Construction Ground Freezing: 6.1 Design Considerations
Construction Ground Freezing: 6.1 Design Considerations
Construction Ground
Freezing
l
ria
ate
Controlled ground freezing for construction and mining applica- tural underpinning, containment of hazardous waste, and a
tions has been in use for more than a century. Frozen ground may variety of special projects. Planning and executing these
be used to provide ground support, groundwater control, or projects involves getting information on the geometry of the
structural underpinning during construction. The frozen earth excavation or frozen barrier; soil and groundwater conditions
dM
wall, which is constructed prior to excavation, for practical pur- at the site; proximity of adjacent streets, utilities, and struc-
poses eliminates the need for sheeting of the earth, site dewater- tures; and known characteristics of the freezing method. These
ing, soil stabilization, or concern for movement of adjacent topics are covered in subsequent sections.
ground. It is a versatile technique that involves the use of refriger-
ation to convert in situ soil pore water into ice. The ice becomes a
Ground Freezing Applications
bonding agent, fusing together adjacent particles of soil or blocks
of rock to increase their combined strength and make them
hte
regardless of structure, grain size, or permeability. The around each freeze pipe. With continued heat extraction, the
mechanical properties of frozen ground are more dependent on frozen soil columns increase in diameter until they merge and
time and temperature than on the geology of the strata. Hence form a frozen wall. This frozen barrier (Fig. 6-1a) serves as a
it is less sensitive to advance geologic prediction than other retaining wall and permits soil excavation within the dashed
alternative construction methods. Only lateral groundwater lines. Utility lines in the area of soil freezing are appropriately
py
flow requires additional considerations. Freezing may be used insulated to prevent either freezing of the utility or thawing
for any size, shape, or depth of excavation, and the same refrig- within the frozen zone. The surface view of a shaft 6.7 m in
eration plant can be used from job to job despite a wide varia- diameter by 31.0 m deep immediately after excavation is shown
tion in these factors. The actual direct costs of freezing for a in Fig. 6-2. Protective insulation on the walls has not yet been
Co
specific project—excluding the contractor’s capability—will placed. Coolant supply and return connections to vertical
depend largely on ground conditions, including groundwater freeze pipes are clearly visible.
flow and impurities, spacing of freezing elements, time avail- From the given geometry of the structure to be constructed
able, and type of refrigeration system used. In this chapter, we and the available space on site, the designer must select a struc-
describe the design considerations, freezing methods and tural system for the frozen earth wall. Because of the relatively
installation, structural design, monitoring requirements, and high compressive and low tensile strengths of frozen soil, curved
other construction considerations. arch walls, particularly circular walls, are a good solution, as is
illustrated in, Fig. 6-lb. In general, when possible, a circular,
elliptical, or arched frozen wall should be chosen. An ellipse can
6.1 Design Considerations be employed effectively for rectangular structures if the ratio of
length to width does not exceed about 2.0 (Braun, Shuster, and
A review of typical applications of ground freezing includes Burnham 1979). The sections below provide information on the
shafts, deep excavations, tunnels, groundwater control, struc- structural calculations needed for wall design.
137
l
ria
FIGURE 6-2 Exposed frozen earth immediately after excava-
tion of a 6.7-m-diameter by 31.0-m-deep shaft.
Source: Courtesy of Joseph A. Sopko, Jr., Layne-Northwest.
ate
dM
hte
Elsevier.
If space or other restrictions prevent the use of curved sec- Ground freezing application to tunneling appears to be
tions, the designer may choose other structural elements. A related primarily to the construction of relatively thin frozen
py
gravity wall (Fig. 6-3a) has the advantages that it can be con- soil masses around the outer tunnel contours, as is shown in
structed in a straight line and that the area inside the adjacent Fig. 6-4. Three possibilities include support of the roof portion
excavation is free of any obstructions. A disadvantage of the of an excavation, roof and sides, and the closed circular ring.
gravity wall is that the soil volume that must be frozen for any Special technology has been developed to construct the tunnel
Co
given depth of excavation is from two to more than five times as freeze wall in stages. Freeze pipes are placed starting from a
much as that required for a structurally curved wall. The design shaft or from a cavern location in an excavated tunnel section
of a gravity wall is governed by overturning and sliding criteria, (Fig. 6-5). This leads to a fanlike placement of freeze pipes sur-
not stress in the frozen wall. An anchored wall (Fig. 6-3c) has rounding each tunnel section. The freeze pipes are slightly
the same advantages as the gravity wall. Its design is highly inclined outward with reference to the tunnel axis. To avoid
sophisticated, and it requires very careful field control. The fro- gaps in the frozen wall, the drilling tube direction is controlled
zen wall may be damaged or destroyed by unanticipated water by a boring gauge, as is shown in Fig. 6-5. This equipment per-
flow during drilling of the anchors through the frozen soil, mits the placement of freeze pipes with sufficient accuracy for
which together with localized thawing may result in unloading tunnel sections up to 60 m in length. Jessberger (1980) reported
of anchors and unacceptable movement of adjacent unfrozen that precise borings of up to 115 m are possible with a steerable
soil. Braun, Shuster, and Burnham (1979) indicated that the drilling bit (Fig. 6-5). A permanent liner of concrete or prefab-
tieback anchor design is too sensitive to be used reliably in the ricated structural elements is placed in the tunnel immediately
field for most projects. after excavating a working section.
FIGURE 6-4 Tunnel supported by frozen soil: (a) roof; (b) roof
and sides; (c) closed ring.
l
Source: Reproduced with permission from Jessberger 1980. Copyright 1980
ria
Elsevier.
ate
FIGURE 6-6 Typical effect of thermal properties on the frozen
zone.
dM
Source: Reproduced from Shuster 1972. Copyright 1972 American Society of
Civil Engineers.
5, with lower values for silts and clays and higher values for
rock. For any given refrigerant temperature, the relatively thin-
ner frozen zones will occur in the silts, clays, and organic soils.
These are also the weaker strata; hence, structural analysis and
design will often be dictated by these soils. When the freeze
rig
Elsevier.
effects may result in the frozen zone having a conical shape near
the surface, and it may be difficult to obtain closure between
Soil Conditions
adjacent frozen soil columns at shallow depths. Surface insula-
The site investigation should include borings that extend well tion around the top of the freeze pipes will reduce this effect.
Co
below the planned excavation depth. These borings must pro- During the winter months, this will not be a problem.
vide samples for classifying the soil as well as undisturbed soil Soil conditions below the bottom of the excavation are
samples needed for both frozen and unfrozen strength tests. extremely important. Ideally, a frozen soil wall should be tied
Soil type, density, and water contents are needed to estimate into an impervious layer to develop a closed bottom condition.
soil thermal properties. In situ permeability tests can provide This eliminates the need for any significant pumping to control
information on the order of magnitude and variability for nat- groundwater. In situations where an impervious layer does not
ural pervious soil stratum. Ground temperatures and water exist at a reasonable depth and the excavation has an open bot-
table measurements should be made after in situ conditions tom, extreme caution must be used to minimize water move-
have recovered from the disturbances caused by the boring ment under the bottom of the frozen earth wall, as well as to
operation. Standard methods for investigating and sampling satisfy the usual concerns for piping or heaving at the bottom
unfrozen soils and rocks are appropriate. of the excavation. Deep wells outside the excavation have been
The soil profile shown in Fig. 6-6 illustrates the effect of used to collect seepage for the open bottom construction
thermal properties on the probable shape of the frozen zone (Braun, Shuster, and Burnham 1979). Continuous open sump
pumping should not be used within a frozen earth wall because the groundwater flow exceeds 3 m/day, either the permeability
of the free water and relative lack of control. The movement of of the formation or the groundwater gradient must be reduced.
water to the pump brings heat toward the frozen soil and thus This can be accomplished by grouting prior to, or during, the
aggravates the condition. installation of the freeze pipes. An alternative solution includes
In addition to seepage concerns, the location of the water intercepting the water flow with wells. The relative cost and
table is significant because of the importance of water content degree of success for these methods depends largely on the uni-
and degree of saturation in determining the uniaxial compres- formity of the ground and the accuracy of the available subsur-
sive strength of frozen soil. Above the water table, the soil is face data used to plan the work. Extreme care must be exercised
normally unsaturated. Unconfined compression tests on frozen in the handling of water around a frozen earth structure
silica sand (Alkire and Andersland 1973) indicated that the because poorly controlled pumping inside or outside a frozen
strength approached zero for an ice saturation close to 15%. In earth cofferdam, or indiscriminate flooding of the excavation,
general, the required degree of water saturation should be on may cause thermal erosion damage to the frozen earth wall.
l
ria
the order of 50–70% (Borkenstein, Jordan, and Schäfers 1992).
