SOIL-CEMENT: RECENT RESEARCH BY THE OVERSEAS UNIT OF TRRL* H-. E. BOFINGER, ME., Ph.D.,1F.I..E., Principal Scientific Officer, Overseas Unit, Transport and Road Research Laboratory, Department of the Environment, Department of Transport ABSTRACT ~ This paper describes some of the research which has been con- ducted recently by the Transport and Road Research Laboratoly (TRRL) into the behaviour of soil-cement and considers the imn- plications for its use in pavement layers. The laboratory studies have concentrated on the volumetric changes in the material and on the effectiveness of curing membranes. Shrinkage in clay-ce- ment mixtures -has been shown to be anisotropic. When it is measured in the direction appropriate for estimating the horizon- tal shrinkage in pavements, shrinkage is inversely proportional to the cement content provided that the material is properly cured. As the cement content is increased up to 15 per cent the shrinkage of horizontally compacted specimens decreases. The effects of the method of compaction, density and pretreatment moisture content on the shrinkage of clay-cement is considered. Curing methods commonly used in the field have been studied under controlled conditions and their effectiveness on three different types of soil- cement (sand-cement, clay-cement and gravel-cement mixtures) has been evaluated. It is apparent that some common methods for curing soil-cement pavement layers are ineffective. TRRL, have concentrated on finding out more about the shrinkage characteristics of soil-cement, and on the factors that influence shrinkage, in particular the effec- tiveness of different methods of curing soil-cement layers. It should be recognised however that whilst the shrinkage characteristics of cement-stabilised materials do indeed contribute to the initial cracking of soil-ce- menit pavement layers, several other important factors contribute to the long-term performance of such layers. These factors are the strength of the material, its modulus and strain capacity, the restraint of the un- derlying layer, the stresses induced by traffic or fluctua- tions in termperature or moisture content, and the creep properties of the material. A considerable volume of knowledge exists about the influence of these factors on pavement performance, and hence the recent in- vestigations by the Overseas Unit have concentrated on the shrinkage characteristics of soil-cement, about which relatively little quantitative information is availa- ble. INTRODUCTION 1. Many road pavements with soil-cement bases or sub-bases have given excellent service in a wide variety of climatic conditions. In tropical countries soil-cement is often used for the main structural layer of a pave- ment, sometimes because sources of traditional materials such as crushed rock are scarce or are com- pletely absent in a region, and sometimes because the cost of winning and transporting such materials subs- tantially exceeds the cost of cemnent-stabilisation of locally occurring materials. 2. In some tropical countries roads with soil-ce- menit bases have earned a reputation for premature failure. Often such failures can be attributed to poor construction techniques and inadequate construction control, but some failures do occur where the quality of the soil-cement, in terms of accepted standards, is high. Engineers often attribute such failures to the 'shrinkage' of the soil-cement used, and suggest that this can be minimised by placing an upper limit on the cement con- tent. Recent investigations by the Overseas. Unit, 'ACKNOWLEDGEMENTS: The Investigation of curing techniques was undertaken In the Overseas Unrit by C.G. Ouffell, C.R. Jones and Mrs E.BC. Buchan. The Investigation of shrinkage characteristics was undertaken in the Overseaa tUnit by H.O. Htassan as part of a program of work for a Ph.D. thesis for the Unkiversity of Surrey. The work described in this paper forms part of the program of the Transport and Road Reaearch Laboratory, and Is published by permission of the 'Director. Crown Copyright. Any views expressed In this paper are not necessariiy those of the Oepartment a of the Environmnent or Transport, or of the Ministry