You are on page 1of 12

Arab J Sci Eng (2012) 37:1535–1546

DOI 10.1007/s13369-012-0268-4

R E S E A R C H A RT I C L E - C I V I L E N G I N E E R I N G

Ömür Çimen · S. Nilay Keskin · Hüseyin Yıldırım

Prediction of Swelling Potential and Pressure


in Compacted Clay

Received: 22 February 2010 / Accepted: 25 December 2010 / Published online: 18 April 2012
© King Fahd University of Petroleum and Minerals 2012

Abstract In this study to predict free swelling potential and pressure in compacted clay, simple relationships
are proposed. Three different clay samples were prepared with initially at constant dry unit weight and varying
water contents and at constant water content and varying dry unit weights by standard compaction. In the
experiment part, the free swelling technique was used. Values of free swelling potential and pressure were
measured using the oedometer analysis. Analyses were repeated at least two times. To predict free swelling
potential and pressure, the multiple regression analyses were performed for the different values of dry unit
weight, initial water content and plasticity index of three samples. The proposed relationships and the equations
in the literature were compared with the test results of these samples in optimum water content and maximum
dry unit weight and of another compacted sample for the known values of the dry unit weight, initial water
content, and plasticity index. The comparison was also done with a relation given in the literature. The results
found that the estimated values of swelling potential and pressure obtained from the proposed relationships
were in consistent with the measured ones.
Keywords Swelling · Compacted clay · Soil · Prediction · Regression

Ö. Çimen (B) · S. N. Keskin


Faculty of Engineering-Architecture, Department of Civil Engineering,
Suleyman Demirel University, 32260 Isparta, Turkey
E-mail: omurcimen@sdu.edu.tr

S. N. Keskin
E-mail: nilaykeskin@sdu.edu.tr

H. Yıldırım
Faculty of Civil Engineering, Istanbul Technical University, 34000 Istanbul, Turkey
E-mail: yildirimh@itu.edu.tr

123
1536 Arab J Sci Eng (2012) 37:1535–1546

1 Introduction

Swelling soils are problematic soil types which exist in many regions in the world [1–3]. It was stated that
expenditures due to damages caused by swelling soils were fairly high, and the problem was so important
that it could not be neglected [4,5]. The swelling behavior of expansive soils often causes accidents such as
differential settlement, ground heaving, and cracking of pavements and canal linings [6]. The damage ranges
from minor cracking of pavements or interior finishes in buildings, which is very common, to irreparable
displacement of footings and superstructure elements [7]. In the United States, the cost of damages arising
from expansive soil problem amounts to $2.3 billions annually. Consequently, expansive soil damages now
exceed the combined average annual damages from floods, hurricanes, earthquakes, and tornados [4,8]. In
Jordan, more than 40 % of buildings were severely damaged by the swelling of expansive soils [9]. China is
one of the countries with a large distribution of expansive soils, which have successively been found in more
than 20 provinces and regions, occupying nearly 600,000 km2 . Based on incomplete estimation, the buildings
destroyed by the expansive soil amount to a floor space of 10 million square meters and the projected loss
exceeded 1 billion US dollars annually in China [10,11]. In Saudi Arabia, expansive soils were responsible
for millions of dollars worth of damage to man-made structures [12].
Swelling soils are described as soils which show considerable amount of volume increase when subjected
to water. The conventional one-dimensional oedometer swell tests are performed using three different pro-
cedures: free swell, swell under certain overburden pressure and constant volume swell [13]. Now, the most
common method of swelling soil identification is the free swell procedure in a standard oedometer by the
ASTM D 4546-90 standard [14]. Swelling potential is known as an indicator of magnitude of the swelling.
It can be defined as the equilibrium vertical volume change obtained from oedometer-type test, expressed as
a percent of the original height. The swelling pressure is also defined as the pressure required recompressing
the fully swollen sample back to its initial volume in free swell test [15].
Mechanism of swelling in expansive clay is fairly complex and influenced by a number of factors. These
factors can be considered in three main groups: (a) soil characteristics (clay mineralogy, soil water chemistry,
soil structure and fabric, plasticity, dry unit weight, initial water content, etc.), (b) environmental factors (water
content variation, overburden stress, etc.), and (c) state of stress in the soil [15,16]. Researchers have found that
plasticity characteristics and volume change behavior of soils are directly related to the amount of clay-sized
particles in the soil. In general, soils that exhibit plastic behavior over a wide range of water content, have
higher liquid limits, and have greater potential for swelling. The swelling properties of clay minerals follow
the same trend as their plasticity properties; i.e. the more plastic the mineral, the greater is the swell potential
[15,17–19].
Volume expansion occurs when the water content of the clay changes. Slight increase of water content, in
the magnitude of only one to two percent, is sufficient to cause detrimental swelling. The initial water content
of clay also controls the amount of swelling. Dry clay with plasticity index more than 35 % and liquid limit
more than 60 % usually indicates risk while natural water content is less than 15 % [8,20]. Such clay easily
absorbs water as high as 35 % and results in damage by expanding structures. Conversely, clay with water
contents above 30 % indicates that most of the expansion have already been taken place and further degree
of expansion could be lesser [8]. Erguler and Ulusay [21] stated that swell pressure of Ankara (Turkey) clay
increases when the water content is less than 30 %, while it decreases sharply or ceases out at higher values
than 30 or 40 %. Rahimi and Barootkoob [14] showed that soils of even low to medium plasticity could swell
appreciably and be applied high swelling pressure to the adjacent structures, if they were originally compacted
at a higher density and lower water content. Swelling potential and pressure of clay decrease as the initial
water content at the same dry unit weight increases. Until it reduces to the shrinkage limit, the initial water
content has a small effect on both swelling potential and swelling pressure. For values of initial water content
exceeding the shrinkage limit, the rate of decrease of swelling potential and swelling pressure by increasing
water content becomes steeper [22].
The dry unit weight of the clay is another index of expansion. When dry unit weight increases, swelling
pressure rapidly increases. Soils with dry unit weights in excess of 17.62 kN/m3 generally exhibit high swelling
potential [8]. Soil engineers are interested in dry unit weight ranging from 16.0 to 20.83 kN/m3 at embankments
where compacted soils are used and swelling must be taken into account [8]. Swelling potential can be reduced
considerably by decreasing soil density to 92–95 % of maximum dry unit weight and by increasing compaction
water content by 2–3 % with respect to optimum water content [14]. Maximum swelling pressure increases
exponentially with increasing initial dry unit weight, whereas it is almost independent of initial water content.
It was stated that the curve of swelling deformation versus time is strongly dependent on the initial dry unit