It is possible to entrain water into coarse-grained soils at a site ■ EXAMPLE 6.1-1: A proposed circular frozen earth coffer-
during construction, but questions may arise about water dam 20 m deep in a gravelly sand deposit will consist of a single
retention up to the point of freezing. The presence of lenses of row of freeze pipes spaced at 1.0 m. Groundwater flow across
less pervious materials will limit this form of water entrain- the site is close to 0.9 m/day. Information about the site and the
ate
ment. Water combined with a drilling mud additive has been freezing system to be used includes ground temperatures close
used to form a viscous suspension for injection into the perme- to 10 °C, average soil density of 1,968 kg/m3 with a water con-
able soil pores (Borkenstein, Jordan, and Schäfers 1992). Exper- tent of 23.0%, freeze pipe diameter equal to 76 mm, and freeze
imental results have shown that a high degree of saturation (Sr pipe surface temperatures close to –20 °C. Should the contrac-
> 70%) was obtained even in the more pervious gravel/sand tor be concerned that the freezing soil columns may not merge?
layers. If a problem exists, what options are available to the contractor?
dM
Solution: Compute the solid dry density [1,968/(1 + 0.23)] =
1,600 kg/m3. Using Fig. 2-26, estimate the frozen soil thermal
Groundwater Flow
conductivity (kf = 3.2 W/m · °C). Determine Vs = To – TS = 0 –
Groundwater flowing through a proposed site adds heat, which (–20) = 20 °C and Vo = Tg – To = 10 – 0 = 10 °C. Using Eq. (6.1-
may cause problems relative to the formation of a continuous 1), compute the critical flow velocity for the freeze pipe spacing
frozen wall. If the water flow velocity is too large—greater than of 1.0 m:
1 to 2 m/day—the freezing columns will not merge, leaving
hte
water (Sanger and Sayles 1979). The freezing soil columns will velocity. The contractor should be concerned that wall closure
not merge for a critical groundwater velocity: may not be attained. Grouting prior to freeze pipe installation
is recommended. If necessary, liquid nitrogen (LN2) can be
kf Vs used to ensure wall closure.
uc = (m/day) (6.1-1)
Ê S ˆV
py
4S ln Á ˜ o
Ë 4ro ¯ Groundwater Quality
where ro (m) is the freeze pipe radius, kf (W/m · °C) the frozen Contaminants in groundwater include a variety of constituents
soil thermal conductivity, S (m) the freeze pipe spacing, Vs (°C) (inorganic, organic, radionuclides, and bacteriological). Many
Co
the difference between the freeze pipe surface temperature and substances have a very low solubility in water. Other substances
the freezing point of water (°C), and Vo (°C) the difference (immiscible fluids) exist in a liquid state in contact with water
between the ambient ground temperature and the freezing in a manner that does not lead to mixing in a dissolved form.
point of water. These seepage velocities agree very well with The presence of contaminated water in ground to be frozen can
those obtained by Khakimov (1957) and Hashemi and lead to several problems, including lower freezing temperatures
Sliepcevich (1973) using the finite-difference method, and with (Fig. 2-3), reduced ice content (Fig. 5-31), and lower strength
field observations. (Fig. 5-33). The salinity of groundwater should be determined
The magnitude and direction of water flow can be measured when there is doubt as to its freshness. Frozen strength proper-
in single bore holes with devices using fluorine ion or radioac- ties should be determined by laboratory tests. The extrapola-
tive solutions (Grisak, Merritt, and Williams 1977; Drost et al. tion of existing strength data obtained for soils containing fresh
1974). If the groundwater seepage is greater than 1.5 m but less water is not recommended. For partially ice saturated granular
than 3.0 m/day, either reduced freeze pipe spacing or a second soils, liquid contaminates with a low solubility in water and vis-
row of freeze pipes on the upstream side is a feasible solution. If cosities similar to water will flow through continuous soil voids
(Appendix C.7). This can create problems when frozen soil bar- These two phenomena occur simultaneously; however, they
riers are intended to contain liquid contaminants. differ in predictability and magnitude. Vertical displacements
When soil pore water contains a significant quantity of sol- (heave) due to a change in the phase of the pore water were esti-
utes (often sodium chloride), the freezing point is depressed. mated by Sanger and Sayles (1979) on the assumption that half
The process is limited by their solubility in water, as is shown in of the volume expansion is in the vertical direction. This
Fig. 2-12 for NaCl. Frozen soil barrier deterioration (i.e. thaw- change in soil column height (DH), based on a constant water
ing of frozen soil) will involve ice erosion by contact with liquid content and only phase change, is
contaminants having a freezing point lower than the frozen soil
temperature. A contaminant with a small freezing point nH È 1 ˘ nH (6.1-2)
DH = Í - 1˙ ª
depression, say –5 °C, may cause erosion only in the outer por- 2 Î 0.917 ˚ 22
tion of the frozen barrier where frozen soil temperatures are
above –5 °C. When ice erosion reaches the –5 °C isotherm, where n is soil porosity, H the frozen soil column height, and
l
0.917 the specific gravity of ice. Equation (6.1-2) assumes that
ria
equilibrium conditions develop between the liquid contami-
nant (brine) and ice. A colder average frozen wall temperature the effect of unfrozen water content will be very small for tem-
and/or a thicker wall may be needed to avoid distress. If the peratures below –10 °C. This method is limited by several fac-
freezing point of the contaminant is lower than the freeze pipe tors. For partially water saturated soils, a portion of the ice
temperature and if little or no dilution of the contaminant expansion will be absorbed by the available pore voids. In clean
ate
occurs, slow ice erosion may lead to the complete disappear- granular soils with a high permeability, the excess pore water
ance of the frozen barrier. This behavior has been demon- will drain ahead of the freezing front so that no volume change
strated using bench-scale frozen soil barriers (Andersland, occurs. In cohesive soils, with a low permeability, drainage may
Wiggert, and Davies 1996). not occur during the freezing period and some heave can be
During slow freezing and the formation of a frozen wall in expected. In these soils, part of the water will remain unfrozen;
saline soils, solutes are excluded by the ice, forming a band in the amount is dependent on soil type, temperature at each
dM
front of the frozen–unfrozen interface. This mechanism point, and particle surface area characteristics. Unfrozen water
appears to be responsible for the ice banding in saline soils contents can be estimated using the methods described in Sec-
observed by Chamberlain (1983). Andersland, Wiggert, and tion 2.3.
Lehner (1994) observed a similar banding in front of a vertical Frost expansion with the formation of ice lenses is more dif-
frozen–unfrozen interface with decane as the contaminant. The ficult to predict and may be much greater. It is related to soil
increase in contaminant concentration will lower the freezing permeability; hence, this parameter can be used (as is shown in
Fig. 6-7) to provide a relationship with the rate of unconfined
hte
point and may prevent freezing in the band. With more heat
extraction and lower temperatures, the freezing front will pass expansion due to ice segregation and the combined effective
through the contaminant-enriched band and continue with the intergranular and potential expansion pressures. These curves
formation of a series of bands until closure between freeze (Fig. 6-7) do not represent a precise relationship; however, they
pipes occurs. Baker and Osterkamp (1988) have shown that in do approximate frost expansion behavior and provide a conve-
uniform medium sand saturated with 35 parts per thousand nient means for discussing this problem. As a rule, fine silts and
rig
(ppt) of sodium chloride solution, significant salt rejection lean silty-clay mixtures usually represent the worst combina-
occurred at a rate that increased with decreasing freezing rate. tion of potential combined pressures and permeability. Note
that as the soil becomes more clayey the combined pressures
increase exponentially. These pressures are also affected by tem-
Ground Movement perature, being higher for colder temperatures (Hoekstra
py
The design of temporary ground support systems involves pos- 1969b). Note that for clay soil with a high combined pressure,
sible subsidence adjacent to the excavation and frost expansion. the low permeability and/or confining pressure will limit the
Subsidence after excavation can occur due to creep of the fro- rate of frost expansion. This is illustrated by the unconfined
zen barrier under prolonged loading. The amount of creep that expansion rate curve in Fig. 6-7. The location and magnitude
Co
will occur under any given stress can be predicted by finite ele- of the expansion rate curve is dependent on the freezing rates.
ment methods using parameters based on laboratory creep For higher freezing rates, the curve would be decreased in mag-
data and will normally be accounted for during design. Thaw nitude and displaced to the right (higher permeability). For
subsidence involves a volume change corresponding to the very high freezing rates, such as those attained with a liquid-
phase change of ice to water with no change in total water con- nitrogen system, frost expansion is greatly reduced during
tent. Additional consolidation, which occurs with drainage, active freezing and may approach that corresponding to only
will be small if the soil was initially in a relatively dense state phase change of the pore water.
before freezing. (Details on computing thaw settlement were A technique for determining potential frost expansion con-
given in Chapter 4.) Ground movement due to frost expansion ditions at a freezing site is illustrated in Fig. 6-8. The three
results from two different phenomena: (1) expansion due to the intergranular stress curves represent the lateral (s3¢) and verti-
conversion of soil pore water into ice during freezing (about cal confining pressures (s1¢) of undisturbed soil at rest and the
9%), and (2) frost expansion due to pore-water migration and lateral pressure (s3¢) required to mobilize the passive strength
the formation of ice lenses with time at the freezing isotherm. of the undisturbed soil. Appropriate data for the potential com-
l
ria
ate
dM
FIGURE 6-7 Typical effect of soil type on frost expansion pressures and rates.
