Cracking in Cement-Treated Bases and Means for Minimizing It

Transcription

1 Craking in Cement-Treated Bases and Means for Minimizing It KALANKAMARY P. GEORGE, University of Mississippi This report desribes the auses and ontrol of raking of pavements, with speifi referene to ement-treated bases. In order to study the variables influening raking of ement-treated bases, analytial expressions for both rak spaing and rak width were derived. The rak spaing (L) is influened by the tensile strength, the oeffiient of sliding frition (µ) and speifi weight of the material (y). The rak width (ot), primarily a funtion of the total maximum shrinkage, is to some degree influened by u, y, L, and the modulus of elastiity in tension (Et). A simple expedient to minimize raking would be to ontrol the shrinkage of the ement-treated soil. A searh for treatments to redue shrinkage led to several promising additives; lime and fly ash proved to be the best and sulfates in appropriate onentrations, partiularly those of magnesium and sodium, appear to be effetive. lviixing ement and soils redues shrinkage beause the ement matrix tends to restrain the movement of the soil; nevertheless, the resulting produt undergoes some shrinkage due to moisture loss. Results of a study on the shrinkage harateristis of ement-treated soil are reported elsewhere in this RECORD (1). A ement-treated base that is trying to ontrat due to internal. hanges, if fully or partially prevented from doing so, will be stressed in tension and usually in shear. When the ultimate tensile strength of the material is exeeded, raks begin to form. This study is onerned with the problem of building pavement bases with fewer raks and minimizing the rak width. A simplified theoretial analysis of the rak-spaing and rak-width problem is presented. It is possible that the intensity of raking an be ontrolled by reduing the shrinkage of the soil-ement through treatment with trae additives. SIMPLIFIED THEORETICAL ANALYSIS OF CRACKING Craks in ement-treated bases may be due to two fators: ambient temperature and hanges in moisture ontent. Calulations indiate that the ontration and expansion due to hanges in temperature are insignifiant ompared to shrinkage and swelling due to drying and wetting. For example, for a temperature differential of about 3 F, the strain is only about. 2 perent whereas the shrinkage for a typial sand- lay topping due to drying out is O. 2 perent. For this reason, emphasis in this study is on the auses and ontrol of transverse raking aused by drying out. The analytial disussion that follows will onentrate on rak spaing and rak width, whih determine to an important degree the damage due to raking in a ementtreated base. Crak Spaing As a result of linear shrinkage, tension stresses an be set up in ement-treated base slabs. If the slab is free to move (no frition between the slab and the subgrade), Paper sponsored by Committee on Soi I-Portland Cement Stabilization and presented at the 47th Annual Meeting. 59

2 6 stresses will not result. However, if frition exists between the slab and subgrade, restraint results from the frition fores. Figure la shows the fores ating on a ontrating slab. The stress distribution due to subgrade frition is shown in Figure lb. The movement of a ontrating slab is inreased from zero at the enter to a maximum toward the free end, as is the fritional resistane. For equilibrium onditions, the summation of the frition fores from the enter of the slab to the free end must be equal to the total tension in the slab. Balaning total fores in Figure la, CY bh where a tensile stress at enter of slab, psf; b breadth of pavement, ft; h depth of pavement, ft; µ. oeffiient of sliding frition; y unit weight of material, pf; and L length of slab, ft. (la) I Cente of Slob I {==:: (o).?'.:-y..