123
Arab J Sci Eng (2012) 37:1535–1546 1537

Table 1 Summary of some empirical relationships for estimating swelling potential and swelling pressure

Reference Emprical relationships Remarks


Seed et al. [17] SP = 0.00216PI2.44 for undisturbed ±35 % approximation soils
SP = 0.0036PI2.44 for disturbed soils
Komornik and David [39] LogPs = −2.132 + 0.0208LL For undisturbed soils, γd = kg/m3 ,
+0.000665γd − 0.0269w  0 Ps = kg/cm2
Vijayvergia and Ghazzaly [40] LogSP = (0.44LL − w0 + 5.5) 12 For undisturbed soils
Vijayvergiya and Sullivan [41] LogSP = 0.0526γd + 0.033LL − 6.8 For undisturbed soils, γd = lb/ft3
SP = (0.0229PI1.45 C) w0 + 6.38

Nayak and Christensen [38] For soils compacted to the maximum
standard AASHTO unit weight at
optimum water content by free
swell test, Ps = Psi
1.12 C 2 w 2 + 3.7912

Ps = 0.035817PI
 0
Schneider and Poor [42] LogSP = 0.9PI w0 − 1.19 For undisturbed soils
Johnson [43] SP = 23.82 + 0.7346PI − 0.1458H − 1.7w0 For undisturbed soils, at PI ≥ 40 %
+(0.0025PI)w0 − (0.00884P I )H
SP = −9.18 + 1.5546PI + 0.08424H + 0.1w0 For undisturbed soils, at PI ≤ 40%
−(0.0432PI)w0 − (0.01215PI)H
Weston [44] SP = 0.00411(LLw )4.17 q −3.86 w0−2.33 LLw = (% < 0.425 mm/100)LL
Chen [8] SP = 0.2558e0.0838PI Compacted soils with initial
conditions at γd = 15.7 − 17.3 kN/m3 ,
w0 = 15 − 20 % by free swell test
Basma [29] SP = 0.00064PI1.37 C 1.37 For soils compacted to the
maximum standard AASHTO unit
weight at optimum water content
by free swell test
Erguler and Ulusay [21] Ps = −227.27 + 2.14w0 + 1.54LL + 72.49γd For remoulded samples,
ASTM Method B,
Ps = N/cm2 , γd = gr/cm3
Erzin and Erol [30] LogPs = −4.812 + 0.01405PI + 2.394γd
−0.0163w0
LogPs = −5.020 + 0.01383PI + 2.356γd For constant volume swell test,
γd = gr/cm3 , Ps = kg/cm2
Sabtan [28] SP = 1.0 + 0.06(C + PI − w0 )
Ps = 135.0 + 2.0(C + PI − w0 ) For undisturbed samples,
ASTM Method A, Ps = kPa
SP swelling potential (%), Ps swelling pressure, PI plasticity index (%), LL liquid limit (%), w0 initial water content (%), γd dry
unit weight, C clay content (%), H depth of expansive layer (feet), L L w weighted liquid limit (%), q surcharge load

