Anda di halaman 1dari 8

Bull Eng Geol Environ (2012) 71:71–78

DOI 10.1007/s10064-011-0410-7

ORIGINAL PAPER

Prediction of expansive soil swelling based on four micro-scale


properties
Botao Lin • Amy B. Cerato

Received: 23 January 2011 / Accepted: 12 November 2011 / Published online: 15 December 2011
Ó Springer-Verlag 2011

Abstract A comprehensive study of expansive soil réalisées, basées sur chacune des propriétés de micro-échelle,
behavior includes understanding the surface phenomena of afin de prévoir le potentiel de gonflement et la pression de
clay particles within the soil matrix. This research studies gonflement. La succion matricielle a été identifiée comme le
four micro-scale properties of four remolded expansive meilleur paramètre pour prévoir le comportement gonflant des
soils––matric suction, pH, surface conductance and per- sols étudiés qui avaient été initialement compactés à la teneur
centage of montmorillonite––in order to predict soil en eau optimum. La conductivité de surface, qui intègre la
swelling. An approach to approximate surface conductance valeur de surface spécifique, la capacité d’échange de cations
is presented. Linear regression analyses were undertaken in et la mobilité cationique a donné également de bonnes pré-
an attempt to predict percent swell and swell pressure visions, sauf pour l’un des sols de forte acidité.
based on each of these micro-scale properties. Matric
suction was found to be the most accurate predictor of the Mots clés Sols gonflants  Succion matricielle 
swelling behavior of the studied soils, which were initially Conductivité de surface  Type d’argile  Potentiel de
compacted at optimum moisture content. Surface conduc- gonflement  Pression de gonflement
tance, which is a combination of specific surface area,
cation exchange capacity and cation mobility, also gave
good predictions, except for one soil with high acidity. Introduction

Keywords Expansive soils  Matric suction  Surface Expansive soil, the type of clayey soil that experiences
conductance  Clay type  Swell percent  Swell pressure volumetric swelling or shrinkage with adsorption or
desorption of water, covers a quarter of the United States
Résumé Une étude approfondie du comportement des sols and causes more than 9 billion dollars in loss every year
gonflants inclut la compréhension des phénomènes de surface (Nelson and Miller 1992). Swell potential, in terms of
concernant les particules argileuses au sein de la matrice du volumetric expansion and swell pressure, is difficult to
sol. Cette recherche s’intéresse à quatre propriétés de micro- predict based on common geotechnical properties because
échelle de quatre sols gonflants remaniés : succion matricielle, the individual clay platelets are largely controlled by sur-
pH, conductivité de surface et pourcentage de montmoril- face physicochemical forces rather than mechanical forces,
lonite, afin de prévoir le gonflement du sol. Une méthode due to their infinitesimal size and the diffuse double layer
permettant d’approcher la valeur de conductivité de surface formed around each clay platelet (or aggregate). Conse-
est présentée. Des analyses de régression linéaire ont été quently, to more fully understand the identification of
expansive soil and predict quantitative movement requires
an appreciation of the surface phenomena of individual
clay platelets within the soil matrix and the interaction of
B. Lin  A. B. Cerato (&)
contacting platelets or aggregates caused by surface forces.
University of Oklahoma, 202 W. Boyd Street, Room 334,
Norman, OK 73019, USA The conventional study of expansive soil characteristics
e-mail: acerato@ou.edu can be divided into the study of plasticity and soil heave