--:-:-r.. t::: :: : :: :: ::: : :: :::...: Stress--- (b) Fore r r () Figure l. Stress resulting from shrinkage; (o) fore oting on shrinking slob, (b) voriotion of stress with length, ond () fore diogrom. The slab length (Lmax) at whih tensile stress will beome ritial is as follows: where au = ultimate tensile strength, psf. In other words, for a speifi slab plaement the spaing of raks is diretly related to the tensile strength of the material. Crak Width Let us onsider a slab with raks at L-ft intervals, as in Figure 2a. The slab ontrats from both ends while the enter portion of the slab is assumed to remain stationary (Fig. 2b). The rak width is thus influened by two opposing fators: the tendeny of soil-ement to shrink, ompensated to some exlenl uy Lhe exleusiuility of the material. Aordingly, the width of rak (o'er) will be the differene between the ontration due to shrinkage of the slab (o ), 1 assuming no frition, and the elongation of the same setion of the slab (o J due to fritional resistane. To make the derivation more general, it is assumed that at the time of raking, the material has not attained Center of Slab I.t=a=r I t-rd I r==jl : u } (a) Figure 2, Transverse roks in o pavement bose; (o) roked bose, ond (b) setion onsidered for rok width olulotion. (b) (lb) its maximum shrinkage. Let o'er denote the rak width immediately after raking while o r refers to the subsequent widening of the rak. Then L/2 2 /dx Et

3 61 And where e-r Et shrinkage at raking, in./in.; modulus of elastiity of ement-treated soil in tension, psf; and 6" er = (Ee - Er)L where Ee = total shrinkage, in. /in. In Eq. 2b, it is taitly assumed that the oeffiient of sliding frition remains unhanged from that before raking; hene the narrowing of the rak due to the extensibility of the slab beomes zero. Total width of rak, therefore, will be obtained by ombining Eqs. 2a and 2b. Thus, where 6T = total rak width, ft. Fators Affeting Crak Width and Crak Spaing Narrow raks, at the widest spaing possible, is the objetive in a soil-ement base. From Eq. lb it is evident that rak spaing is influened byµ, O'u, and y. Eq. 2 reveals that rak width is primarily a funtion of the total maximum shrinkage (Ee) and is to some extent influened by u, y, L and Et. The effet of sliding frition (u) on rak spaing an be seen from Eq. lb, whih indiates that the rak spaing is inversely proportional to the frition oeffiient. To evaluate its influene on rak width, however, the partial derivative of fjt with respet to u. is determined. Substituting L = 2 ' in Eq. 2 and performing the difu. y ferentiation, we get (2a) (2b) (2) O' Sine 2 E > > 2 Et, for normal values of O' and Et, the slope of OT vs µ. is al ways negative. It is therefore onluded that the rak width always dereases with inrease in subgrade frition. Similar reasoning may be advaned to interpret how other fators, i.e., tensile strength and speifi weight, might influene rak spaing and rak width. Table 1 summarizes these results. The signifiane of these fators is emphasized by an index number suh as 1 or > 1. The use of this index number an best be illustrated by an example. From the results in Table 1, the hange in sliding frition appears to have greater influene on rak spaing than on rak width. Also, the rak width inreases with rak spaing, though not linearly. TABLE 1 DEPENDENCE OF VARIOUS FACTORS ON CRACKING Craking Coeffiient of Sliding Frition (I') Tensile Strength (au) Speifi Weight (y) Crak spaing Inreases with Crak width dereases with Derease >1 Inrease 1 Inrease >1 Derease 1 Derease Derease

4 62. e :> U) 1 8 Sieve Sizes, U.S. Standard. 8 Q :!; i...- _...-::. v If '/.' /,/ Kl5 M3---/ // K25-j.' J 6 Upl. vpl. /.l---k32 "17-( Soil MolA. '""""It / I :.1 K M /v /.JJ Kl K f K J?nn. < - -- :,"":'!" ,... V' M K ,2 = K N Partiia Sile millimelers C lo l Silt Sand l Grovel K27 i' K31-f, f- -K3 Figure 3. Partile size distributions of soils. One of the limitations of Eqs. lb and 2 is the fat that they are appliable only to elasti materials. Sine ement-treated soil exhibits reep, it is only partly elasti. It may be asserted that the higher the reep, the greater the rak spaing, and the smaller the rak width. INFLUENCE OF ADDITIVES ON SHRINKAGE A positive appraoh to minimize raking in ement-treated pavements would be to ontrol the shrinkage of the treated material. Some of the fators that regulate the shrinkage of the soil-ement mix are disussed elsewhere in this RECORD (1). This report evaluates the ompetene of various additives