weight, vertical pressure, and initial water content. The early swelling–deformation rate increases as the initial
dry unit weight decreases for the standard water content when vertical pressure is relatively low; however, it
decreases when vertical pressure is relatively high. The relationship between swelling deformation and time is
independent of the initial water content for a low initial dry unit weight, but dependent on initial water content
for a high dry unit weight [6].
Some empirical studies of predicting the swelling potential and swelling pressure are recorded in the lit-
erature due to prolonged time in conducting the oedometer tests. In these studies, researchers have concluded
many results by collecting oedometer test data, and correlating it with physical and index properties of soils.
Some of these studies were carried out on the undisturbed samples on site or in the laboratory [21,23–28]
while the others were performed on compacted or remolded samples [14,16,21,24,29,31]. It was found that the
undisturbed specimens have showed less swelling than the remolded specimens. Chen [8] combined the studies
which have been performed in order to classify swelling soils, and obtained relationship between plasticity
index and swelling potential for different surcharge pressures. Variation of swelling potential versus plasticity
index was found out to be fairly dissimilar and the differences were due to variations in the experimental
methods and different soil types. In general, the soil characteristics such as clay percent, plasticity index,
activity, dry unit weight, water content, liquid limit and shrinkage limit have been used for the prediction of
swelling in the clayey soils. The empirical relationships that were reported in the literature based on estimation
of swelling potential and swelling pressure are showed in Table 1.
In the present study, new relationships are proposed to estimate the free swelling potential and pressure on
both dry side and wet side of the optimum water content by the multiple regression analysis. The equations

123
1538 Arab J Sci Eng (2012) 37:1535–1546

Fig. 1 Geological maps showing the distribution of the soil and rock units in Zekeriyakoy (Istanbul) (modified and simplified
from [32–34]), Seki Stream-Gacak (Fethiye, Mugla) (modified and simplified from [35]) and Arapsuyu (Antalya) (modified and
simplified from [34])

are obtained with the aid of the engineering properties, such as the plasticity index, dry unit weight and initial
water content of compacted clay. To estimate the free swelling potential and pressure from these factors on
the compacted clayey soils, such relations have not been previously reported in the literature.

2 Experimental Study

Four different disturbed clay samples with high plasticity from different regions in Turkey were used for the
current study. Sample 1 was taken from the Zekeriyakoy—Istanbul district, Samples 2 and 3 were taken from
the Arapsuyu—Antalya district, and Sample 4 was taken from the Seki Stream—Fethiye district. The modified
and simplified geological maps of the studied areas were presented in Fig. 1. In the Zekeriyakoy area, geologi-
cal units are from bottom to top as follows: Basement rock units, lower Paleozoic in age; igneous rocks, lower
Paleozoic in age; Sariyer Formation, upper Cretaceous to Eocene aged clastics and Plio- Quaternary sediments
[32–34] (Fig. 1a). Around the Seki Stream area, the following geological units are observed: Basement rock
units (mainly carbonates platform deposits), clastics of Seki Formation, Plio-Quaternary in age and alluvial
deposits [35] (Fig. 1b). In the Arapsuyu area, four geological units were defined. The basement rock consists of
carbonates, Mesozoic in age. Pleistocene is represented by Antalya travertine. In the Quaternary time, recent
alluvial and beach sediments were developed on eroded surfaces [34] (see Fig. 1c).
Mineralogy of all the soil samples was determined by X-ray diffraction (XRD) method in Mineralogy and
Petrography Laboratory at Mining Exploration Institute (MTA) of Turkey. X-ray diffraction analysis permits
reproducible and accurate calculation of the mineral contents of rocks, including the major clay mineral fami-
lies. X-ray analyses were performed on two parts, as whole rock and clay fractions. The proportions of mixed
layer clay minerals were determined by comparing areas of selected basal peaks on X-ray diffraction traces

123
Arab J Sci Eng (2012) 37:1535–1546 1539

Table 2 Mineralogical composition of samples used in the experimental program

Samples Mineralogical composition


Sample 1 Montmorillonite + chlorite, quartz, calcite, feldspar, illite
Sample 2 Montmorillonite + chlorite, quartz, feldspar, calcite
Sample 3 Montmorillonite + chlorite, calcite, illite, quartz
Sample 4 Lizardite, smectite, chrysotile, little quartz, very little calcite

(a) 18 (c) 18
Compaction values Compaction values
17 100% saturation line 17
Dry Unit Weight, kN/m3

100% saturation line

Dry Unit Weight, kN/m3


Initial conditions in Table 4 Initial conditions in Table 4
16 16
15 15

14 14

13 13

12 12

11 11
5 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45
Water Content (%) Water Content (%)

(b) 18 (d) 13,5


Compaction values
17 13
100% saturation line
Dry Unit Weight, kN/m3
Dry Unit Weight, kN/m3

Initial conditions in Table 4 12,5


16
12
15 11,5
14 11
10,5
13
10 Compaction values
12 100% saturation line
9,5
Initial conditions in Table 5
11 9
5 10 15 20 25 30 35 40 45 10 20 30 40 50 60
Water Content (%) Water Content (%)