Source: Reproduced from Shuster 1972. Copyright 1972 American Society of Civil Engineers.
hte
rig
py
Co
l
ria
sion will continue until freezing is discontinued or the pressure
distribution is changed such that the intergranular pressure
exceeds the potential combined pressures. This pressure crite-
rion illustrates that freezing in the top 12 m (Fig. 6-8) would
cause no expansion. Freezing would cause frost expansion
ate
between 12 and 17 m and to a lesser extent in the deeper clay to
about 25 m. The frost expansion potential of the hard clay is
great; however, the clay is probably so impervious that the rate
of expansion would be too slow to be of concern during the
expected duration of freezing. For the data and soils shown in
Fig. 6-8, it is probable that short-term freezing for temporary
dM
ground support of a tunnel would cause minimal frost expan-
sion. These computations are not precise; however, they would
enable the designer to evaluate whether or not a potential prob-
FIGURE 6-9 Refrigeration methods: (a) primary plant and
lem exists, and the order of magnitude of movements that may
pumped loop secondary coolant; (b) expendable liquid refrig-
be expected. erant.
Source: Reproduced with permission from Jessberger 1980. Copyright 1980
hte
Elsevier.
6.2 Freezing Methods and System Installation
The cooling system normally consists of a primary refrigeration Cooled liquid ammonia leaves the condenser and passes
source and a secondary distribution system needed to circulate through an expansion valve, where ammonia is sprayed at con-
coolant or refrigerant to the freeze pipes that are extracting heat stant high pressure through a fine nozzle into another coil at a
rig
from the soil. Alternative expendable refrigerants include the constant but lower pressure (Fig. 6-9a). With the decrease in
use of liquid nitrogen or carbon dioxide. Both systems are dis- pressure to about 0.15 MPa, the ammonia evaporates, causing a
cussed, along with installation of a refrigeration plant, coolant drop in temperature to about –25 °C. The brine cooler, or evap-
distribution manifold, and freeze pipes. orator, includes a series of coils for circulating the ammonia
gas. The coils are surrounded by brine, which gives up its heat
py
or two-stage ammonia refrigeration plant (two-stage for tem- brine tank, a brine pump, an insulated coolant supply mani-
peratures below –25 °C). These plants are commonly available fold, a number of parallel connected freezing elements in the
in a wide range of capacities and can frequently be rented com- ground with inner supply and outer return lines, and an insu-
pletely assembled in portable modules for field use. These lated return manifold. This system is simple but rather cum-
plants use either diesel or electric prime movers, they have high bersome and thermally inefficient. Heat transfer occurs
thermal efficiency, and their technology is well understood between the coolant and the freeze pipes by convection; no
(Shuster 1972). A schematic diagram (Fig. 6-9) illustrates the phase change occurs. Because of this, large quantities of coolant
cooling plant in which the refrigerant is circulating. As the must be circulated to cause freezing, and an inherent thermal
compressor liquefies ammonia gas at about 0.8–1.2 MPa, the gradient exists in the system during active freezing periods.
temperature of the liquid ammonia rises to about 100 °C. Liq- Large flows require large coolant volumes and large fixed
uid ammonia is then pumped to a condenser under high pres- plumbing systems. The flexibility to increase the capacity of
sure, where it is cooled to about 15 °C during passage through a individual freezing elements independently is desirable so as to
system of coils. Circulating water removes the excess heat. facilitate controlling localized wall conditions, including unex-
pected water flows. Different types of coolant have been used gen (LN2) throughout a series of freeze pipes represents the
with the system (diesel oil, propane, glycol-water mixtures, and fastest, thermally most efficient means of ground freezing. Dif-
brines); the most common is calcium chloride brine. Calcium ficulties encountered with the use of expendable refrigerants
chloride is added to water in sufficient quantities to depress its involve control of the system. The unconfined venting of LN2 in
freezing point below that attainable by the refrigeration plant a series of vertical or horizontal freezing elements frequently
during on-line operation. These brines have high specific heat; results in a waste of refrigerant and a very irregular frozen zone.
however, they are also dense, relatively viscous, and corrosive. This irregularity occurs because the heat transfer coefficient
Other fluids may have more attractive properties under the varies by an order of magnitude dependent on the quality of
same conditions, but flammable or toxic coolants should be the liquid-vapor mixture and its velocity (Flynn, Draper, and
avoided for obvious reasons. Ross 1962). A supply and exhaust manifold with appropriate
valves at each end of a series of freezing elements permits
reverse flow, which tends to even out the irregular freezing
l
Expendable Liquid Refrigerant
ria
characteristics. This is particularly true if no more than about
Expendable refrigerants, such as liquid nitrogen or carbon four freezing elements are connected in series (Shuster 1972).
dioxide, are very attractive for single projects of short duration When a series of open vertical freezing elements is filled with
(a few hours to a day or two) or for projects where the cost of LN2 and allowed to boil (nitrogen boils at atmospheric pres-
delay is high. The characteristics of and relevant differences sure), the fluid may be ejected after a short period of boiling
ate
between LN2 freezing and brine freezing are given in Table 6-1. (Murphy 1965). The sudden geysering of supercold LN2 is
No refrigeration plant is required. Liquid nitrogen is supplied wasteful and may be dangerous. This situation may be con-
directly or through a storage tank (Fig. 6-9b) into the freeze trolled by placing a supply tube with a slightly smaller diameter
pipes. After circulating through a few freezing pipes, nitrogen is inside the freezing element (Shuster 1972). To ensure success,
released into the atmosphere. Uniform boiling of liquid nitro- the supply tube should be nearly as large as the freezing ele-
ment, and the supply of LN2 should be regulated to match the
dM
rate of boil-off.
TABLE 6-1 Characteristics of Liquid Nitrogen and Brine Sublimating carbon dioxide (CO2) is thermally less efficient
Freezing than LN2 and, except for the simple case of a row of vertical
Item LN2 Brine freeze elements, is harder to control. If liquid CO2 is used as
“snow,” it has a tendency to plug orifices, valves, and piping.
Site installation Solid dry ice, even in pelletized form, is bulky and difficult to
hte
Electric power Not required Required handle. However, dry ice used with a mixing tank and a circu-
Water for cooling Not required Required lating secondary coolant is probably one of the most efficient
Refrigeration plant Not required Required ways to use expendable CO2 on many projects (Shuster 1972).
When using expendable refrigerants, it is necessary to pro-
Storage tank Required Required
vide positive exhaust for the vapor. Neither carbon dioxide nor
Circulation pumps Not required Required nitrogen is flammable or toxic. However, they are heavier than
rig
Pipe system for distribution Supply only Supply and return air and in large quantities can cause suffocation. In addition to
of coolant adequate ventilation, emergency oxygen should be available at
Low-temperature material for Required Not required the site and protective gloves should be provided.
surface pipes, valves, etc.
Low-temperature material for Not required Not required
py
freeze pipes
Installation of the Cooling System
Execution of freezing Most ground freezing projects employ the circulating coolant
freezing system. The remainder employ expendable refrigerant
Physical condition of coolant Liquid/vapor Liquid
systems, particularly when rapid formation of the wall is
Co
Minimum temperature –196 °C –34 °C MgCl2 important and only small earth volumes are to be frozen. The
achievable (theoretical) –55 °C CaCl2
temperature, geometry, and performance of the frozen earth
Reuse of coolant Impracticable Standard structure formed by a given freeze pipe installation is depen-
Control of system Difficult Easy dent on the specific refrigeration plant and coolant distribution
Shape of freeze wall Often irregular Regular system employed. Construction operations necessary to com-
Temperature profile in freeze Great differences Small differences plete any ground freezing project include site preparation, sub-
wall surface installation of refrigeration pipes, and installation of
Frost penetration Fast Slow the refrigeration plant and coolant distribution manifold.