123
72 B. Lin, A. B. Cerato

measurement (Nelson and Miller 1992). The most com- pressure, X-ray diffraction (XRD) provides a straightfor-
monly studied plasticity properties are plasticity index (PI) ward way to study the mineralogical make-up of expansive
and activity (Skempton 1953; Raman 1967; Chen 1988), soils (Al-Rawas 1999), while scanning electron micro-
with the latter defined as PI over percentage of clay con- scopy (SEM) has been frequently used to investigate the
tent. Soil heave has been investigated in terms of free swell microstructure of soils (Al-Rawas 1999; Al-Rawas and
tests (Holtz and Gibbs 1956); swell index and potential McGown 1999; Du et al. 1999; Shi et al. 2002). However,
volume change (Lambe 1960); percent swell in the Cali- to date SEM micrographs are still difficult to quantify.
fornia bearing ratio (CBR) test (Yoder and Witczak 1975; In this study, matric suction, pH, surface conductance (a
Kassiff et al. 1969) and coefficient of linear extensibility combined property of SSA, CEC and cation mobility) and
(COLE) (Brasher et al. 1966). All of these indices are percentage of montmorillonite (derived from clay size
related to percent of swell or swell pressure of expansive fraction and XRD diffractogram), are taken as micro-scale
soils under various test conditions. However, plasticity parameters because they can be quantified and are closely
alone is far from a quantitative prediction of soil expansion related to the development of the diffuse double layer
as it reveals little about the underlying physicochemical which determines the macroscopic soil behavior of highly
mechanisms in soil microstructure; whereas soil heave clayey expansive soils. Four field-sampled expansive soils
measurement is simply a demonstration of soil expansion were tested to obtain their micro-scale properties in order
under various confining conditions. In other words, the to predict the percent of swell and swell pressure. Linear
traditional methods used to study expansive soil behavior regression analyses were performed based on each micro-
are phenomenological and limited by various boundary and scale property and the predicted and observed values were
environmental conditions. compared.
An expansive soil is a type of unsaturated clayey soil
within which the voids are partially filled with water.
Therefore, before soil expansion even takes place, the soil Tested materials and soil properties
behavior is significantly affected by surface tension forces
at the air–water interface, which are defined as matric Four remolded expansive soils––Carnisaw, Hollywood,
suction (Nelson and Miller 1992; Fredlund and Rahardjo Heiden and Eagle Ford––were studied; the first three
1993). Mckeen (1992) created a classification model of sampled from Oklahoma and the last from north Texas.
swell potential based on suction by correlating the soil
volume response to suction change with the slope of Common geotechnical properties
the suction water content relationship. Garbulewski and
Zakowicz (1995) presented a classification scheme of swell Common geotechnical laboratory tests were performed
potential by evaluating the relationship between swelling following ASTM standards (ASTM 1999); the results are
pressure and suction. Delage et al. (1998) investigated given in Table 1. The swell potential was assessed by
volume changes of an expansive clay during wetting–dry- performing the standard test method for one-dimensional
ing circles under constrained suction values. In the field swell test of cohesive soils (ASTM-D4546 1999). Com-
where expansive soils are encountered, e.g., a landfill pacted samples at OMC were submerged in de-ionized
barrier, the soil is usually compacted at optimum moisture water in an oedometer cell and allowed to swell under a
content (OMC). As a result, it would be interesting to learn seating load of 1 kPa. After full expansion, the samples
the effect of matric suction at OMC on the swell potential were consolidated to their original height so that the swell
of expansive soils, which will be investigated in this study. pressure could be deduced. The swell test results are given
Along with matric suction, specific surface area (SSA) in Table 2.
and cation exchange capacity (CEC) have been recognized
as important factors controlling many physical and chem- Measurement of matric suction
ical properties of fine grained soils (Mitchell and Soga
2005; Peterson et al. 1996). pH is derived from the acidity The matric suction of each soil sample at OMC was first
of the soil and is indicative of the way hydrogen is bounded measured by the filter paper method (ASTM-D5298 1999).
on the soil mineralogical unit, which can both affect the Two specimens were prepared for each soil sample and the
degree of effective negative charge of the soil particle and process of moisture equalization was maintained for
determine the charge type of ionized alumina exposed at 2 weeks. Pressure plate tests were also performed to
the edge of clay platelets (Mitchell and Soga 2005). In deduce the soil water characteristic curves (SWCC) of
addition to the physicochemical parameters of soil, such as these four soils with a measuring range from 1 to
matric suction, SSA, CEC and pH, in assisting in under- 1,000 kPa. Soil specimens were compacted at OMC and
standing the mechanisms controlling soil swell and swell then saturated in a humidity-controlled chamber. The