in reduing shrinkage Of ementtreated soil. Materials and Proedures Eight soils with partile-size distributions (Fig. 3) were used. The preeding paper (1, Table 1) lists ompositional data, physial properties, and lassifiation of these soils. Eah soil is identified by a 1-letter, 2-digit system; for example, K3 means No. 3 soil, with kaolinite as predominant lay mineral. Various additives tested are listed in Table 2. A desription of the proedure for preparation and testing of beam speimens is given elsewhere in this RECORD (1). Harvard miniature samples were used for un- - onfined ompression testing. The TABLE 2 shrinkage result reported for a spe Lime Material Calium hloride Fly ash Pozzollth Expansive ement Gypsum Sodium sulfate Magnesium sulfate Sodium hydroxide Cationi emulsion SS-K ADDITIVES TESTED Soure United States Gypsum Company, New Orleans Reagent grade Detroit Edison Co., Detroit The Master Builders Co., Cleveland Penn-Dixie Cement Corp., Nazareth United States Gypsum Co. Reagent grade Reagent grade Reagent grade Chevron Asphalt Co., Tuson ifi variable is the average of two speimens and the strength results is the average of three speimens. Effet of Additives The additives investigated and found benefiial are broadly grouped aording to the prinipal mehanism responsible for their effetiveness. They are lime and alium hloride, widely known for the ation exhange

5 'E CD CD "". U) K3 w u " N :;, M7 Kl5 K25 K27 M3 K31 K32 Soll Figure 4. Effet of lime and alium hloride on shrinkage of soi I-ement. 63 properties; fly ash; pozzolith 8; and, to some extent, alium hloride. These improve the workability of the mix and thereby inrease the density and/or derease the optimum moisture. Expansive ement and sulfates of alium, sodium, and magnesium expand and partly ompensate for the shrinkage. Lime-Depending on the lay ontent (2-µ) lime proportions were varied from 2 to 3 perent. In the soil-ement blends, lime replaed an equal amount of ement. In virtually all soils studied, shrinkage was redued by blending trae amounts of lime (Fig. 4). Typially, 3 to 4 perent redution in shrinkage was observed; in a few sand soils it was as muh as 6 perent. This finding is in general agreement with the reported results (2, 3). Some other advantages in using lime with ement-treated soil are the improved workability and inreased ompressive strength ( 4). The ompressive strength was slightly inreased in two soils, and dereased in two others (Fig. 4). The slight redution in strength of M3, a friable loess, is in agreement with the reported findings. Pinto et al ( 4) observed that for a friable loess-ement, addition of lime aused a gradual but small redution of strength. One of the important reations of lime with lay, repeatedly doumented in the literature is aggradation aused by floulation. Lime floulates lay more effetively than ement. The smaller shrinkage on drying for floulated lay than for dispersed lay is illustrated by Seed and Chan (5), and others. In two ement-treated soils quiklime was substituted for hydrated lime. Results indiate that the former was nearly as effetive as the latter. Calium Chloride-As little as. 5 perent alium hloride, substituted for 1 perent of ement, redued the shrinkage in four out of six soils tested (Fig. 4). A omparison shows that lime redued shrinkage somewhat more than alium hloride. The theory that alium hloride assumes the role of an aelerator for ement hydration and, in so doing, beomes less effetive is in keeping with the results reported by the author (1) that shrinkage inreased with the rate of ement hydration. Repeatedly ited in the literature (6, 7) (but not onsidered in this study) is the hypothesis that under the same ompative -effort the dry density of a hloride-treated soil is often inreased with a orresponding derease in optimum moisture. This hypothesis is substantiated in two soils, K3-6 and M3-1, where it is observed that with. 5 perent alium hloride, the inrease in dry density is 1. 2 and 1. 1 pf and the derease in optimum moisture is. 9 and. 8 perentage points, respetively. So far as shrinkage is onerned this result is signifiiant, sine the writer's study (!)

