PDF content (text-only)
1 1 1 1 1 F ~~~~ InternationalDevelopment
Expansive soils: TRI~s research
C S Gourley, D Newill and H D Schreiner
Berkshire RG45 6AU
PAl 301193 GOURLEY, C 5, D NEWILL and H D SCHREINER, 1993. Expansive
soils: TRL's research strategy. In: Proceedings of the First International
Symposium on Engineering Characteristics of Arid Soils, City University,
London, 5-8 July 1993. EXPANSIVE SOILS: TRL'S RESEARCH STRATEGY
C.S.Gourley, D.Newill and H.D.Schreiner*
Overseas Centre, Transport Research Laboratory,
Crowthorne, Berkshire, UK.
* University of Natal, Durban, Natal, S.Africa.
ABSTRACT This paper describes some aspects of a TRL programme of
research on expansive soils. The aim of the research programme was
to provide practising engineers with guidance on design and
construction of roads built on-or with expansive soils. Particular
emphasis was placed on the development of improved identification
procedures, both in the laboratory and in the field, for expansive
soils. A further aspect of work was the development of new
techniques whereby the influence of suction on the soil behaviour
could be investigated both in the field and the laboratory. The
influence of stress, density and moisture content on the swelling
characteristics of compacted expansive soils was investigated
experimentally using a range of swell test procedures.
Recommendations were then made on appropriate test procedures for
assessing the swell characteristics of compacted expansive soils.
The practical application of the research programme is to provide
road engineers with guidance on; laboratory and field identification
of expansive soils, laboratory swell testing of expansive soils,
assessment of the potential for volume change in the field and the
options available for road design.
Expansive soils are generally characterised by the presence of a clay
mineral of the smectite group. These soils can give rise to problems
in civil engineering works because of their capacity to undergo large
volume changes with changes in the moisture content or suction.
Some examples of the annual cost of damage to buildings and light
structures, including roads, caused by expansive soils are $1000
million in the USA, $150 million in the UK and at least $4 million in
South Africa. Expansive soils are found on all five continents and
are especially widespread in the wet and dry tropics. There are no
statistics available for the cost of damage in these areas but it is
clear that the figures given above represent only a small proportion
of the cost of the problem worldwide.
The Transport Research Laboratory (TRL) has been involved in various
aspects of research on these soils. This has involved collation of
existing information and an extensive programme of laboratory
research. The laboratory research carried out with Imperial College
London concentrated on methods for identification and evaluation of
swell of expansive soils. Particular reference was placed on their
behaviour under conditions of partial saturation, where large
negative pore water pressures can be present.
The aim of the research programme was improve the understanding of
these soils as construction materials, clarify the design
requirements for their use and to make this information available to highway engineers. The overall strategy and outputs of the research
programme is illustrated in Figure 1. The programme was structured
to include information on:
(a) identification and engineering significance of expansive
(b) potential for volume change in the field,
(c) laboratory evaluation and geotechnical properties of
expansive soils, and
Cd) use of expansive soils in road engineering
A series of reports will be published by TRL to describe each element
of the work in more detail. -Practical guidelines for highway
engineers will be prepared encompassing each of the four elements
given above. This paper briefly reviews some the main research
findings from the study.
IDENTIFICATION AND ENGINEERING SIGNIFICANCE
Damage caused by expansive soils is almost entirely restricted to
light structures and is a particular problem when they are
encountered in road construction. Early identification during site
investigation and laboratory testing is extremely important to ensure
that the correct design strategy is adopted.