123
Prediction of soil swell 73

Table 1 Some geotechnical properties of the samples tested


Test sample LL (%) PL (%) SL (%) PI (%) CF (%) A Gs USC

Carnisaw 59 32 20 27 51 0.5 2.68 MH


Hollywood 54 20 16 34 62 0.6 2.78 CH
Heiden 70 21 19 49 50 1.0 2.77 CH
Eagle ford 92 35 25 57 66 0.9 2.71 CH
LL liquid limit, PL plastic limit, SL shrinkage limit, PI plasticity index, CF clay size fraction (\2 lm), A activity, Gs specific gravity, USC unified
soil classification

Table 2 Summary of measured matric suction and swell potential


Test sample OMC cd (kN m-3) (ua-uw)OMC (kPa) S (%) SP (kPa) Swell potentiala
FP PP Averaged

Carnisaw 26.2 16.2 407 469 438 2.3 75 Medium


Hollywood 20.6 16.7 565 758 662 5.6 141 Medium
Heiden 24.2 15.5 771 1,120 946 9.3 230 High
Eagle Ford 27.1 14.2 1,040 1,126 1,083 12.7 263 Very high
OMC optimum moisture content, cd maximum dry unit weight, (ua-uw)OMC matric suction at OMC, FP filter paper test, PP pressure plate test,
S percent of swell, SP swell pressure
a
Based on Chen (1988) classification scheme

SWCCs were then determined following the drying path the averaged matric suction and the quantified swell
according to ASTM-D2325 (1999). The SWCC model of potential (percent of swell and swell pressure), which will
Fredlund and Xing (1994) was used to fit the measured data be discussed later.
points of volumetric water content versus matric suction
and the matric suction at OMC was determined from the Specific surface area and cation exchange capacity tests
fitted equations. The matric suction values of each soil
sample obtained using each of the measuring methods are The SSA values given in Table 3 were obtained following
listed in Table 2. the ethylene glycol monoethyl ether (EGME) surface area
The magnitude of matric suction predicted from the determination method proposed by Cerato and Lutenegger
filter paper tests is consistently lower than that from the (2002). The CEC and the pH of each soil were determined
pressure plate tests. During the process of moisture by Harris Laboratory, Inc., Lincoln, Nebraska using a 1 N
equalization, the water vapor dissipated to the sandwiched ammonium acetate extraction method (Rhoades 1982).
filter paper may contain infinitesimal particles or salts In Table 3, a more descriptive soil property, surface con-
existing in highly clayey soils that would deposit on the ductance ks was obtained from the measured SSA, CEC, in
paper resulting in underestimation of matric suction. On the combination with the mobility of counterions, ucat (m2 V-1
other hand, in the pressure plate tests, the time period for s-1), following an approximate equation (Santamarina
moisture stabilization at a specific matric suction was set to et al. 2001; Ning 2007) defined as:
be 24–48 h. It is possible that the stabilization was
CEC
incomplete after this period such that the corresponding ks  ucat ð1Þ
water content is overestimated and hence the matric suction SSA
at a given water content is over-predicted. As a result, the Considering the cations as counterions, the mobility for
measured values from these two different methods are each cation of K, Mg, Ca, Na and H in each soil is 7.6, 5.5,
averaged to approximate the actual magnitudes as given in 6.2, 5.2, 36.2 (910-8 m2 V-1 s-1), respectively, according
Table 2. to the mobility chart of cations given by Santamarina et al.
As indicated in Table 2, soils with higher averaged (2001). Electron mobility is an ion property determining
matric suction at OMC show a greater swell potential in the drift velocity of the electrons; it can mobilize given an
terms of percent of swell, swell pressure and the classifi- applied electrical field. Equation 1 assumes that the anionic
cation scheme of swell potential. It is desirable to correlate deficit in the diffuse double layer is zero, while in reality