of Overseas Develop- ment. Extracts from the text may be reproduced, except for conmnericai purposes, pro- vided the source Is acknowledged. SHRINKAGE OF SOIL-CEMENT 3. Studies of the shrinkage of soil-cement under various conditions have been conducted in a number of institutions. Among the more important studies are those by Nakayarna and Handy (1965), George (1968), Wang (1973), Pretorius and Monismith (1971), and Dunlop (1973). Each of these investigations was carried out using specimens that were unsealed against loss of moisture during part or all of the period of observation. Nakayarna and Handy (1965), George (1968), and others have shown that the shrinkage is maximised when the material is allowed to dry, particularly within the first few days. It is important, therefore, that effi- cient methods are used to cure soil-cement layers in order to minimise the loss of moisture and the effects of drying shrinkage. Hypotheses to explain the mechan- ism of drying shrinkage have been proposed by George (1968) and Wang and Kremmydos (1970). 4. An investigation of the volumetric changes in hardening sand-cement mixtures has been carried out at TRRL (Bofinger and Duffell 1973). When these mix- tures were sealed to prevent loss of moisture, expansion occurred, indicating that the hydrating gel expands and its structure has sufficient strength to retain the ex- panded dimensions of the specimen when the volume of the cement paste decreases due to recrystallisation (Bernal 1952). 49 VOLUME 9, PART 4, 1978 f.,
Z . S a a. BOFINGER -SOIL-CEMENT: RECENT RESEARCH BY TRRI, 5. Recently, studies at the TRRL (Bofinger, Hassan and Williams 1978) have concentrated on ex- amining the anisotropic nature of shrinkage and the relationship between shrinkage and the cement content of specimens moulded from tine-grained soil. Addi- tional tests were made to assess the effect on shrinkage of the method used to compact the specimen, the average density of the specimen and the pretreatment moisture content of the soil, i.e. the moisture content of the soil before it is processed into soil-cement. 6. Some properties of the fine-grained soil that was used in these tests, a brickearth from Littlehampton, are listed below. (a) Particle sizes (b) Atterberg limits Linear shrinkage CompactionBS (2.5 kg ramnmer) (1975) BS (4.5 kg rammer) (1975) sandsilt clay liquid limit plastic limit plasticity index 0MGMDD 0MGMDD (e) CBR at 0MG 23% 51% 26% 39%19%20% 12.4% 17% 1.78 Mg/rn 3 13%1.97 Mg/rn 3 13% 7. In the TRRL studies specimens of Lit- tlehampton brickearth were moulded by one of three methods of compaction. Static compaction was used to make cylindrical specimens 100 mm x 50 mm diameter in which the compacting force was applied along the longitudinal axis of the cylinder (i.e. moulded ver- tically), and also to make 150 mm x 38 mm x 38 mm bars in which the compacting force was applied at right angles to the axis (i.e. moulded horizontally). Kneading compaction was carried out in the same cylindrical moulds using a tamper based on the Harvard miniature compaction apparatus. The standard compaction ham- mer, BS 1377 (1975), was used to mould another type of specimen in the standard compaction mould. 8. All of the specimens were sealed in wax im- mediately they were extruded from the mould and measurements of axial shrinkage were commenced within two minutes. A minimum of five specimens was used in each group of tests. RESULTS 9. The results of the tests to examine the effects of anistrqpy and the method of compaction on autogenous shrinkage are listed in Table L. TABLE1I Total Shrinkage Strains (x 1 O-5) after 28 days of Curing for Specimens Moulded at OMC to BS (2.5 kg) Density by Four Methods of Compaction Cement Content(%) Method of Compaction 0 4 6 8 10 1s Vertical static compaction 1850 1300 300 400 600 - - 960 50 500 - - - 4500 3900 4090 - - ANISOTROPY 10. The autogenous shrinkage of specimens moulded vertically was similar in magnitude to the shrinkage of specimens moulded horizontally but there was, however, a significant difference in the effect of the cement content on the shrinkage of each type of specimen. When specimens were moulded vertically, there