An Overview of the 35 Years of Research on the Volume
Change Behavior of Ankara Soils
Erdal Çokca Associate Prof. of Civil Engineering, Dept. of Civil
Engineering, Middle East Technical University, 06531 Ankara
TURKEY
Email: ecokca@metu.edu.tr
Introduction :
(The following has been extracted from Dr. Erdal Çokça's Ph.D.
thesis, amended and edited by the thesis supervisor Dr. Altay
Birand. Questions with respect to any item may be addressed to Dr.
Erdal Çokça at his e-mail given above)
The purpose of this work is to make the reader aware of what has been
done in the Geotechnical Section of the Civil Engineering Department of
the Middle East Technical University with regard to the partly saturated
soil behavior. The main emphasis has been on the volume change
characteristics . The following is a short account of the research that has
been going on since 1963.
General Considerations:
The volume change potential of a soil is generally a result of geological
settings which determines the mineralogical composition and history
involving preconsolidating agents like erosion-sedimentation and/or
desiccation. The environmental conditions are usually climatic,
characterized by the rate of evaporation exceeding the rate of rainfall.
The regions with the most severe problems are usually those with local
climates that produce desiccation. In this connection the geology of
Ankara region is described briefly.
In a broad sense, soil may be thought of as an incidental material in the
vast geological cycle that has been going on continuously and
relentlessly throughout hundreds of millions of years of geological time.
This cycle may be considered as consisting of a number of phases.
The first step in the cycle is represented by igneous rocks- that is, rocks
that have solidified from molten magma. Igneous rocks include the
oldest rocks found on earth and represent the original or primordial
sources for soils.
Igneous activity which involves uplift and exposure to the atmosphere
initiates the other step in the cycle, slow chemical degradation or
weathering. The gradual breakdown of hard rock into soil result in
"residual soils ".
1
In addition to exposing rocks to weathering, geologic uplift also initiates
the forces of erosion. Erosion and transportation are followed by
deposition in a different locale (Surgel, 1976; After Spangler and Handy,
1973).
It should be noticed that the most important transporting agent which
may be active in soil formation is running water, contributing to the
formation of "alluvial soils " (Surgel,1976; After Krynine ,1941).
Aggregates of particles that come to rest in some place after having been
transported laterally or vertically for some distance are called
"sediments".
When first deposited, the particles are unconsolidated or essentially
unconsolidated. Such deposits are called "recent sediments". With time
the sediments consolidate and harden into rock. Such consolidated
sediments are called "ancient sediments" (Surgel, 1976; After Trask,
(1959).
After deposition, most sediments immediately reenter the weathering erosion - deposition cycle (Surgel, 1976; After Spangler and Handy,
1973) As a result of this, terraces are formed by running water in
floodplains and deposition by floods on terraces are called "terrace
deposits " (Surgel 1976; After Lohnes, 1974).
Fig.1. Regional Geology:
2
Several researchers have investigated geology of Ankara region and
geotechnical properties of Ankara soils.
The Geology is quite complex. Formations encountered range from
Paleozoic to Quaternary in age and include sedimentary, metamorphic
and igneous rocks (Ordemir et al 1977).
The regional soils may be classified in three main groups; i) residual
soils, ii) recent alluvium deposits, iii) deposits of Pliocene - Pleistocene
age which are mostly terrace formations in the flood plains (also called
as Ankara Clay). The surface area of Pliocene and Quaternary age
deposits are very large compared to other formations. By disintegration
and weathering process a soil layer having a thickness about 0.5-4.0
meters has been formed on the outcrops, which is called as "residual
soils". One would expect that since the residual soils have weathered in
place or been moved small distance downslope, there is little reason to
suspect that their overconsolidated behavior is a result of overburden
pressure.
The second group of Ankara soils is called as recent "Alluvial Deposits"
formed in Quartenary age by floodwaters. These formations are
encountered along the local streams. These are normally consolidated
soft deposits and ground water level is close to the surface. Most of the
alluvial soils are seen along the Ankara River 200-2000 meters in width
and 10-45 meters in depth (Furtun 1989; After Ordemir et. al, 1977;
Birand, 1978; Kasapoglu, 1980). The alluvium, which has been
deposited by the floodwaters, has not been in place long enough to show
any appreciable effect of soil forming factors. The parent material within
the alluvium will vary, depending upon the nature of rocks and soils in
the areas drained by the streams (Surgel, 1976; After Lohnes, 1974).