123
74 B. Lin, A. B. Cerato

Table 3 Some physicochemical properties of the samples tested


Test sample SSA (m2 g-1) CEC (meq/100 g) pH K (%) Mg (%) Ca (%) Na (%) H (%) ks (910-8 S)

Carnisaw 107.5 27.3 4.4 0.9 9.1 1.8 0.2 88 8.27


Hollywood 145.5 26.4 7.2 1.3 4.6 93.2 1.0 0 1.12
Heiden 229.0 50.7 8.7 0.8 11.0 82.2 6.0 0 1.34
Eagle Ford 213.5 49.6 7.7 1.5 13.2 77.8 7.5 0 1.41
SSA specific surface area (total), CEC cation exchange capacity, ks surface conductance (weighted estimation)

surface conductance depends on both the concentration of


excess cations and the deficit of anions, in which the latter
is much smaller than the former and can be neglected in the
estimation of surface conductance (Klein and Santamarina
2003). If the surface conductance is weighted based on the
percentage of each cation present in the soil, the original
surface conductance definition can be modified as:
P
ui Pi CEC
ks ¼ ð2Þ
SSA
where ui is the mobility of each cation and the percentage,
Pi, is based on the count and not the mass of each cation. It
is thought that the weighted surface conductance property
describes the electrical conductivity of the diffuse double
Fig. 1 X-ray diffraction patterns of the Carnisaw soil
layer and may help in predicting the volumetric swelling
and swell pressure of expansive soils.

Mineralogical studies

Mineralogical studies were undertaken on only the clay


portion of each soil sample. It is believed that the physi-
cochemical effects only take place in micropores between
clay platelets (aggregates), which suggests that the swell
potential of the clay portion accounts for the majority of
the soil expansion. Approximately 10 g of each soil that
passed the #200 sieve (0.075 mm) were poured into a
centrifuge bottle filled with 200 mL of 10-3 mol L-1
sodium hexametaphosphate solution (dispersive agent) and Fig. 2 X-ray diffraction patterns of the Hollywood soil
mixed to prepare suspensions for XRD tests.
The soil suspension was then put into a centrifuge bottle. Report 2001). Untreated, glycolated and heated (at 550°C
The bottles were placed on a shaker table and shaken for for 2 h) sample slides were prepared and the corresponding
15 min, and sonified in an ultrasonic dismembrator for diffractograms of each soil are given in Figs. 1, 2, 3 and 4.
about 1 min. Next, the bottles were placed in a centrifuge The mineral symbols in these figures represent: V, ver-
for 10 min at 95 rcf to separate out the[2 lm size fraction. miculite; P, palygorskite; M, montmorillonite; K, kaolinite;
After this was accomplished, the remaining clay suspen- I, illite; GO, goethite. The X-ray generator used 40 kV and
sion was decanted into a high-speed centrifuge bottle and 30 mA to generate Cu–K alpha radiation and the data were
centrifuged at 16,278 rcf for 30 min. For the preparation of collected from 4–34° 2h at a rate of 0.05° 2h per second.
oriented XRD samples, the clear liquid was removed and A summary of the results and observations from the
the remaining slurry was thoroughly mixed with pure water XRD tests is given in Table 4, together with the geological
to make a thick suspension, which was then mounted on a classification and the geotechnical description. The per-
glass slide following the method of ‘‘Oriented aggregate centages of cations determined from the CEC tests are also
mounts for X-ray powder diffraction’’ (USGS Open-File presented in Table 4. The percentages of montmorillonite,

