• Back to Encyclopedia Arctica homepage

    Soil Mechanics in Permafrost Regions

    Encyclopedia Arctica 2a: Permafrost-Engineering

    Unpaginated      |      Vol_IIA-0211                                                                                                                  
    EA-I. (Karl Terzaghi)




    Introduction 1
    Physical Properties of Soils 1
    Soil Exploration by Boring and Sounding 2
    Soil Testing 7
    Geophysical Methods 8
    Footing and Raft Foundations 8
    Pile Foundations 10
    Retaining Walls 11
    Stability of Side Slopes of Opencuts 11
    Slope of Fills 12
    Bibliography 13

    Unpaginated      |      Vol_IIA-0212                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics in Permafrost Regions



    Fig. 1 Sketch map of an area with discontinuous permafrost

    east pf Fairbanks, Alaska. Progressive degradation

    has reduced the permafrost to strips located beneath

    Fig. 2 Penetration diagram for test rails which were driven

    through sand and gravel without striking permafrost

    001      |      Vol_IIA-0213                                                                                                                  
    ( EA-I. (Karl Terzaghi)





            Soil mechanics deals with those physical properties of soils, including

    unconsolidated sediments and residual soils, which have a significant influence

    on the bearing capacity of the ground, the stability of slopes, and the earth

    pressure on artificial supports such as retaining walls or bulkheads. Soil

    mechanics also deals with the methods of subsoil exploration and with the

    application of the laws of mechanics and hydraulics to the solution of engineer–

    ing problems involving soils.



            In earthwork and foundation engineering many important decisions and all

    the semiempirical rules are partly based on experience. If experience is not

    correlated with an adequate description of the soils associated with the

    observed phenomena, and records of the observations are worthless of they

    may even be misleading. Therefore, one of the foremost tasks of soil mechanics

    was to establish simple and reliable procedures for discriminating between

    different soils and between different states of the same soil. The properties,

    such as grain-size characteristics, porosity, relative density, consistency, etc.,

    002      |      Vol_IIA-0214                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics

    on which the distinction is based are known as index properties and the

    soil test required for determining the index properties are called

    classification tests . Since the index properties also have a vital bearing

    on permafrost phenomena, they are described in detail in the article on

    “Physics of Ground Frost.”

            If an engineering problem such as the design of the foundation of an

    important structure on bad ground calls for elaborate computations, the

    classification tests must be supplemented by others, such as consolidation,

    permeability, or shear tests , to determine the numerical values of the soil

    constants which appear in the equations. General information on the values

    of these constants for unfrozen soils and on the methods of testing can be

    found in any textbook on soil mechanics (2). The properties of frozen soils

    are described in the article “Physics of Ground Frost.”



            The techniques of making test borings had already been developed at the

    beginning of the nineteenth century. However, up to about 1910, the boring

    tools such as augers, wash bits, and bailers were so constructed that they

    destroyed the structure of the soil and made the samples unfit for classifi–

    cation tests. When samples of mix-grained soils such as silty sand were hauled

    out of the hole, the finest soil fraction escaped from the sample and the

    material which was recovered gave an incorrect impression of the grain-size

    characteristics of the material in situ . Hence, as soon as the practical

    importance of the index properties of the soils was recognized, various tools

    for recovering complete and relatively undisturbed soils were constructed

    and only these should be used.

    003      |      Vol_IIA-0215                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics

            Samples of fine-grained, cohesive soils much as silt or clay are

    obtained by means of thin-walled, seamless steel sampling tubes with an

    inner diameter of about 2 inches and a length of 2 or 3 feet.

            The methods of securing samples of cohesionless soils (clean sand or

    gravel) depend on the position of the site of the sampling operation with

    reference to the water table. Above the water table, samples can most

    conveniently be secured on the bottom of test shafts. In order to get

    satisfactory samples from a test shaft at elevations located beneath the

    water table it is necessary to lower the water table by pumping from well

    points or filter wells to a level below that of the bottom of the shaft.