was an optimum cement content at which the shrinkage was a mninimumn, confirming the results of George (1968). In contrast, when specimens were moulded horizontally, the magnitude of the shrinkage progressively decreased as the cement content in- creased. These horizontally-moulded specimens most closely model soil-cement in a pavement layer. Two other soils were also tested to confirm that this effect was not peculiar to soil-cement made from Lit- tlehampton brickearth. METHOD OF COMPACTION 11. Vertically moulded specimens in which knead- ing compaction or dynamic compaction (using the BS compaction method (1975)) was used, also exhibited an optimum cement content at which shrinkage was minimised, but the magnitude of the shrinkage of dynamically compacted specimens was very large. In this type of specimen the shrinkage strains were up to six times greater than those in the other types of speci- men. DENSITY 12. It has often been shown that there is a miarked increase in the strength and stiffness of soi!-cement when its density is increased. Both of these properties contribute to the initial spacing of cracks in a stabilised road base. It is important, therefore, to know how the shrinkage characteristics of soil-cement are related to the density of the material. 13. In the initial tests, specimens containing various percentages of cement were moulded vertically by static compaction at optimum moisture content (0MC) to maximum dry density (MDD) for the BS 4.5 kg ram- mer method (1975) and their shrinkage was compared with previous results obtained from specimens com- pacted to the lower maximum dry density given by the BS 2.5 kg rammer method. Two additional sets of speci- mens containing 8 per cent of cement were moulded horizontally at the 0MG to the MIDD for the BS 4.5 kg rammer method and also to 95 per cent of this MIDD and their shrinkage was compared with similar speci- mens moulded to the MD3D given by the BS 2.5 kg ram- mer method. A summary of the total autogenous shrinkage strains after 28 days is given in Table II. 14. The autogenous shrinkage of all specimens was greater at the higher density than at the lower density, even though the higher density specimens were moulded at a lower moisture content. It should be noted that when specimens were compacted horizontally at the same moisture content but to different densities, the specimens with the lower density had the smallest shrinkage strains, thus indicating that this phenomenon is not dependent'on the moisture content. PRETREATMENT MOISTURE CONTENT 15. Rallings (1971) has shown that the pretreatment moisture content can have a significant effect on the strength and swell potential of clays stabilised with ce- ment. The effect oF pretreatment moisture content on 50 ~~~~~~~~~~~~~~~~~~~ARRB PROCEEDINGS (c) (d) Kneading compaction Dynamic compaction Horizontal static compaction 2100 1230 640 580 4.40 330 NOTE: -. denotes mixtures that were not tested 50 - , b BOFINGER - SOIl-CEMENT: RECENT RESEARCH BY TRRL TABLE 11 Total 28 day Shrinkage Strain (x 1 0-6 ) at Different Densities and Optimum Moisture Contents Moulding Conditions Cement Type)C Q/ conteni Comipaction % BS (2.5 kg) 95% BS BS (4.5 kg) MD)D~OmC (4.5 kg) at MDDl&OMC (/7%) 13% MC (/3%) 4 1300 - 1700 6 300 - 1850 8 410 - 1950 CI)Il~lii~i~iil 10 600 - 2150 ~~t~itic ~ 8 580 1200 2100 Coil]pac I on NOTE: -den,,Is mi~iiures that cre not, tested the autogenous shrinkage of specimens of Lit- tlehampton brickearth stabilised with 6 per cent of ce- ment and moulded vertically by static comipaction was evaluated in this investigation.. 