6 64 8 ;: : Q) ti Q) > ;;; Q) Ci. u 6 u u,,!! - 4 >- 2 > " () " > ;;; w *.,. u,.., () " > w *... u K3 M7 M3 K32 Figure 5. Unonfined ompressive strengths of soi I-ement with various additives. Speimens 7-day ured, 1-day immersed. Soii points out that inreasing the dry density and/or dereasing the molding moisture tends to redue shrinkage. Diret experimental evidene to this effet an be found in Wood (8), who reports that in field test setions the alium hloride treated setions were found to be free of raks and showed no failure. From the strength results (Fig. 5), it appears that alium hloride does not improve the strength on replaing ement in soil-ement. Clare and Pollard (9), however, report that in soils ontaining ative organi matter, alium hloride results in marked improvement in strength. In onlusion, it is postulated that either a poorly reating sand or a soil that presents ompation problems ould be benefited from alium hloride; a typial example..: " (I) " :;:: u i 2 a: <:..., Cl Clay Content, per ent K32 Figure 6. Redution in shrinkage of fly ash-treated soi 1- ement as perent of untreated mix. Four perent ement and 4 perent fly ash in soi Is K3, K 15, K25, and K27. Seven perent ement and 6 perent fly ash in soi Is M3 and K32. is soil K31. Of the seven admixtures tested in this soil only three were effetive; of the three, alium hloride proved to be the best. Fly Ash-Although the use of fly ash in mass onrete has been extensively studied, there have been only a few reports on its use in soil-ement. The strength results of soil-ement on adding fly ash onflit (1, 11, 12, 13). Some of the benefits in onrete, repeatedly doumented, are dereased shrinkage, improved workability, and permeability. This investigation, therefore, evaluates fly ash as an additive to soil-ement, with partiular referene to shrinkage.

7 Speimens of soil-ement for shrinkage study were prepared in whih one part of the ement was replaed by 2 parts of fly ash. The results show that fly ash redues shrinkage of soil-ement. Figure 6 shows the redution in shrinkage of fly ash-treated soil-ement, expressed as a perentage of the untreated soil-ement, in relation to the - 2µ lay ontent in the soil. The data indiate that the effetiveness of fly ash in reduing shrinkage dereases with the lay ontent. The benefiial effet of fly ash in reduing shrinkage an be due to the fat that fly ash retards the setting-up of the soilement. The observed variation of shrinkage with lay ontent may be expeted sine sand soils, being oarse, will not reat well with ement alone, and a pozzolan suh as fly ash is nearly always highly desired. The 7-day ompressive strength is redued by replaing ement with fly ash (Fig. 5). However, the 28-day strength of fly ashtreated soil-ement, with the exeption of a few, equals that of the untreated mixtures. In summary, so far as shrinkage is onerned, fly ash is benefiial in sand and friable soils. Conerning the proportion of fly ash, a good rule of thumb would be to replae one-fourth of the ement by fly ash (1: 2 ratio). Pozzolith-Pozzolith 8 (sulfonated lignin), one of several basi formulations available, was used in this investigation. It is known to be a water-reduing agent for onrete, whih, when added at the normal rate, has a retarding effet on the setting of onrete mixes. The hypothesis that pozzolith an improve workability of ement-treated soil was substantiated in four soils; K3-6, M7-, M3-1, and K31-3, where it was ob K32 (al 'E 1io1""-----,s"1 s -----,;11""-- ". Ordinary Cement, per ent/expansive ement, pw tnt.. 11 ;:. (/)... u; Q. w " (al K3.'------'-----' ' 6/ 3/3 /6 Ordinary Cement, per ent/expansive Cement, per ent o.osoil K K31.1.2'---- (b) ' Figure 7. Effet of expansive ement on shrinkage; expansive ement replaed ordinary ement in l to l ratio. (Net shrinkage denotes that a 1- lowane has been made for expansion.) (a) Shrinkage re lated to ontent of expansive ement, and (b) expansion during moist uring. 