Residual Soils called Ankara Clay in short is the main focus of attention
as far as Volume change problems are concerned.
Here, two different views for the formation and the cause of
overconsolidation of Ankara clay will be discussed. First the studies of
Kasapoglu (1980) who studied the region falling east of Macun river and
Kiper (1983) who studied the area west of Macun river will be given,
than the studies of Surgel (1976) (After Lohnes ,1974) and Birand
(1977) will be discussed:
At the end of the Miocene age, as a result of Attic and Radonic phase
movements several lakes have been developed. In the Middle Pliocene
age, lakes have been filled by sand, silt, gravel and clay particles
transported from the surrounding old formations,mostly andesites and
graywackes. At the same time as a result of Epirogenic movements
many lakes disappeared and at the beginning of the upper pliocene age
the sediments were subjected to desiccation and preconsolidation.
During upper Pliocene new deposits were formed and the previous
deposits were subjected to a second loading. At the beginning of the
Quaternary age, as a result of uplift and subsequent erosion, sediments
were preloaded once again. By means of cyclic wetting and drying,
calcareous concretions occurred near the surface. This formation is
3
generally called as Pliocene - Pleistocene age Fluvial Lacustrine
Deposits or Terrace Deposits. Montmorillonite, Illite, Kaolinite are the
basic minerals in these formations. According to Kasapoglu (1980),
Pliocene Deposits are not the same everywhere since their parent
materials are different such as graywackes in the south and andesite in
the north. Percentages of granular particles are higher at the sides of the
Ankara basin is placed east of Enguru Plain where deltas of rivers are
believed to exist in the past. Kiper (1983) divided Pliocene clay into
groups: yellow or gray Macun member around Elvan village and reddish
brown Balgat member who are the dominant one. The data suggests that
reddish brown clay layer thickness reduce westward and beyond Macun
River where gray and yellow clay is encountered. Kiper (1983) states
that the upper Pliocene deposits of Etimesgut - Batikent area can be
considered as overconsolidated soils, which have been subjected to very
high level of preloading. It is presumed that these deposits have first
been exposed to atmosphere right after their deposition at the beginning
of Upper Pliocene; and subjected to preloading by drying at the surface.
Later they have been compacted under the weight of fluvial deposits
piled on top of them. Subsequently, they have been unloaded by erosion,
which has occurred during the Quartenary period; and underwent a
second preloading process. Recent surface drying and wetting activities
form the last preloading process that the sediments have undergone since
the beginning of the Quaternary. The researcher states that during the
last preloading stage, the traces of first two preloading stages may have
become obscure due to the fissures in the drying depths.
Kiper (1983) states that the liquidity index IL (IL= W-WP/IP) value for
overconsolidated clays is around zero, for heavily overconsolidated
clays is negative and for normally consolidated clays is around one.
According to the depth - liquidity index diagram given by Surgel (1976)
down to 30 m depth the Liquidity Index value is around zero. If the soil
at these depths were normally consolidated the Index should be around
one. (On the other hand Birand (1977) states that the soil samples after
15 m are normally consolidated as related below).
Kiper (1983) used the overburden pressure versus void ratio chart given
by Rieke and Chilingarian (1974) and determined the average void ratio
of his soil samples as 0.76 and a value of 15 kg/cm2 overburden
pressure value. The average preloading pressure of his samples as a
result of consolidation tests was found as 3.6 kg/cm2 and the difference
between these two pressure values lead him to take this as an indication
of an overburden load which had been there in the geological history of
the soil and than eroded.
Kiper (1983) (After Parcher and Means, 1963) states that the Permian
clays which were under the overburden load of 30 kg/cm2 in their
geological history shows a 4 kg/cm2 preloading pressure due to the
shrink - swell cycle that they exposed later on in their geological history.
If the Permian clays had only 4 kg/cm2 preloading pressure at higher
pressures, the e- logP curve had to be straight, but at 30 kg/cm2 pressure
4
level there is a slight outward curvature. The same behavior was also
observed by Kiper (1983) at 10 - 15 kg/cm2 pressure levels. As a result
Kiper suggests that the preloading pressure value obtained from
consolidation test result (i.e. e - logP curve) gives the preloading
pressure value due to capillary forces depending on surface drying.