Impact on freeze wall in case of None Thawing effect
Site Preparation. Site preparation must include grading to
damage to freeze pipe
ensure that surface water is collected and drained away from
Noise None Little
the planned frozen earth structure. Inadequate surface water
Source: After Stoss and Valk 1979. drainage can result in severe damage to a frozen earth wall if the
water is allowed to pour over or to pond against it. Once the single row of freeze pipes, is exponentially proportional to the
wall is eroded by water, it is generally difficult and costly to relative pipe spacing. For example, doubling the spacing with
repair the affected area. all other factors held constant will increase closure time by a
In urban areas, utility lines may traverse the frozen zone or factor of about 5. Because of this and the desire to maintain
be located in close proximity to it. Although it is unlikely that more or less uniform temperatures in the frozen earth struc-
the amount of energy removed will be sufficient to freeze the ture, it is important that the alignment of all freeze pipes be
moving water in a water main or sewer, the potential must be verified after their installation. The alignment plots of vertical
considered. Most utilities are located close to the ground sur- pipes can be obtained with a wide variety of downhole incli-
face. Exposing the utility line in the affected area and insulating nometer instruments. An example of freeze hole deviations
the pipe with sprayed polyurethane foam (instrumented with determined over a depth of 107 m is illustrated in Fig. 6-10.
thermocouples) have proven effective in preventing freezing of The 1.07-m spacing at the surface varies by a factor approach-
ing 2 near the bottom for some adjacent holes. Note that hole
l
the utility or thawing of frozen ground surrounding the utility.
ria
Steam pipes are the most sensitive utility lines. Any drop in sat- 14 intercepted core hole 7 at the 79.25-m depth, requiring that
urated steam temperature will cause water buildup in the lines. hole 14-A be drilled to get a freezable pattern. Horizontal and
Because of this, extreme caution should be exercised before vertical deviations of horizontally driven pipes may also be
ground freezing is used in proximity to steam lines that must obtained by using deflectometers. It is both unwise and unsafe
remain in service. to assume that pipe deviations are acceptable unless the pipes
ate
Every site must have electric power unless the refrigeration are short and the probability of deviation beyond tolerance
plant is driven by an internal combustion engine. It is better to limits is very small. For pipes longer than 20 m, deviation mea-
provide temporary commercial power than to run a refrigera- surements should be taken and the alignment verified. Where
tion plant with a motor-generator set (Shuster 1982). Routine spacing exceeds the criterion of 13 diameters of the freeze pipe,
starting and stopping of the motor-generator for maintenance an additional pipe should be installed.
and/ or fluctuations in the voltage generated can cause cumula- Installing a freeze pipe into a drilled hole requires a hole
dM
tive damage to electrical motors and even the compressors of a larger than the pipe diameter. In clays, rock, and some other
refrigeration plant. The damage may result in costs of compo- soils—especially above the water table—this results in an annu-
nent replacement and emergency repairs far in excess of the lar void between the pipe and the surrounding soil. Under these
cost of providing temporary power in the first place. Backup circumstances, heat transfer between the pipe and soil is poor.
power is normally not required unless the frozen earth struc- If this condition is left uncorrected, it can lead to unexpected
ture is designed to function in a highly stressed state. The time- collapse or to the loss of a portion of the frozen earth wall.
hte
temperature response of frozen earth routinely permits refrig- Attempts to backfill the annular void must be handled carefully
eration downtime on the order of several days. When a power so as to ensure that the backfill material does not “bridge” near
outage or component failure may result in downtime in excess the surface before the hole has been filled completely. Grouting
of 72 h, backup refrigeration plant capacity should be consid- techniques with bentonite, sand–bentonite, cement–bentonite,
ered. and plain cement have been used with good success. Wet sand
can be used for vertical holes if properly jetted with water dur-
rig
Subsurface Installation of Freeze Pipes. Ordinary ASTM A- ing the backfilling process. Rotary mud drilling procedures
120 or A-53 steel is commonly used for freeze pipes, including provide one method to rapidly and efficiently install refrigera-
liquid-nitrogen applications (Shuster 1982). When blasting is tion pipes. Oil- or saltwater-based drilling muds should not be
anticipated in the vicinity of the pipes, ASTM A-333 or 8% used, because they may prevent freezing of the soil or signifi-
nickel steel pipe should be used. All subsurface connections on cantly reduce the frozen soil strength. Also, lost drill mud circu-
py
a refrigeration pipe should be welded. Threaded couplings are lation below the water table indicates the potential for ground-
not recommended. After the completion of all welding, but water flow problems. When drill mud circulation is lost,
prior to installing the pipe in the ground, the pipe should be remedial grouting is indicated to fill the void and reduce the
pressure tested to at least 125% of its working pressure to soil permeability. If it is necessary to set drill casings outside the
Co
ensure that no coolant leaks into the ground. freeze pipe to advance a hole through soil strata, it is imperative
The techniques of pipe installation vary widely depending to carefully evaluate the potential for voids and groundwater
on requirements of the particular application. Several factors flow problems . It is easier and less expensive to conduct reme-
are basic regardless of the procedure used. The spacing of dial grouting at this stage of construction than later.
refrigeration pipes throughout their length should not exceed
about 13 times their diameter unless careful analysis indicates Refrigeration Plant and Coolant Distribution Manifold.
that a larger relative spacing will be acceptable for a particular Ground freezing refrigeration plants are designed as low-tem-
application. The empirical data upon which this criterion is perature machines. They should be routinely capable of rapidly
based were derived from projects with freeze pipes varying achieving and maintaining suction temperatures between –30
from 50 to 150 mm in diameter (Shuster 1982). More than any and –40 °C under warm, humid, ambient air conditions. A
other controllable variable, the relative freeze pipe spacing con- large refrigeration compressor and motor must have an ade-
trols the time needed to complete satisfactory freezing. The quate condenser. The refrigeration plant must be evaluated and
duration of time required to close a frozen wall, consisting of a sized as a complete operating system, not on the basis of the
l
ria
ate
dM
hte
rig
heat rejection capacity of an individual component. The heat of pipes in series is connected in parallel between an insulated
py
extraction of the system at about –30 °C suction should be on coolant supply manifold and a coolant return manifold. The
the order of 100–230 W/m of freeze pipe in the ground (this manifold configuration around the perimeter of the area to be
assumes adequate insulation of the surface distribution mani- frozen (Fig. 6-11) should be arranged so that multiple sections
fold). The refrigeration plant should be instrumented so as to are of approximately equal length to ensure approximately
Co
shut down automatically on any one of three criteria: (1) high equal coolant distribution. Manual valves should be inserted
head pressure, (2) low suction pressure, or (3) loss of hydraulic into the main supply and return manifold system at a number
pressure in the coolant circulating system. of predetermined locations. The intent of these valves is to
A coolant distribution system is normally assembled in a enable a portion of the system to be disassembled for repair or
series-parallel configuration. The amount of freeze pipe in a correction without the loss of coolant in the entire system or
single series should usually not exceed about 50 m2 of heat the need to shut down the remainder of the system. It should be
transfer area in the ground (Shuster 1982). The parallel supply possible to isolate hydraulically each group of pipes in series, as
and return temperatures of each group of pipes in series are well as each major section of the manifold, without having to
normally monitored by thermocouples to determine energy discontinue refrigeration operations. It is not possible to com-
removal in the group. The difference between inlet and outlet pute the hydraulic flow in each section of the coolant distribu-
temperatures (commonly called “the split”) should vary more tion manifold because the length and configuration of the pip-
or less proportionally to the number of pipes in the series (e.g., ing and hoses vary too widely. In the absence of external
more pipes, larger split; fewer pipes, smaller split). Each group sources, balanced flow is accomplished more effectively by bal-
l
ria
tions of the coolant distribution system. Aside from the imme-
diate need to turn off the nearest valve, isolating that portion of
the system, there is normally no cause for alarm. The damaged
component merely needs to be replaced, additional coolant
FIGURE 6-11 Refrigeration system supply and return mani-
fold; each vertical freeze pipe is individually connected to the
mixed to replace that which was lost at the time of the break,
ate
manifold. and the system put back on line. Damage of this type, although
Source: Courtesy of Joseph A. Sopko, Jr., Layne-Northwest.
not desirable, is certainly common and should not be a cause
for undue concern. The single exception is the breakage of a
freeze pipe deep in the ground, which would allow concen-
ancing the coolant split temperatures. When the split tempera- trated coolant to penetrate the earth, potentially destroying the
tures are more or less equal (proportional to the number of adjacent frozen earth wall and preventing it from being frozen
dM
pipes in each group), so is the hydraulic flow. Although it is except with expendable low-temperature refrigerants. Because
technically possible to measure coolant flow in each series of of this, underground blasting or excavating with mechanical
pipes, it is not practical or necessary for most applications. The equipment in close proximity to the freeze pipes must always be
main supply and return temperatures of the coolant, together monitored carefully by an alert operator prepared to close off a
with the split temperatures for each group of pipes in series, valve immediately if the pipe is inadvertently broken.