Expansiveness is a property of the soil. There is no direct measure
of this property and therefore it is necessary to make use of
comparative values of swell, measured under known conditions, in
order to derive a method for assessing expansiveness. Consideration
of the mechanisms of interaction between water and clay soils show
that the three most important components are the clay minerals, the
change in moisture content or suction and the applied stresses. The
type of clay mineral is largely responsible for determining the soil
property referred to as the intrinsic expansiveness. It is the
change in moisture content or suction that controls the actual amount
of swell which a particular soil will exhibit under a particular
identification of intrinsically expansive soils
It is important that geotechnical and materials engineers should have
a reliable method available for identifying expansive soils. Many
attempts have been made in the past to provide such a method by
comparing swell data with one or more commonly determined soil index
values. Each may have some validity for a particular set of soils
and conditions but none has proved universally reliable. Reasons for
the failures of past attempts to use index tests centres around the
important influence of soil microfabric and stress history on the
A procedure has been developed as part of this study which avoids the
problems inherent in the use of compacted or natural undisturbed
samples and allows soils to be compared purely on the basis of
expansiveness without interference from microfabric and stress
history. The proposed means of estimating the intrinsic
expansiveness is shown in Table 1 using the liquid limit CBS 1377,
part 2, 1990) on the horizontal axis and the difference between the plastic limit (BS 1377, part 2, 1990) and the shrinkage limit (ASTM
D4943-89 (1992)) determined using the American procedure on the
vertical axis. The lines on the graph, shown in Table 1, represent
contours of expansiveness, (%) , determined using:
where: e15 is the void ratio after swelling of samples
reconstituted, consolidated and dried past the shrinkage
limit before swelling in an oedometer under l5kPa
vertical stress, and
eSL is the void ratio determined at the shrinkage limit
of the dried samples.
The soil is reconstituted by mixing to a paste at a moisture content
of approximately 1.2 of the liquid limit. The paste is placed in an
oedometer and consolidated in stages to 100OkPa. After unloading in
stages to 10OkPa, the sample is dried to the shrinkage limit and esL
determined. The sample is then allowed to swell in the oedometer
under 15kPa vertical stress until swelling is complete. The void
ratio, e15, is then determined at the end of this swelling stage.
The graph serves to provide comparative data for soils but does not
provide a means of estimating expansiveness for any other conditions.
Further data is being collected to validate the procedure.
POTENTIAL FOR VOLUME CHANGE OF EXPANSIVE SOILS
The occurrence of changes in the moisture conditions beneath covered
areas has been extensively documented. This part of the research
programme addressed those issues which are of importance in
understanding the changes that are likely to occur beneath roads in
various climatic regions. Surprisingly little useful data are
available, although the general trends of changes are moderately well
defined. The deficiency in reliable field suction data became
apparent during this part of the study and a clear need was
identified for development of suitable and simple methods for in situ
and laboratory measurement of soil suction. A rational approach for
the field and laboratory identification of expansive soils was
developed which gives the engineer a strategy whereby expansive soils
can be identified at an early stage in the site investigation.
Rational approach to identification of expansive soils
The engineering performance of a soil in situ, whether tropical or
temperate, residual or transported, will depend on at least all of
a) Mineralogy and composition (from index tests/grading),
b) Fabric and structure (from visual/microscopic study),
c) Stress history (from geology/laboratory tests), and
d) Applied stress changes (from design, construction and
climate). The composition, state of saturation and engineering properties of
the soils provide valuable information for the engineer in the early
stages of project planning and design. To sati~sfy this need the
identification strategy incorporating the field assessment and
laboratory test procedures shown in Table 1 is suggested. The
identification process is divided into three sections. The f irst
considers the information available through geological and
geomorphological investigation. The second covers the procedures
available for the assessment in the field by the engineer, where
detailed soil profiling, assessment of the moisture condition and
identification of soil class are the minimum requirement. The third
section considers simple index tests which can be carried out in the
laboratory to identify the intrinsic expansiveness of the soil.
Changes to in situ moisture and suction
Expansive soils become a particular problem where the prevailing
(a) arid and the "dry" soils are subjected to unusually high
rainfall, causing the soil to wet and expand,
(b) semi-arid and the moisture condition of the soil reflects the
wet-dry seasonal cycle.