65 served that with. 5 perent pozzolith the inrease in dry density was 2., 2. 2, 6. 3, and 2. 7 pf and the derease in optimum moisture was 1. 2,. 6,. 3, and 2. perentage points, respetively. This result is partiularly signifiant in soilement in that the shrinkage was found to derease with a derease in optimum moisture and an inrease in dry density (1). - The seond phase of the study examined the effet of pozzolith on shrinkage. Making use of the improved moisturedensity results, speimens were molded with varying amounts of pozzolith. The results (Table 3) indiate that the net shrinkage was redued with small perentages of pozzolith. For the four soils studied here, a pozzolith ontent of. 2 perent appeared to be optimum. In greater proportions, although the attainable density inreased, the overall shrinkage tended to remain the same, exept in a few ases where it inreased slightly. With inreasing pozzolith ontent the speimens tended to expand during the first few days of moist uring in 1 perent RH. Similar but even more pronouned expansion was observed with soil-ement treated with expansive ement or sulfates of alium, sodium, and magnesium. Expansive Cement-Expansive ement and onretes are relatively new engineering materials. The primary use of expansive ement in onrete is to expand and ompensate for the shrinkage

8 66 that ours in onventional portland ement onrete as it hardens (14). This ement, known as the "shrinkage-ompensated ement," is used as an admixture to soil-ement. In earlier experiments, a mixture of 25 perent expansive ement and 75 perent portland ement was seleted to fabriate soil-ement speimens. The results were not striking, and expansive ement was inreased to 5 perent. The results (Fig. 7a) indiate that due to the ontrolled expansion made possible by the use of expansive ement, the shrinkae wai:i i etlut.:etl ln five out of seven soils (all five were sandy soils). Furthermore, five of the soils were treated entirely by expansive ement. In sandy soils shrinkage an further be redued by ontrolled expansion (Fig. 7a). A higher proportion of expansive ement will ause a larger expansion of the treated soil. Besides being able to ompensate for shrinkage, the mehanial behavior of a ontinuous base an be signifiantly modified by the expansion. It is hypothesized that if the ends of a pavement base are restrained, as in a ontinuous base, while the expansive soil-ement is uring and tending to expand, a ompressive stress would be built up within the soil-ement. When allowed to dry, the soil-ement, whih would shrink without the prior restraint, would be relieved first of the ompressive stress developed during the uring period. In other words, by prestressing the base material, ultimate tensile apaity is inreased by the same order of magnitude. Aording to Eq. lb for an elasti material, the rak spaing (L) is diretly proportional to the tensile strength. Therefore, rak spaing an either be inreased, or by inreasing the proportion of expansive ement, the shrinkage stress (tensile stress) an be limited to well belowthe ultimate tensile strength, thereby eliminating most of the shrinkage raks. There are two possible drawbaks in using expansive ement in soil-ement onstrution. First, the ost of treating the soil entirely by expansive ement may be prohibitive. Seond, the expansion of the ompated mix takes plae immediately after mixing with water, perhaps due to the rapid set of expansive ement; for example, while the speimen from soil K3 expanded. 89 perent in 2 hours, approximately 8 perent of this total expansion took plae in 1 hour. The expansion shown by the laboratory speimens annot be realized in the field where there is a time lag of two to three hours between mixing and ompating. What is required, therefore, is an., > ;;;..! Q, e u Gypsuin, per ent Sodium Sulfate, in Normality of Molding Water Figure 8. Effet of (a) gypsum and (b) sodium sulfate on ompressive strength of soi I-ement mixtures.