123
Prediction of soil swell 75

percentages of clay minerals can be estimated as the per-


centages in the clay portion multiplied by clay size frac-
tion. In the following regression only the percentage of
montmorillonite (vermiculite) is used as montmorillonite is
much more capable of developing a diffuse double layer
than other minerals (Mitchell and Soga 2005). No differ-
entiation is made here between montmorillonite and ver-
miculite due to the technical limitations of the XRD test
itself; although these swelling minerals may exhibit various
degrees of swell potential. The type of exchangeable cation
in an expansive soil also has a significant influence on its
swelling potential. In Table 4, Ca2? is the predominant
type of cation for all the four soils except Carnisaw.
Fig. 3 X-ray diffraction patterns of the Heiden soil
Nevertheless, divalent cations as Ca2? and Mg2? prevail as
interlayer hydrated cations in the mineralogical unit of
Carnisaw compared to the percentages of Na? and K? in
Table 4. In this study montmorillonite is assumed to be of
one type for the following regression analysis.

Regression analysis

As mentioned above, the measured percent of swell and


swell pressure were from one-dimensional oedometer tests
which are subject to certain boundary and environmental
limitations. At this stage, however, these two parameters
will still be used as the two dependent variables that are
representative of macroscopic swelling. In this study it was
Fig. 4 X-ray diffraction patterns of the Eagle Ford soil
considered desirable to perform single linear regression
analysis based on each of the measured micro-scale prop-
vermiculite, illite and kaolinite were determined from the erties to predict these dependent variables and compare the
XRD diffractograms following the approach of ‘‘Semi- predicted with the observed values. Multi-linear regression
quantitative analysis of clay minerals’’ (USGS Open-File was not appropriate due to insufficient soil sample data; it
Report 2001), and represent the approximate percentages yielded R2 values equal to 1.0, which render this type of
within the clay portion of the soils. regression meaningless. Matric suction, pH, surface con-
Table 4 shows the soils consist of a majority of clay ductance and percentage of montmorillonite (PM) were
weathered from clayey shale or colluvial sediment. Given selected as four independent variables to be used in the
the clay size fraction of each soil (Table 1) the real linear regression analyses.

Table 4 Summary of the geological, geotechnical, mineralogical and chemical results


Test Geological classification Geotechnical description Clay minerals Cations (%)
sample (%)a

Carnisaw Residuum weathered from shale of Red brown silty clay or elastic silt V (24), I (48), Na (0.2), Ca (1.8), Mg (9.1),
Pennsylvanian age KA (28) K (0.9), H (88.0)
Hollywood Clayey colluvial sediments over Yellowish olive fat clay with trace M ? V (37), Na (1.0), Ca (93.2), Mg (4.6),
limestone of Cretaceous age of light grey limestone I (34), KA (29) K (1.3),
Heiden Clayey shale of Cretaceous age Olive grey highly plastic fat clay M (73), I (10), Na (6.0), Ca (82.2), Mg (11.0),
with trace of organic root KA (17) K (0.8),
Eagle Ford Fossiliferous clayey shale with sandy Yellowish tan highly plastic pure M (42), I (41), Na (7.5), Ca (77.8), Mg (13.2),
shale lenses clay KA (17) K (1.5),
M montmorillonite, KA kaolinite, I illite, V vermiculite, Na sodium, Ca calcium, Mg magnesium, K potassium, H hydrogen
a
The mineralogical description is related to the clay portion only

123
76 B. Lin, A. B. Cerato

In the input of the regression, the matric suction and the


swell pressure were normalized by one atmosphere (100 kPa),
while the surface conductance was normalized by 10-8
S cm-1. As a result, both the independent variables and the
dependent variables become unit-less. The regression equa-
tions and the corresponding R2 values are listed in Table 5. As
shown, the predictions based on (ua-uw) and ks2 rendered the
highest R2 magnitudes (highlighted in bold face). The com-
parison between the predicted values and the observed values
are given in Figs. 5 and 6. Note that in Table 5 the predictions
based on surface conductance are divided into two categories:
the first one (ks1) includes the data of all the four soils and the
second (ks2) includes three soils, neglecting Carnisaw. As
shown in Table 3, Carnisaw has an exceptionally high surface
conductance which is mainly due to the effect of hydrogen,
Fig. 5 Comparison between predicted and observed percent of swell
while no hydrogen contributes to the surface conductance
value for the other three soils. As a result, it is desirable to
study the effect of surface conductance on soil swell data with
and without the inclusion of Carnisaw.