(c) predominantly wet and they are subjected to a prolonged period
of drought and exhibit drying shrinkage,
To better assess the behaviour of the soil in the laboratory or
field, due consideration should be given to the influence of climatic
factors and the potential for the moisture content or suction of the
soil to vary. Moisture conditions beneath the ground surface are
described by the moisture content and the pore water pressure. Where
the pore water pressure is negative, i.e. there is a water deficit in
the soil, the pore water pressure is usually referred to as the
various factors will affect the moisture content and the suction of
the soil and a change in one will be associated with or will cause a
change in the other. Water can be supplied to the soil by rainfall,
rising ground water level and by local phenomena such as, irrigation,
leaking pipes etc. These increase the soil moisture and are
associated with a decrease in suction toward a value of zero. Acting
in opposition to this supply of water are those processes which act
to extract water from the soil. These include, for example,
evaporation, transpiration and the lowering of the groundwater level.
These decrease the available soil moisture and are associated with an
increase in the suction of the soil.
The equilibrium suction in compacted soils beneath sealed road
surfaces has been related to the prevailing local climate by use of
the Thornthwaite Moisture Index (TMI) by Russam and Coleman (1961).
Regions with a TMI of -20 to -40 are considered semi-arid, with those
showing values of TMI below -40 arid. This is illustrated in Figure
2 for a heavy clay material, such as expansive clay, where the data
were obtained from samples taken at a depth of 45cm in road sub-
grades. The sites where the data were collected had deep soil
profiles and deep water tables. Different relationships were found
between suction and TMI for different soil types. Figure 2 is
assumed to be representative of the "equilibrium" conditions beneath a sealed surface where the above conditions apply and there is good
drainage, uniform vegetation (without trees) and no change in
condition caused by infiltration.
Problems arise, particularly for the expansive soils group, where the
.local conditions modify the equilibrium suction from this reference
line. Differential wetting can be caused in a road sub-grade by run-
off from the sealed surface, lateral infiltration, leakage from
culverts, fluctuation in the ground water table, and the influence of
vegetation. In arid areas it may be the influence of these factors,
rather than those of the ambient climatic regime, which dictate the
volume change and consequent heave and shrinkage of expansive soils
used in road construction.
Seasonal movement (alternating heave and shrinkage) can occur where
alternate wet and dry seasons are a feature of the climate. Design
and repair are more difficult where this wetting and drying regime
exists. This form of alternating heave and shrinkage is generally
accepted as the primary cause of longitudinal cracking in sealed
roads over expansive soils (Dagg and Russamn 1966).
It is recognised that the determination of soil suction, although
difficult in practice, can give a fundamental insight into the
performance of the soil as an engineering material. In temperate (or
wet climates) , where the science of soil mechanics developed, there
is generally an excess of rainfall over evaporation. Soils generally
remain saturated and pore water pressures remain positive. For these
conditions the effective stress theory was developed and can be
effective stress = total stress -pore water pressure
The effective stress is the component which controls the behaviour of
saturated soils. If the effective stress changes then the soil will
undergo a volume change. For example, if the effective stress is
increased then the volume will decrease and vice versa. It is also
the effective stress that controls the strength of a saturated soil.
Increasing the effective stress will increase the strength of the
In dry climates there may be an excess of evaporation over rainfall,
leading to a water deficit. This is seen in the form of a negative
pore water pressure (or positive suction) and results from
evaporation and vegetation removing water from the soil. If the
negative pore water pressure is sufficiently large air enters the
soil and the effective stress equation is no longer adequate. To
describe the state of stress in the soil it becomes necessary to
investigate soil behaviour in terms of total applied stress (u-ua,)
and suction (uw-ua) . Changes in both total stress and suction will
cause, separately, changes in volume and strength.