Kiper (1983) states that the preloading pressure value obtained from the
consolidation test result is not accurate if the geological history of the
clay is complex .
According to Surgel’s (1976) (After Lohnes, 1974) study, the Pliocene Pleistocene terraces are the result of a cut and fill sequence and as such
their parent surfaces are depositional rather than erosional and that the
soils which lie beneath these surfaces have not experienced any
overburden pressure greater than that which exist at present.
The clay found in METU campus area has mainly the same properties as
"Ankara Clay " defined earlier by Ordemir et. al (1965). This clay is
described as friable with haircracks and slickenslides as well as traces of
lime formations (Surgel, 1976; After Lohnes ,1974).
Birand (1977) states that in Ankara soils the natural water content values
are near to plastic limit values, this indicates the existence of preloading,
According to Birand (1977) who studied the cause of this overburden
pressure, if a soil was preloaded by a geological load this effect should
be seen along depth. In other words if the unit weight of a soil is  and if
this soil was preloaded by a geological load  , than the variation of
preloading load with depth z should be =z. Consolidation tests on
several samples at depth , which have been loaded horizontally and
vertically with respect to their in situ condition has allowed some
findings to be made:
In Fig. 2 below, the variation of vertical preloading pressure with depth
is shown. This figure suggests that after 15 m depth , the soil is normally
consolidated . If a soil has been preloaded by capillary forces, due to the
isotropic character of capillary forces, the preloading effect with depth is
expected to be isotropic.
In Fig.3 Kop =(Lateral Preloading Load) / (Vertical Preloading Load)
value is depicted and seen to be around 1.0; suggesting that near the
surface the preloading is caused by capillary forces due to desiccation.
This indicates that the soil has been subjected to preloading loads which
are isotropic in character. On the other hand it is also possible to agree
with Kiper’ suggestion that earlier cracking of the soil due to desiccation
after the fill and cut process may have obscured the earlier very high
preconsolidation loads due to overburden.
5
Geotechnical Properties of Ankara soils and Research at METU:
The geotechnical properties of Ankara soils have been investigated by
several researchers. The following is a chronological outline of the
studies mainly concentrating on the volume change behavior .
6
Birand (1963) studied the swelling properties of Ankara clays, primarily
concentrating on the factors responsible for the expansion of soils and
the available methods of classification and identification.
The effect of climate, the precompression of the clay, the chemical
compounds within the clay were components of this study. The
conclusions that may be drawn from this study are:
The relationship of the form S=K*Cx as proposed by Seed et al (1962)
for artificially prepared soils is valid for natural Ankara Clay within a
reasonable degree of approximation .Herein C is the clay content K and
x are constants and S is the swelling potential. However, there is a
distinct variation in the behavior of natural and artificial clay samples
as far as their swelling tendencies are concerned. The use of artificially
prepared samples can not reproduce all the variables that may affect
swelling in the field.
Skempton’s original definition of activity as A=PI/C for natural London
clays is found to be equally applicable to natural Ankara clays.
The effect of precompression on the clays investigated on swelling is to
reduce the swelling potential i.e. The higher the precompression
pressure the less is the swelling potential.
It has been verified once more that the mineralogical composition of the
clay effect the swelling of the clays. Clays with high ion exchange
capacity have tendency to swell more upon soaking than those with
lower ion exchange capacities.
The swelling potential of a clayey soils found to increase with the
Plasticity Index. The "Swelling Potential Classification Chart " given by
Seed et. al,(1962) for artificially prepared samples, as far as the natural
clays used in this investigation are concerned, is not applicable
satisfactorily.
The Potential Volume Change method proposed by Lambe et. al(1960)
should be used in estimating the potential heave of Ankara clays. Using
this method, one would be on the safe side and also take the climatic
factors into account.
To investigate climatic conditions prevalent in the Middle Anatolian
region in general and Ankara in Particular, a study was made by
Özmelek(1984) in which the whole region was classified according to
the Thorntwaite Climatic Index. Based on this it is found that Ankara
has a semi-erid climate.
