Академический Документы
Профессиональный Документы
Культура Документы
Steve Burroughs
Steve Burroughs and Associates Pty. Ltd., 38 Blackman Crescent, Macquarie ACT 2614, Australia.
Email: sburroug@bigpond.net.au
The influence of stabilizers and soil properties on the strength of stabilized, compacted earth is investigated. This is
achieved by establishing relationships between natural soil properties (two measures each of gradation and plasticity
and one of shrinkage), stabilizer treatments (0–6% of lime and/or cement and/or asphalt), and stabilized unconfined
compressive strength (UCS). A total of 219 strength tests were performed on 104 different soils. The soils can
be categorized into two groups according to whether single soil properties or combinations of them are “favourable”
or “unfavourable” in their predisposition to stabilization. The mean UCS for favourable soils ranges from
2.98–3.25 MPa (90% 2 MPa) and for unfavourable soils from 2.32–2.48 MPa (60% 2 MPa), having controlled
for stabilizer treatment variation. Soil linear shrinkage is the best single discriminator of stabilization suitability based
on the categorization results. In linear modelling (analysis of covariance), the UCS of favourable soils is a function of
variation in soil properties but not of variation in the quantity of cement or lime used. Conversely, the UCS of
unfavourable soils is a positive function of the quantity of stabilizer used. The results of the study stress the
importance of selecting a soil whose characteristics are favourably predisposed to stabilization in order to attain
satisfactory strengths of compacted earth. Future research should be aimed at developing a system that uses key soil
properties to predict the likelihood of successful stabilization.
Keywords: compacted earth; stabilization; soil properties; cement; lime; compressive strength
L’auteur examine l’influence des stabilisateurs et des propriétés du sol sur la résistance de la terre stabilisée et compactée.
A cet effet, il définit les relations qui existent entre les propriétés naturelles du sol (deux mesures de gradation et de
plasticité et une mesure de retrait), les traitements au moyen de stabilisateurs (0–6% de chaux et (ou) de ciment et
(ou) d’asphalte), et la résistance à la compression non confinée (UCS). Au total, on a procédé à 219 essais de
résistance sur 104 sols différents. Les sols peuvent être divisés en deux groupes selon que les propriétés de sols
uniques ou en combinaisons sont ‘favorables’ ou ‘défavorables’ en ce qui concerne leur prédisposition à la
stabilization. L’UCS moyenne pour les sols favorables était comprise entre 2,98 et 3,25 MPa (90% 2 MPa) et
2,32–2,48 MPa (60% 2 MPa) pour les sols défavorables, après avoir contrôlé les variations de traitement des
stabilisateurs. Le retrait linéaire du sol est le meilleur discriminateur unique de la pertinence et de la stabilisation basé
sur les résultats de la répartition en catégories. En modélisation linéaire (analyse de covariance), l’UCS des sols
favorables est fonction de la variation des propriétés des sols mais non de la variation de la quantité utilisée de
ciment ou de chaux. Réciproquement, l’UCS des sols défavorables est une fonction positive de la quantité utilisée de
stabilisateur. Les résultats font ressortir l’importance du choix du sol dont les caractéristiques sont prédisposées
favorablement à la stabilisation afin d’obtenir des résistances satisfaisantes de la terre compactée. A l’avenir, la
recherche devra porter sur le développement d’un système utilisant les principales propriétés des sols afin de prévoir
la vraisemblance d’une bonne stabilisation.
Mots clés: ciment, terre compactée, résistance à la compression, chaux, propriétés du sol, stabilization
Building Research & Information ISSN 0961-3218 print ⁄ISSN 1466-4321 online # 2006 Taylor & Francis
http: ⁄ ⁄www.tandf.co.uk ⁄journals
DOI: 10.1080/09613210500279612
Burroughs
Research problem and rationale soils as possible. At each construction site, soil was
Studies that have made detailed examinations of stabil- sampled either by test pits or hand borings to a depth
ization with respect to the type of stabilizer, the quan- of 1-3 metres. Forty kilograms of soil, as recommended
tity (percentage) of stabilizer, and soil type include by SAA (1977), were collected at each pit or bore.