Since suction is one of the two stress variables which controls soil
behaviour, we must be able to measure the suction in order to make
progress in the understanding of road performance in non-temperate
climates. As part of TRL's wider study of expansive soils, a suction
probe was developed to measure suction in subgrade soils and embankment fills under experimental road pavements constructed by TRIL
in developing countries. The system uses filter paper which acts as
a disposable sensor sealed within an easily installed sensing
chamber. Data are being collected from a range of field sites to
allow the influence of suction on the behaviour of these soils to be
evaluated. In the laboratory, measurement of suction in soils cannot
routinely be made except for the very low range CO to 10OkPa) of
suction. A comparative experimental study was undertaken to evaluate
a range of laboratory suction measurement techniques, including
filter paper, pressure plate and psychrometers in addition to
pressure plate control tests. All the methods investigated were.
shown to be reliable and to give comparable results. The f ilter
paper method was favoured as a routine measurement system because it
is cheap, simple to use and could be used over a wide range of
suction. Additionally, the filter paper had the advantage that it
could be used for measurement of both matrix and total suction. The
two measurement systems, both based on the filter paper technique,
can be used in tandem for reliable measurement of soil suction.
LABORATORY EVALUATION OF EXPANSIVE SOILS
As part of the research programme, information has been compiled
summarising the geotechnical properties of highly plastic or
expansive soils. This has included laboratory test information from
classification data, strength parameters, compaction characteristics
and swell test data. Further laboratory tests were carried out to
investigate the effect of stress and fabric on the engineering
behaviour of the soil. These tests showed the importance of
distinguishing between swelling of compacted, undisturbed and
reconstituted samples. Compacted soils may have a microfabric that
is very different from that of the undisturbed soil. This may in
turn have a microfabric that differs from that in a sample of the
same soil which has been reconstituted (i.e. consolidated from a
moisture content above the liquid limit) . The intrinsic
expansiveness will not vary between these samples, but the
microfabric present in the soil before wetting may have a significant
modifying effect on the measured swell if the microfabric is
susceptible to alteration.
Very few comprehensive suites of testing have been performed on
expansive soils. The test suite required includes liquid and plastic
limits, shrinkage limit, clay size fraction, the fraction passing the
4251Am sieve (if so separated for index tests) and an appropriate
swell test together with CBR, compaction or shear strength tests if
desired. Very often some of the classification tests are omitted.
It is common to find soils described as expansive without reference
to any swell test or field heave data.
Laboratory testing of highly plastic or expansive soils can often be
a problem. Those procedures where particular guidance will be of
benefit have been summarised in Table 2. Some field investigation
procedures are also considered which can provide reliable design data
and avoid some of the problems associated with laboratory testing. Index tests
Many correlations have been made between routinely determined soil
properties, such as Atterberg Limits, and the magnitudes of swell
obtained from case histories or measured in laboratory swell tests.
Generally these are empirical relations which fail to distinguish
between the three major components involved (expansiveness, suction
change and applied stress) . Attempts to correlate the Atterberg
Limits, which should reflect the clay mineralogy and thus intrinsic
expansiveness, with the observed swell cannot be made. The observed
swell will reflect the particular moisture change and applied stress
used in the test procedure. A limited degree of success has however
been achieved with some of these methods.
The one situation where index data may be used on its own for the
prediction of field behaviour is where the engineering processes
obscure or destroy the effect of factors such as microfabric and
previous stress history. This condition can arise where the soil is
used for fill and road formations. The natural stress history,
fabric and structure of the soil are replaced by those imposed during
the engineering processes. Those factors remaining which dictate the
engineering performance include the soil type, the design,
construction and climatic factors.
Measurement of swell for design purposes should simulate the expected
sequence and magnitude of loading and wetting changes that are
expected in the field. The use of standard oedometer test procedures
where the soils are soaked to zero suction were investigated as part
of the study. These are "quick" tests which provide data on volume
change and vertical stress only and have been used to investigate the
influence of parameters such as vertical stress and compaction
moisture content on swelling. The swell data obtained for a number
of expansive soils, compacted at a range of initial moisture contents
and allowed to swell under a range of vertical stress, was compared
with Brackley's (1983) swell prediction equation. The results showed
that although Brackley's equation gave similar patterns swelling, the
values calculated can be so different that the equation must not be
universally applied. For example, a sample of expansive clay from
Kenya compacted at G6% moisture content and subjected to swelling
under 64kPa vertical stress gave a measured swell of 921k with 115%
swell predicted using Brackley's equation. For the same soil
compacted at 35% moisture content, under a 1.5kPa vertical stress,
the comparison was good, with 34% swell measured and predicted.