normally provide adequate control over coolant distribution in
the system.
hte
Each freeze pipe and each group of pipes in series should be 6.3 Structural Design of Frozen Earth Walls
provided with positive air-bleed valves to allow relief of trapped
air in the system during operation. A temperature difference On the basis of required excavation limits and the available
between inlet and outlet pipes of 4–5 °C often indicates the space on site, the designer must select the frozen earth wall
presence of air pockets. If this is not checked continuously dur- geometry. Because of the relatively high compressive and low
rig
ing the initial operation of the system, some freezer pipes prob- tensile strengths of frozen soil, curved arch walls, particularly
ably will not be properly cooled—even though the refrigeration circular walls, are the best solution from a structural point of
plant is functioning normally and the main coolant supply and view. Space permitting, they normally will result in the most
return temperatures are reasonable. One air bleeding is usually economical design. An ellipse may be employed effectively for
not sufficient. In addition to air entrapped in the system at the rectangular structures if the ratio of length to width does not
py
time it is started, there is also air in solution in the coolant. exceed about 2. If space or other restrictions prevent the use of
Although air in the latter situation is not normally a problem curved structural elements, straight walls provide an alterna-
on start-up, it is a problem subsequently if the compressor load tive. Complex structures may involve combinations of curved
is reduced and the coolant is allowed to warm, causing air to or straight walls, along with some form of bracing system. The
Co
come out of solution and forming airlocks in the system. structural sections associated with tunneling involve relatively
Inspectors should routinely check and bleed any suspicious thin frozen soil masses to support the roof portion of the exca-
freeze pipes to ensure adequate coolant flow. The tops of indi- vation, roof and sides, or a closed circular ring.
vidual freeze pipes should not be insulated, to enable the
inspector to see the amount and quality of ice that has formed
Curved Walls
on the pipe. To the trained eye, this is a very quick and effective
indicator of the performance of the system. The normal procedure for designing a frozen earth wall is to
The frozen earth mass formed by a vertically installed determine the thickness based on the creep strength of the fro-
ground freezing system tapers toward the refrigeration pipe as zen soil and then compute the amount of deformation
it nears the ground surface, much like the neck of a bottle (Fig. expected during the life of the structure. Design loads have
6-6). This results in the frozen mass being somewhat thin, even been assumed to vary between at-rest and passive pressures
discontinuous, near the ground surface. This situation can within about 2 to 4 h after excavation to at-rest and active pres-
cause problems because the thin top of the wall has little resis- sures for periods greater than one work shift. Ladanyi (1982c)
l
ria
tions show that during the excava-
tion process, due to freezing pre-
stress, ground pressures can be close
to the at-rest ground pressures.
Installing a rigid permanent lining
ate
will stop the displacements and
keep the ground pressure from FIGURE 6-12 Ground pressure variation with time after excavation on a frozen clay shaft
decreasing with time. If the frozen wall at a depth of 200 m. Php , freezing prestress pressure; Ph¢, horizontal effective pres-
wall deformation predicted is exces- sure; Uw , hydrostatic water pressure.
sive for the loading period, the Source: Reproduced with permission of B. Ladanyi from Ladanyi 1982c.
design can be modified by changing
dM
the structural geometry, increasing
the wall thickness, or lowering the wall temperature. It is essen-
tial to know the strength and deformation behavior of frozen
soil that is representative of the site for the expected loads in
order to determine the most economical design and to avoid
possibly unsafe conditions.
hte
d sr (6.3-1)
r = sq = sr
dr
The condition for plastic equilibrium in frozen soil, as based on
the geometry of the Mohr–Coulomb diagram (Fig. 6-13b),
becomes
FIGURE 6-13 (a) Thick wall cylinder section (P = external
pressure); (b) Mohr–Coulomb diagram showing the stress
sq = 2c N f + sr N f (6.3-2)
relationships.
where the cohesion c is given by Eq. (5.3-46) and the flow value
Nf by Eq. (5.3-47). Equations (6.3-1) and (6.3-2) may be solved
using the boundary conditions sr = 0 at r = a and sr = p at r =
b: thus
1 /(N f -1)
b Ê p+Hˆ (6.3-3)
=Á ˜ for f > 0
a Ë H ¯
where
H = c(t, q) cot f (6.3-3a)
and
b È p ˘ (6.3-4)
= exp Í ˙ for f = 0
a Î 2c (t , q) ˚
For a circular frozen wall without lining, Sanger (1968) pre-
l
ria
sented the design chart shown in Fig. 6-14. In cases where pri-
mary creep dominates the deformation behavior, Sanger and
Sayles (1979) recommended using for p the at-rest lateral earth
pressure to determine the required wall thickness, d = b – a.
Klein and Gerthold (1979) have extended this solution to
ate
the case where the internal pressure acting on the wall pi > 0, in
which case, for the Mohr–Coulomb failure condition, Eqs.
(6.3-3) and (6.3-4) become
1 /(N f -1)
b Ê pe + H ˆ (6.3-5)
= for f > 0
a ÁË pi + H ˜¯
dM
and
b È p - pi ˘
= exp Í e ˙ for f = 0 (6.3-6)
a Î 2c (t , q) ˚
where pe and pi are external and internal pressures acting on the
hte
Now compute pe : Using Eq. (6.3-5), compute the ratio b/a, thus
l
t
Ë wsc q ˜¯
ria
Ê ph ˆ
d2 = b - a = 3 Á (6.3-11)
where B is the parameter b in Table 5-3, Ë s fu ˜¯
n The frozen wall behavior is more closely characterized by inter-
ˆ Ê e ˆ
B
Ê ˆ Ê
K =Á 3˜ Á 3˜Á c ˜ (6.3-8)
mediate conditions of fixity. Vialov (1962) recommended using
Ë n ¯ Ë 2 ¯Ë B ¯
ate
1.3 ph
Êaˆ
2
n d3 = b - a = (6.3-12)
w =1- Á ˜ (6.3-9) s fu
Ëb¯
Considering displacement of the inner frozen wall surface, the
The reference stress sc q = sco f (q), a function of the reference equation for determination of the safe wall thickness (Vialov
dM
strain rate, ėc , is defined as in Chapter 5. 1962) becomes (in the notation of Chapter 5)
For creep closure of frozen circular shafts in a frictional fro-
zen soil (f > 0, c > 0), Klein (1985) has given closed-form solu- n
Ï 1 ¸ (n -1)
tions for cases where the creep exponent n = 1 or 2. Ô È Ê h 3 ˆ n +1 ˘ n Ô
Í Á ˜ ˙ Ô
b ÔÔ Ê n - 1ˆ Ê p ˆ Í Ë a ¯ ˙ Ô
■ EXAMPLE 6.3-2: An 8.0 m diameter circular shaft, pro- = Ì1 + K Á (6.3-13)
a Ô Ë n ˜¯ ÁË sc q ˜¯ Í D Ê B ˆ B ˙ ˝Ô
tected by a 6.0 m thick frozen wall, is excavated deep in Callov- ÍÊ ˆ ˙
ÍÎ ÁË a ˜¯ ÁË ec t ˜¯ ˙˚ Ô
hte
3.7 0.37
Ê 3ˆ Ê 3 ˆ Ê 10-5 ˆ
K =Á ÁË ˜Á ˜ = 0.00107
Ë 3.7 ˜¯ 2 ¯ Ë 0.37 ¯
Co
2
Ê 4ˆ 3.7
w = 1- Á ˜ = 0.3906
Ë 10 ¯
3.7
Ê 4.9 ˆ
ui = 4.0(0.00107)Á (12)0.37 = 1.028 m
Ë 3.65 ¥ 0.3906 ˜¯
This displacement is clearly unacceptable. A lining will be
required with placement closely following the shaft excavation.
The stability of frozen earth walls for deep (mine) shafts
with unsupported wall heights (h) of 1.5–6 m below a perma- FIGURE 6-15 Diagram of ice-soil retaining structure: 1, frozen
nent shaft lining have been described by Vialov (1962) and soil; 2, surrounding soils; 3, liner; 4, refrigeration pipes; 5, opera-
Vyalov, Zaretsky, and Gorodetsky (1979). The unlined portion tional unsupported part of shaft; 6, lateral soil pressure diagram.
is short in comparison with the shaft diameter, and the wall is Source: Reproduced from Vialov 1962.
where D is the permissible deflection of the unsupported wall solution, which is based on the power-law creep equation, is
–
section at mid-height, K is a coefficient describing the degree of able to follow the buildup of pressure on the lining with time. It
fixity at the end planes of the unsupported wall section, the creep takes into account the time of lining installation (or the gap left
parameters are defined as in Chapter 5, and p is the lateral soil between the lining and the frozen wall) and the elastic rigidity
–
pressure. Vialov (1962) reported that K falls within the limits of the lining. Some essential results of this solution, modified
so as to be applicable to a frozen wall of finite thickness, are
2- ( ) < K <1 (6.3-14)
1+1
n presented in the following.