those of Croft (1968), Akpokodje (1985), Bryan
(1988a), Walker (1995), and Ngowi (1997). The Locations of sample pits or bores at each construction
numbers of soils tested ranged from three to fourteen. site were considered representative of the area sur-
Of these investigations, Croft (1968), Ngowi (1997), rounding the site within a radius of 250 m, which rep-
and Akpokodje (1985) examined all three effects resents a practical, economic distance within which soil
(stabilizer type, stabilizer quantity, and soil type) on would be transported to a building site during construc-
the strength of stabilized samples. Bryan (1988, tion. At each site, the surrounding area was surveyed
1988a) and Walker (1995) each studied different soils and the vegetation, topography, rock outcrops, drai-
types but used only cement as a stabilizer. nage, and soil colour were mapped. These landscape
indicators were assumed to represent a broad variation
Previously therefore, the effects of soil type and of in soil properties around the site. Sampling was per-
stabilizer type/amount have been assessed by experi- formed to cover the range in soil properties as indicated
ments involving small numbers of different soils. by the mapping. Between 1 and 12 soil sample were
What has been lacking is a comprehensive investi- taken from each site, depending on the amount of vari-
gation of stabilization using a large number and wide ation in soil type as indicated during mapping.
range of different soil types. This would provide a suf-
ficient number and spread of data to quantify relation-
ships between soil properties, stabilizer treatments, and Experimental design
stabilized strength across the naturally-occurring spec- Soil properties measured
trum of soil properties, and would also obviate the pro- Sample particle size distribution was determined
blems of inter-study comparisons (e.g. differences in because numerous studies (e.g. Croft, 1968; Spence,
experimental aspects such as sample compaction, 1975; Bryan, 1988; Walker 1995) have shown its
curing time, and curing environment). There is a par- importance in determining the strength of stabilized
ticular need to be able to assess the contributions of earth. The three textural variables measured were %
stabilizer type/amount and soil properties in influen- gravel (the proportion of soil retained on the
cing stabilized strength, and whether the importance 2.36 mm aperture sieve), % sand (0.075 –2.36 mm
of stabilizer treatments on the one hand and of soil aperture sieves), and % clay/silt (,0.075 mm aperture
properties on the other vary across the spectrum of sieve). The clay and silt fractions were combined so
soil types. that size distributions could be obtained by sieve analy-
sis alone and the applicability of the study thereby
The overall research project from which this paper is widened.
drawn aims to establish improved guidelines for com-
pacted earth stabilization and construction. This par- The importance of liquid limit (LL), plastic limit (PL),
ticular paper focuses on defining the contributions of and plasticity index (PI) to stabilized strength has been
stabilizers and soil properties to the strength of stabil- reported, amongst others, by Croft (1968), CSIRO
ized soil. The research objective is to determine the (1987), Bryan (1988), and Walker (1995). The
effects of soil properties (gradation, plasticity, and measures of plasticity used in this study were LL, PL,
shrinkage), stabilizer types (asphalt, cement, lime), and PI, all measured on the natural soil (i.e. on the
and stabilizer quantities (0–6% by weight of dry soil) soil prior to adding stabilizer and compacting).
on the unconfined compressive strength (UCS) of Linear shrinkage (LS) has been used as a variable in
stabilized soil. This objective is addressed through the several previous studies of stabilization, including
analysis of data derived from 219 stabilization experi- those of Croft (1968), Bryan (1988), and Walker
ments on 104 different soils collected from compacted (1995), and in this study was determined for the
earth construction sites in New South Wales, Australia. natural soil.
density (MDD) under a given compactive effort. The The textural, plasticity and shrinkage properties
aim during actual rammed earth construction is to measured defined the variables by which the soils
compact the material under conditions of OMC and were characterized. These five soil property variables,
MDD (CSIRO, 1987; UN, 1992). OMC was deter- together with the stabilizer types and amounts, were
mined in order to identify the moisture content at regarded as the predictor variables (e.g. Fitzmaurice,
which each soil should be compacted and strength- 1958; Spence, 1975; Bryan, 1988; Walker, 1995).