    In drill holes, undisturbed samples of cohesionless soil can be obtained

    only if the sampling operation is preceded by a solidification of the soil

    located beneath the bottom of the drill hole. This can and has been accom–

    plished by the injection of bituminous emulsions which solidify in the voids

    of the w s oil one or two hours after injection or by the freezing method s — — —

    (2, pp. 271-74).

            All these procedures, as well as the lowering of the water level for

    unwatering the site for test shafts, are cumbersome and expensive. However,

    in connection with cohesionless soils it is commonly sufficient to determine

    these grain-size characteristics and their degree of compactness. This

    information can be obtained without the recovery of undisturbed samples.

    The grain-size characteristics can be reliably determined by the mechanical

    analysis of disturbed samples, provided none of the constituents of the soil

    was lost during the transport of the sample from the bottom of the drill hole

    to the surface. Complete though disturbed samples of fine sand can commonly

    be obtained by driving sampling tubes with a diameter of 1½ or 2 in. and a

    004      |      Vol_IIA-0216                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics

    length of 22 in. into the sand below the bottom of the drill hole. In clean,

    coarse sand it may be necessary to use the scraper bucket, which is a tube

    with a drive point. The upper half of the bucket is provided with a vertical

    slit and one lip of the slit is flared out into a cutting edge. Complete

    samples of gravel have been obtained by means of a cylindrical sampling

    tube with a slightly constricted lower end. An inch or so above the cutting

    edge the inner walls are provided with an annular ridge, causing a local

    reduction of the clearance of the sampling spoon. If such a spoon is driven

    into gravel and then the spoon is pulled, the pebbles or likely to plus the

    space above the ring and to prevent the balance of the sample from slipping

    out of the spoon during recovery.

            Information on the degree of compactness of a deposit of cohesionless

    material is obtained by counting the number of blows of the drop weight,

    which are needed to drive the sampling spoon to a depth of one foot into the

    soil located beneath the bottom of the drill hole. A weight of 140 lb. and

    a drop of 30 in. are considered standard, and the penetration test performed

    under standard conditions is called the standard penetration test . Experience

    has disclosed the following approximate relation between the number of blows,

    N, per foot of penetration, and the corresponding degree of compactness of


    Number of blows, N ; : 0-4 4-10 10-30 30-50 over 50
    Degree of Compactness: very loose loose medium dense very dense

            Very satisfactory cores of frozen silt and sand, with a diameter of

    about 6 inches, have been obtained by the U.S. Army Engineers with rotary

    drills equipped with a saw-toothed cutter. When the equipment was used in

    frozen sand and gravel, difficulties were encountered. The outer layer of

    005      |      Vol_IIA-0217                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics

    the core melted on account of the heat developed by the rotating tool and

    the core slipped out of the barrel. Better results were obtained when the

    wash water was mixed with alcohol and cooled to about 32°F. On routine

    jobs, drill holes in frozen ground are still made with the chopping bit,

    which breaks the material up into angular fragments and rock powder. The

    fragments are recovered with a bailer or other suitable tools. By the time

    the fragments arrive at the surface, the ice may already have melted. There–

    fore, the drilling foreman cannot always be sure whether he has encountered

    frozen ground or a dense and slightly cemented layer of gravel which also

    delays the progress. Furthermore, if two layers of permafrost are separated

    by a layer of unfrozen ground or talik (see “Physics of Ground Frost”), the

    presence of the talik may easily escape the foreman’s attention. Hence, the

    technique of sampling of frozen ground still leaves a wide margin for


            If the geological structure of the subsoil is complex or errafic, the

    number of drill holes required for obtaining a reasonably accurate soil

    profile is prohibitive. In such instances one of several methods of sub–

    surface soundings can be used to advantage. All of them are based on the

    fact that a change of the type of soil is commonly associated with a change

    of the resistance of the soil against the penetration of a drive point.