– An increase of pressure on the lining, pc , at a time t after the
A value of K = 1 is recommended when soil inside the radius r
lining-frozen wall contact has been established, is given by
£ a remains unfrozen and p corresponds to the condition of
elastoplastic equilibrium in the unfrozen soil surrounding the pc = pe [1 - (1 + t *)-1/(n -1) ] (6.3-17)
frozen wall. The lower value
l
where the dimensionless time t* is defined by
ria
K = 2- ( ) (6.3-15)
1+1
n
pe(n -1)
corresponds to the condition where soil inside r £ a is frozen, t * = K sc K (n - 1)t B (6.3-18)
and p corresponds to the hydrostatic pressure of unfrozen soil. (sc q w)n
On the basis of experimental work, Vialov (1962) used where K and w are defined by Eqs. (6.3-8) and (6.3-9), and Ksc
ate
is the elastic rigidity of the concrete lining, given by
1- x (6.3-16)
K= 1
2n È Ê r ˆ2 ˘
2Gc Í1 - Á i ˜ ˙
where x represents a coefficient taking partial fixity into consid- ÍÎ Ë re ¯ ˙˚
eration. The computed wall thickness depends, to a large K sc = (6.3-19)
2
dM Ê ri ˆ
extent, upon the accuracy of lateral earth pressures. The safe 1 - 2mc + Á ˜
working height h depends on several factors, including wall Ë re ¯
thickness as determined by the freezing method, lateral earth
where ri and re refer to the internal and external radii of the lin-
pressures on the frozen wall, the frozen soil creep parameters,
ing, and Gc and mc are the elastic constants of the lining mate-
and the time t required for soil excavation and placement of the
rial. The corresponding creep closure after installation of the
shaft liner. Prefabricated permanent lining segments are nor-
lining is then
mally used to save time during construction. Klein (1988) has
hte
sandy loam include: B = 0.37, n = 3.70, w = 0.89, and sc q = 0.31 where Kss is the elastic rigidity of the frozen wall, given by
–
MPa for ė c = 10–5 h–1. It is also assumed that K in Eq. (6.3-13) is
equal to 0.75. For a freezing temperature of –15 °C, sc q = 3.656 È 2˘
2G Í1 - ÊÁ a ˆ˜ ˙
MPa. Now compute the ratio b/a using Eq. (6.3-13)
K ss = Î Ëb¯ ˚ (6.3-22)
Co
2
Êaˆ
Ï
3.7
1 - 2m + Á ˜
˘ 3.7 ¸
1 2.7
È Ê2 3ˆ
4.7 Ëb¯
Ô Í ˙ Ô
b Ô ÁË ˜¯
Ê 2.7 ˆ Ê 4.9 ˆ Í
= Ì1 + 0.75 Á 4 ˙ Ô where a, b, G, and m now refer to the frozen soil cylinder.
Ë ˜¯ ÁË ˜¯ ¥ Í 0 .37 ˙ ˝ Finally, the solution shows that the lining will fail in compres-
a Ô 3.7 3.656
Í 0.0125 ÊÁ 0.37 ˆ˜ ˙ Ô sion when the external pressure attains the value
Ô Ë - ¯ ˙˚ Ô
Ó ÎÍ 12 ¥ 10 5
˛
1
l
ature of –15 °C. The shaft will be continuously lined with a [18.8(0.74536)]1.28
ria
0.20-m-thick concrete lining. The lining will be installed at a
distance of 10 m above the shaft bottom, 24 h after excavation. t* = 0.22537(t0.45 – 240.45)
Determine the shaft closure at the 50-m level, when left unlined
Now compute the increase in pressure on the lining using Eq.
for 24 h, and the buildup of pressure on the lining after installa-
(6.3-17)
ate
tion. The lateral ground pressure acting on the frozen wall is
assumed to be isotropic, with pe = 1.0 MPa. Properties of the
frozen soil correspond to those of frozen Ottawa sand (Table 5- {
pc = 1.0 1 - [1 + 0.22537(t 0.45 - 4.18)]
-357
}
3). They include G = 200 MPa, m = 0.30, B = 0.45, n = 1.28, w =
1.0, sc q = 1.05 MPa for ė c = 10–5 h–1, and f = 30 degrees. The The resulting time-dependent wall closure and the correspond-
properties of the concrete include Ec = 26 GPa, mc = 0.20, and ing buildup of pressure on the lining (Fig. 6-16) tends very
dM
the compressive strength fc¢ = 30 GPa. quickly toward the original ground pressure of 1.0 MPa.
Solution: At T = –15 °C and with f = 30 degrees, use Eq. According to Eq. (6.3-23), the lining would fail only when pc
(5.3-53) to compute sc q = 1.05(1 + 15) + 1.0(3 –1) = 18.8 MPa. attains about 2.3 MPa; hence, there is sufficient safety against
Equation (6.3-8) gives lining failure.
1.28 0.45 Shaft Bottom Heave. As has been shown by Ladanyi (1981a,
Ê ˆ Ê 3 ˆ Ê 10-5 ˆ
K =Á 3 ˜ ÁË ˜Á ˜ = 0.01027 1992), the shaft bottom heave due to frozen soil creep can be
hte
Ë 1.28 ¯ 2 ¯ Ë 0.45 ¯ estimated by comparing the bottom of the circular shaft with a
contracting hemispherical cavity. For the power-law type of
Eq. (6.3-9) gives
creep equation, as used previously, one gets the average time-
2 dependent bottom heave
Ê 2.50 ˆ 1.28
w = 1- Á = 0.74536
Ë 6.00 ˜¯ n
Ê e ˆ Ê 3 ˆ Ê p ˆ
B n
rig
s = a Á c ˜ Á ˜ Á o ˜ tB (6.3-25)
Compute the time-dependent closure with Eq. (6.3-7): Ë B ¯ Ë 2n ¯ Ë sc q w ¢ ¯
1.28 where po is the average total ground pressure acting at the bot-
È 1.0 ˘
ui = 2.50(0.01027) Í ˙ t 0.45 tom level, and
py
Î18.8(0.74536) ˚
ui = 0.0000875t 0.45 Êaˆ
3
n
w¢ = 1- Á ˜ (6.3-26)
For t = 24 h, ui = 0.00366 m. To determine the instantaneous Ë b¢ ¯
closure, compute Kss using Eq. (6.3-22):
Co
takes into account the limited thickness of the frozen soil below
the bottom, b¢ ≥ b.
È 2˘
2(200) Í1 - ÊÁ 2.5 ˆ˜ ˙
Î Ë 6.0 ¯ ˚ = 576.3 MPa ■ EXAMPLE 6.3-5: Under the same conditions as in Example
K ss = 6.3-4, and assuming that the frozen soil thickness below the
Ê 2ˆ
1 - 2(0.30) + Á 2.5 ˜ bottom is equal to the thickness of the wall, i.e., b¢ = b = 3.5 m,
Ë 6.0 ¯
compute the bottom heave.
ape 2.5(1.0) Solution: Using Eq. (6.3-26) compute
ui = = = 0.00434 m
kss 576.3 3
Ê 2.5 ˆ 1.28
w¢ = 1- Á = 0.87151
Total closure before installation of the lining is then Ë 6.0 ˜¯
ui = 4.34 + 3.66 = 8.00 mm Now compute the bottom heave using Eq. (6.3-25)
l
ria
ate
dM
FIGURE 6-16 Wall closure and pressure build-up curves for the freeze-shaft, example 6.3-4.
hte
0.45 1.28 1.28 anchored and braced walls must incorporate steel or concrete
Ê 10-5 ˆ È 3 ˘ È 1.0 ˘
s = 2.5 Á ¥Í t 0.45 bearing elements to distribute concentrated stresses. For the con-
Ë 0.45 ˜¯ Í 2(1.28) ˙
Î ˚
˙
Î18.8(0.8715) ˚ struction of deep, wide trenches, such as those required for sub-
s = 0.00069t 0.45 ways or large sewers, a combination of soldier beams, bracing,
and frozen earth can be an attractive construction alternative.