tested. It was not used as a predictor variable of UCS These variables were viewed as causing variation in
because one of the aims of the research is to enable the dependent variable, being the UCS of the stabilized,
suitable soils for stabilization to be identified on the cured sample having been compacted under conditions
basis of easily-measured textural and plasticity data, of OMC and MDD.
rather than on complex and time-consuming compac-
tion/strength tests. OMC and MDD determinations
were made on the sample after the addition of stabilizer Experimental procedures
and compacting, but before curing. The tests on the natural soil included particle size dis-
tribution, Atterberg limits, and LS. The tests on the
Saturated unconfined compressive strength (UCS) was soil-stabilizer mixture included OMC and MDD
used as the dependent variable because it is an (before curing), and UCS (after curing). The procedures
outcome of stabilization that can be compared both for the determination of soil particle size distribution,
with the results of previous studies (e.g. Akpokodje, LL, PL, PI, and LS followed those described in the
1985; Bryan, 1988; Walker, 1995; Bell, 1996), and methods of testing soil for engineering purposes
with construction-related standards and building regu- AS1289.C6.1, AS1289.C1.1, AS1289.C2.1, AS1289.
lations (e.g. CSIRO, 1987; UBC, 1997). UCS was C3.1, and AS1289.C4.1, respectively (SAA, 1977).
measured for each sample after 28 days of curing in a The procedure to measure the moisture content of
humidity cabinet at a temperature of 228C. each sample followed that described in AS1289.B1.1
(SAA, 1977).
Bivariate relationships between UCS and For values of LL below 50% (Figure 3), there is a very
soil properties wide range of values of UCS ranging between about 1.5
For clay/silt content, there is a range of values of UCS to 5 MPa. For values of liquid limit in excess of about
for any given % value (Figure 1). For example, for a 60%, all samples have a UCS below 2 MPa. For values
clay/silt content of 16%, UCS varies from 1.2 MPa of PI between 0 and 35% (Figure 4), there is a wide
to greater than 5 MPa. The highest values of UCS are range of values of UCS for any one value of PI.
associated with clay/silt contents of 10 –35%, Above a PI value of 35%, no sample tested has a
although for clay/silt contents of 5 –20% the data UCS .2 MPa. The relationship between UCS and
are almost bimodal with one group of data linear shrinkage (Figure 5) differs from those of the
,2.3 MPa and a second group .2.6 MPa. These two plasticity properties. Although for any one value of
groups of samples correspond primarily to differences linear shrinkage there exists a range of values of
in plasticity index, with the weaker samples having UCS, there is a distinct negative slope to the data
higher values of plasticity index. Sand content with higher values of linear shrinkage generally being
(Figure 2) ranges from 30 –94%, and the relationship associated with lower values of UCS. For values of
with UCS appears rather complex. Samples containing linear shrinkage up to 3.5%, no sample has a UCS
less than 48% sand exhibit high values with only a few lower than 2 MPa. As linear shrinkage values increase,
samples below a UCS of 2 MPa. However, this may the maximum UCS decreases and a greater proportion
reflect the lower number of samples tested with sand of samples have a UCS below 2 MPa.
Figure 1 X-Y Scatterplot of compressive strength versus CLAY/ Figure 3 X-Y Scatterplot of compressive strength versus liquid
silt content (n ¼ 219). limit (n ¼ 219).