    As a rule the drive point is given a conical shape. The entire tool, drive

    point included, is called a penetrometer .

            In soft, cohesive soil (silt or clay), the drive point is forced into

    the ground by static pressure and the pressure which acts onto the sounding

    rod while the drive point penetrates the ground is measured by suitable means

    at the upper end of the sounding rod (Swedish, Danish, and Dutch methods).

    006      |      Vol_IIA-0218                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics

    In sand and gravel the sounding rod is driven into the ground by means of a

    miniature pile driver. Whenever the drive point crosses the boundary between

    loose and dense material or vice versa, the number of blows per foot of pene–

    tration changes into (Ohio Swiss, and various other methods; 2, pp. 274-81).

            The diameter of the drive point of the customary penetrometers is equal

    to or small than 2 inches. Drive points with so small a diameter are

    inevitably attached to very slender sounding rods and they cannot be driven

    into coarse gravel. If the subsoil consists of coarse gravel or if it

    contains layers of coarse gravel, adequate information on the location of

    the boundaries between loose and dense material can be obtained by driving

    steel rails into the ground by means of a full-sized pile driver and by

    plotting in a diagram the number of blows per foot of penetration against

    total penetration. This method has also been successfully used along the

    southern margin of the permafrost zone in Alaska for determining the boundaries

    Fig 1 and 2 of the permafrost sheet. Figures 1 and 2 illustrate the results of such an

    investigation. Both the frozen and the unfrozen ground consisted of sand and

    gravel. The numerals that appear next to the small circles in Figure 1

    indicate the depth at which permafrost was encountered. Figure 2 shows the

    penetration diagrams for two of the rails which were driven entirely into

    unfrozen ground. Experience has led to the following tentative conclusions

    regarding the resistance of frozen ground against the penetration of test

    rails. If the number of blows per foot of penetration of a test rail

    rail with a weight of about 70 lb. per yard, hammer with a weight of 2,200 to 2,500 lb. and

    ( 70-lb. rail, 2200-ft. hammer a 10-ft. drop) is smaller than 10, the ground

    is probably unfrozen, and if it is greater than 12, it is probably frozen.

    If the number of blows per foot of penetration is between 10 and 12, one

    cannot know whether or not the ground is frozen. As a rule the increase

    006a      |      Vol_IIA-0219                                                                                                                  

    Fig 1

    006b      |      Vol_IIA-0220                                                                                                                  

    Fig 2

    007      |      Vol_IIA-0221                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics

    of the number of blows at the boundary of the permafrost is so conspicuous,

    that there cannot be any doubt regarding the location of this boundary.

    Otherwise, a few supplementary borings are required. However, before this

    procedure is standardized it will be necessary to collect reliable data

    regarding the resistance against the penetration of rails into frozen ground

    at temperatures close to the freezing point.



            The laboratory tests for determining the index properties of cohesionless

    soils consist of mechanical analysis, which furnished the grain-size character–

    istics, and determination of the moisture content. Cohesive soils required the

    determination of the natural water content, the Atterberg limits, and the

    unconfined compressive strength of relatively undisturbed samples (see “Physics

    of Ground Frost”). The necessary equipment includes a small drying oven with

    temperature control, accurate scales, and a few pieces of simple apparatus.

            To get reliable results, all the samples must be carefully protected

    against loss of water by evaporation prior to testing and samples of cohesive

    soils must be protected against freezing. In order to prevent the deformation

    of samples of cohesive soils prior to testing, the samples are taken with thin–

    walled tubes or with samplers containing liner tubes. As soon as the samples

    come out of the drill hole both ends are sealed with metal disks and paraffin,

    and the samples are kept in the tubes until the tests are started.