giving s = 2.87 mm after 1 day, 8.10 mm after 10 days, and 40.9
rig
as illustrated in Fig. 6-3. They are a massive, relatively rigid elasticity and Poisson’s ratio (Sopko 1990). Time and tempera-
structure formed before excavation takes place. The design of ture are considered indirectly in the modeling technique
the gravity wall is governed by overturning and sliding criteria, through selection of the elastic modulus used for the frozen
not stresses in the wall. Conventional earth pressure theories soil. Next, the action of gravity forces and applied loads is
Co
may be used in design for wall sliding and overturning. The allowed to act through the unfrozen soil to develop pressures
cantilevered and anchored walls involve highly sophisticated on the frozen wall. Soil excavation is represented by forcing
designs and require careful field control. Unanticipated water normal and shear stresses to zero on the inside surface of the
inflows during drilling of the anchors through the frozen earth wall. New stresses in the wall are computed and compared with
together with localized thawing can seriously damage or the long-term frozen soil strength. Wall displacements are com-
destroy the frozen wall. As a practical matter, the gravity wall is puted and compared with permissible displacements. Braun,
favored when a straight wall is required on a project. Shuster, and Burnham (1979) stated that these procedures have
A braced wall has all the advantages of a gravity or anchored been used successfully on numerous projects. More recently, a
wall. It is much less sensitive to field installation procedures and time-dependent finite element program (TEMP-W) designed
it can be readily packed or rebraced should the need arise. Tech- for geothermal solutions of problems with phase change of
nically, bracing is more desirable than anchors for use with fro- water to ice is being used (Sopko 2001). Its introduction signif-
zen earth, but it may be more awkward and difficult for the con- icantly advances the science of structural analysis and design
tractor to deal with during other subsurface work. Both involving frozen soils.
l
Finite-element models are suitable for structural analysis and
The stresses that act on an annular frozen tunnel wall are
ria
for designing frozen earth support systems with complex wall
illustrated in Fig. 6-17. The static system includes both vertical
loads and horizontal soil and water pressures. These loads may
be combined, giving the radial pressure distribution shown in
Fig. 6-17. Bending moments within the structure will alter the
ate
stress distribution. Consider a wall section of thickness d located
at point A. Normal stresses across the section are linear when lin-
ear elastic behavior is assumed for the frozen wall. Recognizing
that temperature will vary from warm (0 to –5 °C) at the wall
surface to cold (–20 to –25 °C) at the freeze pipe, some stress
redistribution will occur across the wall section with time. A
dM
mean temperature of –15 °C is shown for the section in Fig. 6-17.
A system for static calculations (Fig. 6-18) is based on the
assumption that unfrozen soil is bedded against the annular
frozen soil structure. It is assumed that the frozen soil behaves
like an elastic mass. The relevant Young’s modulus of elasticity
must be selected for different stages of performance and for the
effective service life of the annular structure. Time-dependent
hte
geometry and that involve multiaxial stress states. Details of the shear stresses were all less than 450 MPa. A comparison with
finite-element method (FEM) are given in various textbooks. strength results, from laboratory tests on frozen soils represen-
Applications to frozen earth structures have been described by tative of the site, showed acceptable compressive stresses at all
Klein and Jessberger (1979); Klein (1981); Jessberger (1982); locations except point A in Fig. 6-20. High principal stresses,
Soo (1983); Soo, Wen, and Andersland (1987); Meissner both in tension, indicated the need for a change in the tunnel
(1988); and Sopko, Shuster, and Andersland (1991). Its use for cross section so as to reduce or eliminate any tensile stresses.
analyzing frozen earth structures has increased with the avail- Another example, reported by Sopko, Shuster, and Ander-
ability of FEM programs that can handle the time-dependent sland (1991), involved the construction of a drop shaft and
behavior and material properties, which vary with temperature approach structure intended to transfer storm runoff water
and stress (tensile and compression). from a surface collector system to a deep large-diameter tunnel.
A feasibility study of a tunnel support system (Jones and The overall dimensions of the structure, which are shown in
l
Brown 1979) involved the use of the finite-element method to Fig. 6-21, indicated that a rectangular excavation would be
ria
calculate stress levels and wall displacements. The plane strain appropriate. Lateral earth support using straight frozen walls
finite-element model, which is shown in Fig. 6-20, illustrates generally involves high tensile stresses and bending moments
the 23-m-wide tunnel with a frozen earth roof in the form of an that are not acceptable for frozen soils. More conventional fro-
arch. Linear elastic stress-strain moduli corresponding to a very zen walls are typically circular or elliptical in cross section so as
long loading time were used in the analysis to model the frozen to create a structure with only compressive stress. The use of a
ate
soil behavior. The first step in analyzing the tunnel involved the single elliptical cofferdam would require a large number of
determination of geostatic stresses and the stress increase due freeze pipes, making the single elliptical cofferdam economi-
to surface loads prior to excavation. Next, the finite-element cally unfeasible.
method was used to compute the stress induced during excava- To minimize the frozen wall length, reduce the required
tion by forcing the final shear and normal stresses around the refrigeration capacity, and limit the excavation volume, a series
inside tunnel perimeter to equal zero. Calculated maximum of four separate but interconnected frozen soil structures were
dM
hte
rig
py
Co
l
ria
Loads applied at nodal points were those obtained from
boundary element forces for the previous analysis. The results
of this analysis indicated that load transfer to the reinforcement
members eliminated the high tensile stresses. Shear stresses at
the interface between frozen soil and reinforcement members
ate
were acceptable with no danger of adfreeze bond failure. Fur-
ther analysis of the reinforced section considered the time-
dependent properties of the frozen soil. It was confirmed that
most of the displacement at the sides of the wall opening
between cells occurred during the initial elastic phase. The total
wall displacements were very small, less than 2 mm, and did
dM
not interfere with construction.
both internal stresses for individual cofferdams and forces Soil freezing is achieved by a system of freeze pipes placed in a
transferred to buttress sections (Fig. 6-22). A four-node shell certain pattern (e.g., in a circle around a shaft, in a semicircular
element was used for the analysis because it could model the screen above an adit, or in lines to form a straight wall). For all
three-dimensional geometry of the cofferdam while limiting these cases, the distance between pipes (boreholes) is of consid-
the total number of nodes and required computer time. More erable importance because deviations may result in poorly fro-
py
details are given by Sopko, Shuster, and Andersland (1991). zen zones or in frost gaps. Borehole inclinometers can be used
Required wall thicknesses were significantly less than those to trace the freeze pipe locations, as is shown in Fig. 6-10. Com-
indicated by conventional analysis (Sanger and Sayles 1979). parison with repeated measurements provides information on
High compressive stresses at the cell 1 wall/drop shaft intersec- displacements associated with freezing and subsequent thaw-
Co
tion indicated a potential problem. To compensate for these ing. According to different requirements for vertical and hori-
high stresses, an additional row of freeze pipes was installed so zontal boreholes, a number of instruments and techniques have
as to increase the wall thickness at critical areas. been developed.
The next concern (Sopko, Shuster, and Andersland 1991) To describe the trace of a vertical borehole in space, both the
involved buttress sections (Fig. 6-22) and the potential for high coordinates of the borehole collar and the deviation of the
tensile stresses, when openings were excavated through the wall actual borehole axis from the vertical (including the azimuth)
between two cells. A three-dimensional elastic model was used must be known. These data can be obtained using a measuring
to determine (1) nodal deflections, (2) the magnitude and loca- device positioned within the borehole by means of centralizing
tion of maximum compressive and tensile stresses, and (3) devices arranged at both ends of this instrument and having a
reaction forces on the buttress section. The eight-node brick measuring base of 1 m. By moving this instrument downhole
(type 5) element available in the SAP IV program was used. along the borehole axis in increments of 1 m, a vector sequence
The results of the analysis showed (1) excessive tensile stresses is obtained that shows the simplified borehole trace as an open
near the face of the opening through the buttress sections, (2) polygon (Fig. 6-23). Measurement accuracy of deviation
l
ria
ate
dM
hte
rig
FIGURE 6-22 Three-dimensional illustration of the proposed frozen earth structures; frozen
walls formed by installing 76-mm-diameter steel freeze pipes on approximately 0.9-m centers.