16. The results illustrated in Fig. I show that the pretreatment moisture content has a profound effect on the shrinkage of soil-cement. Samples prepared from air-dried soil actually expanded rather than shrank, even though the moulding moisture content of all of the specimens was the same. 501) '100 1 00 '00 7n 11 1-1 7. AE~~~~~~~~* 100 - 0 0 ~~~~10 1 10, Fig. 1 -Influence of pre-treatment moisture content on shrinkage of specimens containing 6% cement DISCUSSION OF SHRINKAGE RESULTS 17. The results of these tests raise several issues that have important implications for the construction of road bases from fine-grained soil-cement mixtures. The reduced shrinkage of horizontally-moulded specimens when higher quantities of cement are incorporated is contrary to the increased shrinkage at higher cement contents when specimens are moulded vertically. It is likely, however that the horizontally-moulded speci- mens most closely resemble the shrinkage conditions in the field, which implies that the adoption of an upper limit for cement content will not reduce the incidence i'f shrinkage cracking. The effects on shrinkage of the VOLUME 9, PART 4, 1978 method of compaction, the density and the pretreat- menit moisture content, should be considered carefully when soil-cement layers are being designed and con- structed. Laboratory compaction procedures embody- ing dynamic techniques are recognised as being poor replicates of field compaction processes. Nevertheless the possibility should not be ignored that the shock loading associated with vibratory rollers may cause larger shrinkage strains in soil-cement layers than comn- paction by dead weight rollers. The most suitable roller for minimising shrinkage in a soil-cement layer at a par- ticular target density is likely to be a pneumatic-tyred roller. 18. The results of this investigation indicate that the selection of the target density for a stabilised base is a complex matter. The Highway Research Board (1961) and others, have shown that the strength and stiffness of soil-cement increase progressively as the density is raised, but unnecessarily high densities appear to cause difficult cracking problems. As the density, and hence the strength and stiffness, of fine-grained soil-cement are increased, the spacing between shrinkage cracks in a pavement layer of the material will increase. Corres- pondingly for the same shrinkage strain, the cracks formed will be wider and will have a greater probability of being reflected through the bituminous surfacing of a pavement. If, however, the density is kept at a low value to reduce the cracking problem, the strength and durability of the soil-cement will be adversely affected, and its ability to resist crack propagation due to traffic and temperature stresses will be significantly impaired. The best compromise is probably to aim for a durable material, accepting that it will crack at regular intervals, and to take steps to seal the cracks when they appear. 19. The tests to assess the effects of the pretreat- menit moisture content of the soil show that for best results it should be processed in the driest practicable condition. In addition to reducing the autogenous shrinkage, this practice will improve the transport of ce- mentitious material into lumps of soil and will increase their stability. 20. 'The foregoing study was confined to the autogenous shrinkage of a fine-grained soil-cement and the results are appropriate only when the material is cured under ideal conditions without loss of moisture. This ideal is not possible in practice because there will always be a delay between the completion of the comn- paction and the placing of a curing seal on the layer. Re- cent work by May Loh (1977) on a laboratory scale indi- cates that shrinkage increases significantly when there is a delay before a curing membrane is applied. She has shown that if one day elapses the shrinkage will be ap- proximately the same magnitude as that of completely uncured material. Some consideration should be given, therefore, to arranging the construction procedure so that a curing seal is applied immediately after the layer has been finally trimmed and compaction has been completed. 