9 additive that will reat slowly and ause gradual expansion of the base. Sulfates of alium, sodium, and magnesium were investigated for this purpose. Sulfates-Effets of sulfates of alium, sodium, and magnesium in soil-ement have been studied, with onfliting results (15, 16, 17, 18, 19). The objetive of this study is to eluidate the effet of sulfates onstrengthandshrinkage in soil-ement. Gypsum-Small amounts of gypsum, in proportions up to 1 perent, inreased the 7-day soaked strength of ement-stabilized soils (Fig. 8a). In onentrations greater than about 1 perent, however, the strength tended to derease with gypsum. The results of shrinkage study appear in Figure 9 and Table 3. The plots below and above the absissa (Fig. 9) represent the expansion during moist uring and shrinkage on air-drying. Due to the ontrolled expansion, however, the net shrinkage of soilement is slightly dereased with gypsum (Table 3). Far more signifiant is the fat that the sulfate-treated soil expanded during the 7-day moist uring (Fig. 9), and the expansion inreased with the ontent of gypsum. It an be asserted that gypsum, like expansive ement, would inhibit shrinkage raking. Sodium Sulfate-The strength results with sodium sulfate were even more signifiant, sine in the soils tested (with one exeption-m3, silty lay) the strengths were substantially improved (about twofold) when the normality of the molding water was inreased from to 1. 5 (Fig 8b). Stated differently, for the sand soil M7 (optimum moisture = perent), 1 normal solution is equivalent to. 98 perent of salt by dry weight of soil. The expansions and shrinkages were similar to those observed with gypsum. Again, it was observed that there is an optimum onentration for maximum strength and minimum shrinkage. fl i a "" i i "., "ii a.2... LIJ ".3 K6 M r'--l '-'l'K K K3-6 M7-6 Figure 9. Expansion and shrinkage observed with gypsum. 67 Magnesium Sulfate-Magnesium sulfate was at least as effetive as the other two sulfate salts. For example, when the normality of the molding water was inreased from to 2, the initial expansion was slightly inreased, and the net shrinkages were signifiantly redued (Table 2). This result is in keeping with the finding of Uppal and Kapur (2), who reported that shrinkage dereased with inreasing quantities of magnesium sulfate. Unlike the other two sulfates, magnesium sulfate did not improve the ompressive strength. Up to onentration about 1 normal solution, the strengths remain unhanged, and from there onward they gradually derease with the sulfate ontent. Besides being able to redue the overall shrinkage of the ement base, the expansion has other impliations in the performane of the base. Another benefit is that the ementtreated soil ould beome muh stronger if ementation took plae under a ompressive fore. To substantiate this point, the soaked ompressive strength of sodium sulfate-treated (1 normal onentration) ement. mixtures of two soils were determined. Respetively, the ompressive strengths of soils K3-6 and M7-6, when onfined in their respetive aluminum molds during

10 68 TABLE 3 EFFECT OF ADDITIVES ON SHRINKAGE (%) Additive Soil Conentration (%) K3 M7 (2) K15 K25 K27 K31 M3 K32 Cement, Cement, 4; lime, 2.152@ o Cement, 5; alium hloride, Cement, 4; fly ash, Cement, 6; pozzolith, Cement, 3; expansive ement, Cement, 6; gypsum, o Cement, 6; normal sodium sulfate, Cement, 6; normal magnesium sulfate, Cement, 6; sodium hyd,.oxide, '.la Cement, 6; SS-K emulsion, Cement, o Cement, 7; lime, Cement, 9; alium hloride, Cement, 8; fly ash, 4 o Cement, 1; pozzolith, Cement, 5; expansive ement, Cement, 1; gypsum, Cement, 1; normal sodium suliate, 1 o Cement, 1; normal magnesium sulfate, Cement, 1; sodium hydroxide, Cement, 1; SS - K emulsion Shrinkage redued to 7 perent or less of the ontrol. the 7-day moist uring, were inreased from 38 to 45 psi and from 215 to 35 psi. In summary, sulfates in small quantities inreased the strength of soil-ement and dereased the overall shrinkage. The prestressing of soil-ement bases, as aused by the initial expansion, resulted in inreased rak spaing and muh higher ompressive strength. However, large amounts of sulfates had a detrimental effet on strength and durability of soil-ement mixtures. Tentatively, the sulfate ontent should not exeed 1 perent, based on the dry weight of soil solids. Sodium Hydroxide-Shrinkage results with sodium hydroxide indiate that stabilized sand soils (those with kaolinite as the predominant lay mineral) were benefited from approximately. 5 perent of the alkali ompound (Table 3). The shrinkages of both

