.9 11.2
sizes)
Colloidal clay 0.01 10Å — — 12 — 0.60 92 37 2.0 14.1 16.7 11.2 20.1 1.3 10.4
(–0.0002 mm: 50%)

Organic soils
Organic silt — — — — 3.0 — 0.55 75 35 6.3 — 17.3 13.7 20.6 3.9 10.8
Organic clay — — — — 4.4 — 0.70 81 41 4.7 — 15.7 12.7 19.6 2.8 9.7
(30–50% clay sizes)
a
Tabulation is based on Gs = 2.65 for granular soil, Gs = 2.7 for clays, and Gs = 2.6 for organic soils.
b
Granular materials may reach emax when dry or only slightly moist. Clays can reach emax only when fully saturated.
c
Granular materials reach minimum unit weight when at emax and with hygroscopic moisture only. The unit submerged weight of any saturated soil is the unit weight
minus the unit weight of water.
d
Applicable for very compact glacial till. Unusually high unit weight values for tills are sometimes due not only to an extremely compact condition but also to unusu-
ally high specific gravity values.
e
Applicable for hardpan.
Source: After Department of the Navy 1982.
24 FROZEN GROUND ENGINEERING

r¢ = rsat - rw and the total degree of saturation (ice and unfrozen water) for
(2.1-11)
frozen soil is
Unfrozen soils below the water table are normally assumed
to be saturated. Typical values for submerged unit weights of Êwr ˆG
Srf = Sri + Sru = Á i w + wu ˜ s (2.1-15)
several soils are listed in Table 2-1. For unfrozen soils the degree Ë ri ¯ e
of saturation Sr is defined as
From Eqs. (2.1-1) and (2.1-15), one can derive a relationship
Vw between dry density of a frozen soil (rdf ), ice and water con-
Sr = (2.1-12) tents (wi and wu), and the degree of saturation:
Vv
rw
where Vw is the volume of water and Vv is the volume of voids. rdf =
The degree of saturation tells what percentage of the total voids Ê 1 ˆ Ê 1.09wi + wu ˆ (2.1-16)
volume is filled with water. For dry soil, Sr = 0%, and for fully ÁË G ˜¯ + Á Srf
˜
s Ë ¯
saturated soil, Sr = 100%. The corresponding degree of satura-
tion with ice, Sri, defined as the ratio of ice volume to pore vol- Note that rw/ri = 1.09.
ume in frozen soil, is given by
Ice Phase
V wG r
Sri = i = i s w (2.1-13) The formation of ice in soil pores involves the cooling of a
Vv eri
soil-water system, as is illustrated in Fig. 2-3. The pore water
where ri is the density of ice (916.8 kg/m3). On the same vol- does not start to freeze until the temperature drops to Tsc . The
ume basis, the degree of saturation with unfrozen water is supercooled water is in a metastable equilibrium state until an
abrupt transformation of free water to ice is triggered by nucle-
Vuw wuGs ation centers. These nuclei can be aggregations of water mole-
Su = = (2.1-14) cules or soil particles. The formation of ice releases latent heat,
Vv e causing a rise in temperature to Tf , the initial freezing tempera-

FIGURE 2-3 Cooling curve for soil water and ice.

Source: Reproduced with permission of V. Lunardini from Lunardini 1981.
 PHYSICAL AND THERMAL PROPERTIES
ture. For cohesionless soils with small specific surface areas, Tf
will be close to 0 °C. For fine-grained soils (silts and clays), the
temperature depression (DT) can be as much as 5 °C.
Free water in the soil pores will now freeze at the tempera-
ture Tf . As free water changes to ice, the release of latent heat
will slow the rate of cooling until a temperature Te is reached.
All the free water and most of the bound water (unfrozen water
film on the soil particles) is frozen at Te (about –70 °C). A sig-
nificant amount of unfrozen water can exist at higher tempera-
tures for fine-grained soils with high-specific-surface areas.
Frozen soil contains ice in several forms, ranging from coat-
ings on individual soil particles and small lenses to large ice
inclusions and massive ice deposits. In soil with relatively low
ice content, the ice may not be visible or will be revealed only
by crystalline reflections on a fractured or trimmed surface.
Pore ice bonds or cements soil particles together and is
included in the phase diagram shown in Fig. 2-1. The relative
ice content is defined by an iceness ratio (ir):
Mi w - wu
ir = = (2.1-17)
Mw w Fine soil particles in the dry state tend to cohere or form
where Mi is the mass of ice, Mw the total mass of water, w the
total water content, and wu the unfrozen water content.
Ice lenses form in all soil types by the addition of water dur-
ing stationary or slow movement of the freezing front. The sup-
ply of water and ease of movement will often determine the ice
lens size or thickness. Ice lens formation is associated with frost
heave in seasonally frozen ground. These lenses are composed
of nearly pure transparent ice interlayered with soil. For slow
freezing, foreign ions contained in the pore water remain in the
melt. In perennially frozen ground, segregated ice may range
from hairline streaks to pure ice masses several meters in thick-
ness. Soil particles and small stones may appear as if suspended
in clear ice. Larger ice masses may form as a result of processes
associated with ice wedges or pingos (see Chapter 1).
Particle Size and Size Distribution
Natural soils exist in a wide range of particle sizes and shapes.
For the purposes of describing and explaining soil behavior, it
is convenient to make an arbitrary division of soils on the basis
of particle size. Bulk or gravity forces determine the engineer-
ing characteristics of coarse particles. This includes all particle
sizes that can be distinguished individually by the unaided eye,
down to about 0.05 mm. The behavior of fine particles, those
smaller than 0.05 mm, is determined predominantly by the
effects of surface, or short-range, interparticle forces. These
forces are responsible for the adhesion or cohesion that devel-
ops when water is added to fine particles. The finer-grained
materials are called cohesive soils and their complex behavior is
best explained in terms of the surface chemistry of the mineral
particles. The reader is referred to texts on clay mineralogy for
more information on this topic.
The use of particle size to classify soil requires a technique
for measurement. Each method has limits within which the
technique can operate, depending on the mechanical and
25
chemical nature of the soil particles, the sensitivity of the
apparatus, and the lower size range of the particles. The direct
method of observing individual particles and measuring their
size by comparison with a scale would quickly become very
tedious. Smaller particles could be examined using the appro-
priate electronic techniques. Seeing the particles, either
directly or electronically, permits measurement of typical
dimensions of a few particles but does not give information on
the range of particle sizes present in the soil or the proportion
of particles in any given size range. The most direct method for
separating a soil into particle-size fractions is the use of
woven-wire square mesh sieves. (Information on U.S. Stan-
dard sieves is given by ASTM Designation E-11.) These sieves
determine an equivalent diameter for the nonspherical parti-
cles that slip through the square openings. These equivalent
diameters are adequate for soil identification and comparison
with other coarse materials, especially because most cohesion-
less soils consist of roughly equidimensional, blocky particles.
(Detailed procedures for the sieve analysis are given by ASTM
Designation D-422.)
lumps that are unsuitable for sieve analysis. Typically, the soil
is washed through a nest of sieves, which must then be dried
in an oven to obtain the mass fraction remaining on each
sieve. A grain-size analysis of materials finer than 0.074 mm
(No. 200 sieve) is generally impractical. Most particle-size
analyses of fine soil, silts, and clays use the settling velocity of
particles in a fluid to determine the grain-size distribution.
The hydrometer test, which is based on Stokes’ law for spheres
falling in a viscous fluid, is used to determine the terminal fall
velocity. This velocity depends on particle diameter, density
of the suspension, and density of the fluid. Measurements of
distance and time of fall permit calculation of the equivalent
particle diameter. The percentage of particles of a certain
diameter is determined by the density of the suspension.
(Detailed procedures for the hydrometer test are given by
ASTM Designation D-422.)
The results of a sieve analysis and/or the hydrometer test are
presented in the form of a graph to make the effect of the varia-
tions in the size distribution more apparent to the eye. Equiva-
lent particle sizes are plotted to a logarithmic scale on the
abscissa, and the percentage by weight (or mass) of the total
sample—either passing (finer than) or retained (coarser
than)—is plotted arithmetically on the ordinate (Fig. 2-4). Sev-
eral typical grain-size distributions are shown in Fig. 2-4. The
gravelly sand represents a well-graded soil with a good distribu-
tion of particle sizes over a wide range. This expression implies
that the major portion (by weight) of the soil consists of sand
but that sufficient amounts of gravel are present to be notice-
able. The silty fine sand represents a uniform gradation with
most of the particles close to the same size. The particle diame-
ter whose size is greater than 10% of the particles by weight is
called the effective size (Dl0). The coefficient of uniformity (Cu) is
an approximate shape parameter and is defined as
D60
Cu = (2.1-18)
D10
26 FROZEN GROUND ENGINEERING
FIGURE 2-4 Typical grain-size distribution curves for several soils.
where D60 is the grain diameter (in mm) corresponding to 60%
passing and Dl0 is the grain diameter (in mm) corresponding to
10% passing by weight (or mass). A soil with only one grain
size would have Cu = 1. A well-graded soil may have a Cu of 15
or more. The coefficient of curvature Cz , used for soil classifica-
tion, is defined as
(D30 )2
Cz = (2.1-19)
(D10 )(D60 )
where D30 is the grain diameter (in mm) corresponding to 30%
passing by weight (or mass). The other terms were defined previ-
ously. A Cz value between 1 and 3 is typical for a well-graded soil.
Consistency of Cohesive Soils
The water content of a cohesive soil serves as an indicator of
soil behavior under certain test conditions. Identification of
parametric water contents with particle mineral composition
continues to be the most important item in the description of
fine-grained soils. In these soils, where the ratio of the surface
area of an individual particle to its volume or weight (specific
surface) is large, water and its dissolved electrolytes play an
important part through their modification of interparticle
forces. The thin film of water on the surface of these particles,
which remains unfrozen at freezing temperatures, can repre-
sent a significant portion of the total soil water content. Simple
empirical tests have been developed for the purpose of obtain-
ing the water content at certain stages of soil behavior that are
considered limiting or critical. These water contents serve to
identify qualitatively the soil and its mechanical properties.
The change in behavior of a cohesive soil with increase in
water content can be illustrated with several diagrams. The
horizontal line shown in Fig. 2-5a represents qualitatively the
water content of one cohesive soil, dry on the
left and wet on the right. At very high water
content, the soil behaves like a slurry,
deforming continuously at all levels of
applied stress. With a decrease in water con-
tent (and increase in solids concentration),
the response of the soil to applied stress grad-
ually changes. The boundary between liquid
and plastic behavior will be arbitrary,
depending on the limitations of equipment
available for stress measurement. The water
content at which the soil exhibits a small
shearing strength is taken to be the liquid
limit (wl or LL). (The liquid-limit device and
test procedure are described by ASTM Desig-
nation D-4318.)
With further reduction in the water con-
tent, the yield stress will increase until little
or no plastic flow will occur in the soil. A
small deformation corresponding to each
level of shear stress will occur, until at a high
value the sample will rupture (Fig. 2-5b).
The water content representing a change in
soil characteristics from those of a plastic to those of a brittle
material is labeled the plastic limit (wp or PL). At water con-
tents lower than the plastic limit, a clay will crumble upon
being reworked or remolded. The plastic limit test defines wp as
the water content at which a thread of soil just crumbles when
it is carefully rolled out to a diameter of 3 mm. (The test proce-
dure is described by ASTM Designation D-4318.)
Liquid limits may range from 0 to 100 or more. Most plastic
limits are less than 40. These limits will increase with lower
temperatures, as is shown in Table 2-2. Casagrande (1948) rec-
ommended that these limits (water contents) be reported with-
out the percent sign. The limits, in practice, are reported both
ways using the symbols wl (or LL) and wp (or PL) with a per-
cent sign. A. Atterberg used the plasticity index (Ip or PI),
defined as
Ip = wl – wp = LL – PL (2.1-20)
to describe the range of water contents over which a soil is plas-
tic. Engineering classification of fine-grained soils requires the
use of the plasticity index, and many engineering properties
correlate empirically with Ip. Ctori (1989) reported that a
decrease in temperature from 35 °C to 6 °C changes the classifi-
cation of a clay from low to medium plasticity, as is shown in
Table 2-2.
TABLE 2-2. Dependence of Atterberg Limits on Temperature
Temperature Liquid limit Plastic limit Plasticity index
(°C) (%) (%) (%)
6 41 20 21
20 36.5 18 18.5
35 26 15 11
Source: After Ctori 1989.
 PHYSICAL AND THERMAL PROPERTIES
FIGURE 2-5 Cohesive soil behavior over a range of water contents: (a) Atterberg limits; (b)
brittle solid; (c) plastic solid; (d) suspension of particles in water.
2.2 Soil Classification
A soil classification system that uses simple methods of identifi-
cation and gives an approximate but fairly accurate separation
into soil groups or soil types is of great convenience on any soil
engineering project. Such a system enables geotechnical engi-
neers to describe and discuss a soil (frozen or unfrozen) with
brevity and the assurance that their descriptions will mean the
same to other engineers. A classification system does not elimi-
nate the need for detailed soil investigations or soil testing for
engineering properties. By knowing the soil classification, an
engineer can anticipate the way a soil will behave in the engi-
neering situation, during construction, under structural loads,
and so on. Many soil classification systems have been proposed
during the past 60 years. From this experience, the classifica-
tion system that has evolved and is most generally used by U.S.
government engineering agencies and many geotechnical engi-
neering consulting firms and soil testing laboratories is the
Unified Soil Classification System.
Unified Soil Classification System
Since its original development by Casagrande (1948), the Uni-
fied Soil Classification System (USCS) has undergone a num-
ber of modifications to make it more applicable to dams, foun-
dations, and other construction. The USCS (Table 2-3)
separates soils into granular or cohesionless soils, fine-grained
cohesive soils, and highly organic soils, because these are the
basic divisions between soil types for engineering purposes.
Classification is performed on material passing the 75-mm
sieve. Oversized material is noted on drill logs, with particles
larger than 300 mm termed boulders and materials between 75
27
and 300 mm labeled cobbles.
Materials having 50% or more
material retained on the No.
200 (0.074 mm) sieve are
coarse-grained soils (sands and
gravels). Materials with more
than 50% passing the No. 200
sieve are fine-grained soils (silts
and clays). A secondary classifi-
cation groups the fine-grained
soils on the basis of their liq-
uid-limits and plasticity indices
(Fig. 2-6). The organic soils
and peat can generally be iden-
tified visually or by their
organic content.
The gravelly soils (G) have
more than half of the coarse
fraction larger than the No. 4
sieve (4.76 mm), and the sands
(S) have more than half smaller
than the No. 4 sieve but larger
than the No. 200 sieve (0.074
mm). Both gravels and sands
are subdivided into four sec-
ondary groups (GW and SW, GP and SP, GM and SM, GC and
SC) depending on grain-size distribution and fines in the soils.
Poorly graded (P) soils are uniform, skip, or gap graded. Well-
graded (W) soils have a good representation of all particle sizes.
The coefficient of uniformity Cu [Eq. (2.1-18)] and the coeffi-
cient of curvature Cz [Eq. (2.1-19)], along with supplementary
criteria for visual identification of soil types, are given in the
last two columns of Table 2-3.
Fine-grained soils have more than 50% passing the No. 200
sieve. They are subdivided into silts (M) and clays (C) on the
basis of their liquid-limits and plasticity indices. Organic clays
(O) and highly organic soil (Pt) are included with the fine-
grained fraction passing the No. 200 sieve. No grain size is spec-
ified, although undecomposed organic particles may be larger
than the No. 200 sieve. Using the plasticity chart (Fig. 2-6),
fine-grained soils are silts (M) if their liquid-limits and plastic-
ity indices plot below the A-line and clays (C) if the LL and PI
plot above the A-line. The organic clays (OL and OH) are
exceptions and plot below the A-line. Further subdivision of
the silt, clay, and organic fractions are based on relatively low
(L) or high (H) liquid limits, with the dividing line arbitrarily
set at 50. Note that several soil types tend to plot in about the
same area on the LL-PI chart (Fig. 2-6), indicating that these
soils tend to have about the same engineering behavior.
Changes in soil compressibility, permeability, toughness, and
dry strength are a function of LL and PI, as is indicated in Fig.
2-6.
Dual symbols are assigned to soils having between 5% and
12% passing the No. 200 sieve. Whether the coarse fraction is
well graded or poorly graded is indicated by the first symbol.
The classification GW-GC indicates a well-graded gravel with
some clayey fines. A poorly graded sand with between 5% and
28 FROZEN GROUND ENGINEERING