Fig. 2 Compaction curves and saturation lines of soil samples. a Sample 1, b Sample 2, c Sample 3, d Sample 4

of untreated samples with those of samples that were heated at 550 ◦ C and saturated by ethylene glycol. The
analysis showed that dominant clay minerals and other minerals in the samples were determined by considering
abundances of order (Table 2). According to the Mitchell’s [19] clasification, Samples 1, 2 and 3 have “high”
swelling potential while Sample 4 has “medium” swelling potential.
Three (Samples 1, 2 and 3) of the four examined samples were evaluated for obtaining multiple regression
relationships in order to estimate swelling potential and pressure, and the fourth sample was used to compare
with these equations. The physical properties of clay used in the tests, such as particle size, hydrometer tests,
Atterberg Limits, maximum dry unit weight, and optimum water content, were determined using the relevant
ASTM standards while soil types were determined according to the Unified Soil Classification System (USCS)
[36]. Standard compaction tests were carried out in accordance with ASTM D698 and the results were given in
Fig. 2. The results of the above-mentioned experiments were also presented in Table 3. Samples were labeled
according to the values of plasticity index and liquid limit on modified plasticity card proposed by Dakshana-
murthy and Raman [18]. Figure 3 shows that Samples 1 and 2 have “very high” swelling potential, Sample 3
has “high” swelling potential, and Sample 4 has “extra high” swelling potential.
To determine swelling potentials and pressures of compacted clay, free swelling method as described in
ASTM D4546 was employed as experimental procedures [13]. The samples were oven-dried for approximately
24 hr, crushed and sieved through ASTM sieve No. 40 (0.425 mm). Then, they were mixed with certain amounts
of distilled water calculated as necessary to obtain the initial water contents (15, 20, 25, 30, 35 and 40 % at both
dry and wet sides of optimum water content) for three samples only. Samples were compacted with standard
proctor hammer (2.5 kg) in the proctor mould. Material needed was calculated according to initial dry unit
weights of 11.5, 13.0, 14.0, 15.0, 16.0 and 17.0 kN/m3 , which were randomly determined between upper and

123
1540 Arab J Sci Eng (2012) 37:1535–1546

T ab l e 3 Properties of clay samples used in the experiment [45]

Properties Sample 1 Sample 2 Sample 3 Sample 4


Liquid limit (%) 75 73 66 108
Plastic limit (%) 21 26 28 43
Plasticity index (%) 54 47 38 65
Specific gravity, G 2.79 2.82 2.86 2.70
Max. dry unit weight (kN/m3 ) 16.1 16.0 15.2 12.0
Optimum water content (%) 23 23 27 37
Gravel (%) 1 1 0 0
Sand (%) 6 3 2 2
Silt (%) 38 46 46 38
Clay (%) 55 50 52 60
Activity 0.98 0.94 0.73 1.08
Soil type (according to USCS) CH CH CH CH
Nonplastic

Swelling

Low Medium High Very High Extra High


120

100 U Line PI = 0.9 (LL - 8.1)


Plasticity Index, PI, (%)

80 A Line PI = 0.73 (LL - 20)

4
60
1
2
40 3

20

0
0 20 40 60 80 100 120 140 160
Liquid Limit, LL, (%)
F i g. 3 Plasticity chart

lower limits of maximum dry unit weight. Samples prepared in different initial water contents and different
dry unit weights were compacted in three layer. Compaction was carried out until all materials were placed
into the compaction mould. Samples were collected from the top and bottom parts of mould with 75 mm
diameter and 20 mm height oedometer ring. Collected samples were weighed, and these samples showed that
they satisfied to initial dry unit weight and water content conditions. Then the samples were transferred to the
oedometer cell and the free swell tests were performed. During the compaction, a homogeneous compaction
was provided and water content was fixed.
Soil samples prepared under these conditions and placed into oedometer cell were allowed to get water
through capillary action enable them to swell freely. A seating pressure of at least 1 psi (7 kPa) was applied by
the weight of top porous stone, and load plate until swelling was completed freely. It should be noted that a
certain surcharge load was not used in this study. It is known that increasing the applied surcharge load would
reduce the magnitude of swelling. The amount of swelling was measured from the vertical deformation dial
gauge. The durations of completing swelling varied between 2 and 7 days. At the end of this period, final
swelling value did not changed and this value was used to determine the swelling potential. Figure 4, as an
example, shows the results of free swelling versus time for γd = 11.5 kN/m3 and w0 = 30 %. After completion
of swelling, the samples were loaded in small increments until their initial heights were obtained. The free
swelling pressure was taken as the pressure that brings the sample back to its initial height. The durations
which bring the samples back to their initial heights also varied during 2–7 days. The results of swelling tests
obtained for three samples were given in Table 4. It is seen from this table that for any constant dry unit weight,
swelling pressures and potentials decrease when the initial water content increases; therefore, constant initial
water swelling pressures and potentials increase when the dry unit weight increases.

123
Arab J Sci Eng (2012) 37:1535–1546 1541

10

Free Swelling (%)


7
Sample 1
6 Sample 2
Sample 3
5
Sample 4
4

0
0 8 16 24 32 40 48 56 64 72
Time (hr)
Fig. 4 Change of free swelling with time for γd = 11.5 kN/m3 and w0 = 30 %

Table 4 Initial conditions of samples and measured values of swelling potential and pressure [45]

γd (kN/m3 ) w0 (%) Sample no.


