Soil and Environmental Analysis: Physical Methods - Chapter 9 doc

The li-quidity index, LI, is related to the percentage gravimetric soil water content, w%, the plastic limit, PL, and the plasticity index, PI, by w% ⫺ PL PI The activity, A, is the rati

Engineers found the limits, particularly the plastic limit, to be useful in thedesign and control testing of earthworks and soil classification (Dumbleton, 1968)

as a result of the development by Casagrande of apparatus to measure the limits(Casagrande, 1932) Although his apparatus was based on that of Atterberg, Casa-grande appreciated the need, where empirical tests were concerned, to specifyclosely every detail of the test procedure so that both the repeatability of the test

by one operator and the reproducibility between operators were optimized wood, 1970) Consequently, the Casagrande tests became widely adopted as the

(Sher-official standard by engineers in the United Kingdom (BSI, 1990), the UnitedStates of America (Sowers et al., 1968), and elsewhere.

Soil scientists have made less use of the Atterberg limits, which do not ture in soil survey or land capability classification systems but have been usedmainly as indicators of the likely mechanical behavior of soil (Baver et al., 1972;Archer, 1975; Campbell, 1976a) This has generally been done by establishingsimple correlations between the plasticity limits or plasticity index and other prop-erties considered important in determining soil behavior An example is shown inTable 1 It has been suggested, however, that liquid and plastic limit values would

fea-be a useful addition to soil particle size distributions in the classification of soils

in the laboratory (Soane et al., 1972) This is particularly relevant as the Atterberglimits are related to the field texture, as determined in the hand, a method oftenpreferred by soil scientists concerned with practical problems of soil workability

in the field (MAFF, 1984)

Two further index values can be derived from the Atterberg limits The

li-quidity index, LI, is related to the percentage gravimetric soil water content, w%, the plastic limit, PL, and the plasticity index, PI, by

w% ⫺ PL

PI The activity, A, is the ratio of the plasticity index to the percentage by weight of

soil particles smaller than 2mm, C, thus

Table 1 Relation Between Potato Harvesting Difficulty,

as Indicated by the Number and Strength of Clods in Potato

Ridges, and Plasticity Index of Soil

30 – 45 mm

diameterclods (N) (A)⫻ (B)

Plasticityindex

II THEORIES OF PLASTICITY

In attempting to explain the mechanism behind the existence of the liquid andplastic limits, two basic approaches have been adopted Traditionally, soil behav-ior is considered in terms of the cohesive and adhesive forces developed as a result

of the presence of water between the soil particles (Baver et al., 1972) The criticalstate theory of soil mechanics that is used in the second approach has been detailed

by Schofield and Wroth (1968) and is mathematically complicated However, thebasic concepts and their importance have been discussed by Kurtay and Reece(1970)

A Water Film Theory

Cohesion within a soil mass is due to a variety of interparticle forces (Baver et al.,1972) Bonding forces include Van der Waals forces; electrostatic forces betweenthe negative charges on clay particle surfaces and the positive charges on the par-ticle edges; particle bonding by cationic bridges; cementation effects of sub-stances such as iron oxides, aluminum, and organic matter; and the forces associ-ated with the soil water Taken together, these forces will determine whether a soilwill, when stressed, undergo brittle failure, plastic flow, or viscous flow

At low water contents, most of the soil water forms annuli around the particle contact (Haines, 1925; Norton, 1948; Schwartz, 1952; Kingery andFrancl, 1954; Vomocil and Waldron, 1962) These annuli provide a tensile forcethat increases with decreasing particle size, through this relationship breaks down

inter-at higher winter-ater contents because the individual annuli of winter-ater start to coalesce(Haines, 1925) Just above the plastic limit, the soil becomes saturated, and, in acohesive soil, the soil water tension and other bonding forces are in equilibriumwith the repulsive forces due to the double layer swelling pressure Nichols (1931)

showed that, for laminar clay particles, the interparticle force F was related to the particle radius r, the surface tension of the pore water T, the angle of contact

between the liquid and the particlea, and the distance between the particles d, by

where k is a constant He also showed that, for each of three soils, the product of

the cohesive force and the water content was a constant at low water contents Athigher water contents, however, the cohesive force decreased rapidly with increas-ing water content