Discussion

For highly clayey expansive soils, swelling is largely depen-


dent on electrical, chemical and physical forces that act on the
surface of clay platelets or aggregates. These surface forces
are related to the interaction between the diffuse double layers
of adjacent particles and the interaction between the diffuse
double layer and bulk fluid or air. However, these forces are
difficult to identify and quantify. Instead, the four micro-scale
properties, as proposed in this study, are investigated from
direct laboratory measurements. Fig. 6 Comparison between predicted and observed swell pressure
(a) Matric suction determines how the soil particles are
bonded together by the surface tension force as well (c) Surface conductance (weighted) is a comprehensive
as the degree of development of a diffuse double layer electrical property combining SSA, CEC and coun-
around clay platelets. terion mobility. It is believed to be a more direct
(b) pH is derived from the acidity of the soil and is illustration of the effect of electrical forces.
indicative of the way hydrogen bonds on the soil (d) Percentage of montmorillonite is used to describe the
mineralogical unit; as a result, it affects the develop- influence of expansive minerals of the clay portion on
ment of the diffuse double layer. soil expansion. In this case, calcium montmorillonite

Table 5 Regression equations for the prediction of percent of swell and swell pressure
Prediction Percent of swell, S (%) R2 Swell pressure, SPa R2

Based on (ua-uw)a S = -4.64 ? 1.55 (ua-uw) 0.99 SP = -0.54 ? 0.30 (ua-uw) 0.99
Based on pH S = -6.17 ? 1.95 pH 0.63 SP = -1.03 ? 0.40 pH 0.75
Based on ks1 S = 10.39 - 0.96 ks1 0.55 SP = 2.35 - 0.19 ks1 0.60
Based on ks2 S = -20.11 ? 22.71 ks2 0.92 SP = -3.33 ? 4.22 ks2 0.99
Based on PM S = -0.85 ? 33.31 PM 0.59 SP = 0.02 ? 7.03 MI 0.73
ks1 surface conductance of all the soils, ks2 surface conductance of the soils excluding Carnisaw
a
(ua-uw) and SP were normalized by 1 atmospheric pressure (100 kPa)