1 2 3
Ps (kPa) SP (%) Ps (kPa) SP (%) Ps (kPa) SP (%)
11.5 15 208 18.0 163 14.0 86 12.0
20 161 13.5 128 12.1 75 11.0
25 157 10.7 103 11.0 52 9.0
30 112 9.1 63 8.6 41 7.5
35 69 3.6 48 3.1 22 2.2
40 42 1.5 20 0.9 10 0.5
13.0 15 321 18.5 202 14.2 110 13.0
20 260 15.0 187 12.5 96 11.6
25 202 12.1 128 11.4 88 9.5
30 130 10.1 92 9.0 71 8.0
35 100 4.8 68 3.6 24 2.6
40 48 2.8 32 1.1 15 0.8
14.0 15 509 20.0 340 15.0 203 13.4
20 300 15.8 271 13.0 126 12.0
25 281 13.0 152 11.8 98 10.0
30 173 11.8 102 9.7 85 8.4
35 103 5.0 58 4.1 48 3.0
15.0 15 640 20.6 376 16.0 217 14.0
20 410 16.8 250 14.2 183 12.8
25 290 13.7 186 12.3 122 11.0
30 145 12.2 122 10.4 94 9.0
16.0 15 680 21.8 398 17.6 250 15.0
20 478 18.2 282 15.8 204 13.9
25 292 15.1 210 13.2 132 11.6
17.0 15 848 22.1 482 18.0 260 15.6
20 526 20.2 326 16.4 218 14.0

3 Results and Discussion

The literature on swelling soil behavior indicates that there are three primary factors controlling total swelling,
namely, the soil characteristics, environmental factors, and state of stress in the soil. The properties of the soil
influence the volume change indices. For example, soils compacted at various densities and water contents
produce different soil structures which have different swelling. In addition, various densities and water contents
affect the magnitude of the stress state of compacted soils [37]. Plasticity characteristics and volume change
behavior of soils are also related to the amount of clay-sized particles in the soil. The swelling properties
of clay minerals follow the same trend as their plasticity properties [15,17–19]. It can be concluded that the
density and water content conditions in a soil affect both the volume change indices and the stress state. The
dry unit weight, water content and plasticity index of a soil can be determined easily in a laboratory. Therefore,

123
1542 Arab J Sci Eng (2012) 37:1535–1546

( a) 25

Predicted Swelling Potential, SP*, (%)


20

15

10 Sample 1

Sample 2

Sample 3

5
5 10 15 20 25
Measured Swelling Potential, SP, (%)

(b) 900
Predicted Swelling Pressure, PS* , (kPa)

800

700

600

500

400

300
Sample 1
200
Sample 2
100 Sample 3

0
0 100 200 300 400 500 600 700 800 900
Measured Swelling Pressure, PS, (kPa)
Fig. 5 Comparison of measured and predicted swelling potentials and pressures (Equations 1a and 1b)

these parameters were considered to estimate the swelling potential and pressure on a compacted clay soil in
the current study.
The dry unit weight, water content and plasticity index affected the swelling potential and swelling pressure
of a soil in a nonlinear manner. Therefore, the multiple nonlinear regression analyses were applied to correlate
the measured swelling potential and pressure from the soil properties (dry unit weight, initial water content,
and plasticity index). This was done for the measured values of Samples 1, 2, and 3 given in Table 4 by the
Statistical Package for the Social Sciences (SPSS v.18.0) program. The following relationships were obtained
separately for the values on the dry and wet sides of the optimum water content. Samples on the dry side
have flocculated structure and those on the wet side have dispersed structure. The expressions of the swelling
potential (the coefficient of determination, R 2 = 0.946) and pressure (R 2 = 0.977) on the dry side were
determined as;

SP∗ = (0.3139γd0.3552 − 0.1177w00.4470 )PI0.9626 (1a)


LogPs∗ = 0.0276PI − 365.2118γd−2.4616 − 0.0320w0 + 2.2292 (1b)

In addition, the proposed relations for swelling potential (R 2 = 0.952) and pressure (R 2 = 0.974) on the wet
side were:

123
Arab J Sci Eng (2012) 37:1535–1546 1543

(a) 20

Predicted Swelling Potential, SP*, (%)


15

10

5 Sample 1

Sample 2

Sample 3

0
0 5 10 15 20
Measured Swelling Potential, SP, (%)

(b) 350
Predicted Swelling Pressure, PS*, (kPa)

300

250

200

150

100
Sample 1
Sample 2
50
Sample 3

0
0 50 100 150 200 250 300 350
Measured Swelling Pressure, PS, (kPa)
Fig. 6 Comparison of measured and predicted swelling potentials and pressures (Equations 2a and 2b)

SP∗ = (0.4768γd0.3888 − 0.0033w01.6045 )PI0.7224 (2a)


LogPs∗ = 0.0239PI − 1285.3723γd−3.2768 − 0.0396w0 + 2.3238 (2b)