11 the montmorillonite soil-ements (M7-6, M3-1) were inreased with the addition of the sodium ompound. It is believed that the relatively high shrinkage of these speimens was primarily from partial onversion of the montmorillonoid omponent of the soil into the highly swelling sodium form. The strength results reported by Lambe et al (16) and Norling and Pakard (18) are somewhat in agreement with this finding. For example, Lambe reported that!perent sodium hydroxide signifiantly inreased the strength of a silty loam (7 perent -2µ illite lay), whereas the same onentration was detrimental in a lay (36 perent -2µ montmorillonite lay). Emulsion-The SS-K grade ationi emulsion was investigated in ement-stabilized soils. Emulsion (1 perent based on the dry weight of soil solids) was dissolved in water before mixing with the dry-mixed soil and ement. Four soils (K3, M7, M3, and K31) were investigated for shrinkage and ompressive strength. In three sand soils neither the shrinkage nor the soaked ompressive strength was influened. In silty lay (M3), however, the indiation was that shrinkage ould be slightly redued. Another benefiial effet of emulsion may be in the ontrol of the rak width. That is to say, beause of the exessive extensibility of emulsion-treated soil, the raks ould be narrower. In summary, 9 of the 1 additives tested appeared to be benefiial in reduing the shrinkage (Table 3). Emulsion did not appear useful. With those 9 additives, the 7- day soaked ompressive strength in some ases was inreased and in others it was unhanged or slightly dereased. Insofar as the onentration of materials is onerned, a word of aution is in order, sine a few of the additives (speifially, sulfates and pozzolith) were benefiial only at a ritial optimum amount. Other levels of onentration, espeially those above the ritial optimum, may impair effetiveness. Influene of Soil Texture on Response to Additives-In general, well-graded soils were responsive to pratially all the additives. Typial examples were K3, K25, and K27, with very high uniformity oeffiient values. M7 and K31, in this order, were less responsive. As expeted, the uniformity oeffiient values of these soils are the lowest. Interestingly enough, two samples of M7 with different uniformity oeffiient values shrank differently, in that shrinkage dereased with inrease in uniformity oeffiient. Conerning the texture of the soil, the writer has reported (1) that the shrinkage of sands and sandy soils was probably due to the shrinkage of ement. The results of this study reinfore this hypothesis. For instane, lime, espeially fly ash and expansive ement, when added as replaements to ement signifiantly redued the shrinkage in sand soils, but not in lay soils. In onlusion, well-graded soils shrink less and are more suseptible to improvement by trae additives than uniformly graded soils. CONCLUSIONS Analytial expressions for both rak spaing and rak width were presented and disussed. Suh refinements as reep and theory of failure appliable to ementstabilized soil have been omitted, beause muh basi researh information on these subjets in not available or is inomplete. Results show that rak spaing is primarily a funtion of tensile strength of treated soil. In simple terms, rak width is the subjet of two opposing influenes. The tendeny of the rak width to inrease is to some extent ompensated by the extensibility of the treated soil. Of all fators, total shrinkage exerts the most influene on raking of pavements. A searh for treatments to redue shrinkage, therefore, led to several promising additives. Lime and fly ash proved to be the best. Sulfates of magnesium, sodium, and alium; and expansive ement (in this order), by virtue of their ability to expand and ompensate for the shrinkage, are the seond best additives. Pozzolith 8, although less effetive than fly ash, improves the workability and thereby enables better ompation, whih, in turn, redues shrinkage. Calium hloride provides improvement in poorly reating uniformly graded sands, sodium hydroxide only in kaolinite soils, and emulsion none at all. 69