Source: Reproduced with permission from Sopko, Shuster, and Andersland 1991. Copyright 1991 A. A. Balkema.
py
between the borehole direction and the vertical depends on the To describe the trace of a horizontal or an inclined borehole
exactness with which the inclinometer is centralized within the in space, a number of vectors are required. Starting from a
Co
borehole casing and on the sensitivity of the instrument. The known reference section outside the borehole—which may be
inclinometer may be centralized with a skid, with springs, or the connecting line of two guide sockets for the drilling pipe or
with three-point centralizers. The sensor consists of a mechani- an outside guide tube—an open polygon trace is continued
cal pendulum or an accelerometer. To control the deviation along the axis of the borehole casing by measuring the bend
direction (i.e., of the azimuth), there are several possibilities. angles in both the horizontal and vertical directions in relation
The magnetic compass measures the angle of distortion of the to the preceding element of the polygon. For this purpose, a
inclinometer in relation to magnetic north. The gyrocompass two-armed measuring device with a pivoting point in between
(north-seeking gyroscope) will orient the instrument sus- is used, one arm of which is placed (at the beginning of a mea-
pended from its framework in the direction of the meridian. surement) in the guide sockets or the guide tube mentioned,
Accuracies for the casing axis of 10 mm per 100 m of depth while the second arm rests in a measuring section of equal
have been reported by Heinrich, Müller, and Voort (1979). The length within the borehole casing. By moving the instrument
inert gyroscope maintains its direction, as preset at the surface, again and again for the length of the measuring distance (i.e.,
even after insertion into the borehole. half the instrument length along the casing), the vector
l
ria
ate
dM
hte
sequence that follows gives the borehole trace as an open poly- sors may be used either to monitor cooling liquid at the refrig-
gon (Fig. 6-24). Two different ways of determining the amount eration plant and in the freeze pipes or to measure soil temper-
of horizontal and vertical bending of the two arms in relation atures in specific observation holes. Sensors may be stationary
to each other have been reported (Heinrich, Müller, and Voort as well as movable, and they may have a direct readout with an
py
1979). The optical probe for horizontal deviation measure- eventual recording system. Examples of temperature monitor-
ments uses a light beam traveling parallel to the longitudinal ing of a freezing shaft are given in Fig. 6-25.
axis of one arm toward the second arm, where its image on Temperature sensors for the cooling liquid may be used for
cross wires is photographed, together with a pinpoint-shaped both stationary and mobile measurements. They can either be
Co
pendulum end. The maximum accuracy obtainable is 3 mm placed inside a flange welded to the feed and return pipes or
on a 10-m-long borehole, 10 mm on a 20-m-long one, and 30 screwed into the pipes with fittings. More complicated sensor
mm on a 30-m-long one (Heinrich, Müller, and Voort 1979). versions allow them to be placed in the center of the pipe, with
In the electrical probe, the position of a tensioned wire or of a an insulating spacer to prevent thermal flow through the steel
rod on both sides of the pivoting point between the two pipe. Remote indication and automatic recording of tempera-
instrument arms is determined with the use of inductive trans- ture data guarantee constant control of the freezing progress
ducers. and allow an estimate of thermal energy released by the soil.
l
ria
ate
dM
hte
(a) Determine the average heat removal per meter of freeze Soil temperature measurement is almost always carried out
pipe per day. in observation boreholes. The use of a movable probe involves
rig
(b) Would a temperature drop of 4–5 °C between brine inflow lowering it by cable into a vertical borehole. In a horizontal
and return be indicative of heat removal? Explain. borehole, it is inserted with positioning rods or placed at the
position required with a cable, pully, and retracting cord. When
Solution: (a) Heat removal (J/m day) = brine flow rate (m3/
greater measurement accuracy is required, stationary probes
s) × brine specific gravity × water density (kg/m3) × time con-
are used so as not to cause any disturbing flow of the liquid in
py
l
ria
ate
dM
hte
rig
py
■ EXAMPLE 6.4-2: Temperature data at the 10-m depth for Frost Boundary Location and Wall Thickness
Co
two freeze pipes and a probe hole are given in Fig. 6-27. Esti-
Seismic measurements provide a means for observing the frost
mate the radius of the frozen soil columns (location of the zero-
advance in water-bearing soil or rock on the basis of an increase
degree isotherm).
in the velocity of the compression waves after water has
Solution: Assume a linear variation in temperature between changed into ice. The test procedure is similar to that used for
the freeze pipes and the probe hole. The calculations are as crosshole seismic testing (ASTM D-4428). In water, these waves
given in Table 6-2. travel at a velocity of about 1,500 m/s, in comparison with
Comment: More accurate methods for calculating the extent about 4,000 m/s in ice. This change in velocity gives a direct
of freezing around a freezing element are available. They indication of the change in state, from liquid to solid, for water
require accurate information on the frozen soil properties. in the soil pores. Soil and rock subjected to freezing may be
Without this information, it is normally assumed that a linear tested in situ by measurements between boreholes by placing a
temperature variation gives a reasonable estimate of the loca- seismic (energy) source in one borehole and a receiver (geo-
tion of the zero-degree isotherm. phone) in a second one. The increase of frozen soil is deter-
l
between adjacent boreholes is shown by lower velocities at
ria
those particular locations. Subsequent closure corresponds to
higher wave velocities, which will approach an upper limit for
the soil type and temperatures at the site.
ate
pipe.
6.5 Other Construction Considerations
Source: Reproduced with permission from Lacy and Floess 1988. Copyright
1988 Transportation Research Board.
The exposed frozen wall surface is susceptible to deterioration
dM and possibly unstable conditions due to several factors, includ-
hte
rig
py
Co
FIGURE 6-27 Probe hole location relative to freeze pipes for Example 6.4-2.
l
ria
unsaturated soils, particularly coarse clean sands and gravels,
will become unstable and slough due to sublimination of the
small amount of ice bonding them together. For most applica-
tions, these materials must be protected and/or mechanically
stabilized immediately after excavation.
ate
Flooding of an excavation supported by a frozen wall may
result in severe damage and possible wall collapse due to ther-
mal undermining. Shuster (1982) reported this problem to be
particularly bad in cohesionless soils. Extreme care must be
exercised in removing water from an excavation protected by
frozen earth. When pumping is required from inside the exca-
dM
vation, steel or plastic piping should be employed. Rubber or
fabric discharge hoses should not be used, except for temporary
daytime applications that are closely supervised. Discharge
hoses usually hang on the face of the excavation, and because of
their potential to leak or break, they represent a distinct threat
to a frozen earth wall. A small hole in a discharge line can emit
hte
ing (1) thermal load from sun, rain, and moving ambient air; danger to the frozen earth. Poorly controlled blasting may
(2) sloughing of partially saturated frozen granular soils due to result in the fracturing of refrigeration pipes and a loss of cool-
sublimation of the ice; and (3) improper construction methods ant into the soil. If a relief hole or free face is available for the
involving water and soil removal from the excavation. The blast to relieve into, controlled blasting may be completed with-
placement of concrete against frozen earth involves possible out difficulty. Water jetting requires closer supervision than
py
freezing and a reduced curing rate, which can prevent the con- blasting. Improperly supervised workers can inadvertently jet a
crete from developing adequate strength. These topics are dis- hole through a frozen earth wall in a matter of minutes.
cussed in this section.
Concrete Placement against Frozen Earth
Co
l
ria
ate
FIGURE 6-29 Temperature variation of concrete lining and soil adjacent to shaft. Concrete data: 7 sack-mix, type II cement, W/C
ratio = 0.45, no additives, placing temperature = 16.1 °C, concrete wall thickness = 355 mm; shaft data: shaft diameter = 9.1 m,
frozen shaft depth = 249.3 m, circulating brine temperature = –25 to –26 °C, brine circulation maintained continuously during
placement and curing of concrete.
Source: Reproduced from Shuster 1984.
dM
concerns for any application involving concreting against frozen some situations, it may be desirable to use a concrete mix mod-
earth include the following: (1) that the ground and concrete ified by one of the previously mentioned methods to hasten
will refreeze before the concrete hydrates sufficiently to achieve hardening and to increase the heat of hydration. For example,
its initial set; and (2) that the reduced curing rate of the concrete concrete may be subjected to higher working stresses before the
when frozen will prevent it from developing adequate strength structural frozen earth wall has thawed and been allowed ade-
before it must carry working stresses. quate normal curing, or for thinner sections. Cores taken from
hte
Normally, neither freezing nor the reduced curing rate pre- initial pours may be tested to confirm this approach.
sents any problem for ordinary concrete placed in sections
thicker than about 250 mm (Braun, Shuster, and Burnham
1979). For marginal or thinner sections, the heat of hydration Problems
and/or rate of set must be increased by using more or less, in
order of economy and desirability, any of the following: (1) a 6.1 A cross section of a frozen wall in a layered soil strata is
rig
lower water–cement ratio; (2) a richer mix design; (3) high shown in Fig. 6-6. Relative values of thermal conductivity and
early or regulated-set cement; (4) accelerating additives; (5) strength are given for the rock and for each soil type. Briefly
aluminous cement; or (6) high concentrations (9–15%) of cal- describe five problems that a contractor may encounter during
cium chloride. The use of high concentrations of a calcium the design and freezing of this wall. Suggest a solution for each
py