In the proposed relationships, SP∗ is the estimated swelling potential (%), Ps∗ is the estimated swelling pressure
(kPa), γd is the measured dry unit weight (kN/m3 ), PI is the measured plasticity index (%), and w0 is the mea-
sured initial water content (%).
In the literature, there are several error methods for comparing the estimated values according to the real
(measured) values. The root mean square error (RMSE) and mean absolute error (MAE) are often used as:

 N 

RMSE =  (X i,me − X i,es )2 N (3)
i=1

N
1   
MAE = X i,me − X i,es  (4)
N
i=1

123
1544 Arab J Sci Eng (2012) 37:1535–1546

Table 5 Comparison of swelling potential and pressure

Test results Nayak and Proposed equations


Christensen [38]
SP Ps SP∗ (%) Ps∗ (%) SP∗ (kPa) Ps∗ (kPa)
Sample 1: PI = 54 %, C =55 %, wopt = 23 %, 15.3 310 24.2 149 16.5∗ 378∗
γd max = 16.1 kN/m3
Sample 2: PI = 47 %, C = 52 %, wopt = 23 %, 13.4 225 20.1 120 14.6∗ 246∗
γd max = 16.0 kN/m3
Sample 3: PI = 38 %, C = 50 %, 10.0 120 14.7 76 10.2∗ 95∗
wopt = 27 %, γd max = 15.2 kN/m3
Sample 4: PI = 65 %, C = 60 %, 5.6 109 22.2 96 6.5∗ 109∗
wopt = 37 %, γd max = 12.0 kN/m3
Sample 4: PI = 65 %, C = 60 %
γd = 10.5 kN/m3 , w0 = 25 % 5.3 192 29.8 179 12.7 127
γd = 10.5 kN/m3 , w0 = 30 % 3.4 70 25.9 132 10.3 88
γd = 10.5 kN/m3 , w0 = 35 % 2.9 65 23.1 104 8.2 61
γd = 11.5 kN/m3 , w0 = 25 % 4.1 209 29.8 179 14.0 213
γd = 11.5 kN/m3 , w0 = 30 % 3.8 140 25.9 132 11.6 148
γd = 11.5 kN/m3 , w0 = 35 % 3.2 74 23.1 104 9.5 102
γd = 12.5 kN/m3 , w0 = 25 % 5.7 239 29.8 179 14.6 262
γd = 12.0 kN/m3 , w0 = 30 % 5.5 169 25.9 132 12.3 181
γd = 12.0 kN/m3 , w0 = 35 % 4.5 165 23.1 104 10.1 125
γd = 13.0 kN/m3 , w0 = 25 % 6.3 179 29.8 179 15.8 365
Mean error −18.4 29 −5.6 −17
* These are average values of Equations (1a) and (2a) or Equations (1b) and (2b)

where N is the number of data, X i,me and X i,es are the measured and estimated swelling potential or swelling
pressure values, respectively. These expressions indicate the mean magnitude of the differences between all
the measured and estimated data. The smaller the error, the better would be the accuracy of the model. The
estimated values of swelling potentials and pressures obtained from these relationships were compared with
experimental results (see Figs. 5, 6). According to the graphs, the estimated swelling potentials and swelling
pressures were correlated significantly. The values of RMSE and MAE for Equation (1a) were calculated as
0.76 and 0.60 % while those of Equation (1b) were found as 26.77 and 21.91 kPa, respectively. Similarly, the
values of RMSE and MAE for Equation (2a) were calculated as 0.96 and 0.84 % while those of Equation (2b)
were found as 12.19 and 9.62 kPa, respectively.
To validate the proposed relationships, swelling tests were also carried out for the Sample 4 (Table 3).
The values of initial water content and dry unit weight of the Sample 4 were determined from the compaction
curve in Fig. 2d, as w0 = 25–35 % and γd = 10.5–13.0 kN/m3 . All the necessary experiments were performed
with the same procedure mentioned in Sect. 2. The test results were compared with the values obtained from
the equations proposed for the swelling potential and pressure (Table 5). Some of the relationships in Table 1,
used by the free swelling method on the compacted soils, were also compared with the proposed equations.
Among the relationships, Chen [8] used only the plasticity index of compacted soil in order to predict the
swelling potential while Basma [29] considered the plasticity index and the clay content. Nayak and Chris-
tensen [38] took into consideration the plasticity index, the clay content, and the water content of compacted
soil in order to predict both the swelling potential and the swelling pressure by the free swelling procedure.
For this reason, comparisons were also made with the relationships obtained by Nayak and Christensen [38].
It was seen in Table 5 that the mean error (−5.6 %) of Equations (1a) and (2a) was superior to the ones
(−18.4 %) of the relationship by Nayak and Christensen [38] and that the mean error (−17 kPa) of Equations
(1b) and (2b) was lesser than the ones (29 kPa) of the relationship by Nayak and Christensen [38]. These
results show that the proposed relationships would be used reliably for the compacted samples in the ranges
38 ≤ PI ≤ 65%, 11.5 ≤ γd ≤ 17 kN/m3 , and 15 ≤ w0 ≤ 42 %.
Sample 4 has “extra high” swelling potential according to the plasticity chart given in Fig. 3 while it has
“medium” swelling potential according to the Mitchell’s [19] mineral clasification. As it is seen in the test
results and in Fig. 4, Sample 4 had less swelling potential than the ones of the other specimens at the same
initial water content, and dry unit weight. Mineralogical composition of Sample 4 given in Table 2 shows the
lizardite mineral as a dominant characteristic of clay. This is one of the serpentine group minerals and it is
nonswell in water.