59
Burroughs
Soil categorization and UCS statistics for soil groups ANCOVA: Effects of soil properties and stabilizer
The mean UCS of the 219 samples is 2.66 MPa, and treatments on UCS
73% of the samples equal or exceed the criterion A linear ANCOVA model for the whole data set, with
value of 2 MPa. The soils were categorized into two UCS as the dependent variable and eight predictor vari-
groups using four of the five individual soil properties ables (three stabilizers and five soil properties), is sig-
(% sand, LL, PI, and LS) and one combination invol- nificant at p , 0.000 (Table 2). Although the model
ving PI and LS (Table 1). Clay/silt was not used as a is statistically significant, it explains only 31.1% of
categorizing variable as it shows no particular ability the variation in UCS. The tabled data show that there
in this data set to distinguish sample strength for two are no statistically significant differences in UCS
different groups of soils of higher and lower clay/silt between different percentages of stabilizer treatment
contents (Figure 1). This may be due to competing for any of the three stabilizers. Clay/silt, LS, and
60
Strength of compacted earth
Table 2 Model signi¢cances and R-square values for selected Of particular interest is the pattern of significant vari-
ANCOVA analyses, with UCS as dependent variable and soil ables in the six soil group models (Table 3). No stabil-
properties and stabilizer types/amounts as independent
variables izer assumes significance in the three models for
“favourable” soil groups, whereas clay/silt content
ANCOVA data group Model R-square
and LS are significant variables in all three models,
signi¢cance adjusted for and sand content is significant in two models. In con-
(p-value) d.f. (%) trast, for the “unfavourable” soil groups, the amount
of both cement and lime are highly significant determi-
All data (n ¼ 219) 0.000 31.1 nants of UCS, with the only exception being lime when
Soil samples with LL 30 0.000 31.7 soils are categorized using LL. For the “unfavourable”
(n ¼ 93)
Soil samples with LL . 30 0.000 40.4 soil group models, sand content is the only significant
(n ¼ 126) variable, being so in each of the three models. Sand
Soil samples with LS , 6.0) 0.000 33.7 content is the most consistent significant variable
(n ¼ 100) across the six group models (Table 3).
Soil samples with L.S. 6.0) 0.000 30.4
(n ¼ 119)
Soil samples with PI , 15 0.000 31.5 The ANCOVA analysis for the unfavourable soil group
and LS , 6.0) (n ¼ 96) with LS 6.0 shows that there is a significant increase
Soil samples with PI 15 0.000 30.9 in stabilized strength with greater quantities of cement
or LS 6.0) (n ¼ 123) (Table 3 and Table 4). Using the mean values from
Table 4, there is a 0.48 MPa strength increase per %
cement added between the 2% and 6% stabilizer
levels. The data for lime also show that there is a
general increase in stabilized strength with increments
sand content are the soil properties that have signifi- of this stabilizer (Table 3 and Table 4), with a strength
cant effects on variation in UCS for the whole data increase of 0.28 MPa per % lime between the 2% and
set (Table 3). 6% stabilizer levels. Asphalt has no significant effect on
UCS variation when used with either lime and/or
The apparent lack of effect of the stabilizers for the cement, for any of the models (Table 3).
whole data set may be due to competing or different
trends in stabilization across the wide spectrum of
soil property values measured, thereby dampening
the effect in the pooled data set. Therefore, “favour- Discussion
able” and “unfavourable” soil groupings as categor- As in previous studies, this investigation has found that
ized in Table 1 for selected properties were used for soil properties and the type and amount of stabilizer
three additional sets of ANCOVA analyses. All six (lime and cement) all influence the strength of com-
resulting soil group models are highly statistically sig- pacted, stabilized earth. The large size of the dataset
nificant (Table 2). The R2 values for these models used, the wide range of variation in the properties
range between 30 and 40%, and lie mainly around of the soils studied, and the statistical treatments
the 30% value. For the six categorized groups the employed, have enabled new findings to be made
average R2 is 33%. regarding soil-stabilizer-strength relationships, which
Table 3 Signi¢cance of independent variables for selected ANCOVA analyses (UCS as dependent variable). FAV ¼ ‘‘Favourable’’ soil
group; Unfav ¼ ‘‘Unfavourable’’ soil group
61
Burroughs
Table 4 Mean UCS (and upper and lower 95% con¢dence properties of the stabilized material. More advanced
intervals) for different percentages of cement and lime, using schemes of categorization, using more complex com-
ANCOVA of UCS on soil properties and stabilizer treatments for
soil samples with linear shrinkage 6.0% (n ¼ 119) binations of soil properties and extending the simple
bi-fold classification of soils used here, may improve
Stabilizer type Mean UCS 95% Con¢dence
the ability of soil property data to distinguish soil
and quantity (MPa) interval (MPa) suitability for stabilization.