Although the existence of a relationship between water content and sion, which exhibits a maximum, has been demonstrated experimentally (Nichols,1932; Campbell et al., 1980), the relation is valid only for dry soils that have beenrewetted When puddled soil is allowed to dry, cohesion increases with decreasingwater content and reaches a maximum when the soil becomes dry This effectprobably arises because, in puddled soils, the number of interparticle contacts aremaximized, and hence cohesive forces other than those due to soil water are large.Baver (1930) suggested that when a soil at the plastic limit is stressed, thelaminar clay particles, which are each surrounded by a water film and which werepreviously randomly orientated in the friable state, are rearranged so that theyslide over each other Thus the cohesive forces associated with the tension effects

cohe-in the water films are overcome, and the soil deforms When the stress is removed,the particles remain in their new position under the action of the cohesive forcesand there is no elastic recovery The soil has undergone plastic deformation orflow Before the soil reaches the liquid limit, the water films have completely coa-lesced, and the soil water tension has greatly decreased Thus cohesion decreasesand the soil is capable of viscous flow As the water content and particle separationfurther increase, the liquid limit is reached, and the viscosity of the outermostlayers of water is reduced to that of free water, allowing the soil to flow like aliquid (Grim, 1948; Sowers, 1965)

The liquid limit is related to clay content and its surface area for most types

of clay mineral Montmorillonite is an exception in that the liquid limit is trolled essentially by the thickness of the diffuse double layer, thereby giving alinear relation between the liquid limit and the amount of exchangeable sodiumions present (Sridharan et al., 1986)

con-Although the interparticle forces associated with soil water may not provide

a comprehensive explanation of the mechanism of plasticity, it is clear the soilparticle sizes, their specific surface, and the nature of the clay minerals are allimportant This is consistent with the common experience that, generally, the liq-uid and plastic limits are both dependent on both the type and the amount of clay

in a soil (DSIR, 1964)

B Critical State Theory

If a relatively loose sample of soil is subjected to a progressively increasing axial (deviatoric) stress while the confining stress (spherical pressure) is kept con-stant, then the soil volume will decrease This will occur for both unsaturated soil

uni-and soil that is saturated but allowed to drain as it is compressed Eventually, apoint will be reached where the soil can be compressed no further However, if thedeviatoric stress is maintained and the soil continues to distort without any change

in volume, then the soil is said to be in the critical state In terms of the dimensional relationship of spherical stress, deviatoric stress, and specific vol-ume, the point describing this critical state is one of the many possible criticalstate points that together form the critical state line The critical state line is anextremely important concept in that it allows, within the confines of a singletheory, the stress–strain behavior of a soil with any particle size distribution to beexplained, be it wet or dry, dense or loose, confined or unconfined

three-As the line describes all conditions under which a soil will undergo ous remolding without a change in volume, it follows that soil being prepared foreither the liquid or the plastic limit test must be described by a point on this line.Thus the liquid and plastic limit tests can give more than simple qualitative infor-mation about soil behavior

continu-During the liquid limit test, the soil water content, and hence the specificvolume, is adjusted by adding water and remolding the soil until, in effect, the soilhas a fixed undrained shear strength determined by the conditions of the test Be-cause the soil is continuously remolded as water is added, it is in the critical stateand under the action of a negative pore water pressure

When soil is prepared for the plastic limit test, it is continuously remoldedand hence once again is in the critical state However, since the soil is much drierthan in the liquid limit test, the pore water pressure (matric potential) is even morenegative This negative pore water pressure acts in the same way as if the soil weresubject to an additional externally applied stress and serves to increase the shearstrength of the soil It is reasonable to speculate that the plastic limit should, likethe liquid limit, correspond to a state in which the soil has a fixed undrained shearstrength Atkinson and Bransby (1978) reported that the undrained shear strengthdata obtained for four clay soils by Skempton and Northey (1953) revealed thatall four soils had very similar undrained shear strengths at the plastic limit Per-haps more remarkably, the undrained shear strength of each soil at the plastic limitwas almost exactly 100 times the undrained shear strength at the liquid limit.Knowing the ratio of the shear strengths at the liquid and plastic limits, it ispossible to define the slope of the critical state line on a plot of the logarithm ofthe spherical pressure versus the specific volume in terms of the plasticity index(Schofield and Wroth, 1968; Atkinson and Bransby, 1978) Thus the plasticityindex can be used as a direct indicator of soil compressibility