TABLE 2-3 Unified Soil Classification System

Primary divisions for field and Group Typical names Laboratory classification criteria Supplementary criteria for visual
laboratory identification symbol identification
Coarse- Gravel Clean gravels GW Well-graded gravels, Cu = D60 /D10, greater than 4 Wide range in grain size and
grained soils (more (less than gravel-sand mixtures, Cc = (D30)2/(D10 × D60), between 1 and 3 substantial amounts of all
(more than than half 5% of little or no finesa intermediate particle size
half of of the material is
GP Poorly graded gravels, Not meeting both criteria for GW Predominantly one size
material coarse smaller than
gravel-sand mixtures, (uniformly graded) or a range of
finer than 3 fraction is No. 200 sieve
little or no finesa sizes with some intermediate sizes
in. sieve is larger than size
missing (gap graded)
larger than No. 4 sieve
No. 200 sieve size, about Gravels with GM Silty gravels, and Atterberg limits Nonplastic fines or fines of low
size) ¼ in.) fines (more gravel-sand-silt below A-line or PI Atterberg limits plasticity
than 12% of mixtures less than 4 above A-line with
material is GC Clayey gravels, and Atterberg limits PI between 4 and 7 Plastic fines
smaller than gravel-sand-clay above A-line, and is borderline case
No. 200 sieve mixtures PI greater than 7 GM-GC
sizea)
Sands Clean sands SW Well graded sands, Cu = D60 /D10, greater than 6 Wide range in grain sizes and
(more (less than gravelly sands, little or Cz = (D30)2/(D10 × D60), between 1 and 3 substantial amounts of all
than half 5% of no finesa intermediate particle sizes
of the material is SP Poorly graded sands Not meeting both criteria for SW Predominantly one size
coarse smaller than and gravelly sands, (uniformly graded) or a range of
fraction is No. 200 sieve little or no finesa sizes with some intermediate sizes
smaller missing (gap graded)
than No. 4
Sands with SM Silty sands, sand-silt Atterberg limits Nonplastic fines or fines of low
sieve size)
fines (more mixtures below A-line or PI Atterberg limits plasticity
than 12% of less than 4 above A-line with
material SC Clayey sands, sand- Atterberg limits PI between 4 and 7 Plastic fines
smaller than clay mixtures above A-line, and is borderline case
No. 200 sieve PI greater than 7 SM-SC
sizea)
Toughness
Dry Reaction near plastic
strength to shaking limit
Fine-grained Silts and clays (liquid limit ML Inorganic silts, very Atterberg limits None to Quick None
soils (more is less than 50) fine sands, rock flour, below A-line or PI slight to slow
than half of silty or clayey fine less than 4 Atterberg limits
material is sands above A-line with
smaller than CL Inorganic clays of low Atterberg limits PI between 4 and 7 Medium None Medium
No. 200 sieve to medium plasticity; above A-line, and is borderline case to high to very
size)b gravelly clays, silty PI greater than 7 ML-CL slow
clays, sandy clays, lean
clays
OL Organic silts and Atterberg limits below A-line Slight to Slow Slight
organic silt–clays of medium
low plasticity
Silts and clays (liquid limit MH Inorganic silts, Atterberg limits below A-line Slight to Slow to Slight to
is greater than 50) micaceous or medium none medium
diatomaceous fine
sands or silts, elastic
silts
CH Inorganic clays of high Atterberg limits above A-line High to None High
plasticity, fat clays very high
OH Organic clays of Atterberg limits below A-line Medium None Slight to
medium to high to high to very medium
plasticity slow
Highly organic soils Pt Peat, muck, and other High ignition loss, LL and PI decrease Organic color and odor, spongy
highly organic soils after drying feel, frequently fibrous texture

a
Materials with 5–12% smaller than No. 200 sieve are borderline cases, designated GW-GM, SW-SC, and so on.
b
Visual: more than half of particles are so fine that they cannot be seen by the naked eye.
Source: After Department of the Navy 1982.
 PHYSICAL AND THERMAL PROPERTIES
FIGURE 2-6 Utilization of Atterberg Plasticity Limits.
Source: Reproduced from Department of the Navy 1982.
12% silty fines would be classified as an SP-SM soil. Dual sym-
bols are also used for fine-grained soils. In Fig. 2-6, the soils
defined by PI between 4 and 7 and LL between 10 and 25 are
classified as CL-ML. Other possible dual symbols are indicated
in Fig. 2-7. (ASTM Designation D-2487 gives step-by-step
charts for classifying both fine- and coarse-grained soils.)
The USCS symbols do not completely describe a soil or soil
deposit. Descriptive terms used for coarse-grained soils may
include information on stratification, degree of compactness,
cementation, moisture conditions, and drainage characteris-
tics. Along with a typical name, the description may include
approximate percentages of sand and gravel, maximum size,
angularity, hardness of the coarse grains, surface condition, and
other information. Descriptive terms used for fine-grained soil
include the degree and character of plasticity, amount and
maximum size of coarse grains, color in the wet condition,
local or geologic name, and other information. The structure,
stratification, consistency in the undisturbed and remolded
states, and moisture and drainage conditions can be added for
undisturbed soils. Definitions of soil components and fractions
used in the visual identification of soil samples are listed in
Table 2-4.
■ EXAMPLE 2.2-1: The description for a coarse-grained soil
sample reads as follows:
Gray coarse to fine sand, trace silt, trace fine gravel (SW).
29
Explain the meaning of the terms (1) coarse to fine and (2)
trace.
Solution: Refer to Table 2-4. (1) Coarse to fine means particle
sizes ranging from the No. 4 sieve (4.76 mm) to the No. 200
sieve (0.074 mm). (2) Trace refers to a percentage of 1% to
10%.
Frozen Soil Classification
The system for describing and classifying frozen soil (Table 2-5)
involves three parts. In part I, the soil phase is identified inde-
pendently of the frozen state using the Unified Soil Classifica-
tion System. Part II involves adding characteristics resulting
from the frozen state to the soil description. In part III, ice
strata found in the soil are described. This approach provides a
frozen soil description and classification that is independent of
the geologic history and mode of origin and is flexible enough
to provide any desired degree of detail. (The visual manual pro-
cedure for classifying frozen soils, presented in Table 2-5, is
similar to ASTM Designation D-4083-83.)
A description of the soil phase applies to both the thawed and
frozen states and would normally be completed independent of
the frozen state. Soil samples required for adding a description of
the frozen state depend on the subsurface conditions and design
requirements for the proposed structure or facility. For tempo-
rary structures or those that can tolerate some movement, infor-
mation derived from auger holes and cuttings may be adequate.
30 FROZEN GROUND ENGINEERING

may be shipped to a laboratory. For

undisturbed block or core samples
special precautions (see Section 10.1)
are required to prevent thermal and
mechanical disturbance or sublima-
tion of specimens (Baker 1976b).
Usual procedures involve sample
placement in de-aired and tightly
sealed plastic bags with transporta-
tion in well-insulated boxes. Crushed
ice and additional insulation or
refrigeration may be used to main-
tain temperatures close to field con-
ditions.
On examination of the frozen
samples, they are divided into two
groups: soils in which segregated ice
is not visible to the unaided eye
(designation N) and soils in which
segregated ice is visible (designation
V). Soils with the N designation, on
FIGURE 2-7 Guide for borderline cases of soil classification.
close inspection, will normally
Source: Reproduced from Howard 1977.
reveal the presence of ice within the
soil pores by crystalline reflections
For soil with high ice content, partially disturbed drive samples or by a sheen on trimmed or fractured surfaces. Two types of
may be adequate. For soil with low ice content and when struc- segregated ice include the poorly bonded or friable material
tures have heavy foundation loads or are sensitive to small move- (Nf) and the well-bonded frozen soil (Nb), in which ice
ments, undisturbed samples will be required for special consoli- cements the material into a hard solid mass. The Nf condition
dation and creep tests. As a standard procedure, samples should exists when the degree of soil saturation is low. A subgroup of
be taken to determine water and organic contents, bulk- and the Nb soils includes Nbn with no segregated ice and Nbe with
dry-unit weights, grain-size analysis, and other classification uniformly distributed excess ice. On thawing, appreciable set-
tests. Some of these tests may be completed at the site, or samples tlements may occur in the Nbe soils.