Table 5 Summary of results of previous work concerning the effect of quantity of cement and lime on stabilized compressive strength
Akpokodje (1985) 3 Soils ranging: sand 15^ 59%, silt 27^75%, and 0.28 0.14
clay 10^ 37%.
Croft (1968) 6 Soils ranging: LL 40^68%, PI 15^ 38%, and LS 0.64 0.28
4 ^14%.
Bryan (1988a) 5 Soils ranging: sand 30^78% and PI 6.4 ^ 26.2%. 0.32 ç
Ngowi (1997) 2 Soils ranging: sand 27-63%, clay/silt 27^73%, 0.47 0.29
LL 31^ 50%, and PI 12 ^ 26%.
Walker (1995) 10 Soils ranging: clay 15^100%, sand 0^85%, LL 0.25 ç
22 ^60%, PI 0^ 35%, and LS 1^17%.
This study 72 soils Soils with linear shrinkage 6.0%. 0.48 0.28
(119 UCS tests)
Symons, 1999). Therefore, the values in Table 5 per- Traditionally, much has been made of matching speci-
taining to the work of Croft (1968) and Akpokodje fied quantities of cement or lime to a soil dependent
(1985) are probably minimum values when compared upon its characteristics. In particular, stabilizer quan-
with the results of this study. In addition, the samples tities have been viewed in terms of requiring less stabil-
tested here were compacted using the modified izer for “good” soils, and more stabilizer for “poor”
Proctor test (compaction pressure of 2700 kN/m2), soils. As a general guide, for favourable soils, the rec-
whereas Croft (1968) and Akpokodje (1985) ommendation from this study is to use a minimum
compacted their samples using the standard Proctor amount of stabilizer, in the region of 2 –3%.
test (600 kN/m2), and Bryan (1988a), Walker Unfavourable soils will require 4 –6% as shown by
(1995), and Ngowi (1997) used intermediate compac- the data indicating much better stabilized strengths
tion pressures of 2000 kN/m2. Increased compaction for these quantities than for 2% (Table 4). The greatest
pressure generally increases the strength (and density) effectiveness of cement is with low clay content soils
of a sample and decreases the OMC (e.g., Bryan such as sands, silty sands, and clay soils of low to
1988a), although the effect diminishes with increasing medium plasticity. These soils respond very well to
compaction pressure. If differences in strength between the addition of cement, and can require as little as 2 –
samples did in fact increase under higher compaction 2.5% of cement to stabilize satisfactorily (CSIRO,
pressures, the values in Table 5 pertaining to Croft 1987; UN, 1992).
(1968) and Akpokodje (1985) would be greater, and
the values pertaining to Bryan (1988a), Walker Concerning asphalt as a stabilizer, the debate amongst
(1995), and Ngowi (1997) would be slightly over- rammed earth practitioners has not been about the
estimated. Overall, however, Table 5 shows that additive’s waterproofing effect, but rather the potential
there is close similarity in the results between the it may have to alter the strength of compacted earth.
current investigation and other studies, even after The addition of asphalt to soil provides a membrane
taking into account any reported differences in that impedes the penetration of water, and on this
experimental technique. basis it has been suggested that asphalt stabilization
can improve compressive strength characteristics
A major finding in this study concerns the increase (Winterkorn, 1975; Al-Homoud et al., 1996). Although
in compressive strength of “unfavourable” soils asphalt has been viewed as being better suited as a
with increments of cement or lime added, compared stabilizer for soils with high clay contents (CSIRO,
with the non-effect for “favourable” soils. Soils with 1987; UN, 1992), the results of this study suggest
unfavourable characteristics are evidently more depen- that asphalt can be used in combination with either
dent on additives than on soil properties for their lime and/or cement on a wide range of soil types.