The description of soil behavior at the liquid and plastic limits offered bycritical state theory is, at first sight, quite different from that given by the waterfilm theory and may give the impression that soil water content is irrelevant How-ever, the water content is important in critical state theory, but only insofar as itaffects the pore water pressures

III DETERMINATION OF THE LIQUID AND PLASTIC LIMITS

The methods initiated by Atterberg (1911, 1912) and subsequently developed byCasagrande (1932) were adopted by the British Standards Institution and theAmerican Society for Testing and Materials as the standard tests in civil engineer-ing However, in 1975, a new test for the liquid limit, based on a procedure in-volving a drop-cone penetrometer, was introduced and is included in the currentBritish Standard (BSI, 1990) The Casagrande tests were retained, but the conepenetrometer method was described as the preferred method for the determination

of the liquid limit Although various other methods of determining the liquid andplastic limits have been suggested, usually, but not always, based on correlation

of the limits with other soil rheological properties, by far the most widely usedmethods are the Casagrande and, to a lesser extent, drop-cone tests

A Casagrande Tests

In the Casagrande liquid limit apparatus (BSI, 1990) (Fig 1), the sample is tained in a cup that is free to pivot about a horizontal hinge and which rests on arubber base of specified hardness A rotating cam alternately raises the cup 10 mmabove the base and allows it to drop freely onto the base The test soil is mixedwith distilled water to form a homogeneous paste, allowed to stand in an air-tightcontainer for 24 hours and remixed, and then a portion is placed in the cup Thesample is divided in two by drawing a standard grooving tool through the sample

con-at right angles to the hinge The crank is then turned con-at two revolutions per seconduntil the two parts of the soil come into contact at the bottom of the groove over alength of 13 mm The number of blows to the cup required to do this is recordedand the test repeated If consistent results are obtained, a subsample of the soil istaken from the region of the closed groove for the measurement of water content.More distilled water is added to the test sample and the procedure repeated This

is done several times at different water contents to give a range of results lyingbetween 50 and 10 blows The linear relation between the water content and thelog of the number of blows is plotted, and the percentage water content corre-sponding to 25 blows is recorded, to the nearest integer, as the liquid limit ofthe soil

A simplified test procedure for liquid limit determination using the grande apparatus is that known as the ‘‘one point method.’’ Essentially the methodinvolves making up a soil paste such that the groove cut in the sample in the cupcloses at a number of blows as close as possible to 25, and certainly between 15and 35, blows A correction factor, which varies with the actual number of blows,

Casa-is applied to the water content of the soil to give the liquid limit (BSI, 1990) Themethod has the advantage of speed, but this is at the expense of reliability (Nagarajand Jayadeva, 1981)

For the Casagrande plastic limit test (BSI, 1990), the sample is mixed withdistilled water until it is sufficiently plastic to be molded into a ball A subsample

of approximately 10 g is formed into a thread of about 6 mm diameter, and thethread is then rolled between the tips of the fingers of one hand and a flat glassplate until it is 3 mm in diameter The thread is then remolded in the hand to drythe sample and again rolled into a thread The operation is repeated until the threadcrumbles as it reaches a diameter of 3 mm A second subsample is similarly tested,and the mean of the two water contents (expressed as percentages) at which thethreads crumble on reaching a diameter of 3 mm is recorded, to the nearest integer,

as the plastic limit of the soil Where the plastic limit cannot be obtained or where

it is equal to the liquid limit, the soil is described as nonplastic

Fig 1 The Casagrande grooving tool and liquid limit device, showing a soil sampledivided by the tool prior to testing

Both these tests are undertaken on air-dried material passing a 425 mmsieve, although it has been susggested that, when the bulk of the soil materialpasses 425 mm, it may be more convenient to test the whole soil (BSI, 1990).However, it is generally agreed that the results for soils tested in the natural con-dition may be different from tests conducted on material that has previously beenair-dried, and this is certainly the case when soils are at above-ambient tempera-tures (Basma et al., 1994) This is particularly true of organic soils Where anappreciable proportion of the soil is retained on the 425mm sieve, removal of suchmaterial can influence the plasticity characteristics of the soil (Dumbleton andWest, 1966) Because of these various aspects of the test procedures and becausethe tests are conducted on remolded soil, the results should be interpreted withcaution in relation to the likely behavior of soil in the field.