123
Prediction of soil swell 77

and vermiculite are two major swelling minerals in will result in excellent correlations, as can seen in Table 5
the soils studied. and Figs. 5 and 6. PM provides a relatively poor prediction
probably because of the variation in swelling potential
Another possible expansive mineral is swelling chloride,
among the swelling minerals and the estimation involved in
which was not found in the four soils studied. Percentage of
the quantification of the mineralogy from the XRD analy-
montmorillonite is able to somehow describe the degree of
sis. Moreover, ks, pH and PM have nothing to do with the
development of the diffuse double layer. For example, soil
initial compaction condition of the soil and the prediction
with a large amount of sodium montmorillonite can
based on these properties should include such information
develop a well-formed and thick diffuse double layer
in future research.
around clay platelets. At the same time, the formation of a
diffuse double layer is stronger around wide and platy
montmorillonite platelets than small and thicker ones. Conclusions
Plastic index (PI) and activity are not considered micro-
scale properties because they are determined from an Four micro-scale properties were studied to predict the
empirical specification for plasticity (e.g., blow number) macroscopic behavior of expansive soils in terms of soil
and not directly from an underlying physicochemical expansion. These properties are described as micro-scale
mechanism. In other words, PI is largely a phenomeno- because they quantify or semi-quantify (in respect to
logical and user-defined concept. Furthermore, the deter- the case of PM) the intrinsic microscopic behavior of
mination of PI of identical specimens of soil has been expansive soils. Plasticity is not considered a micro-scale
found to produce highly variable results between different property because it is defined from phenomenological
laboratory testers and is also sensitive to the specimen observations, i.e., obtaining the moisture content according
preparation method (Armstrong and Petry 1986). Conse- to a specific blow number to determine liquid limit.
quently, neither PI nor activity were used for regression, Moreover, it cannot explain the development of a diffuse
though they are to some extent indicative of the relative double layer, which is the predominant determinant of the
swell potential, as seen from Tables 1 and 2 i.e., percent of behavior of fine-grained soils.
swell and swell pressure increase with the increase of An approach to approximate the surface conductance of
plasticity index or activity (except for the case of Eagle expansive soils is proposed. Surface conductance is the
Ford where the activity is smaller than that of Heiden). electrical conductivity of a diffuse double layer and is
Single linear regression analyses were performed for the representative of the electrical forces between adjacent
predictions of swell percent and swell pressure based on diffuse double layers. Percentage of montmorillonite is
each of the four micro-scale properties discussed. The derived from semi-quantitative approximations from XRD
regression analysis attests to the exceptional performance test results and the clay size fraction is determined from the
of the matric suction variable in predicting percent of swell hydrometer test.
or swell pressure. The matric suction value used for pre- Single regression analyses were performed to predict
diction represents the state of the air–water interface of the swell percent and swell pressure. Matric suction of soil
soil in its initial unsaturated and compacted condition. The samples at OMC was found to be the most significant
matric suction values reveal the formation of a meniscus on variable in the predictions, probably because the magnitude
or in between clay platelets or aggregates from which the represents the change of diffuse double layers from the
diffuse double layer begins to expand until stabilization initially compacted state of the soil. For the four soils
during a saturation process. The one-dimensional swell studied, surface conductance provides good predictions for
tests were performed in such a way that the soil underwent soil swell, except for the sample from Carnisaw which had
this same process of development of diffuse double layers. a low pH. pH and PM, if measured more accurately, are
As a result, matric suction at OMC most closely reflects the expected to play a more significant role in predicting
state of the soil samples for the swell tests and thus gives macro-scale soil behavior when used in combination with
the best prediction of the swelling behavior. From Figs. 5 other micro-scale properties for multi-linear regression
and 6, the prediction using pH shows poor performance as analysis using a larger database of soil properties. Other
the pH values of the tested soils are all close to neutral, possible micro-scale properties, such as dielectric permit-
except for Carnisaw which has a value of 4.4. At the same tivity and directional conductivity, should also be included
time, the results show that both the percent of swell and the to possibly strengthen the predictions. The choice of micro-
swell pressure show a good correlation with the increase scale properties should include, but not be limited to, those
of surface conductance, excluding Carnisaw which has a capable of reflecting information of the initial state of
high content of hydrogen resulting in an exceptionally the soil specimen, e.g., the property of matric suction, as
large conductance (Table 3). The exclusion of Carnisaw this could help to explain the underlying microscopic