However, for the same compaction moisture content and a vertical
applied stress of 256kPa, 14% swell was predicted using Brackley's
equation where no swell was measured.
Advanced evaluation of swell tests
In general, and particularly in arid regions where these soils are
compacted at a moisture content close to the optimum moisture
content, they do for the most part dry back and exhibit a negative
pore water pressure. It is clear that information on the volume
change at an applied vertical stress under soaking in the standard
types of oedometer may not reflect the response of the soil in situ.
TRL commissioned the design and fabrication of a hydraulic oedometer for determination of the full stress state during swelling and
compression of unsaturated soils, where the matrix suction can be
measured or controlled.
Tests can be carried out in the stress path oedometer for detailed
study of the volume change characteristics of compacted expansive
soils under controlled suction and applied stress. The tests are
"9slow"l in comparison with the standard oedometers but provide data on
axial (vertical) and radial total stresses, suction and void ratio
(or volume) . The programme of tests carried out using the stress
path oedometer allowed evaluation to be made of the swell measured in
three standard oedometer procedures. Investigation of alteration of
the microfabric during swell- and swell pressure testing was also
EXPANSIVE SOILS IN ROAD ENGINEERING
The design engineer needs to address the influence of expansive soils
both as naturally occurring undisturbed soils beneath the road and as
compacted soil in the road formation. It is therefore necessary to
identify the source of expected heave or shrinkage before selecting
the design solution. There may be little benefit importing non-
expansive fill materials if the undisturbed soil beneath the road
formation is expansive and heaves with consequent damage to the
entire road structure. Those issues which are of importance in
understanding the moisture changes that are likely to occur beneath
roads in various climatic regions have been reviewed. Generally
where a strength design is based on a permanently wet condition it is
conservative or over-designed. No guidance has been f ound on the
engineering implications of changes in moisture content or suction
during the life of a road, nor on their effect on CBR values as
determined in the laboratory. However, in areas with seasonal
variations of moisture content and suction in the soil, damage in the
form of longitudinal cracking can occur, whilst spatial variation of
density, suction and soil type can cause differential heave.
For the road design engineer working in areas of expansive soils six
options were identified which should be considered where both in situ
and compacted expansive soils are present. These are:
(a) Alter the route/alignment to avoid the expansive soil.
(b) Remove the expansive soil and replace it with non-
(c) Design for the low strength and allow for maintenance to
repair heave deformations.
(d) Provide non-expansive material as a cover or surcharge
(e) Control moisture movement.
(f) Improve the expansive soil by stabilisation.
The choice of which option or combination of options depends
primarily on the size of the project, economic considerations and the
element of risk acceptable to the client. It seems sensible to
suggest and base judgement on the argument that if a method is
effective in preventing damage then it is justifiable to use that
method in preference to another which is seen as less reliable. Cost
considerations, though important, should not totally dictate design. Generally a compromise must be sought between quality, performance
and cost. The selection of one design method over another must be
made through reasoned engineering judgement and understanding of all
those factors which can affect change and should not be based solely
on short term cost considerations. It is also important to note that
though there is an apparent increased cost of using a preventative
design approach, the cost of remedial work may be considerably
higher. Careful evaluation of each case is required by the client
and engineer working together.
TRL's research programme on expansive soils has been briefly
described. The TRL, research programme has:
(a) shown that existing procedures for identifying expansive soils
using index test data are frequently incorrect. An improved
procedure has been developed.