            To avoid long-distance transportation and delays the tests are commonly

    made in a small soils laboratory attached to the headquarters of the resident


    008      |      Vol_IIA-0222                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics



            Since about 1920 civil engineers have made an increasingly extensive

    use of geophysical methods for engineering purposes. Some of these methods,

    such as the electric potential method , are based on the fact that the geometry

    of every field of force depends on the location of the boundaries between the

    substances which occupy the field and on the difference between the physical

    properties of these substances. Others, known as seismic methods , are based

    on the fact that the rate of propagation of elastic waves is a function of

    the elastic properties of the materials through which the waves travel. (1). The

    equipment re a q uired for both electric and seismic surveys is compact enough to

    be accommodated, together with the field personnel, on a single car or jeep.

            Both methods have been successfully used for determining the boundary

    between sediments and rock. The application of geophysical methods to the

    determination of the upper boundary of the permafrost zone is still in an

    experimental sta t g e.



            If the bearing capacity of the top layers of the ground is adequate,

    structures can safely be established on spread footings, continuous footings,

    or rafts. The design of such foundations is commonly based on semiempirical

    rules (2, pp. 413-55). However, all these rules presuppose that the base of

    the foundation is located below the base of the active zone and that the

    subsoil does not contain permafrost. In temperate zones, such as the United

    States, this condition is easily satisfied. However, going north, the depth

    of the active layer increases and it become a maximum at the fring s e of the — —

    permafrost, which coincides approximately with the 0°C. isotherm of the

    009      |      Vol_IIA-0223                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics

    mean annual temperature (see “Physics of Ground Frost”). At that fringe

    it may attain a depth up to 15 feet.

            If the active layer is occupied by a coarse-grained soil such as clear,

    coarse sand and gravel, and if, in addition, the water table is permanently

    located beneath the base of the active layer, the base of foundations can

    safely be installed within the active layer. If the active layer is occupied

    by coarse-grained sediments with part of it located below the water table,

    alternate freezing and thawing may be associated with a slight heave and

    settlement of the surface and of the base of foundations located above the

    base of the active layer; but these movements are not necessarily objectionable.

    On the other hand, if the soil conditions within the active layer are conducive

    to the formation of ice layer (see “Physics of Ground Frost”) the ground move–

    ments associated with freezing and thawing can be very important and make it

    necessary to establish the base of foundations below the bottom of the active

    layer. In the estimate of the depth of the active layer, adequate allowance

    must be made for the effect of the future structure on this depth. On the

    north side of a large building the thickness of the active layer is likely

    to increase and on the south side it will decrease.

            If an important structure with provisions for heating is constructed

    on footings or a raft above a permafrost layer, the thickness of this layer

    will be decrease or the layer may completely disappear on account of the increase

    of the mean annual temperature over the area covered by the building

    (see “Physics of Ground Frost”). The ground movements associated with the

    local degradation of the permafrost depend on the thickness of the layer

    and the soil conditions within it. If the results of the subsoil exploration

    indicate the possibility of important movements, the engineer faces the problem

    either of preventing the movements, for instance, by thawing the frozen ground,

    010      |      Vol_IIA-0224                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics

    or else of eliminating the harmful effects of the movements by adequate

    design of the superstructure (see “Excavations and Foundations”).



            A structure is founded on piles if the bearing capacity of the soil

    located immediately beneath the structure is inadequate or if an estimate

    of costs indicates that a pile foundation may be cheaper than a shallow one.

    Piles may transfer the load either by side friction onto the surrounding

    ground or else through their point onto a firm stratum such as a layer of

    dense sand and gravel.

            South of the southern boundary of the permafrost region the depth to

    which piles can be driven is limited only by the subsoil conditions. However,

    in wintertime and within the active layer, the soil surrounding the piles may

    freeze and the frozen soil adheres to the piles. If the soil conditions are

    such that the freezing is associated with the formation of ice layers, and if,

    in addition, the permanent load on the piles such as those supporting a trestle

    is light, the frost may lift the piles many inches and the first application

    of the live load after the thaw would cause a settlement by an equal amount.