21. The efficiency of common curing methods 'IS considered in the next section of the paper. CURING EFFICIENCY 22. Various methods are used to reduce loss of moisture from the upper surface of newly constructed 51 ROFINGER - SOIL-CEMENT: RECENT RESEARCH BY TRRL soil-cement pa~,ement layers. Some methods attempt to seal the surface with an impermeable membrane, while other methods reduce the loss of moisture from the layer by keeping the surface wet. 23. The Overseas Unit of TRRL has investigated the efficiency of some of the common methods of curing soil-cement. In the investigation, specimens were made from three different types of soil, viz a sand, a clay and a gravel. Some of the properties of the three soils are listed below. Soil Type Particle Size Percentage Chertsey Sand > 3001.cm 2 300-15Ogm 88 < 150jom 10 Iver Brickearth' Sand 1 5 Coarse silt 48 Medium silt 14 Fine silt 5 Clay 1 8 Malawi Gravel t 200m7 10-5 mm 7 5-2.36 mm 1 1 2.36 mm-425 gam 33 425-75 jum 22 < 75jum 20 MOID (11% cement) is 1826 kglmn' at OMC of 15.5%. (BS 2.5 kg ramnmer (0975)) LL- 37%. PL =(9%. PI -18% tMDD 0% /cementt) is 2162 kglm' at OMC ofl%. (BS 4.5 kg rammer (1975). ILL =35%, PI, -9%, PI -(6% 24. Specimens were compacted statically in three layers in steel moulds with internal dimensions of 300 x 75 x 75 mm and the top surface of the ~pecimens was treated with one of several curing techniques. The loss of moisture from the samples was then measured at- regular intervals during exposure to different condi- tions of temperature and humidity. 25. One of the most common curing techniques is to apply a bituminous prime such as a fluid cut-back, on the assumption that this material will also act both as a curing membrane and a prime for subseciuent surface dressing. Alternatively, the soil-cement layer is sprayed with water at regular intervals (say twice each day) to keep the surface moist. A modification of this method uses a layer of loose sand spread on the soil-cement layer to retard the rate of drying and to prevent water from. being sprayed directly onto the soil-cement sur- face. Each of these curing techniques was examined in the investigation and, in addition, the sealing efficiency of 55 per cent anionic bituminous emulsion was tested. One set of specimens was left completely unprotected to measure the rate of loss of water under the worst con- ditions. 26. Pairs of specimens were subjected to relative humidities of 33, 50, 75 and greater than 96 per cent at a temperature of 250C, and also to a relative humidity of 50 per cent at a temperature of 45"C. The loss of moisture expressed in terms of the percentage loss after seven days of the total water in the mixture when the specimen was moulded is shown in Table III. DISCUSSION OF RESULTS OF THE INVESTIGATION OF CURING TECHNIQUES 27. The results shown in Table III indicate that some curing techniques in common use, such as spraying with water or spraying a thin cut-back bitumen on to the surface of the soil-cement, have only a very limited effect in preventing loss of moisture. Some of the results obtained did not follow the general trends (these are marked with an asterisk in the table), but the general conclusions that can be drawn are clear. These are that the best methods for curing soil-cement layers (from among those studied) are by sealing the surface with bitumen emulsion or by spreading a layer of loose sand on the soil-cement which must then be kept damp by spraying with water. TABLE Ill Average Percentage Loss of Moisture After 7 Days Curing Dryingt Contditiirns (Relative Humnidity & Temtp.) soil Typ e Spray 50%/14SC 33%,'125 0C 5M%25 0C 7S%/25 0C 96+%/25 0C RateI/rm 21day _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Chertsey sand Nil - 81 72 68 59 46 + MCO 1.1 69 40 31 23 19 6% cement MC2 1.1 36 26 3'22 ( 3 55% 1.1 2.7* 6.2' 10 7.4 6.5 Em ulsionDamp 2.2 - 241 113 6.9 6.S Sand Iver Nil - 46 38 36 32 1 9 Brickcarth MCO 0.54 28 8.9 6.6 8.0 2 (0 +I MC2 0.54 1 3 6.8 4.8 2.4 2.0 8'% cemient 5 5%". 0.54 1 2 4.4 4.8 4.8 3.0 EmulsionWater 2.2 40 3)1 2 3 lB 9.6 Spray Malawi grdve( Nil - 37 29 ( 9 + MCO 0.54 37 19 1 3 4`4cement MCO 1.1 403 18 ( 4 5 5 0.54 . 33Y 9. 7.2 EmulisonDamp 2.2 II1 95 5.6 Sand For espi;,.ittiienl asicrisks see prao 2 7 52 ~~~~~~~~~~~~~~~~~~~ARRB PROCEEDINGS J - 52 BOFINGER -SOIL-CEMENT: RECENT RESEARCH BY TRRL 28. In tropical and sub-tropical countries the results obtained under conditions of 50 per cent relative humidity and a temperature of 450C are likely to be more appropriate than the other values listed in the Ta- ble. As the relative humidity drops, there is a marked increase in the loss of water through the curing membrane. Hence in very dry regions, particular atten- tion must be given to the speed with which the curing membrane is applied and to the quality of the curing method. The most difficult material to cure with bituminous membranes is sand-cement, whereas it is the easiest of the materials to prime for subsequent sur- face dressing (see Highway Research Board 1949). 