123
Arab J Sci Eng (2012) 37:1535–1546 1545

4 C onclusions

New relationships to predict the swelling potential and pressure in compacted soils were proposed by the free
swelling oedometer technique in the present study. The proposed expressions were obtained by the multiple
regression technique from the swell results of three samples repeated at least two times, depending on the
initial water content, dry unit weight and plasticity index. The test results of the Sample 4 were compared with
the proposed equations. It was concluded that the estimated values for swelling potential and pressure were
close to the experimental values. At any constant dry density, swelling pressure and potential decrease as the
initial water content increases. On the other hand, at any constant initial water content, swelling pressure and
potential increase as the dry unit weight increases. Furthermore, swelling pressure and potential increase as the
plasticity index increases. The mineralogical composition of clay affects its swelling potential. As a result, the
proposed relationships could be used reliably to predict the swelling potential and pressure of the compacted
clay. The relationships are valid for 38 ≤ PI ≤ 65 %, 11.5 ≤ γd ≤ 17 kN/m3 , and 15 ≤ w0 ≤ 42 % and it
should be noted that three samples considered to the proposed relations have similar mineralogical structure.

Acknowledgments The authors would like to thank Prof. Muhittin Görmüş for his valuable help to modify and simplify geolog-
ical maps of the studied areas and two reviewers for their constructive comments and suggestions, which resulted in a significant
improvement of the paper.

References

1. Aitchison, G.: Status of the art of dealing with world problems on expansive clay soils. In: Engineering Effects of Moisture
Changes in Soils. International Research and Engineering Conference on Expansive Soils, pp. 1–6 (1965)
2. Dhowian, A.; Ruwaih, I.; Erol, A.: The distribution and evaluation of the expansive soils in Saudi Arabia. In: Proceedings
of Second Saudi Engineering Conference King Fahd University of Petroleum and Minerals, vol. 4, pp. 1969–1990 (1985)
3. Abduljauwad, S.N.; Hameed, R.A.; Al-Sulaimani, G.J.; Basunbul, I.A.; Safar, M.M.: Expansive soils in eastern province of
Saudi Arabia. In: Proceedings of Seventy International Conference on Expansive Soils, Dallas, vol. 1, pp. 426–431 (1992)
4. Jones, D.E.; Holtz, W.G.: Expansive soils–the hidden disaster. Civil Eng. 43(8), 49–51 (1973)
5. Al-Shamrani, M.A.; Dhowian, A.W.: Experimental study of lateral restraint effects on the potential heave of expansive soils.
Eng. Geol. 69, 63–81 (2003)
6. Komine, H.; Ogata, N.: Experimental study on swelling characteristics of compacted bentonite. Can. Geotech. J. 31, 478–490
(1994)
7. Al-Mhaidib, A.I.: Swelling Behaviour of Expansive Shales from the Middle Region of Saudi Arabia. Geotech. Geol. Eng.
16, 291–307 (1999)
8. Chen, F.H.: Foundations on expansive soil. Elsevier, New York (1988)
9. Tuncer, E.R.; Basma, A.A.; Taqeddin, S.: Geotechnical engineering properties of some selected irbid clays. Jordan University
of Science and Technology, Research Report No. 14/87 (1987)
10. Shi, B.; Jiang, H.; Liu, Z.; Fang, H.Y.: Engineering geological characteristics of expansive soils in China. Eng. Geol. 67,
63–71 (2002)
11. Li, S.L.; Qin, S.J.; Bo, Z.Z.; Shi, B.: Studies on the Engineering Geology of Expansive Soils in China. Jiangsu Science and
Technology Publishing House, Nanjing (1992)
12. Dhowian, A.W.; Erol, A.O.; Youssef, A.: Evaluation of Expansive Soils and Foundation Methodology in the Kingdom of
Saudi Arabia. Saudi Arabia, Final Report: SANCST, AT-5–88 (1985)
13. ASTM.: Standard Test Method for One-Dimensional Swell or Settlement Potential of Cohesive Soils. Annual Book of
ASTM Standards, D 4546-90, 04.08: 853–859 (1993)
14. Rahimi, H.; Baroothoob, S.H.: Concrete Canal Lining Cracking in Low to Medium Plastic Soils. Irrigat. Drainage 51,
141–153 (2002)
15. Nelson, J.D.; Miller, D.J.: Expansive Soils Problem and Practice in Foundation and Pavement Engineering. Wiley, London
(1992)
16. El-Sayed, S.T.; Rabbaa, S.A.: Factors affecting behavior of expansive soils in the laboratory and field. Geotech. Eng. 17,
89–107 (1986)
17. Seed, H.B.; Woodward, R.J.; Lundgren, R.: Prediction of swelling potential for compacted clays. J. Soil Mech. Found. Div.,
ASCE 88(6), 107–131 (1962)
18. Dakshanamurthy, V.; Raman, V.: A simple method of identifying an expansive soil. Soils Found. Jap. Soc. Soil Mech. Found.
Eng. 13(1), 97–104 (1973)
19. Mitchell, J.K.: Fundamental of Soil Behavior. Wiley, New York (1993)
20. Snethen, D.R.: Evaluation of expedient methods for identification and classification of potentially expansive soils. In: Pro-
ceedings of Fifth International Conference on Expansive Soils, Institution of Engineers, Adelaide, pp. 22–26 (1984)
21. Erguler, Z.A.; Ulusay, R.A.: A simple test and predictive models for assessing swell potential of Ankara (Turkey) clay. Eng.
Geol. 67, 331–352 (2003)
22. El-Sohby, M.A.; Rabba, E.A.: Some factors affecting swelling of clayey soils. Geotech. Eng. 12, 19–39 (1981)
23. Holtz, W.G.; Gibbs, H.J.: Engineering properties of expansive clays. Transact. Am. Soc. Civil Eng. 121, 641–677 (1956)
24. Noble, C.A.: Swelling measurements and prediction of heave for a lacustrine clay. Can. Geotech. J. 3(1), 32–41 (1966)
25. Erol, A.O.; Dhowian, A.: Swell behavior of arid climate shale from Saudi Arabia. Quat. J. Eng. Geol. 23, 243–254 (1990)