    In a region with a temperate climate, where the depth of the active layer did

    not exceed about 4 feet, timber piles were driver late in the fall through a

    stratum of saturated silt to refusal in dense sand. The construction of the

    foundations was postponed until next spring. After the frost went out of the

    ground it was found that the heads of the piles were located from 1 to 4 inches

    above their original elevation and it was necessary to redrive them.

    011      |      Vol_IIA-0225                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics

            If the point of a pile strikes permafrost, the pile cannot be driven

    any further. Yet after the building is constructed and heated, the perma–

    frost inevitably starts to degrade, whereupon the points of the piles lose

    their firm support and settlement ensues. If the degradation of the perma–

    frost involves the melting of pockets or layers of clear ice located below

    the level of the points of the piles, the settlement may even be catastrophic,

    though, at the outset, the bearing capacity of the piles was perfectly




            Retaining walls are designed to withstand the active earth pressure of

    backfills in an unfrozen state. If the backfill and ground-water conditions

    are such that winter temperatures may lead to the formation of ice layers,

    layers develop not only parallel to the top surface of the backfill, but also

    parallel to the back of the wall. During the growth of the vertical layers

    the wall is acted upon by a horizontal pressure far in excess of the active

    earth pressure and, as a consequence, the wall moves out. To prevent an

    excessive outward movement due to the freezing of the backfill, it is necessary

    to intercept the free communication between the backfill and the adjoining

    aquifers by adequate drainage provisions (2, p. 320). The drains convert the

    backfill from an open into a closed system (see “Physics of Ground Frost”).



            In the theories of the stability of slopes, it is assumed that the soil

    beneath the slope is acted upon by no forces other than those that are produced

    by the force of gravity. During the formation of ice layers the soil is also

    acted upon by the forces produced by the freezing of the water contained in

    012      |      Vol_IIA-0226                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics

    the soil. If the freezing is associated with the formation of ice layers,

    the process of freezing causes the body of soil to split open along the

    seats of layer formation and the growth of the layers is associated with an

    increase of the average water content of the soil next to the slope, which

    could not possibly be produced by the force of gravity. Furthermore, while

    the ice layers melt, the structure of the soil between the layers disintegrates.

    Therefore, the stability conditions that prevail on slopes during the thawing

    period are entirely beyond the scope of the conventional theories of the

    stability of slopes, and the soil movements that occur during the period of

    thawing have little resemblance to slope movements caused by gravity alone.



            The slopes of fills cannot be expected to be stable throughout the year

    unless the fills were made out of unfrozen ground. If a fill of fine-grained

    material such as silt or silty sand is located above an aquifer, ice layers

    may form parallel to its slopes during the winter season, whereby the water

    that enters the layers is drawn by capillarity out of the aquifer. Subsequent

    melting would cause a failure of the slopes by solifluction in spite of all

    the precautions which may have been observed during the construction of the

    fill. This danger can be eliminated or at least greatly reduced by inserting

    a well-drained layer of clean gravel between the base of the fill and the

    natural ground. This layer intercepts the capillary flow of water from the

    ground into the fill and converts the fill into a “closed system” (see “Physics

    of Ground Frost”). Further improvements can be achieved by covering the crest

    and the slopes of the fill with a relatively impermeable layer of soil.

    013      |      Vol_IIA-0227                                                                                                                  
    EA-I. Terzaghi: Soil Mechanics

    The presence of such a cover reduces the maximum degree of saturation of

    the fill material to less than 100 per cent. Hence, this precaution

    eliminates entirely the possibility of ice-layer formation within the fill.


    Karl Terzaghi


    1. Heiland, C.A. Geophysical Exploration . N.Y., Prentice-Hall, 1940.

    2. Terzaghi, Karl, and Peck, R.B. Soil Mechanics in Engineering Practice .

    N.Y., Wiley, 1948.

    Back to top