B Drop-Cone Tests

Most of the shortcomings of the Casagrande liquid limit test are related to itssubjectivity and to the tendency for some soils to slide in the cup or liquefy fromshock, rather than flow plastically (Casagrande, 1958) After reviewing five alter-native cone penetrometer tests, Sherwood and Ryley (1968) concluded that amethod developed by the Laboratoire Central des Ponts et Chausse´es, 58 Boule-vard Lefebre, F-75732 Paris Cedex 15, France (Anon., 1966) offered the possi-bility of a suitable method for liquid limit determination The new method, whichused apparatus already available in most materials testing laboratories, was shown

to be easier to perform than the Casagrande method, to be less dependent on thedesign of the apparatus, to be applicable to a wider range of soils, and to be lesssusceptible to operator error Largely as a result of the work of Sherwood andRyley (1968), the drop-cone penetrometer test was adopted as the preferredmethod for liquid limit determination by the British Standards Institution (BSI,1990) in the United Kingdom

The apparatus used in the drop-cone penetrometer test is shown inFig 2.The mass of the cone plus shaft is 80 g, and the cone angle is 30⬚ The test soil,which is prepared to give a selection of water contents in exactly the same way as

in the Casagrande test, is contained in a 55 mm diameter, 50 mm deep cup Ateach water content, the soil is pushed into the cup with a spatula, so that air is nottrapped, and then levelled off flush with the top of the cup The cone is lowereduntil it just touches the soil surface, and the cone shaft is allowed to fall freely for

5 s before the shaft is again clamped and the cone penetration noted from the dialgauge Usually, the 5 s release is automatically controlled via an electromagneticsolenoid clamp as shown in Fig 2 A duplicate measurement is made, and theprocedure is then repeated for a range of water contents The linear relation be-tween cone penetration and water content is plotted, and the percentage watercontent corresponding to a penetration of 20 mm is recorded, to the nearest inte-

ger, as the cone penetrometer liquid limit Typical test results for four soils areshown inFig 3.

Attempts have been made to develop a one-point cone penetrometer liquidlimit test analogous to the one-point Casagrande test As with the latter, themethod is a compromise between speed and accuracy but has been shown to be

a satisfactory alternative (Clayton and Jukes, 1978) The one-point cone trometer test has been shown to be theoretically sound and not based simply onstatistical correlations (Nagaraj and Jayadeva, 1981)

Fig 2 The drop-cone penetrometer, showing the cone position at the start of a test

The drop-cone liquid limit method has been compared with the Casagrandemethod for a range of soils used in civil engineering (Stefanov, 1958; Karlsson,1961; Scherrer, 1961; Sherwood and Ryley, 1968, 1970a, b) and agriculture(Towner, 1974; Campbell, 1975; Wires, 1984) Generally, the two tests giveequivalent results (Littleton and Farmilo, 1977; Moon and White, 1985; Sivapul-laiah and Sridharan, 1985; Queiroz de Carvalho, 1986) A comparison of the twomethods is shown inFig 4, which also shows the reproducibility of the drop-conemethod.

With the widespread adoption of the drop-cone method for measuring theliquid limit, there were obvious advantages in using the same apparatus to measurethe plastic limit, if that were possible Scherrer (1961) proposed a method of plas-tic limit determination that involved extrapolation of the linear relation between

Fig 3 The results of cone penetrometer liquid limit tests on four arable topsoils of trasting texture The horizontal broken line indicates the cone penetrometer liquid limit.(From Campbell, 1975.)

con-water content and cone penetration found in the region of the liquid limit butconceded that the necessary extrapolation implied possible sources of inaccuracy

in the method In fact, Towner (1973) showed that, although the water content /cone penetration relation is linear in the region of the liquid limit, it becomesnonlinear at lower water contents, tending to show a minimum penetration Camp-bell (1976b) made detailed measurements of the water content /cone penetrationrelations for 18 soils and found a pronounced minimum in the curve for each soil

in the region of the Casagrande plastic limit Results for three of the soils areshown inFig 5 The water content corresponding to the minimum of the curvewas always numerically less than, but correlated closely with, the plastic limit Itwas suggested that the plastic limit be redefined as the water content correspond-ing to the minimum of the curve and that it be referred to as the cone penetrometerplastic limit The possibility of the establishment of a fixed penetration value cor-responding to the plastic limit was considered (Towner, 1973; Campbell, 1976b;Allbrook, 1980) but was dismissed because variation in penetration between soilswas unacceptably high (Campbell, 1976b) The cone penetrometer plastic limitwas shown to offer reduced operator errors and to be a good indicator of soilbehavior in an examination of the variation with water content of soil cohesion,soil–metal friction, susceptibility to compaction, implement draught, and theslope and intercept of the virgin compression line of critical state soil mechanics