12 7 So far as the soils in relation to the response to additives are onerned, wellgraded soils shrink less and are more suseptible to improvement by trae additives than uniformly graded soils. ACKNOWLEDGMENTS This report is part of a researh projet of the Engineering Experiment Station, University of Mississippi, under the sponsorship of the Mississippi State Highway Department and the U. S. Bureau of Publi Roads. The author wishes to express his appreiation to Russell Brown, of the Mississippi Highway Department, for many suggestions made in the ourse of the study. The ontribution of graduate student Lee Pan Shyong is aknowledged here. Sidney Boren, now with the Southern Bell Telephone Co., reviewed the manusript. The opinions, findings, and onlusions expressed in this publiation are those of the author and not neessarily those of the State or the Bureau of Publi Roads. REFERENCES 1. George, K. P. Shrinkage Charateristis of Soil-Cement Mixtures. Presented at the 47th Annual Meeting and published in this RECORD. 2. Wang, J. W. H., Mateos, M., and Davidson, D. T. Comparative Effets of Hydrauli, Caliti and Dolomiti Limes and Cement in Soil Stabilization. Highway Researh Reord 29, p , Spangler, M. G., and Patel,. H. Modifiation of a Gumbotil by Lime and Portland Cement Admixtures. Pro. HRB, Vol. 29, p , Pinto, C. desousa, Davidson, D. T., and Laguros, J. G. Effet of Lime on Cement Stabilization of Montmorilloniti Soils. HRB Bull. 353, p , Seed, H. B., and Chan, C. K. Struture and Strength Charateristis of Compated Clays. Pro. ASCE, Vol. 85, p , Allen, H. Use of Calium Chloride in Road Stabilization. Pro. HRB, Vol. 18, Part II, Sheeler, J. B., and Hofer, D. W. Density-Compative Energy-Calium Chloride Content Relationships for an Iowa Dolomite. HRB Bull. 39, p. 1-8, Wood, J. E. Use of Calium Chloride for Soils Base Stabilization in Maryland. HRB Bull. 241, p , Clare, K. E., and Pollard, A. E. The Relationship Between Compressive Strength and Age for Soils Stabilized With Four Types of Cement. Magazine of Conrete Researh, Vol. 7, p , Lilley, A. A. Soil-Cement Roads: Experiments With Fly Ash. Teh. Rept. TRA/ 158, Cement and Conrete Asso., Great Britian, Davidson, D. T., Katti, R. K., and Welh, D. E. Use of Fly Ash With Portland Cement for Stabilization of Soils. HRB Bull. 198, p. 1-11, Wright, W., and Ray, P. N. Use of Fly Ash in Soil Stabilization. Magazine of Conrete Researh, Vol. 9, No. 25, p , O'Flaherty, C. A., Mateos, M., and Davidson, D. T. Fly Ash and Sodium Carbunale al:l Addilivel:l tu Suil-Cemenl Mixlurel:l. HRB Bull. 353, p , Klein, A., and Troxell, G. E. Studies of Calium Sulfoaluminate Admixtures for Expansive Cements. Pro. ASTM, Vol. 58, p , Sherwood, P. T. Effet of Sulfates on Cement-and-Lime Stabilized Soils. HRB Bull. 353, p , Lambe, T. W., Mihaels, A. S., and Moh, Z. C. Improvement of Soil-Cement With Alkali Metal Compounds. HRB Bull. 241, Kozan, G. R. Soil Stabilization: Investigations of a Chemially Modified Cement as a Stabilizing Material. U. S. Army Eng. Waterways Exp. Sta., Teh. Rept , Report 3, Norling, L. T., and Pakard, R. G. Disussion of the paper, Improvement of Soil-Cement With Alkali Metal Compounds. HRB Bull, 241, 196.

13 19. Cordon, W. A. Resistane of Soil-Cement Exposed to Sulfates. HRB Bull. 39, p , Uppal, I. S., and Kapur, B. P. Role of Detrimental Salts in Soil Stabilization With and Without Cement -3, Effet of Magnesium Sulfate. Indian Conrete Journal, Vol. 31,