Ordemir et. al, (1965) reported on the general properties of Ankara clay :
The semi-arid climate of Ankara is prone to alternate wetting and
drying which results in swelling and shrinking cycles of the soil which
in turn causes damage and cracking of light buildings. The structures
7
resting on deep foundations are not harmed because the seasonal
changes are not felt below a certain depth.
Furthermore Ankara clay as a highly plastic and expansive which results
in a very poor performance of this clay as a subgrade material.
Because of its influence on the swelling potential ,the nature and degree
of preconsolidation of Ankara Clay has been a subject of research as
somewhat exposed in earlier sections. Arda (1966) , independent of the
Geologists’ studies until then ,investigated this phenomenon using soils
laboratory testing methods. According to the results of his study, Ankara
clay has been preconsolidated in the past. The reason for this
preconsolidation is the existence of overburden which has covered the
basins in Ankara region during upper Pliocene and in early Quaternary.
However , an effect of desiccation is detected whose influence is said to
be relatively small in comparison to the geologic loading except at
shallow depths from the surface.
For determination of the preconsolidation pressure of Ankara clay,
conventional test of 24 hour duration of load application with pressure
increment ratio of 1.0 is recommended. Schmertmann method is
suggested for evaluating the preconsolidation pressures.
It is also recommended that laboratory consolidation test loads should go
up to 32 kg/cm2 for clays which are located at the middle parts of the
basin since they have greater preconsolidation pressures. Otherwise
virgin portion which is extremely important in the determination of
preconsolidation pressure can not be obtained satisfactorily with load
increments only up to 16 kg/cm2.
Elias (1967) investigated the equilibrium water content-suction
relationship under covered surfaces: In this study, vacuum gage type
tensiometer placed at five different depths to a depth of 1.5 m under two
old pavements in Ankara region to measure the natural soil suction. In
both cases’ the ground table was at a shallow depth.(3.2-4.5 meters). The
results were compared with suction-water content relations obtained by
direct and indirect laboratory methods to determine soil suction at
different water contents. The direct methods with wetting-drying cycles
were performed by Sandbox, Pressure Plate Extractor, Ceramic Plate
extractor and Bar Ceramic Plate extractor so as to cover a wide range of
soil suction measurements. High air entry values porous stones were
used during testing. The indirect methods were shrinkage and
consolidation tests. Shrinkage test was done in a drying cycle from a
saturated condition. Consolidation tests were run in the oedometer on
saturated samples in the usual manner. Utilizing this data, Croney's
(1958) suction-total stress relationship was used to prepare charts
relating dry density, moisture content, suction, and total stress. The
equilibrium water content-depth relationships in the field were hence
obtained by using these charts.
Comparison was made by :
a. Direct water content determinations on undisturbed samples,
8
b. By utilizing the suction-water content relationship obtained and
assuming a simple capillary suction distribution,
c. By the Projections as described above.
The results were within bounds of experimental accuracy.
Doruk (1968) conducted a study using compacted and undisturbed
samples on the three kinds of clayey soils (residual, Pliocene-Pleistocene
deposits, alluvial) recognized in the campus of METU. The factors
affecting the swelling of soils were examined. swelling tests under
different conditions were made and the soils were classified by the
USBR and Seed et. al, (1962) methods. The results indicated that:
-Soil structure has a paramount effect on swelling such that an increase
in the degree of orientation of the particles due to preconsolidation
decreases the swelling amount of soil.
-Increases in the initial moisture content of the soil decreases swelling.
-Increase in percentage of soil finer than 2 micron increases swelling
amount and swelling potential.
-Increase in plasticity index as a rough measure of clay type and its
amount, increases swelling.
-Although the soils can be classified as to their Swell Potential by means
of the classification methods mentioned , the actual amount of swell that
occurs under natural conditions can not be predicted.
-Increase in dry density increases swelling for soils having small amount
of initial water content. For high initial water contents, increase in initial
dry density may increase or even decrease swelling.
-It is very difficult, almost impossible to simulate natural conditions in
laboratories. The best thing to do is to found stations at different places
and to record the swelling of the soils under natural conditions for a long
period of time, from which the actual swelling behavior of soils can be
predicted.