TABLE 2-4 Visual Identification of Samples: Definitions of Soil Components and Fractions
3. Fine-grained soils. Identify in accordance with plasticity charac-
1. Grain size teristics, dry strength, and toughness, as described in Table 2-3.
Material Fraction Sieve size Descriptive term Thickness
Boulders 12+ in. (300+ mm) Stratified soils
Cobbles 3–12 in. (75–300 mm) Alternating thick
Gravel Coarse ¾ to 3 in. (19–75 mm) and thin with:
Fine No. 4 to ¾ in. (4.76–19 mm) Parting 0 to 1/16 in. (0–1.6 mm)
Sand Coarse No. 10–No. 4 (2.0–4.76 mm) Seam 1/16 to ½ in. (1.6–12.7 mm)
Medium No. 40–No. 10 (0.42–2.0 mm) Layer ½ to 12 in. (12.7–300 mm)
Fine No. 200–No. 40 (0.074–0.42 mm) Stratum Greater than 12 in. (> 300 mm)
Fines (silts and Varved clay Alternating seams or layers of
clay) Passing No. 200 (0.074 mm) sands, silt, and clay
Pocket Small, erratic deposit, usually less
2. Coarse- and fine-grained soils
than 1 ft (300 mm)
Descriptive adjective Percentage requirement Lens Lenticular deposit
Trace 1–10 Occasional 1 or less per foot (300 mm) of
Little 10–20 thickness)
Some 20–35 Frequent More than 1 per foot (300 mm) of
And 35–50 thickness
Source: After Department of the Navy 1982.
 PHYSICAL AND THERMAL PROPERTIES 31

TABLE 2-5 Description and classification of frozen soils by the Unified Soil Classification Systema
Major group Subgroup
Description Designation Description Designation
(1) (2) (3) (4) (5) Field identification (6) Pertinent propertiesc (7)
Part I: Segregated ice Poorly bonded or friable Nf Identify by visual examination. To In-place temperature
Description is not visible determine presence of excess ice, use density and void ratio
of soil phaseb by eyed no excess procedure under footnote e and hand a. In frozen state
n
(indepen- ice magnifying lens as necessary. For soils b. After thawing in place
N
dent of Well bonded Nb not fully saturated, estimate degree of
Water content (total
frozen state) ice saturation: medium, low. Note
H2O, including ice)
excess ice e presence of crystals or of ice coatings a. Average
around larger partricles. b. Distribution
Part II: Segregated ice For ice phase, record the following as
Strength
Description is visible by Individual ice crystals or applicable:
Vx a. Compressive
of frozen soil eye (ice inclusions location b. Tensile
thickness is 1
orientation c. Shear
in. or less)d thickness d. Adfreeze
length
Ice coatings on particles Vc Elastic properties
spacing
Plastic properties
size
V Thermal properties
shape
Random or irregularly pattern of arrangement Ice crystal structure
Vr hardness (per Part III) (using optical
oriented ice formations
structure (per Part III) instruments)
color (per part III) a. Orientation of axes
b. Crystal size
Estimate volume of visible segregated
Stratified or distinctly c. Crystal shape
Vs ice present as percentage of total sample
oriented ice formations d. Pattern of
volume
arrangement
Part III: Ice (thickness Designate as ICEf and use descriptive Same as Part II, as
Description is greater than terms as follows, usually one item from applicable, with special
of 1 in.) Ice with soil inclusions ICE + each group, as applicable emphasis on ice crystal
substantial soil type Hardness (of mass, not individual structure.
ice strata crystals): hard, soft
ICE Structure: clear, cloudy, porous,
candled, granular, stratified
Color (for example): colorless, gray,
Ice without soil inclusions ICE
blue
Admixtures (for example): contains few
thin silt inclusions
a
Definitions: Ice coatings on particles are discernible layers of ice found on or below the larger soil particles in a frozen soil mass; they are sometimes associated with
hoarfrost crystals, which have grown into voids produced by the freezing action. Ice crystal is a very small individual ice particle visible in the face of a soil mass. Crys-
tals may be present alone or in a combination with other ice formation. Clear ice is transparent and contains only a moderate number of air bubbles (see footnote g).
Cloudy ice is translucent but essentially sound and nonpervious (see footnote g). Porous ice contains numerous voids, usually interconnected and usually resulting
from melting at air bubbles or along crystal interfaces from the presence of salt or other materials in the water, or from freezing of saturated snow. Although porous,
the mass retains its structural unity. Candled ice is ice that has rotted or otherwise formed into long columnar crystals, very loosely bonded together. Granular ice is
composed of coarse, more or less equidimensional ice crystals weakly bonded together. Ice lenses are lenticular ice formations in soil occurring essentially parallel to
each other, generally normal to the direction of heat loss and commonly in repeated layers. Ice segregation is the growth of ice as distinct lenses, layers, veins, and
masses in soils, commonly but not always oriented normal to direction of heat loss. Well-bonded signifies that the soil particles are strongly held together by the ice and
that the frozen soil possesses relatively high resistance to chipping or breaking. Poorly bonded signifies that the soil particles are weakly held together by the ice and that
the frozen soil consequently has poor resistance to chipping or breaking. Friable denotes a condition in which material is easily broken up under light to moderate
pressure.
b
When rock is encountered, standard rock classification terminology should be used.
c
These are the properties of frozen materials that may be measured by physical tests to supplement field identification.
d
Frozen soils in the N group may, on close examination, indicate the presence of ice within the voids of the material by crystalline reflections or by a sheen on frac-
tured or trimmed surfaces. However, the impression to the unaided eye is that none of the frozen water occupies space in excess of the original voids in the soil. The
opposite is true of frozen soils in the V group.
e
When visual methods may be inadequate, a simple field test to aid evaluation of volume of excess ice can be made by placing some frozen soil in a small jar, allowing
it to melt, and observing the quality of supernatant water as a percentage of total volume.
f
Where special forms of ice, such as hoarfrost, can be distinguished, more explicit description should be given.
g
The observer should be careful to avoid being misled by surface scratches or frost coating on ice.
Source: Adapted from Linnell and Kaplar (1966).
32 FROZEN GROUND ENGINEERING

N-type materials (Fig. 2-8a) contrast with V soils (Fig. 2-8b), several soils are listed in Table 2-6. For those soils not listed in
where the ice is visible as separate ice inclusions of measurable Table 2-6, Tice, Anderson, and Banin (1976) developed a sim-
dimensions. Frozen V soils include four subgroups: Vx, with ple procedure for calculating a and b on the basis of on liquid-
individual ice crystals; Vc, with ice coatings on particles; Vr, with limit data. Application of this procedure to a given soil requires
random or irregularly oriented ice formations; and Vs, for strati-
fied or distinctly oriented ice formations. When ice inclusions
become 25 mm or more in thickness, the frozen soil is given the TABLE 2-6 Unfrozen Water Content Parameters, a and b
designation ICE. Two categories include ice with soil inclusions Specific
(ICE + soil type) and ice with no soil inclusions (ICE). Because surface
the proportions of ice and soil may vary widely, it may be diffi- Soil (m2/g) a b Source
cult to decide which classification to assign a given material. An West Lebanon gravel 15 2.10 –0.408 M. Smith and Tice
estimate of excess ice present can be obtained by placing a frozen 1988
soil lump in a jar, allowing it to melt, and observing the relative Manchester silt 18 2.50 –0.515 ≤
volume of free water standing above the soil after thawing.
Kaolinite (KGa-1) 23 5.80 –0.864 ≤
Descriptive information on the ice may include hardness, struc-
ture (clear, cloudy, porous, candled, granular, or stratified), color Chena silt 40 3.20 –0.531 ≤
(colorless, gray, blue), and the presence of any soil inclusions Leda clay 58 10.80 –0.649 ≤
(silt, clay, etc.). When more details and specific information are Morin clay 60 9.50 –0.479 ≤
desired, frozen soil and/or ice samples may be tested, and proper- O’Brien clay 61 10.40 –0.484 ≤
ties may be measured as listed in column (7), Table 2-5. Goodrich clay 68 8.64 –0.456 ≤
Tuto clay 78 12.80 –0.603 ≤
2.3 Water-Ice Phase Relationships Sweden VFB 478 clay 113 27.10 –0.472 ≤
Suffield silty clay 148 11.10 –0.254 ≤
Frozen ground is defined as soil or rock having a temperature Frederick clay 159 14.0 –0.297 ≤
below 0 °C. The definition is based solely on temperature, rec-
Ellsworth clay 184 11.2 –0.293 ≤
ognizing that the water–ice phase composition of the soil or
Regina clay 291 21.1 –0.238 ≤
rock will vary with particle mineral composition, specific sur-
face area of the particles, the presence of solutes, and tempera- Niagara silt 37 6.60 –0.41 ≤
ture. The availability of a prediction equation for unfrozen Norway LE-1 clay 52 9.90 –0.523 ≤
water contents, an understanding of water-ice phase relation- Kaolinite #7 72 19.8 –0.689 ≤
ships, and information on the effect of solutes on freezing con- Athena silt loam 83 6.0 –0.301 ≤
tribute to our knowledge of frozen soil behavior relative to
Sweden CTH 201 clay 106 19.7 –0.492 ≤
engineering construction problems.
Hectorite 419 38.4 –0.369 ≤
Volcanic ash 474 3.1 –0.097 ≤
Unfrozen Water in Frozen Soil
Fairbanks silt 40 4.8 –0.326 Anderson, Tice, and
The cooling curve for soil water (Fig. 2-3) shows an ice phase McKim 1973
forming in soil pores at a freezing temperature Tf . Data Hawaiian clay 382 32.42 –0.243 ≤
reported by Anderson and Tice (1972) show that part of this Umiat bentonite 800 67.55 –0.343 ≤
water remains unfrozen in the form of thin, liquidlike layers on Wyoming bentonite 800 55.99 –0.29 Anderson and Tice
the particle surfaces. Current practice neglects the vapor phase 1972
and divides the total water content w (dry weight basis) into
Basalt 6 3.45 –1.13 ≤
two categories: unfrozen water, wu , and ice, wi . Thus the water
Morin clay 60 13.1 –0.505 Oliphant, Tice, and
content w is expressed as
Nakano 1983
w = wu + w i (2.3-1) Caen silt — 9.5 –0.227 M. Smith 1984
Tice, Anderson, and Banin (1976) have summarized experi- Calgary silt — 9.6 –0.364 Patterson and
mental unfrozen water contents for several soils with varying Smith 1981
total water contents and different physical properties. The Allendale clay — 15.7 –0.187 ≤
experimental data have been conveniently represented by a Inuvik clay — 14.5 –0.254 M. W. Smith 1985
simple power curve of the form Tomokomai clay 54 19.5 –0.305 Kay et al. 1981

wu = aqb (2.3-2) Japanese clay (45%) — 12.8 –0.402 Akagawa 1988

Note: Parameters are for the equation defined by wu = aqb where the unfrozen
where a and b are characteristic soil parameters and q is tem- water content (wu) is defined as weight of water divided by dry weight of soil
perature, expressed as a positive number in degrees Celsius expressed in percent. The term q is temperature expressed as a positive number
below freezing. Typical values for the a and b parameters for in degrees Celsius below freezing.
 PHYSICAL AND THERMAL PROPERTIES 33

FIGURE 2-8 Ice inclusions in frozen soil: (a) ice not visible; (b) visible ice less than 25 mm
thick; (c) visible ice more than 25 mm thick.
Source: Reproduced with permission from Pihlainen and Johnston 1963. Copyright 1963 National Research Council of
Canada.
34 FROZEN GROUND ENGINEERING

measurement of the water contents (wN=25 and wN=l00) corre-

sponding to N = 25 and N = 100 (where N = the number of
blows required to close the standard groove in the liquid-limit
test) and the following empirical relationships:
wu,q=1 = 0.346wN=25 – 3.01 (2.3-3)

wu,q=2 = 0.338wN=100 – 3.72 (2.3-4)

each with a correlation coefficient of 0.98. In these equations, q

FIGURE 2-9 Unfrozen water contents at –1 °C and –2 °C,
from phase composition data, versus unfrozen water contents
at –1 °C and –2 °C, from Eqs. (2.3-3) and (2.3-4).
Source: Reproduced from Tice, Anderson, and Banin 1976.
= 1 and q = 2 refer to temperatures of –1 and –2 °C, respectively.
With the two computed values of wu , the soil parameters a and
b can be evaluated for Eq. (2.3-2); thus, the complete phase-
composition curve for the given soil has been determined.
The agreement between measured and predicted wu values
for several soils, shown in Fig. 2-9, is excellent. Methods used
for measurement of the unfrozen water contents (Anderson
and Morgenstern 1973) included dilatometry, adiabatic and
isothermal calorimetry, X-ray diffraction, heat capacity mea-
surements, nuclear magnetic resonance, and differential ther-
mal analysis. The results obtained by the various methods were
remarkably consistent, although each technique involved dif-
ferent approximations, assumptions, and sophisticated equip-
ment not available to field engineers. Phase composition curves
for several representative soils are shown in Fig. 2-10. Unfrozen

FIGURE 2-10 Phase composition curves for five representative soils.