123
1546 Arab J Sci Eng (2012) 37:1535–1546

26. Erol, A.O.: In-situ and laboratory measured suction parameters for predictions of swelling. In: Proceeding of the Seventh
International Conference on Expansive Soils Dallas, pp. 2–30 (1992)
27. Abduljawad, S.N.; Al-Sulaimani, G.J.: Determination of swell potential of Al-Qatif clay. Geotech. Test. J. 16(6), 469–484
(1993)
28. Sabtan, A.A.: Geotechnical properties of expansive clay shale in Tabuk, Saudia Arabia. J. Asian Earth Sci. 25, 747–757
(2005)
29. Basma, A.A.: Prediction of expansion degree for natural compacted clays. Geotech. Test. J. 16(6), 542–549
30. Erzin, Y.; Erol, O.: Correlations for quick prediction of swell pressures. Electron. J. Geotech. Eng. 9, paper 0476 (2004)
31. Cokca, E.; Yazici, V.; Ozaydin, V.: Stabilization of expansive clays using granulated blast furnace slag (GBFS) and GBFS-
cement. Geotech. Geol. Eng. 27(6), 489–499 (2009)
32. Siyahi, B.; Erdik, M.; Sesetyan, K.; Demircioglu, M.B.; Akman, H.: Liquefaction and slope failure susceptibility and
opportunity maps: Istanbul case. In: Fifth National Conference on Earthquake Engineering Istanbul, AT-081 (2003)
33. http://www.mimoza.marmara.edu.tr
34. http://www.mta.gov.tr
35. Bozcu, M.; Uyanik, O.; Cakmak, O.; Turker, E.: Geotechnical properties of Esen I HEPP Project field. J. Nat. Appl. Sci.
11(1), 75–83 (in Turkish) (2007)
36. ASTM.: Annual Book of ASTM Standards. Soil and Rock: 04.08 (1994)
37. Fredlund, D.G.; Rahardjo, H.: Soil mechanics for unsaturated soils. Wiley, New York (1983)
38. Nayak, N.V.; Christensen, R.W.: Swelling characteristics of compacted expansive soils. Clay Clay Miner. 19(6), 251–261
(1979)
39. Komornik, A.; David, D.: Prediction of swelling pressure of clays. J. Soil Mech. Found. Eng. Div. ASCE 95(1), 209–225
(1969)
40. Vijayvergia, V.N.; Sullivan, R.A.: Simple technique for identifying heave potential. In: Proceedings of Workshop on Expan-
sive Clays and Shales, vol. 2, pp. 149–154 (1973)
41. Vijayvergia, V.N.; Ghazzly, O.I.: Prediction of swelling potential for natural clays. In: Third International Conference on
Expansive Soils, vol. 1, pp. 227–236 (1973)
42. Schneider, G.L.; Poor, A.R.: The Prediction of Soil Heave and Swell Pressures Developed by an Expansive Clay. Contraction
Research Center, University of Texas, Research Report TR-9-74 (1974)
43. Johnson, L.D.: Predicting Potential Heave and Heave with Time in Swelling Foundation Soils. USAWES, Technical Report
S-78-7 (1978)
44. Weston, D.J.: Expansive Roadbed Treatment for Southern Africa. In: Proceedings of Fourth International Conference on
Expansive Soils, pp. 339–360 (1980)
45. Çimen, Ö.: Determination of Swelling and Suction Properties of Clay Soils with Fuzzy Logic. PhD Thesis, Suleyman
Demirel University, Isparta, 2002, (in Turkish)

123

You might also like