29. In general it seems likely that bituminous emul- sions and damp sand offer engineers satisfactory methods for curing soil-cement, and the choice of method will obviously depend on the relative costs of each method for the particular project. It must be stressed, however, that it is essential to commence cur- ing the material as soon as possible so that the shrinkage is kept to a minimum. CONCLUSIONS 30. The shrinkage of soil-cement is anisotropic and contrary to previously published data, the autogenous shrinkage measured in the appropriate direction (e.g. REFERENCES horizontally) is progressively reduced when the cement content is increased. Hence the commonly accepted thesis that increasing the cement content of soil-cement pavement layers above a certain optimum value tends to increase shrinkage, is open to serious doubt. 31. Some methods of compaction induce very high shrinkage strains in soil-cement, and these strains can- not be attributed to the differences in density or moisture content associated with particular compaction methods. 32. Compacting soil-cement to a high density will in- crease its propensity to shrink, but it will also increase the strength and durability of the material. 33. Soil-cement should be processed when the soil is in its driest practicable condition. No attempt should be. made to raise its moisture content over a period of time prior to construction but all of the water needed for compaction should be added during the mixing process. 34. Bituminous emulsion curing seals or blankets of damp sand retain most of the moisture in the soil-ce- menit layer and are satisfactory curing techniques. They should be applied as soon as possible after compaction is complete. If curing is delayed for one day, the total shrinkage of soil-cement will be nearly as great as if no curing measures were taken at all. BERNAL, JiD. (1952). The structure of cement hydration compounds. 3rd Int. Symp. on the Chemistry of Cement. London pp. 216-36. BOFINGER, H.E. and DUFFELL, C.G. (1973). The effect of filler on the characteristics or sand-ce- menit mixtures. Transp. Road Res. Lab. (U.K.) TRRL Lab. Rep. Rep. LR 527. Crowthorne. BOFINGER, H.E., HASSAN, H.O. and WILLIAMS, R.IT. (1978). The shrinkage of soil-cement Transp. Road Res. Lab. (U.K.) TRRL- Spec. Rep. SR 398, Crowthorne. BRITISH'STANDARDS INSTITUTION (1975). British Standard No. 1377. Methods of Test for Soils for Civil Engineering Purposes. British Standards Institution, London. DUNLOP, R.J. (1973). Shrinkage and creep characteristics of soil-cement. Ph.D. Thesis. University of Canterbury, Christchurch, New Zealand. GEORGE, K.P. (1968). Shrinkage characteristics of soil-cement. Highwv. Res. Rec. 255, pp. 42-57. HIGHWAY RESEARCH BOARD (1949). Prevention of moisture loss in soil-cement with bituminous materials. Res. Rep. No. 8-F7. Nat. Res. Council, Washington D.C. -(1961). Soil stabilisation with Portland Cement. Bull. 292 Nat. Res. Council, Washington D.C. pp. 29-74. LOH, M.K.E. (1977). Shrinkage investigations of cement stabilised Littlehampton Brickearth. Report I. M.Sc. Thesis. Trinity College, Dublin. NAKAYAMA, H. and HANDY, R.L. (1965). Factors influencing shrinkage of soil-cement. Highw. Res. Rec. 86, pp. 15-27. PRETORIOUS, P.C. and MONISMITH, C.L. (1971). Prediction of shrinkage stresses in pavement con- taining soil-cement bases. Highwv. Res. Rec. 362, pp. 63-86. RALLINGS, R. A. (1971). The effect of pre-treatment moisture content on the properties of cement and lime stabilised clay soils. Aust. RdRes. 4(6), pp. 10-31. WANG, J.W.H. (1973). Use of additives and expansive cement for shrinkage crack control in soil-ce- menit. Highw. Res. Rec. 442, pp. 11-20. -and K(REMMYD3OS, A.H. (1970). Use of sodium chloride in reducing shrinkage in mrontmorillonitic soil-cement. Highwv. Res. Rec. 31 5, pp. 81-90. Proc. 9th Con-f'. Aust. Rd Res. Bd, 1978, Disc~ussion '75.8 53 VOLUME 9, PART 4, 1978