123
78 B. Lin, A. B. Cerato

mechanism(s) that best affects macroscopic behavior Holtz WG, Gibbs HJ (1956) Engineering properties of expansive
(swelling) of expansive soils. clays. Trans ASCE 121:641–677
Kassiff G, Livneh M, Wiseman G (1969) Pavements on expansive
clays. Jerusalem Academic Press, Jerusalam
Acknowledgments Financial support for this research was provided Klein K, Santamarina JC (2003) Electrical conductivity in soils:
by the National Science Foundation (Grant No. 0746980). The sup- underlying phenomena. J Environ Eng Geophys 8(4):263–273
port is greatly appreciated. The authors are also thankful to Dr. Lambe TW (1960) The character and identification of expansive soils,
Madden at the University of Oklahoma for his help on the XRD tests. soil PVC meter. Federal housing administration, technical
studies program, FHA 701
Mckeen RG (1992) A model for predicting expansive soil behavior.
References In: Proceedings of the 7th international conference on expansive
soils, Dallas, vol 1, pp 1–6
Al-Rawas A (1999) The factors controlling the expansive nature of Mitchell JK, Soga K (2005) Fundamentals of soil behavior. Wiley,
the soils and rocks of northern Oman. Eng Geol 53:327–350 New York
Al-Rawas AA, McGown A (1999) Microstructure of Omani expan- Nelson JD, Miller DJ (1992) Expansive soils: problems and practice
sive soils. Can Geotech J 36:272–290 in foundation and pavement engineering. Wiley, New York
Armstrong JC, Petry TM (1986) Significance of specimen preparation Ning L (2007) Soil and site characterization using electromagnetic
upon soil plasticity. Geotech Test J 9:147–153 waves. PhD thesis, Virginia Polytechnic Institute and State
ASTM (1999) 1999 Annual book of ASTM standards. Volume 04.08, University, Blacksburg, Virginia
soil and rock (I): D420–D4914, West Conshohocken, Pennsylvania Peterson LW, Moldrup P, Jacobsen OH, Rolston DE (1996) Relations
Brasher BR, Franzmeier DP, Valassis P, Davidson SE (1966) Use of between specific surface area and soil physical and chemical
saran resin to coat natural soil clods for bulk density and water properties. Soil Sci 161(1):9–21
retention measurements. Soil Sci 101(2):108 Raman V (1967) Identification of expansive soils from the plasticity
Cerato AB, Lutenegger AJ (2002) Determination of surface area of index and the shrinkage index data. Indian Eng Calcutta
fine-grained soils by the ethylene (EGME) method. Geotech Test 11(1):17–22
J 25(3):315–321 Rhoades JD (1982) Cation exchange capacity. In: Page AL et al (eds)
Chen FH (1988) Foundations on expansive soils. American Elsevier Methods of soil analysis, agronomy 9, 2nd edn. American
Science Publication, New York Society of Agronomy, Madison, pp 159–165
Delage P, Howat M, Cui YJ (1998) The relationship between suction Santamarina JC, Klein K, Fam M (2001) Soils and waves: particle
and swelling properties in a heavily compacted unsaturated clay. materials behavior, characterization and process monitoring.
Eng Geol 50(1–2):31–48 Wiley, Chichester
Du Y, Li S, Hayashi S (1999) Swelling–shrinkage properties and soil Shi B, Jiang H, Liu Z, Fang HY (2002) Engineering geological
improvement of compacted expansive soil, Ning-Liang highway, characteristics of expansive soils in China. Eng Geol 67:63–71
China. Eng Geol 53:351–358 Skempton AW (1953). The colloidal activity of clays. In: Proceedings
Fredlund DG, Rahardjo H (1993) Soil mechanics for unsaturated of the 3rd International conference of soil mechanics and
soils. Wiley, New York foundation engineering, Switzerland, vol 1, pp 57–61
Fredlund DG, Xing A (1994) Equations for the soil–water charac- USGS Open-File Report (2001) A laboratory manual for X-ray
teristic curve. Can Geotech J 31(3):521–532 powder diffraction, http://pubs.usgs.gov/of/2001/of01-041/
Garbulewski K, Zakowicz S (1995) Suction as an indicator of soil Yoder EJ, Witczak MW (1975) Principles of pavement design, 2nd
expansive potential. In: Alonso EE, Delage P (eds) Proceedings edn. Wiley, New York
of the first international conference on unsaturated soils, Paris,
vol 2, pp 505–512

123

Anda mungkin juga menyukai