Source: Reproduced with permission from Anderson and Morgenstern 1973. Copyright 1973 National Academies
Press.
 PHYSICAL AND THERMAL PROPERTIES 35

water contents computed on the basis of Eq. (2.3-2) are consid-

ered adequate for many engineering applications, provided that
the soils being considered do not contain excessive amounts of
soluble salts.

■ EXAMPLE 2.3-1: Unfrozen water contents are required for a

light brown clay. A liquid-limit test provided the following data:
No. of blows, N 10 15 30 50 60 90
Water content (wt %)38.5 35.2 29.3 25.0 23.5 20.0
(a) Determine the soil parameters a and b for this soil.
(b) Estimate the unfrozen water content at a temperature of
–4 °C.
FIGURE 2-11 Liquid limit test data plot for Example 2.3-1.
Solution:
(a) Plot log N versus w (Fig. 2-11) and determine: wN=25 =
30.7% and wN=100 = 19.10%. Using Eqs. (2.3-3) and (2.3-4), entirely liquid until –2.1 °C (point A) is reached. Below –2.1 °C,
compute ice crystals will form that are separate from the liquid. The low-
est temperature at which a solution will remain completely liq-
wu,q=1 = 0.346(30.7) – 3.01 = 7.61% uid is called the eutectic temperature (point C), and the solution
wu,q=2 = 0.338(19.10) – 3.72 = 2.736% that possesses this lowest freezing point is called the eutectic
composition. The eutectic temperature for the H2O–NaCl sys-
Now write Eq. (2.3-2) in the form tem of Fig. 2-12 is –21 °C at about 23.3% NaCl. The eutectic
log wu = log a + b log q temperature for an H2O–CaC12 system is –51 °C; hence, CaCl2
is often used to prevent ice formation on highway surfaces in
and cold climates.
The crystal lattice for ice is very selective; no substitutes for
log 7.61 = log a + b log 1
hydrogen or oxygen atoms are accepted (Pounder 1965). If an
log 2.736 = log a + b log 2 aqueous salt solution is frozen slowly, the foreign ions remain
in the melt and pure ice is formed. During freezing, impurities
Using the first equation, solve for a = 7.61. Substitution into
will be rejected and concentrated at grain boundaries, which
the second equation gives
reduces the cohesion between ice grains. For field freezing con-
ditions, Konrad and McCammon (1990) suggested that all sol-
-0.44430
b= = -1.4759 utes in soil pore water are rejected. The formation of an ice
0.30103 cover on the sea surface is a refining process in which most of
(b) Using Eq. (2.3-2), compute the unfrozen water content at the salt is rejected (Weeks and Ackley 1982). The resulting
–4 °C: brine inclusions affect the properties of sea ice. The combina-

wu = 7.61(4)–1.4759 = 0.98%

Effect of Solutes on Freezing

Water within soil pores may contain dissolved salts which
increase the freezing-point depression and will increase the
unfrozen water content. The presence of dissolved salts results
in a reduced soil frost susceptibility under seasonal temperature
conditions due to a decrease in the freezing index and an
increase in the thawing index. A review by Ladanyi (1989)
showed that increased salinity reduced the ice content, thereby
reducing the frozen soil strength and increasing its creep rate at
a given temperature. It is convenient to illustrate changes in the
water solution using solubility curves. A two-component sys-
tem of water and salt (NaCl) is illustrated in Fig. 2-12. The sol-
ubility of H2O in a brine solution decreases with lower temper-
atures and at an intermediate composition has a freezing
temperature lower than that of pure ice (0 °C). This can be ver-
ified by a simple experiment. A salt and water solution with
3.5% NaCl can be cooled to less than 0 °C and will still be FIGURE 2-12 Solubility of ice in NaCl brine.
36 FROZEN GROUND ENGINEERING
tion of water and salts also affect the weathering of rocks, soils,
and building materials in cold climates (Thurmond and Brass
1987).
Banin and Anderson (1974) described how the presence of
salts in soil pore water lowers the freezing-temperature depres-
sion. As the water freezes, solutes are forced into a smaller and
smaller volume of solution. The freezing temperature, Tn, for a
salinity, Sn, can be estimated (Patterson and Smith 1983) using
the formula:
Sn
Tn = Ti +
Ê wu ˆ (2.3-5)
AÁ ˜
Ë w¯
where Sn is salinity in g/1 (or ppt) of NaCl, A equals –17.04 (g/
1)/°C, Ti is the temperature at which unfrozen water in nonsa-
line soil equals wu, and w is the total soil water content. This
method is useful when the relationship of wu versus tempera-
ture is known for a given soil. (The salinity Sn can be deter-
mined by ASTM Test Method D-4542-85 for “Pore water
extraction and determination of the soluble salt content of soils
by refraction.”) A simpler method for estimating the tempera-
ture shift, DT, due to salinity Sn uses an empirical equation
developed by Velli and Grishin (1983). According to these
authors, DT can be determined from
Ê Sn ˆ heat extraction exceed the rate of heat supply (water flow) to
DT = Tk Á (2.3-6)
Ë 1, 000 + Sn ˜¯
where Sn is salinity in g/1 [or parts per thousand (ppt)] and Tk
is a reference temperature equal to 57 °C for sea salt, 62 °C for
NaCl, and 32.5 °C for CaC12. Note that for Sn equal to 30 ppt
and NaCl, Eq. (2.3-6) gives DT equal to 1.80 °C.
■ EXAMPLE 2.3-2: A nonsaline soil with Ti = –2 °C has a
ratio wu/w = 0.3125. For a salinity Sn = 10 g/l, at what tempera-
ture will the same unfrozen water content occur?
Solution: Substitution into Eq. (2.3-5) gives
10
Tn = -2 - = -3.88 ∞C
17.04(0.3125)
For thaw-settlement problems (Chapter 4), a major portion
of the total settlement is due to the thawing of ice, both pore ice
and ice lenses. Because an increase in salinity will decrease the
ice-fraction and thaw settlement, any change in the ice fraction
becomes important in design. The ice fraction (or iceness ratio,
ir) for a saline soil (Ono 1975) can be estimated from the relation
Sn Ê 54.11 ˆ
ir = 1 - Á1 - ˜ (2.3-7)
1, 000 Ë T ¯
where Sn is salinity in ppt and T is the soil temperature. Because
the ice fraction is defined as the ratio of ice content, wi , to the
total soil water content, one can write
wi w (2.3-8)
ir = =1- u
w w
The unfrozen water content ratio, wu/w, is then
wu S Ê 54.11 ˆ
= n Á1 - ˜ (2.3-9)
w 1, 000 Ë T ¯
The temperature at which all pore water remains unfrozen at a
given salinity corresponds to wu /w = 1 in Eq. (2.3-9).
2.4 Soil Frost Action
Frost action involves a combination of frost heave during a
downward advance of the freezing front followed by a loss of
strength during the spring thaw. Heaving is the result of ice seg-
regation during the freezing process and of the formation of
alternating bands of soil and ice (Fig. 1-16). When ice melts,
the soil mass usually cannot immediately reabsorb all the water.
The result is weaker soils after thawing than before freezing.
These topics are addressed in the following sections.
Frost Action Process
The general unsteady heat flow near the ground surface, cou-
pled with conditions of crystal ice nucleation and growth, is the
necessary condition for the formation of alternating bands of
soil and ice (Fig. 1-16). In addition, it is essential that the rate of
the freezing front. For illustration, consider a homogeneous
fine-grained soil column (Fig. 2-13) with a zone of frozen soil,
an active ice lens, a freezing fringe, and an underlying zone of
unfrozen soil. The fringe is a region of impeded flow caused by
partial filling of soil pores by ice (Nixon 1991). The soil skele-
ton, within the fringe, will expand when pressure in the ice
exceeds the overburden pressure plus any pressure required to
initiate separation of the soil skeleton. With sufficient ice pres-
sure, the soil skeleton separates and a new ice lens forms. Dur-
ing slow freezing, near steady-state temperature profiles are
maintained in the freezing fringe and unfrozen zones (Nixon
1992). With a rapid advance of the frost front (unsteady heat
flow) and a decrease in temperature, the permeability of the
frozen fringe (Fig. 2-13) will decrease, causing a reduction in
the flow of water to the ice lens. This interrelation of heat and
water flow results in finer ice lenses near the surface and thicker
ice lenses at greater depth, where the temperature gradient is
smaller. At some time t > 0, the temperature curve (Fig. 2-13)
with different slopes shows the unsteady heat flow situation
near the ground surface. The temperature gradients
dT dT
Gf = > = Gu (2.4-1)
dz f dz u
show the rate of heat extraction above the frost line to be
greater than the rate of heat supply below the frost line. The
frozen condition is represented by the subscript f, the unfrozen
state by u, and depth by z. With ice forming, the volume of the
soil-water system must increase and/or the soil consolidate.
When water moves to the frost line and freezes, the volume
 PHYSICAL AND THERMAL PROPERTIES
FIGURE 2-13 Frost heave in an idealized one-dimensional soil column.
Source: Reproduced with permission from Nixon 1992. Copyright 1992 Canadian Geotechnical Journal.
increases more than that necessary for a change in density
(water to ice). Frost heave occurs at the frost line, with T £ 0 °C.
Later developments (Konrad and Morgenstern 1980; Nixon
1991, 1992) indicate that the ice lens forms above the 0 °C iso-
therm and that there is a frozen fringe between the ice lens and
the unfrozen soil. In the frozen fringe, liquid water exists in
equilibrium with ice at a temperature below the normal freez-
ing point of water, with the segregation-freezing temperature
located at the base of the growing ice lens.
At the frost line, usually assumed to be at 0 °C, a heat bal-
ance equation can be written:
Ê dT ˆ Ê dT ˆ
kf Á ˜ - ku Á = net heat flux = qz (2.4-2)
Ë dz ¯ f Ë dz ˜¯ u
where kf and ku are the frozen and unfrozen thermal conductiv-
ities and dT/dz is as defined above. The net heat flux qz can
freeze the in situ water and/or the water arriving at the frost
line; hence
dz
qz = Ln + Lv (2.4-3)
dt
where n is the soil porosity, L the latent heat of the water, dz/dt
the rate of frost-front advance, and v the velocity of water
arrival at the frost line. The formation of ice at T < 0 °C induces
a pressure deficiency (suction gradient) across the frozen fringe
and in the unfrozen soil below. The velocity of water flow is
dependent on this pressure deficiency and on the overall per-
meability of the unfrozen soil and frozen fringe (in accordance
37
with Darcy’s law). Heave results when the water freezes and
forms ice lenses. The heave rate (dh/dt) can be expressed as
dh dz
= 1.09v + 0.09n (2.4-4)
dt dt
where the phase expansion is assumed to be 9%. Heave predic-
tion requires a relation between the velocity of arriving water
(v) and the frost-front advance rate (dz/dt). Konrad and Mor-
genstern (1980) described the heat and mass transfer relation-
ships involved in evaluating v and dz/dt relative to experimental
studies on a Devon silt. There is evidence that an advancing
frost front can establish a condition such that all heat extrac-
tion (qz) goes to growing an ice lens. For this case
dz (2.4-5)
= 0 and qz = Lv
dt
This condition requires (1) a low rate of heat extraction, (2) a
readily available water supply, and (3) a highly frost susceptible
soil (silt). A silt soil with permeability equal to 10–5 cm/s can
supply enough water to give a heave rate close to 27 mm/day
(Fig. 2-14). This upper-bound approach to frost heave predic-
tion appears to be justified for some design conditions in highly
frost susceptible soils.
The frost action process also involves thaw weakening dur-
ing the spring warm-up illustrated in Fig. 1-7. Thawing can
proceed from the top down and/or from the bottom up. Dur-
ing a sudden spring thaw, melting will proceed almost entirely
from the surface downward. Downward drainage will be
38 FROZEN GROUND ENGINEERING