Following this last recommendation , Omay (1970) made field
observations and laboratory experiments. In the field, elevation changes
which occur at various depths on two locations of METU campus at
monthly intervals were observed for a year .One of the practical aspect
of the study was presentation by charts of the possible future heave
amount for a given initial water content, for any soil having
approximately identical soil characteristics . By these charts it is
possible have a rough idea about future heave amount for soils with
approximately identical soil characteristics. It is suggested furthermore
that light structures are most sensitive to heaving. Hence; as being done
in some parts of Israel, the zone below which the moisture stays more or
less constant should be determined and foundations should lie below this
zone.
In the laboratory experiments swelling pressures of 1.5-5.0 t/m2 has
been obtained . In some regions of Ankara, the contact pressure for light
structures may be of the order of 1.0-3.0 t/m2. Under these conditions,
damage due to heave is expected.
9
Abbasoglu’s (1971) study was intended to investigate the effects of
inorganic chemicals on the swelling characteristics of Ankara clay.
Throughout the research, 146 samples, compacted under standard
compaction energy and with various initial water contents have been
tested for one dimensional swelling. 32 samples has been tested without
using any additive, and 114 samples have been treated with solutions of
Sodium and Calcium Chloride of various percentages by the dry weight
of the soil specimen. The following observations were made:
-Under a given compactive method and energy, Ankara clay specimens
compacted at low initial water contents exhibit more vertical swelling
than specimens compacted at higher initial water contents.
-Samples compacted on the dry side of the optimum have higher initial
rate of swelling than those compacted on the wet side.
-Under a given initial water content and dry density, remolded clay
samples swell more than undisturbed samples.
-Inorganic chemicals NaCl and CaCl2 produces a net reduction in the
time rate and amount of vertical swelling.
-Addition of NaCl and CaCl2 reduces the optimum water content and
increases the dry density.
-Increasing amounts of NaCl and CaCl2 cause a progressive reduction in
the liquid limit of clay samples. The reduction caused by CaCl2 is more
than that of NaCl due to the bivalence of exchangeable ions.
Kocabayoglu (1971) studied the contribution of desiccation to the
preconsolidation of Ankara clay:.
The purpose of this study was to try and estimate the contribution of
desiccation to the preconsolidation of the upper layers of the Ankara
clay, and hence throw some light on the accuracy of predictions of the
maximum height of geologic overburden in different locations of the
city of Ankara.
Throughout the world there exist areas of surface clays which are
subjected to cycles of drying and wetting due to alternating dry and wet
climatic conditions. The capillary forces in the drying clay subject the
clay to an increase in effective pressures which causes overconsolidation
to considerable depths. Some of these clays have ,in addition, been
preconsolidated by heavy overburden which has subsequently been
removed by erosion.
At this stage , based on the foregoing geological evidence cited and the
research that is to be reviewed subsequently, it is stated that in the upper
layers of the Ankara clay desiccation is responsible for as much as 50 %
of the preconsolidation pressure estimated from laboratory tests. This
percentage is based on the assumption that at the greatest depth
investigated, the effect of desiccation is negligible.
In this study, the predicted values of preconsolidation, in spite of the fact
that they include the additional effect of desiccation, have been found to
be lower than the values calculated on the basis of geomorphologic data
for four of the five sites investigated. This may have been partly due to
10
the inaccuracy of laboratory predictions and partly due to the
overestimate of the past elevation of the ground surface. That the soils
tested have been re-transported after the initial deposition and
preconsolidation is a further possibility.
Yüncü (1972) conducted swell tests on 6 inch diameter , 1 inch thick
samples of statically compacted samples in the laboratory to evaluate the
volume change behavior of Ankara clay. In Ankara, the clay content
decreases and silt content increases as one approaches the quaternary
fluvial deposits. Thus ,it is possible to observe swell damage at one
place and collapse damage at another under the same surcharge: The
influence of varying silt content and that of different surcharge
pressures to the volume change behavior of the Ankara clay were thus
studied.
The following were observed:
-Increasing the surcharge decreases the swelling.
-The value of swelling pressure decreases with increasing silt content
hence Increasing Plasticity Index increases swelling index Cs.
-Wetting after loading or loading and wetting at the same time give
almost the same results for the swelling index and the swelling pressure
for identical samples.