stabilized strength; in other words, stabilizers are the The use of 3% asphalt, as shown, has no overall influ-
major influence on strength. Although this makes intui- ence on the compressive strength of the stabilized soil
tive sense, it has not previously been shown using when used in combination with lime and/or cement,
empirical evidence. In contrast, soils with favourable and no preferential effect for soils regarded as either
characteristics require only a small amount of stabilizer favourable or unfavourable on the basis of their
(2– 3%) to achieve satisfactory strengths (3 MPa), properties. Therefore, its use should be viewed as a
and soil properties (specifically linear shrinkage, clay/ broad-spectrum waterproofing agent, rather than a
silt content, and sand content) are the major influence strength-imparting additive when used in combination
on stabilized strength variation. with either lime or cement.
63
Burroughs
Predicting values of UCS using soil property is favourably or unfavourably inclined to stabilization.
and stabilizer data The amount of lime or cement added to a soil increases
The basic aim of predicting stabilized strength is to its strength if the soil is relatively unfavourable, but the
minimize the chance of building an unsatisfactory com- strength of a favourable soil when stabilized is uniform
pacted earth wall and maximise the probability of within the range of lime or cement used (2– 6%). The
success. Two contrasting methods have been used in key set of variables relating to the compressive strength
this investigation that were able to predict the strength of favourable soils concerns the properties of the soils,
of stabilized material. The first was categorizing two in particular linear shrinkage, clay/silt content, and
groups of soils—“favourable” and “unfavourable” sand content. In contrast, the only soil property that
for stabilization—based on values of soil properties, is a determinant of UCS for unfavourable soils is
and comparing these two groups in terms of mean sand content.
UCS and the percentage of samples passing a 2 MPa
UCS criterion value. The second was using ANCOVA The importance of soil properties in explaining vari-
to quantify the amount of variation in UCS explained ations in stabilized compressive strength suggests that
by soil properties and stabilizer types/amounts. future research should be aimed at better quantifying
the relationship between key soil properties and UCS
There is an obvious efficiency with which threshold in order to predict the strength of stabilized soil.
values of soil properties (linear shrinkage and plasticity Linear models explained an average of only 33% of
index in particular) are able to discriminate between the variation in UCS, whereas soil properties efficiently
favourable and unfavourable soils with respect to the discriminated soils passing or failing a 2 MPa strength
2 MPa UCS criterion value used. This is in contrast threshold. Therefore, a criterion approach to predict-
to the rather poor results of predicting individual ing UCS is probably required rather than one that pre-
UCS values on the basis of soil properties and stabilizer dicts point values using linear methods. Such an
quantities, with only one-third of the variation in com- approach would predict whether UCS should exceed
pressive strength able to be explained using linear stat- a threshold value on the basis of soil properties and
istical models. It seems, therefore, that the way intended stabilizer treatment. The approach would
forward with respect to predicting stabilized strength have the practical advantage of being matched to build-
is not to predict the exact strength of a sample, but ing code specifications which define stabilized strength
rather to predict whether a sample is likely to pass or requirements in terms of threshold values.
fail a criterion strength value. Being able to predict
stabilization success or failure with respect to a cri-
terion value also makes more practical sense in that
building and professional regulations are specified in References
terms of such criterion values. Akpokodje, E.G. (1985) The stabilization of some arid zone soils
with cement and lime. Quarterly Journal of Engineering
Geology London, 18, 173–180.
Al-Homoud, A.S., Khedaywi, T. and Al-Ajlouni, A.M. (1996)
Engineering and environmental aspects of cutback asphalt
Summary and Conclusion (MC-70) stabilization of swelling and collapsible soils.
International Journal of Rock Mechanics and Mining
This study has examined the effect on the strength of Sciences, 33 (7), 317– 325.
compacted earth of soil properties (gradation, plas- Bell, F.G. (1996) Lime stabilization of clay minerals and soils.
ticity, and shrinkage), the type of stabilizer used Engineering Geology, 42(4), 223 –237.
(lime, cement, asphalt), and the quantity of each stabil- Bryan, A.J. (1988) Criteria for the suitability of soil for cement
izer used (0–6%), based on 219 tests of unconfined stabilization. Building and Environment, 23(4), 309–319.