FIGURE 2-14 Frost susceptibility of soils on the basis of soil type and particle size.
Source: Reproduced from Departments of the Army and Air Force USA 1985.
 PHYSICAL AND THERMAL PROPERTIES 39

blocked temporarily by the remaining frozen soil, with only lat-
eral and surface drainage possible. The initial thawing of ice
lenses provides an excess of water that leads to high pore-water
pressures. Before this water can escape into more pervious lay-
ers of the pavement system or be reabsorbed into adjacent drier
areas, the soil strength will decrease, with a severe reduction in
pavement bearing capacity. With continued warm-up (Fig. 1-7)
the frozen layer disappears, any excess pore-water pressures dis-
sipate, and the pavement bearing capacity returns to summer
values. Thaw behavior of frozen ground relative to settlement
and other design problems is described in Chapter 4.
Frost Susceptibility of Soils
A frost-susceptible soil is defined in terms of its frost-heaving
and thaw-weakening behavior. Both can cause considerable
damage to engineering structures, the former during freezing
and the latter during thawing. Frost heave is not necessary for
thaw weakening. Some clay soils develop segregated ice (and
hence thaw weakening) while exhibiting little or no heave.
Shrinkage or consolidation of layers adjacent to an ice lens can-
cels the heave normally associated with ice segregation, particu-
larly where the water supply is limited and soil permeability is
low. Frost-susceptibility index tests permit evaluation of the
potential for frost heaving and thaw weakening of subgrade
soils and unbound base and subbase materials for roads, rail-
roads, and airfields.
In a survey of transportation departments throughout the
world, Chamberlain et al. (1984) found that most agencies have
developed their own frost-susceptibility index criteria based on
laboratory tests and that these criteria fail to discriminate
between marginally frost-susceptible material and that which is
frost susceptible. As part of intensive research on the effects of
frost action on pavements, Johnson et al. (1986) selected the
U.S. Army Corps of Engineers frost design and soil classification
system. This system in its present form is shown in Table 2-7.
Three levels of screening are required: (I) the percentage of
particles smaller than 0.02 mm, (II) soil type based on the Uni-
fied Soil Classification System, and (III) a laboratory freezing
test. Soil frost-susceptibility ratings based on this system
include six categories: negligible, very low, low, medium, high,
and very high. These ratings are shown on the ordinate of Fig.
2-14 with average limits for rate of heave. The simplest rating
(based on level I tests) is the classification of negligible frost
susceptibility given to gravels with less than 1.5% finer than
0.02 mm and sands with less than 3% finer than 0.02 mm. All
soils failing this criterion require complete soil classification
tests (level II). Gravels with 1.5% to 3% finer than 0.02 mm and
sands with 3% to 10% finer than 0.02 mm also require a labo-
ratory frost heave test (level III). The range of possible degrees
of frost susceptibility is very wide for most soils, as is illustrated
in Fig. 2-14. For this reason, the Corps of Engineers procedure
recommends that a freezing test be performed when precise
information on soil frost susceptibility is required.

TABLE 2-7 U.S. Army Corps of Engineers Frost Design Soil Classification System
Amount finer than
Frost susceptibilitya Frost group Kind of soil 0.02 mm (wt %) Typical soil type under USCSb
Negligible to low NFSc a. Gravels 0–1.5 GW, GP
b. Sands 0–3 SW, SP
d
Possibly PFS a. Gravels 1.5–3 GW, GP
b. Sands 3–10 SW, SP
Low to medium S1 Gravels 3–6 GW, GP, GW-GM, GP-GM
Very low to high S2 Sands 3–6 SW, SP, SW-SM, SP-SM
Very low to high F1 Gravels 6–10 GM, GW-GM, GP-GM
Medium to high F2 a. Gravels 10–20 GM, GM-GC, GW-GM, GP-GM
Very low to very high b. Sands 6–15 SM, SW-SM, SP-SM
Medium to high F3 a. Gravels >20 GM, GC
Low to high b. Sands except very fine silty sands >15 SM, SC
Very low to very high c. Clays, Ip > 12 — CL, CH
Low to very high F4 a. All silts — ML, MH
Very low to high b. Very fine silty sands >15 SM
Low to very high c. Clays, Ip > 12 — CL, CL-ML
Very low to very high d. Varved clays and other fine-grained — CL and ML; CL, ML, and SM; CL,
banded sediments CH, and ML; CL, CH, ML, and SM
a
Based on laboratory frost-heave tests.
b
G, gravel; S, sand; M, silt; C, clay; W, well graded; H, high plasticity; L, low plasticity.
c
Non-frost susceptible
d
Requires laboratory frost-heave test to determine frost susceptibility.
Source: Johnson et al. 1986.
40 FROZEN GROUND ENGINEERING

The laboratory frost heave test called

for under level III screening involves soil
specimens 152 mm high, frozen at a pen-
etration rate of approximately 6.3 mm/
day, with free water at 3.3 °C continu-
ously available at the base of the speci-
men. Specimens are compacted to 95% or
better of applicable standard, except
undisturbed clays. Soil saturation before
freezing is generally 85% or greater.
Improvements on the freezing test, rec-
ommended by Johnson et al. (1986), are
illustrated in Fig. 2-15. This equipment
includes a rubber-membrane-lined,
multi-ring freezing cell to minimize side
friction, liquid-cooled cold plates for pre-
cise top and bottom temperature control,
and a data acquisition and control system
for automated temperature control and
data processing. The new test uses two
freeze-thaw cycles to account for changes
in susceptibility to frost heave caused by a
prior freeze-thaw cycle. Four samples are
tested, each 150 mm in diameter and 150
mm in height. Samples are compacted to
field density and moisture conditions.
Water is freely available through the
porous base plates. Five days are required
to complete the test. The heave rate at the
end of the first 8 hours of each 2-day
freeze-thaw cycle is used as an index of
frost-heave susceptibility. A California
Bearing Ratio (CBR) test, conducted after
the second thaw, provides an index of
thaw-weakening susceptibility. Both indi- FIGURE 2-15 Schematic of new freezing test equipment.
ces (frost heave and thaw weakening) are Source: Reproduced from Johnson et al. 1986.
used to determine the frost susceptibility
of a soil.
Preliminary frost-susceptibility classification criteria for gives the uniformity coefficient Cu > 6; therefore, soil 2 has the
the new freezing test are shown in Table 2-8. They are based group symbol SW, a well-graded sand.
on the average frost-heave rate during the first 8 hours of the Using Table 2-7, soil 2 with less than 3% finer than 0.02 mm,
first freeze and the CBR after two cycles of freezing and based on the grain-size curve, gives a frost group designation
thawing. NFS, non-frost-susceptible.
An alternative approach for evaluating frost susceptibility
■ EXAMPLE 2.4-1: Using screening levels I and II, determine and heave involves determining the segregation potential for
the degree of frost susceptibility for the two soils described in the soil in question. Konrad and Morgenstern (1980) have pro-
Table 2-9. posed a detailed model for one-dimensional frost heave. They
Solution: Soil 1 has more than 50% passing the No. 200 suggested that after an ice lens has formed, the frozen soil above
sieve; therefore, it is fine grained. Using Fig. 2-6, LL = 47%, IP = the ice lens does not participate in mass transport, but that
47% – 36% = 11%, and no organic content gives the soil group water is transported to the ice lens from the unfrozen soil
symbol ML. through a thin zone of partially frozen soil referred to as the fro-
Using Table 2-7, soil 1 has a frost group designation F4, low zen fringe (Fig. 2-13). The driving force arrives from suction
to very high frost susceptibility. A laboratory frost-heave test generated at the ice-fringe interface, and the fringe impedes
would be recommended for soil 1 as part of screening level III. flow to the lens because of its low permeability. They give a
Soil 2 has only 4% finer than the No. 200 sieve; therefore, it simple linear analysis of the frozen fringe, assuming that the
is coarse grained. A plot of percent passing versus log particle Clausius–Clapeyron equation is valid at the base of the ice lens,
size gives Dl0 ª 0.18 mm and D60 ª 0.48 mm. Computation that water flow is continuous across the frozen fringe, that the
 PHYSICAL AND THERMAL PROPERTIES 41

TABLE 2-8 Frost Classification System Based on Heave Rates TABLE 2-9 Soil Descriptions for Example 2.4-1
and CBR
Sieve size Soil 1, % finer Soil 2, % finer
Heave rate Thawed Frost susceptibility Probable frost No. 4 97 98
(mm/day) CBRa classification group
No. 10 93 91
<1 >20 Negligible NFS, PFS
No. 20 — 72
<2 >15 Very low S1, PFS
No. 40 80 57
<4 >12 Very low S2, PFS
No. 100 — 29
<6 >10 Low F1
No. 200 55 4
<8 >6 Medium F2
0.005 mm 24 —
<16 >3 High F3
0.001 mm 5 —
>16 <3 Very high F4
Note: Soil 1 has LL = 47% and PL = 36%; Soil 2 is nonplastic.
a
CBR represents the California Bearing Ratio (percentage of initial unfrozen CBR).
Source: Chamberlain, Janoo, and Ketcham 1996.
zen interface pu, the segregation freezing temperature Ts , and
the overall hydraulic conductivity in the frozen fringe (Konrad
fringe can be characterized by an overall permeability kfo , and
1987).
that the temperature in the frozen fringe varies linearly between
The segregation potential, once evaluated at near steady-
the segregation freezing temperature Tso at the lens and the
state conditions and under a negligible overburden pressure,
freezing temperature of bulk water Ti at the bottom of the
may be considered as an index property of a soil that uniquely
fringe (Fig. 2–16).
characterizes its frost-heave susceptibility. Although the SP is
According to Konrad and Morgenstern (1981), when a soil
usually determined at a constant suction at the frost front, the
sample freezes under different cold-side step temperatures but
relationship between SP and pu can be determined by applying
the same warm-side temperature, the water intake flux vo at the
Darcy’s equation to the unfrozen zone once the permeability k
formation of the final ice lens is proportional to the tempera-
and the velocity of moisture migration v are known (Konrad
ture gradient in the frozen fringe (Fig. 2-17):
and Morgenstern 1981; Morgenstern 1981; Konrad 1988),
vo = SP grad T (2.4-6)
lu (2.4-7)
The proportionality factor SP has been termed the segregation pu = v
k
potential. Its value, given by the slope of the straight line in Fig.
2-17, was found to be a function of the total suction potential at
the freezing front pw , the suction potential at the frozen-unfro-

FIGURE 2-17 Relation between water intake velocity and

FIGURE 2-16 Characteristics of the frozen fringe: (a) simpli- temperature gradient across the active system during the for-
fied and (b) actual shape. mation of the final ice lens.
Source: Reproduced with permission from Konrad and Morgenstern 1981. Source: Reproduced with permission from Konrad and Morgenstern 1981.
Copyright 1981 Canadian Geotechnical Journal. Copyright 1981 Canadian Geotechnical Journal.
42 FROZEN GROUND ENGINEERING

FIGURE 2-19 Segregation potential versus suction at the frost

front for Devon silt.
Source: Reproduced with permission from Konrad and Morgenstern 1981.
Copyright 1981 Canadian Geotechnical Journal.

where X is the frost depth and n is the porosity reduced to

account for the percentage of in situ pore water that will not
freeze. For field conditions, Nixon (1987) modified Eq. (2.4-9)
FIGURE 2-18 Conceptual relationship between water intake for prediction of the heave rate (dh/dt) to
flux and temperature gradient for different suctions at the
frost front.
dh 1.09SP* ◊ G f
Source: Reproduced with permission from Konrad and Morgenstern 1981. =
dt L ◊ SP* (2.4-10)
Copyright 1981 Canadian Geotechnical Journal. 1+
kf
where lu is the length of flow in the unfrozen soil. When the where SP* = Vff / Gff is the segregation potential, Vff the pore-
suction at the frost front increases, the segregation potential water velocity in the frozen fringe, Gff the thermal gradient in
decreases with a concomitant decrease in the slope of the rela- the fringe, Gf the thermal gradient in the frozen soil, L the volu-
tionship vo versus grad T. For steady-state conditions, soil freez- metric latent heat of water, and kf the frozen soil thermal con-
ing and formation of the final ice lens is characterized by a set ductivity.
of straight lines (vo versus grad T) passing through the origin Using an analysis similar to that of Gilpin (1980), Nixon
(Fig. 2-18). A correlation between SP and pu is shown in Fig. (1991) has shown that the hydraulic conductivity parameters
2-19 for Devon silt. A detailed procedure for experimentally required for heave prediction can be extracted from a set of
determining the segregation potential has been described by frost-heave test data. The fringe pore-water velocity Vff per-unit
Konrad (1987). temperature gradient (approximately equal to the SP* parame-
The segregation potential of a given freezing soil decreases ter) in a one-dimensional test can be written as follows:
with increasing applied pressure. Konrad and Morgenstern
(1984) expressed the influence of a surcharge by an empirical -1
Ï È Ê 1+ A ˆ ˘
A ¸
power law, Vff Ô
= SP* = Ì L ÍÎ
(1.09 po - pu )ÁË A ˜¯ ˙ Ô (2.4-11)
˚ ˝
SP = SP0 exp(–ape) (2.4-8) G ff Ôk + (1+ A ) Ô
Ó f ko B ˛
where SP0 is the value of SP obtained for zero applied pressure,
pe is the applied pressure (or effective overburden pressure), where L is the volumetric latent heat of water, kf is the thermal
and a is a soil constant. Data available for the SP parameter conductivity of frozen soil, A is the slope of the relationship
(Nixon 1987) show a range of 0 to 70 mm2/s · °C for silty sands, between k and (–T) on a log–log plot, and B is a thermody-
sandy silts, and tills for the range of pressures normally namic constant equal to 12,400 cm water pressure/°C. The a
encountered in engineering practice. For clays and clayey silts, and b in Nixon’s (1991) paper have been replaced with A and B,
the SP parameter ranges from 50 to 200 mm2/s · °C. According respectively, so as to avoid confusion with a and b in Eq. (2.3-
to Konrad and Morgenstern (1982), heave (h) under field con- 2). Nixon (1991) describes the hydraulic conductivity of frozen
ditions can be predicted using the relation soil in the fringe by the two-parameter expression