(b) shown by examination of existing information on suction beneath
roads that many previous investigations have been poorly
interpreted with regard to ground conditions. Extension of the
early TRI, work has led to a better understanding of the inter-
relation between climatic and ground conditions in controlling
the suction and moisture movement beneath roads. A low-cost
field measurement probe using a filter paper sensor has been
developed which is simple, robust and reliable. This is
ideally suited for use in developing countries to obtain soil
suction data beneath roads.
(c) led to clarification of the influence of stress, density,
suction and compaction moisture content on the swelling of
expansive soils. Published procedures for swell testing have
been compared experimentally. Recommendations have been made
on appropriate test procedures.
(d) identified the options available to road design engineers.
Each of the options available have been described so that the
design engineer can make a reasoned choice of which is
appropriate for the project under consideration.
ASTM D4943-89. (1992) . Standard test method for shrinkage factors of
soils by the wax method. Annual Book of ASTM Standards, Volume
Brackley, I.J.A. (1983) . An Empirical Equation for the Prediction of
Clay Heave. Proceedings 7th Asian Regional Conference on Soil
Mechanics and Foundation Engineering, Volume 1, p8-14.
British Standards Institution. (1990) . Methods of test for soils for
civil engineering purposes: BS1377, PART 2, Classification
tests. British Standards Institution, London. British Standards Institution. (1990) . Methods of test for soils for
civil engineering purposes: BS1377, PART 3, Chemical and
electro-chemical tests. British Standards Institution, London.
Dagg, M. and Russam, K. (1966). The relation between soil shrinkage
and the development of surface cracks in an experimental road
in Kenya. RRI, Report No. 12, Road Research Laboratory,
Ministry of Transport, England.
Holtz, W.G. and Gibbs, H.J. (1956) . Engineering Properties of
Expansive Clays. Transactions American Society Civil
Engineers, Volume 121, p641-663.
Jennings, J.E., Firth, R.A., Ralph, T.K. and Nagar, N. (1973). An
improved method for predicting heave using the oedometer test.
Proceedings 3rd International Conference on Expansive Soils,
Haifa, Israel, Volume 2, p149-154.
Jennings, J.E. and Knight, K. (1975) . A guide to construction on or
with materials exhibiting additional settlement due to collapse
of grain structure. Proceedings 6th African Regional Conference
on Soil Mechanics and Foundation Engineering, Durban, Volume 1,
Justo, J.L., Delgado, A. and Ruiz, J. (1984). The Influence of Stress
Path in the Collapse-swelling of soils in the laboratory.
Proceedings 5th International Conference on Expansive Soils,
Russam, K. & Coleman, J.D. (1961) .The effect of climatic factors on
subgrade moisture conditions. Geotechnique Volume 11, No. 1,
Crown Copyright 1993. The views expressed in this paper are not
necessarily those of the Department of Transport.
The work described in this paper forms part of an Overseas
Development Administration funded research programme conducted by the
Transport Research Laboratory, and the paper is published by
permission of the Overseas Development Administration and the Chief
Executive of TRL. 0 00000
a) 30 ( .
>.0 0 U0 0
00 0 0.0 0 0.0000).0'000
300>, '0 00>,
0 0 0.'
'00 0 Oh '0 00) -
2 0 .' 0 .-0W0,0
0 0, 0m ao
0j,c m aN
an0 GaW -0 tot 'D- -
'o n W ot >oC a t
ton-son0n.na 1 too cc: to WoW> Xo~ o
F- 0o-G U F-
F- u n0 0 t o F- t om t o
-10 -t 3 ctotoUtucUP0 -c to t co t aut0a Uo to toa nu n 0co tot
0 '' C
n 0 Futo
o to to- to
-a - >0-- to
--onton to to3totoouE-aotoua toGa3atoo 3 -nGuutoao to"-.Quo
to 0 -to to
a' ato na a-- at- nO nr- Er- - to F- r- to to to n to to -<--to
aato toN CE -too, .0 a C a touG totoanou toC to totoG 0 n- 0 to -3-, to to a u 'to C to