Bryan, A.J. (1988a) Soil/cement as a walling material – I.
compressive strength (UCS) on 104 different soils. Dis- Stress/strain properties. Building and Environment, 23(4),
tinguished on the basis of their soil properties, soils 321–330.
favourably predisposed to stabilization (those with Croft, J.B. (1968) The problem in predicting the suitability of
sand contents .65%, or plasticity indexes ,15%, or soils for cementitious stabilization. Engineering Geology,
2(6), 397–424.
linear shrinkage values ,6.0%) can be differentiated CSIRO (1987) Earth Wall Construction. Bulletin 5, Common-
from unfavourable soils. Having controlled for stabil- wealth Scientific and Industrial Research Organisation
izer quantity, and depending on which particular soil (Division of Building Construction and Engineering),
property or combination of properties is used to Sydney, Australia.
categorize predisposition to stabilization, 86 –94% of DMR (1983) Determination of unconfined compressive strength
of remoulded road materials, Test Method T116. Depart-
favourable soils attain UCS values 2.0 MPa (mean ment of Main Roads, Sydney, New South Wales.
UCS ¼ 3.02 MPa), whereas only 56 –63% of Dumbleton, M.J. (1962) Investigations to Assess the potentialities
unfavourable soils (mean UCS ¼ 2.39 MPa) pass this of lime for soil stabilization in the United Kingdom. Technical
criterion. Paper, No. 64, Road Research Laboratory, Crowthorne,
England.
Fitzmaurice, R. (1958) Manual on stabilized soil construction
ANCOVA models show that the effect of stabilizer for housing. UN Technical Assistance Programme, United
quantity is dependent upon whether the soil of interest Nations, New York.
64
Strength of compacted earth
Metcalfe, J.B. (1963) A small scale trial of lime stabilization for UBC (1997) Uniform Building Code, 1997 Edition. International
building sites. CSIRO Soil Mechanics Section Technical Conference of Building Officials. ICBO: Whittier, CA.
Report, Series 3, Sydney, Australia. UN (1992) Earth Construction Technology. United Nations
Ngowi, A.B. (1997) Improving the traditional earth construction: Centre For Human Settlements, Nairobi.
a case study of Botswana. Construction and Building USACE (United States Army Corps of Engineers) (2000) Soil
Materials, 11(1), 1 –7. engineering and stabilization. http://www.adtdl.army.mil/
Osula, D.O.A. (1996) A comparative evaluation of cement cgi-bin/atdl.dll/fm/5 –410/TOC.htm (Chapters 5, 6, 7, 8,
and lime modification of laterite. Engineering Geology, 42, and 9).
71–81. VITA (1975) Making building blocks with the CINVA-Ram
Ransom, W.H. (1963) Tropical building studies No. 5. HMSO, block press. Volunteers in Technical Assistance, Intermedi-
ate Technology Publications, London.
Building Research Station, London.
Walker, P.J. (1995) Strength, durability and shrinkage character-
SAA (1977) Methods of testing soils for engineering purposes.
istics of cement stabilized soil blocks. Cement and Concrete
Standards Association of Australia, Australian Standard
Composites, 17, 301–310.
1289, Canberra, ACT, Australia. Winterkorn, H.F. (1975) Soil stabilization, in Winterkorn, H.F.
Spence, R.J.S. (1975) Predicting the performance of soil-cement and Fang, H.-Y. (Eds): Foundation Engineering Handbook,
as a building material in tropical countries. Building Van Nostrand Reinhold Company, New York.
Science, 10, 155– 159. Wolfskill, L.A., Dunlap, W. and Gallaway, B.M. (1963) Hand-
Symons, W.G. (1999) Properties of Australian soils stabilized book for Building Homes of Earth, Texas Transportation
with cementitious binders. Structural Materials and Assem- Institute, Report for the Agency for International Develop-
blies Group, University of South Australia, The Levels, ment, Bulletin 21, E 14-63, Texas A&M University,
South Australia. College Station, Texas.
65