dh dX ko
= 1.09SP ◊ G f + 0.09n (2.4-9) k= (2.4-12)
dt dt (-T )A
 PHYSICAL AND THERMAL PROPERTIES
where ko is the hydraulic conductivity at –1 °C in centimeters
per second and T is temperature in degrees Celsius. A log–log
plot of (Gff / Vff – L/kf ) versus (1.09po – pu) has a slope A with an
intercept equal to [(A + 1)/A]A/koB(l+A). The pressures po and pu
have units of centimeters of water or g/cm2. Nixon (1992)
reported that the ratio L/kf is approximately constant for many
soils with a value of about 0.002 (mm2/day · °C)–1. The equa-
tion derived by Nixon (1991) for SP in terms of more funda-
mental parameters helps explain the dependency of SP on
parameters such as overburden pressure, suction, and hydraulic
conductivity for different soil types. The segregation potential
concept has been used successfully to predict the frost-heaving
effect related to chilled pipelines and artificial ground freezing
(Konrad and Morgenstern 1984; Nixon 1982, 1992; Jessberger,
Jagow, and Jordan 1988).
Frost-Heave Forces
Foundations embedded in frost-susceptible soils can be sub-
jected to large uplift forces resulting from frost heaving of the
soils. Forces are transmitted to the foundation by vertical forces
acting on the underside of a foundation (Fig. 2-20a), by lateral
forces acting behind walls (Fig. 2-20b), or by transfer of uplift
forces through soil frozen to the sides of a foundation (Fig. 2-
20c). With frost penetration down both sides of the foundation
wall, tangential adfreeze (side grip) can become considerable,
FIGURE 2-20 Frost heave effects on structures: (a) upward
thrust on the foundation underside; (b) lateral thrust behind
walls; (c) adfreeze and uplift on the sides of a foundation.
Source: Reproduced from Linell and Lobacz 1980.
43
as is indicated in Table 2-10. When frost-susceptible soils
freeze, the heave forces increase as the frost line penetrates the
unfrozen soil. It is difficult to predict the magnitude of these
mobilized heave forces because of the many variables involved.
These variables include soil type and heterogeneity, variation of
soil temperature with time and depth, rate of freezing, avail-
ability of water, foundation surface type (smooth versus rough,
coated or uncoated wood, steel, or concrete), overburden pres-
sure, and foundation loads. Observations based on the weight
of buildings known to have been lifted by frost-heaving soil
indicate forces approaching 760 kPa. Penner (1970) measured
basal forces in excess of 1,800 kPa on a 300-mm-diameter
anchored steel plate. The development of force on the plate rel-
ative to frost penetration and time is illustrated in Fig. 2-21.
Placing the foundation below the depth of seasonal frost pene-
tration and preventing soil adfreeze on the foundation sides
avoids the development of these forces. For this reason, in areas
of seasonally frozen ground, local building codes specify a
required footing depth.
For structures where heaving problems are anticipated and
cannot be avoided, field studies and laboratory experiments to
determine design values for frost-heave forces or heave rates
may be necessary. An indication of the limiting adfreeze heav-
ing forces acting on a pile or column can be obtained using
Dalmatov’s equation (Tsytovich 1959):
F = Lha(c – 0.5bTm) (2.4-13)
where F is the total upward force due to frost heaving (kgf), L
the perimeter of the foundation in contact with frozen soil
(cm), ha the thickness of the frozen zone (cm), Tm the mini-
mum soil temperature (°C) in the frozen zone, and b and c
parameters determined experimentally. For icy silty clayey soils
(loams and sandy loams), Russian workers (Tsytovich 1975)
reported average values of 40 to 70 kPa for c and 10 to 19 kPa
for b. Penner and Irwin (1969) reported that Dalmatov’s equa-
tion agreed closely with data from field experiments for ground
temperatures down to about –12 °C. Good judgment must be
exercised in evaluating local soil conditions and selecting the
appropriate parameters b and c. For design, it is important to
recognize that the heave force will be decreased significantly by
relaxation of stress due to soil creep or by a small displacement
TABLE 2-10 Lifting Force Transmitted by Tangential Adfreeze
(Side Grip) to Unprotected Piles or Piers with Diameters of
200–350 mm in Dry Crust Clay
Lifting force with side grip (kN) for
freezing index, F100 (h · °C)
Material 10,000 30,000 50,000
Steel 50 100 140
Wood 30 50 60
Concrete 30 80 100
Concrete wall (per meter
length) 25 40 50
Note: The values serve as a guide. F100 represents the maximum freezing index
occurring in 100 years.
Source: Torgersen 1976b.
44 FROZEN GROUND ENGINEERING

available to change the soil structure

may arise from phase expansion
(about 9%) of water contained in
the soil pores. Some of this expan-
sion pressure is relieved by heaving
of the soil in the direction of least
resistance. During the process, soil
structural changes occur at the frost
front where pore ice forms.
Hydraulic barriers for waste-
containment structures typically
involve compacted fine-grained
soils. These barriers serve to limit
infiltration of surface water that
would contribute to leachate gener-
ation, the escape of gas that might
kill vegetation or create an explo-
sion hazard, or limit leakage of pol-
lutants from the waste containment
to an acceptable value. Examples
include soil covers or liners for
landfills and hazardous waste sites.
In cold regions, exposure of these
barriers to seasonal surface temper-
atures normally involves the forma-
tion of an active layer to depths of 2
m or more in some locations. Meth-
ods for predicting the maximum
freezing depths are given in Section
3.2. A related increase in permeabil-
FIGURE 2-21 Frost depth and heave force measurements on a 305-mm circular plate in
ity above specified limits [1 × 10–7
Leda clay.
cm/s for municipal wastes (Daniel
Source: Reproduced with permission from Penner 1970. Copyright 1970 Canadian Geotechnical Journal.
and Benson 1990)] during the ser-
vice life of the barrier would be con-
sidered failure.
of the foundation unit. Maximum forces are transmitted when Tentative criteria for placement densities of clay covers and
the foundation is rigidly anchored. liners which allow for freeze–thaw effects on permeability have
been outlined by Chamberlain and Ayorinde (1991). The upper
bound for dry unit weight in Fig. 2-22 is defined by the line of
Freeze-Thaw Effects on Permeability
100% saturation (zero air voids curve). The lower bound is the
Repeated freezing and thawing of clayey soils will produce an line of 80% saturation and a horizontal line that starts at the
increase in the effective void ratio. Konrad (1989) has reported point where the water content corresponding to the plastic
that this leads to a reduction in the segregation potential (SP) limit intercepts the upper bound. This horizontal line extends
after each freeze-thaw event and to an increase in vertical to the left until it intercepts the line of 80% saturation. Data
hydraulic conductivity (permeability) of the thawed soil by fac- points for a silty clay representing modified, standard, and
tors of 2–10. Similar results on change in vertical permeability reduced Proctor densities are shown by triangles, squares, and
have been reported by Benson and Othman (1993), Chamber- circles, respectively, in Fig. 2-22. The shaded portion corre-
lain and Ayorinde (1991), and Chamberlain, Iskandar, and sponds to the water content and dry densities suitable for those
Hunsicker (1990). Konrad (1989) defined the segregation portions of the containment barrier that will be subjected to
potential as the ratio of water-intake rate to the temperature seasonal freezing and thawing.
gradient in the frozen soil near the frost front. The changes in
hydraulic conductivity occurred primarily during the first three
freeze–thaw cycles (Konrad 1989; Chamberlain and Ayorinde 2.5 Thermal Properties
1991). For clayey silts, Konrad (1989) inferred that no struc-
tural changes occurred in the frozen fringe but somewhere in The response of soil materials to thermal changes—the addi-
the colder zone at a temperature between –0.40 and –0.57 °C, tion or subtraction of heat, the raising or lowering of tempera-
dependent upon the overconsolidation ratio of the soil. Forces ture—requires an understanding of their thermal properties;
 PHYSICAL AND THERMAL PROPERTIES 45

The amount of heat transferred by conduction in soil

FIGURE 2-22 Illustration of placement density criteria for silty
clay subjected to freezing and thawing.
Note: Solid symbols represent tests with permeability £1×10–7 cm/s; shaded
area represents acceptable zone (62.43 lbm/ft3 = 1000 kg/m3).
Source: Reproduced from Chamberlain and Ayorinde 1991. Copyright 1991
American Society of Civil Engineers.
thermal conductivity, heat capacity, thermal diffusivity, latent
heat, and thermal expansion (or contraction). These thermal
parameters vary with temperature, soil type, water and/or ice
content, degree of saturation, and soil density.
Thermal Conductivity
Heat conduction in soils involves a transfer
of kinetic energy from molecules in a warm
part of the mass to those in a cooler part.
Considering a prismatic element of soil with
a cross-sectional area A, the rate at which
heat is transferred by conduction is given in
the form
increases as dry density increases and as its degree of saturation
increases. Similarly, the process of freezing and thawing leads to
soil compositional and structural changes with consequent
changes in the soil’s thermal conductivity. Thermal conductivi-
ties of some of the soil constituents are given in Table 2-11.
They vary greatly, with quartz having the greatest thermal con-
ductivity and air the least. Volumetric proportions of the vari-
ous soil constituents will influence the effective soil thermal
conductivity. Various methods for calculating the thermal con-
ductivity of soils have been reviewed by Farouki (1981).
Johansen’s (1975) method generally gives the best results for
unfrozen and frozen soils, coarse or fine, at degrees of satura-
tion above 0.1.
The method developed by Johansen (1975) and summarized
in Table 2-12 is applicable to unfrozen and frozen mineral soils.
The main equation for computing thermal conductivity k is
ku = (ksat – kdry)Ke + kdry (2.5-3)
where ksat and kdry are the saturated and dry thermal conductiv-
ities, respectively. The Kersten number Ke is defined as follows:
coarse-grained unfrozen soil with the degree of saturation Sr >
0.05,
Ke = 0.7 log Sr + 1.0 (2.5-4)
and fine-grained unfrozen soil with Sr > 0.1,
Ke = log Sr + 1.0 (2.5-5)

dT (2.5-1)
Q = -ku A
dx
and

Q dT (2.5-2)
q= = -ku = kui
A dx
where Q/A = q is the rate of heat flow per
unit area (J/m2 · s), ku is the unfrozen ther-
mal conductivity* (J/s · m · K or W/m · K),
dT/dx = i is the thermal gradient (°C/m), and
A is the area (m2). The minus sign indicates
heat flow from a high to a low temperature.
The quantities in Eq. (2.5-1) are illustrated in
Fig. 2-23. The soil element must be large
enough by comparison to represent a homo-
geneous soil.