Fig 4 The relation between the cone penetrometer liquid limit, as determined by twooperators, and the Casagrande liquid limit determined by operator 1 for some arable top-soils (From Campbell, 1975.)

theory For a given soil, all these relations were shown to exhibit turning points at

a water content corresponding to the cone penetrometer plastic limit (Campbell

et al., 1980)

A distinct approach to the use of the cone penetrometer to measure the ticity index was made by Wood and Wroth (1978) They suggested that the plasticlimit be redefined so that the undrained shear strength at the plastic limit is onehundred times that at the liquid limit The proposal was based on the assumptionthat all soils have the same strength at their liquid limits, and this was shown to bereasonable Further, it was shown that the proposal allowed a unique relation to

plas-be developed for remolded soil plas-between strength and liquidity index and also plas-tween compression index and plasticity index (Wroth and Wood, 1978)

Fig 5 Water content /cone penetration relations for three soils of contrasting texture inrelation to the Casagrande liquid (LL) and plastic (PL) limits Results obtained by twoindependent operators are shown (From Campbell, 1976b.)

characteristics There have been attempts to relate the liquid and plastic limits tospecific viscosities (Yasutomi and Sudo, 1967; Hajela and Bhatnagar, 1972), tothe residual water content of a soil paste subjected to a standard stress (Vasilev,1964; Skopek and Ter-Stephanian, 1975), and to various mechanical properties(Sherwood and Ryley, 1970a) However, none of these alternative methods hasbeen widely adopted.

D General Considerations

As both liquid and plastic limit tests are empirical, it is important that the testprocedures be closely specified, if consistent results are to be obtained Mosttest procedures specify that the soil should first be air-dried and then sievedthrough a 425mm sieve (BSI, 1990), although wet sieving through a 425 mm sievefollowed by air-drying has been proposed (Armstrong and Petry, 1986) However,

it has been suggested that in some circumstances either air-drying (Allbrook,1980; Pandian et al., 1993) or removal of any soil particle size fraction ( Dumble-ton and West, 1966; Sivapullaiah and Sridharan, 1985; BSI, 1990) can markedlyaffect the result obtained The development of a practical in situ test might bedesirable, but it is unlikely because of the difficulty in obtaining an appropriatesequence of test water contents without the complication of hysteresis effects asthe soil alternately wets and dries in a random way (Campbell and Hunter, 1986).Such effects, probably together with cementation effects, have led to the need forsamples prepared to a given water content to be thoroughly mixed (Sowers et al.,1968) and allowed to cure for 24 hours before being tested (BSI, 1990), althoughthe latter is not universally agreed to be necessary (Gradwell and Birrel, 1954;Moon and White, 1985) In addition, sample preparation may be complicated bythe fact that some soils undergo irreversible changes on drying (Allbrook, 1980),while other soils may give index values that depend on the number of times thetest sample is remolded and cured prior to the test, especially where the liquidlimit is concerned (Coleman et al., 1964; Davidson, 1983) The latter effect isthought to be due to particularly stable aggregates that break down only withprolonged remolding (Coleman et al., 1964; Sherwood, 1967; Pringle, 1975;Blackmore, 1976)

Although the standard test for the liquid limit using the drop-cone trometer includes a check on the sharpness of the cone used (BSI, 1990), Houlsby(1982) concluded that, in contrast to the work of Sherwood and Ryley (1970b),the effect of variations in cone sharpness was very small compared with the effect

pene-of the roughness pene-of the cone surface Both the cone angle (Budhu, 1985) and thecone mass (Budhu, 1985; Campbell and Hunter, 1986) affect the penetration ob-tained Large variations in temperature affect the Casagrande liquid and plasticlimits appreciably, due to variation in water viscosity (Youssef et al., 1961)