*Other units have been used for thermal con-

ductivity, such as mcal/cm · s · °C, Btu-in./ft2-hr-
FIGURE 2-23 Heat flow through a soil element.
°F, and Btu/ft-hr-°F.
46 FROZEN GROUND ENGINEERING

TABLE 2-11 Thermal Properties of Selected Materials

kdry (W/m ◊ K) = 0.039n -2.2 ± 25% (2.5-7)
Heat Thermal
Density capacity conductivity where n is the soil porosity. For saturated soils, Johansen (1975)
Material (kg/m3) (kJ/kg · °C) (W/m · K) Sourcea observed that variations in microstructure had little effect on
Air, 10 °C 1.25 1.00 0.026 1 thermal conductivity. He proposed the use of a geometric mean
Water, 0 °C 999.87 4.2177 0.56 2, 3 equation based on thermal conductivities of the soil constitu-
ents and their respective volume fractions. For saturated unfro-
10 °C 999.73 4.1922 0.58 3
zen soils, this gives
Ice, 0 °C 900 2.09 2.21 2
–40 °C 900 2.09 0.08 2 ksat = ks1-n kwn (2.5-8)
Snow, loose 85 2.09 0.08 2
and for saturated frozen soils containing some unfrozen water
compacted 500 2.09 0.7 2
wu ,
Polystyrene, foam 30 1.25 0.035 5
n -wu wu (2.5-9)
Polyurethane, foam 32 1.67 0.024 5 ksat = ks1-n ki kw
Rock wool 160 0.84 0.039 2, 3
Taking the thermal conductivity of ice, ki = 2.2 W/m · K and kw
Glass wool 64 0.84 0.042 2, 3
= 0.57 W/m · K, Eq. (2.5-9) reduces to
Straw, compressed 360 0.09 4
Wood ksat = ks1-n (2.2)n (0.269)wu (2.5-10)
Plywood, dry 600 2.7 0.17 5 Johansen (1975) suggested the use of a geometric mean equa-
Fir or pine, dry 500 2.5 0.12 5 tion to determine ks :
Maple or oak, dry 700 2.09 0.17 5
Concrete ks = kqq ko1-q (2.5-11)
Sand and gravel 2,200 0.89 1.3–1.7 4
aggregate
where kq and ko are the thermal conductivities of quartz and
other minerals, respectively, and q is the quartz fraction of the
Lightweight 1,880 0.74 4
total solids content. Johansen (1975) used kq = 7.7 W/m · K and
aggregate
ko = 2.0 W/m · K. For coarse-grained soils with a quartz content
Concrete, asphalt 2,050– 1.05–1.52 2, 4 of less than 20%, Johansen (1975) used ko = 3.0 W/m · K to
2,150
account for the probable mineral composition of such soils.
Quartz 2,660 0.733 8.4 1 Johansen’s (1975) method for computation of the soil ther-
Granite 0.8 1.7–4.0 4 mal conductivity is a technique for interpolation between dry
Limestone 2,640 1.3–5.0 4 and saturated values. It does not take into account possible
Shale 1.5 4 moisture migration at intermediate degrees of saturation. Two
Sandstone 1.8–4.2 4
examples (crushed gravel and a silty clay) of determining ther-
mal conductivity for unfrozen and frozen mineral soils are
Steel 7,500 0.5 43 2, 5
shown in Figs. 2-24 and 2-25, respectively.
Iron, ductile 7,500 51 2, 5 Natural soils will vary in composition over relatively short
Aluminum 2,700 0.88 156–190 2, 4 distances; hence average thermal conductivities are appropriate
Copper 8,950 0.42 386 2, 4 for many thermal problems. General usage describes soils in
Note: Values are representative. Most materials have some variation with den-
terms of cohesionless materials (sands and gravels), fine-
sity and temperature. grained soils (silts and clays), and highly organic soils (peat).
a
1, De Vries 1966; 2, Alter 1969; 3, Weast, Astle, and Beyer 1984; 4, Johnston Sands and gravels normally are high in quartz, with silts and
1981; 5, Lunardini 1981. clays composed of other minerals. Kersten (1949) conducted
tests on natural soils and crushed rock from which empirical
equations for thermal conductivity k were developed. Farouki
For all types of frozen soil, Ke = Sr (with a variation less than (1981) converted these equations to SI units with k (W/m · K)
±Ke = 0.1). For dry natural soils, Johansen (1975) developed expressed in terms of water content w (%) and dry density rd
the following semi-empirical equation for kdry . (g/cm3). For coarse-grained soils (silt-clay content < 20%)
which are predominantly quartz, the thermal conductivity for
0.137rd + 64.7 unfrozen (+4 °C) and frozen (–4 °C) soils, respectively, is
kdry (W/m ◊ K) = ± 20% (2.5-6)
2, 700 - 0.947rd k = 0.1442 (0.7 log w + 0.4 )(10)0.6243rd (2.5-12)
where rd is the dry density (kg/m3) and the solids density is and
2,700 kg/m3. Johansen (1975) observed that crushed rock
materials gave higher thermal conductivity values: k = 0.01096(10)0.8116rd + 0.00461(10)0.9115rd w (2.5-13)
 PHYSICAL AND THERMAL PROPERTIES 47

TABLE 2-12 Method for Calculating the Thermal Conductivity of Mineral Soils
k = (ksat – kdry) Ke + kdry (a) Main equation

Natural (0.137gd + 64.7)/(2700 – 0.947gd) (b)

Dry conductivity
–2.2
Crushed 0.039n (c)

Coarse 0.7 log Sr + 1.0 (d)

Unfrozen
Fine log Sr + 1.0 (e) Kersten number

Frozen Sr (f)

Unfrozen 0.57nks1–n (g)

Saturated conductivity
Frozen 2.2nks1–n · 0.269 wu
(h)

7.7q · 2.01–q (i)

Particle conductivity
q 1–q
q < 0.20 Coarse 7.7 · 3.0 (j)

Source: Johansen 1975.

Eqs. 2.5-12 and 2.5-13 are plotted in Fig. 2-26 with thermal
conductivity values selected on the basis of water content, dry
density, and degree of saturation. Kersten’s (1949) equations do
not take into account variations in quartz content; hence for
soils having a composition between the sand group and the silt-
clay group, interpolation for k values based on quartz content
was suggested.
For fine-grained soils (50% or more silty-clay), the thermal
conductivity for unfrozen and frozen soils, respectively, is
k = 0.1442(0.9 log w - 0.2)(10)0.6243rd (2.5-14)
that of water, measured in the same units. Soil consists of vari-
ous constituents, including solids, water, ice (if frozen), and air.
The heat capacity of the soil can be computed by adding the
heat capacities of the different constituents in a unit mass of
soil. Let ms , mw , mi , and mair represent the mass fractions and
cs , cw , ci , and cair—the heat capacities of solids, water, ice, and
air, respectively—in a soil volume V with total mass m. The
heat capacity of the soil is
1
c (kJ/kg װC) =
m
(cs ms + cw mw + ci mi + cairmair ) (2.5-16)

and Dividing by V and ignoring the very small air term gives the
volumetric heat capacity of the soil:
k = 0.001442(10)1.373rd + 0.01226(10)0.4994rd w (2.5-15)
cv (MJ/m3 װC) = cm rf = rdf (c s + cw wu + ci wi ) (2.5-17)
Unfrozen water contents in the frozen soils are not accounted
for. Kersten (1949) reported that these equations give thermal where rf and rdf are the bulk and dry densities of the frozen
conductivity values that differed from measured values by less soil, respectively, and wu and wi are the unfrozen and frozen
than ±25 %. Farouki (1981) stated that this is sufficient for water contents, respectively.
most practical applications because natural soil properties in Using the specific heat of a material defined as the ratio of its
the field can vary from point to point to a similar extent due to heat capacity to that of water in degrees Celsius, volumetric
lack of soil homogeneity. Harlan and Nixon (1978) summa- heat capacities for mineral (unfrozen and frozen) soils can be
rized much of Kersten’s (1949) work in Figs. 2-26, 2-27, and 2- calculated as
28. These figures can be used to evaluate the soil thermal con-
ductivity on the basis of soil type, dry density, water content, Ê r ˆÊ w ˆ
and degree of water saturation. cvu = Á d ˜ Á 0.17 + 1.0 ˜ cvw (2.5-18)
Ë rw ¯ Ë 100 ¯

Heat Capacity
The heat capacity* [(J/g)/°C] of a soil sample is the amount of *In some references, the heat capacity per unit mass of the sub-
heat required to raise its temperature 1 degree. The specific stance is referred to as the specific heat. Other units used for heat
heat of a material is defined as the ratio of its heat capacity to capacity include (cal/g)/°C or (Btu/lb)/°F.
48 FROZEN GROUND ENGINEERING

FIGURE 2-24 Thermal conductivity of unfrozen mineral soils.

Source: Reproduced from Johansen 1975.
 PHYSICAL AND THERMAL PROPERTIES 49

FIGURE 2-25 Thermal conductivity of frozen mineral soils.

Source: Reproduced from Johansen 1975.
50 FROZEN GROUND ENGINEERING

FIGURE 2-26 Average thermal conductivity for sands and gravels: (a) frozen; (b) unfrozen.
Source: Reproduced with permission from Harlan and Nixon 1978. Copyright 1978 McGraw-Hill.

FIGURE 2-27 Average thermal conductivity for silt and clay soils: (a) frozen; (b) unfrozen.
Source: Reproduced with permission from Harlan and Nixon 1978. Copyright 1978 McGraw-Hill.
 PHYSICAL AND THERMAL PROPERTIES 51

FIGURE 2-28 Average thermal conductivity for peat: (a) frozen; (b) unfrozen.
Source: Reproduced with permission from Harlan and Nixon 1978. Copyright 1978 McGraw-Hill.

Ê r ˆ ÈÊ 1.0wu ˆ Ê w - wu ˆ ˘ 0.96M s 0.04M s 0.13M s

cvf = Á d ˜ ÍÁ 0.17 + + 0.5 Á c (2.5-19) cvu = (1.88) + (2.52) + (4.187)
Ë rw ¯ ÎË 100 ˜¯ Ë 100 ˜¯ ˙˚ vw rmV roV rwV
= 1.72 MJ/m3 ◊ ∞C
where cvw = 4.187 MJ/m3 · °C and rd and rw are the unit mass of
= volumetric heat capacity of the soil
the dry soil and water, respectively. The specific heats 0.17, 1.0,
and 0.5 correspond to mineral soil, water, and ice, respectively. For comparison, neglecting the organic content and using Eq.
For organic soils (peat), replace the specific heat for mineral (2.5-18) gives cvu = 1.66 MJ/m3 · °C.
soils (0.17) with that for organic soils (0.40). The correspond- In fine-grained frozen soils (silt and clays), the liquid-solid
ing mass heat capacities (cm) are given by the relationship phase change is gradual with a decrease in temperature. As the
unfrozen water film on soil particles changes to ice, the release
cv cv of latent heat will indicate a heat capacity different from that
cm = = (2.5-20)
r rd (1 + w ) defined by Eq. (2.5-16). This apparent heat capacity (ca) can be
expressed as the sum of the heat capacities for the different soil
where r is the total soil density. constituents plus a term to account for the latent heat that is
given off on cooling or absorbed on warming:
■ EXAMPLE 2.5-1: A soil with porosity of 49% has a moisture
content of 13% (dry weight basis). The solid material consists 1 T2 ∂wu (2.5-21)
of 4% organic matter and 96% soil minerals (dry weight basis). ca = c s + ci (w - wu ) + cu wu +
DT ÚT
1
L
∂T
dT
Compute the unfrozen volumetric heat capacity of this soil.
Solution: The porosity n = Vv /V = 0.49. Let V = 1.0 m3; then where cs , ci , and cu are the heat capacities of the mineral solids,
Vv = 0.49 m3 and Vs = 1 – Vv = 0.51 m3 = Vo + Vm where Vo and ice, and unfrozen water, respectively; wu = aqb the unfrozen
Vm are the volumes of the organic and mineral solids, respec- water content; T the temperature; and L the latent heat of the
tively. Substitution for Vo and Vm gives liquid-solid phase change. Typical values for a and b for several
soils are given in Table 2-6.
0.04M s 0.96M s
Vs = 0.51m3 = +
ro rm ■ EXAMPLE 2.5-2: For Wyoming bentonite at a water content
of 100%, calculate the apparent heat capacity at a temperature
where ro = 1.3 Mg/m3 and rm = 2.65 Mg/m3 (Table 2-11). Solve of –10 °C.
for Ms = 1.323 Mg for V = 1.0 m3. Note that the unfrozen volu- Solution: Rewrite Eq. (2.5-21) in the form
metric heat capacity can be expressed as cvu = xm cvm + xo cvo +
xw cvw , where xm , xo , and xw are the volumetric fractions of the È Lb ˘
mineral, organic solids, and water, respectively. Substitution ca = c s + ci w + Í(cu - ci ) - ˙ aqb
Î q ˚
into the expression for cvu gives
52 FROZEN GROUND ENGINEERING

TABLE 2-13 Thermal Diffusivities of Several Materials TABLE 2-14 Site Data for Example 2.5-3
Thermal diffusivity, ␣ Observed annual Observed lag
Material [(m2/s) × 10–7] Source Depth (cm) temperature range (°C) (days)
Copper 1,133.00 Terzaghi 1952 0 28.2 0
Iron 173.00 Terzaghi 1952 30 23.4 10.6
Dolomite 78.00 Terzaghi 1952 60 19.5 21.5
Quartzite 45.00 Terzaghi 1952 120 13.5 42.9
Shale 31.00 Terzaghi 1952 300 4.6 106.0
Limestone 27.00 Terzaghi 1952 500 1.3 176.5
Granite 15.00 Terzaghi 1952 700 0.4 247.0
Sandstone 10.00 Johnston 1981
Ice 11.20 Terzaghi 1952
Dense saturated sand 8.00 Terzaghi 1952 Solution: The temperature amplitude at depth z is given by
Eq. (1.2-3). For depths of 60 cm and 120 cm,
Soft saturated clay 4.00 Terzaghi 1952
Fresh snow 3.30 Terzaghi 1952 Ê p ˆ
Az 1 = As exp Á -z1
a p ˜¯
Dry soil 2.50 Terzaghi 1952
Ë
Water 1.40 Terzaghi 1952
Organic solids 1.00 De Vries 1966 and
Air 0.21 De Vries 1966
Ê p ˆ
Az 2 = As exp Á -z2
Ë a p ˜¯
From Table 2-6, obtain a = 55.99 (use 0.5599, so that wu is a Solve for the thermal diffusivity:
weight fraction rather than a percent) and b = –0.29. For bento-
2
nite, cs = 0.71 J/g · °C, and from Table 2-11 obtain ci = 2.094 J/g · È z2 - z1 ˘ È 120 - 60 ˘
°C, and cw = 4.187 J/g · °C. Use L = 333.7 J/g. Substitution gives pÍ ÊA ˆ˙ p Í
a= Í z1 ˙ = Ê 19.5 ˆ ˙ = 9.54 cm2 /h
p Í ln Á ˜ ˙ 8, 766 h Í ln ÁË ˜˙
A ÍÎ 13.5 ¯ ˙˚
Ï
ca = 0.71 + 2.09(1.0) + Ì(4.187 - 2.094) -
333.7(-0.29) ¸ Î Ë z2 ¯ ˚
˝
Ó 10 ˛ An alternative solution involves the equation for time-lag t =
¥ 0.5599(10) -0 .29 = 6.18 J/g ◊ ∞C (z/2)(p/pa)1/2. For depths of 60 cm and 120 cm, solve for a :

2 2
Thermal Diffusivity p Ê z2 - z1 ˆ 365.25 Ê 120 - 60 ˆ
a= = Á ˜
4 p ÁË t 2 - t1 ˜¯ 4 p Ë 42.9 - 21.5 ¯
The rate at which heat is transferred in a soil mass is dependent
on the thermal conductivity k. The rise in temperature that this cm2 cm2
= 228.55 = 9.52
heat will produce will vary with the heat capacity c and the bulk day h
density r of the soil mass. The ratio of these quantities
Latent Heat of Fusion
Ê m2 ˆ k Ê W/m ◊∞C 1 ˆ
aÁ = (2.5-22)
Ë s ˜¯ c r ÁË kJ/kg ◊∞C kg/m3 ˜¯ The amount of heat energy absorbed when a unit mass of ice is
converted into a liquid at the melting point is defined as its
is defined as the soil thermal diffusivity. Typical values for ther- latent heat of fusion.* The same amount of heat (333.7 kJ/kg) is
mal diffusivities, which are listed in Table 2-13, show that a for liberated when the water is converted into ice with no change in
ice is much higher than that of water. For this reason, the diffu- temperature. For soils, the total energy involved in the phase
sivity of frozen soil will be much higher than that of the same change process will depend on the total water contained in a
soil in the thawed condition. This means that the average tem- given soil volume and the fraction of this water that changes
perature of a mass of saturated frozen soil will increase more phase. For a given soil,
quickly than that of a mass of unfrozen soil with equal dimen-
sions at an equal difference between the initial temperature of w - wu (2.5-23)
L = rd L ¢
the soil mass and that of the surrounding soil. 100

■ EXAMPLE 2.5-3: Long-term ground temperature observa-

tions at a site gave the data shown in Table 2-14. Compute the *Latent heat of fusion for water at 0 °C: 333.7 kJ/kg, 79.7 cal/g, or
soil thermal diffusivity for the depth 60 to 120 cm. 143.4 Btu/lb.
 PHYSICAL AND THERMAL PROPERTIES 53

where L (kJ/m3) is the soil volumetric latent heat of fusion, L¢ =

333.7 kJ/kg is the mass latent heat for water, rd (kg/m3) is the
dry soil density, w is the total water content, and wu is the
unfrozen water content (percentage dry mass basis) of the fro-
zen soil. For those soils (sands and gravels) with little or no
unfrozen water, the wu term will be very small. For many prac-
tical problems, the assumption that wu is zero will give accept-
able L values for estimation purposes.

Thermal Expansion (or Contraction)

Decreasing air temperatures will cause an ice cover and frozen
surface soils to contract; the amount is dependent on the coeffi-
cient of thermal contraction, a, and the rate and magnitude of
temperature change. In lakes and reservoirs, the results
observed are long tension cracks that quickly fill with water and
freeze. On warming, thermal expansion of a continuous ice
sheet creates large ice pressures and the potential for damage to
marine and hydraulic structures. Frost cracking, the fracturing
of soil surface layers by thermal contraction, creates major
maintenance problems for highways and the potential for FIGURE 2-29 Thermal expansion coefficient for frozen sand
openings in landfill covers. The design of storage facilities for compared with those for ice and quartz.
liquefied natural gas (LNG) in underground rock and clay for-
mations requires knowledge of their thermal properties at tem-
minerals than of ice. Adhesion between sand grains and pore ice
peratures down to –160 °C (the boiling point of LNG ª –165
is substantially decreased at a temperature of about –60 °C.
°C).
The thermal behavior of Boom clay in the horizontal (paral-
The coefficient of thermal expansion (contraction), a, may
lel to the layers) and vertical directions during cooling at 1 °C
be defined as the rate of change of length L with respect to tem-
per minute from about +2 °C to –140 °C is illustrated in Fig. 2-
perature T, per unit length:
30. The Boom clay is a stiff, overconsolidated, fissured clay
1 dL d e from Belgium, with thin horizontal sand layers. Its typical
a= = (2.5-24) properties are given in Table 2-15. Conversion of water to ice
Lo dT dT
causes an initial large increase in vertical strain with a concur-
where Lo is length at some reference temperature, dL/Lo = de, rent smaller decrease in horizontal strain. Vertical expansion
and e is the strain. If the material is isotropic (i.e., exhibits the continues down to –50 °C as unfrozen water is converted to ice.
same thermal expansion in every direction), a = av /3, where av At this point, the vertical strain begins to show a small decrease
is the volume coefficient of thermal expansion. For ice, Drouin in strain down to –140 °C. The horizontal strain appears to
and Michel (1971) have reported that a varies with tempera- remain relatively constant below about –6 °C. Dilatometer tests
ture q(°C) = 0 – T (°C) according to the relation on kaolinite (Landva and Ladanyi 1991) showed a behavior
similar to the Boom clay and the vertical axis. Peak expansion
a(°C)–1 = (54 – 0.18q ) × 10–6 (2.5-25) occurred at about –6 °C, with no further expansion due to
–6 –1 freezing of water.
Note that for an average a = 52 × 10 (°C) , an ice cover 1 km
long will expand 104 cm for a 20 °C temperature rise. To com-
pensate for this expansion, shore structures must exert consid- TABLE 2-15 Typical Properties of Boom Clay
erable resisting forces. Ice ramparts will form along shore wnat = 25.5% e= 0.65
beaches, and pressure ridges will develop in the ice itself.
wl = 56% gnat = 20.1 kN/m3
Thermal expansion coefficients for frozen sand are compared
with those for ice and quartz in Fig. 2-29. Values for ice (Jakob wp = 30% gd = 16.0 kN/m3
and Erk 1928) are in reasonable agreement with the linear rela- Sr = 96 to 100% Gs = 2.73
tionship of Eq. (2.5-25) for temperatures down to about –80 °C. Mineralogical
The addition of sand (quartz) to ice lowers the coefficient. composition (%) Grain-size distribution (%)
Expansion coefficients reported by Al-Moussawi (1988) for a Vermiculite 30 d > 0.2 mm 0.5
saturated sand (bulk density ª 2.27 Mg/m3) are close to those Illite 25 0.2 mm > d > 0.06 mm 4.5
shown for test SDL-1 with T > –60 °C. At about –60 °C, the
Smectite 20 0.06 mm > d > 0.002 mm 42
coefficient for frozen sand undergoes a significant drop and sub-
Other 25 0.002 mm > d 53
sequently decreases continuously to minimal values at –170 °C.
Below –60 °C, the coefficient is more representative of quartz Source: After Landva and Ladanyi 1991.
54 FROZEN GROUND ENGINEERING
FIGURE 2-30 Strain versus temperature relationship for Boom clay.
Source: Landva and Ladanyi 1991.
Problems
2.1 The water content of a fully saturated silty sand is 30% and
the soil solids have a specific gravity of 2.68. Calculate:
(a) The total density before and after freezing.
(b) The volume increase (%) of the soil sample, assuming that
no water is expelled, all the water is frozen, and the sample
has no access to additional water.
2.2 A frozen soil sample has a mass of 650 g with an unfrozen
water content of 10%. After thawing and oven drying, the sam-
ple’s mass is reduced to 420 g.
(a) Determine the total water content (w), the ice content (w –
wu), and the iceness ratio (ir).
(b) Assuming full saturation, calculate the void ratio and the
degree of ice saturation. The specific gravity of soil solids
is 2.68.
2.3 A frozen fine sand sample has a mass of 448.7 g. After
thawing and drying in an oven at 110 °C, the sample has a mass
of 356.7 g. The specific gravity of the soil solids is 2.65. Assume
100% saturation.
(a) Determine the water content of the soil sample.
(b) For the unfrozen condition and no drainage, determine
the void ratio.
(c) For the frozen condition and an ice density
of 916 kg/m3, compute the void ratio.
2.4 Information on density change with trans-
formation of ice to water is given in Fig. 2-2.
Compute the percentage decrease in the volume
for ice on melting.
2.5 Unfrozen water contents are required for a
gray silty clay. Data from the liquid limit test
include the following:
No. of blows, N 13 17 21 29
Water content
(wt %) 45.5 44.4 43.5 41.8
(a) Determine the soil parameters a and b in
Eq. (2.3-2).
(b) Estimate the unfrozen water content at –4 °C.
2.6 A gravel soil with rd = 1.20 Mg/m3 and w =
15% changes temperature from +10 °C to –10
°C. Discuss quantitatively the changes in ther-
mal conductivity, heat capacity, and latent heat
of the soil.
2.7 The dry mineral and organic matter in soils
have heat capacities of about 0.84 and 1.88 J/g ·
°C, respectively. Weight fractions of mineral,
organic, and water in a soil sample are 0.50,
0.20, and 0.30, respectively.
(a) Estimate the unfrozen mass heat capacity
for this soil sample.
(b) If the soil has a dry density of 1.1 Mg/m3,
estimate the unfrozen volumetric heat
capacity.
2.8 Excavation for new construction in a permafrost area
involves removing frozen sand with a volume of 15 m by 15 m
by 15 m deep. Data on the sand include an average soil temper-
ature of –9 °C, dry density of 1.6 Mg/m3, and a water content
of 10%. How many kilowatt-hours of electrical energy would
be required to raise the temperature of this soil volume to 1 °C?
(Recall that 1 J = 2.778 × 10–7 kWh.)
2.9 Consider the annual surface temperature for Dunbar,
Michigan, to be the sinusoidal curve shown in Fig. 2-31. Calcu-
late the temperature of a homogeneous gravel deposit with rd =
1.60 Mg/m3 and w = 5% at a depth of 0.6 m in January and July.
Consider the system to be without phase-change effects.
2.10 The freeze wall used to protect an excavation adjacent to an
existing power plant has the cross section shown in Fig. 2-32.
The site was underlain by a deep deposit of sand with traces of
gravel. The water table varied with the seasons and river flow.
An average level was close to elevation 269.7 m. Soil properties
included w = 10% and r = 1,602 kg/m3 above the water table,
and w = 22% and r = 1,922 kg/m3 below the water table. The
average ground temperature was close to 10 °C. Compute the
heat to be extracted from the soil in joules for a 1-m wall section
to lower the ground temperature from +10 °C to –10 °C.
 PHYSICAL AND THERMAL PROPERTIES 55

FIGURE 2-31 Annual ground surface temperature, Dunbar, Michigan, for Problem 2.9.

FIGURE 2-32 Freeze wall cross-section for Problem 2.10.