-Increasing the silt content increases the collapse potential. When a
collapse potential classification is made according to the criteria given
by Gibbs (1967) , it is found that samples having high silt content is
close to the region of collapsing soils.
-Though the value of swelling pressure for Ankara clay having various
silt content is obtained between 0.70 - 1.00 kg/cm2, this value can be as
high 2.0 - 2.5 kg/cm2 (Omay, 1970) . So light structures around Ankara
are damaged seriously.
When the structural surcharge is not sufficient to prevent expansion of
the soils, the following measures may be adopted:
i) The natural subgrade expansive clays may be removed and replaced
with non-expansive soils to a sufficient depth.
ii)If the subgrade is a remolded expansive soil, the volume change may
be controlled by compacting the soil at high moisture contents and low
densities as predetermined through proper laboratory tests.
iii)The saturation of the foundation soils may be prevented by cutting off
all sources of water supply, although this may be difficult and even
impossible in some cases.
iv) Piles anchored in an inactive zone below a structure provide a reliable
assurance against uplift if properly incorporated into the structure base.
Söylemez (1972) made an identification of the clay mineral composition
for four samples ( 2 samples from METU campus, one from Yucetepe
and one from Ayranci ) of Ankara clay. A quantitative analysis of the
clay mineral constituents of the samples was also made. X-ray powder
diffraction method was used for the analysis.
11
The analysis showed that the clay fraction of the samples were
composed, on the average, of illite 50 %, kaolinite 26 %,
montmorillonite 12 %, and vermiculate 12 %. The most abundant
mineral in the samples examined is illite. The ratio of its abundance to
that of the total clay minerals is about 0.5. Therefore, the samples
analyzed are to be identified as to be illitic clays. Kaolinite is also quite
abundant in the samples. Its amount varies from 22 to 36 per cent. There
is essentially no chlorite in the samples.
Expansive minerals (montmorillonite and vermiculate ) are present in the
samples in minor amounts. However the researcher believed (Grim,
1953) that montmorillonite affects the physical properties of clay when
present even in minor amounts. The minimum percentage found for
montmorillonite in this study is 9 % . The high values of LL and PI
found for the samples are most probably due to the presence of
montmorillonite.
Kaynar (1972) studied the ratio of swelling index to compression index
in clay soils:
The following general conclusions were reached as a result of his study:
-The ratio of swelling index to compression index increases with
increasing plasticity.
-Rate of increase of swelling index becomes higher for more plastic
soils.
-An increase in plasticity causes soils to have higher void ratios.
-There is a certain proportional linear relationship between LL and the
ratio of swelling index to compression index for clays. That is, This ratio
increases when the Liquid Limit increases.
-The Double Oedometer Test:
At this stage, a deviation from the exhibition of research results is
mandatory to explain what is called a Double Oedometer Test (DOT)
because this test is utilized by some researchers whose studies will be
presented subsequently. The quantitative prediction of partly saturated
soil behavior requires laboratory simulation of pressure and moisture
conditions existing in the field. DOT is a test that may be used for this
purpose.
DOT consists of testing in the oedometer a pair of soil samples obtained
from the expansive soil, adjacent to each other so that their initial
conditions such as void ratio, water content and physico - chemical
properties are presumed to be identical. The pair of samples are first
subjected to a very small seating load of 5 to 10 KN/m2 prior to testing.
Then they are loaded in conventional consolidation testing increments,
one in its natural state and the other after having been soaked and
allowed to swell. The virgin portions of the resulting void ratio-pressure
curves are made to coincide ( if needed, natural moisture content
(unsaturated) curve is displaced upwards so as to comply with this
requirement. It should be considered that the unsaturated sample
12
practically reaches saturation by compression at a critical pressure
increment if the initial degree of saturation is not extremely low. (e.g.
below 40%).
The distance parallel to the pressure scale between these two curves at
any stress level should show the soil suction component of Bishop's
effective stress equation since the curve above is for effective stresses
for saturated soil and the one below is also for effective stresses but of
unsaturated soil. Stress paths may then be envisioned for in situ
conditions of the unsaturated soils when they are inundated under the
applicable overburden. For example, final void ratios in swelling are
estimated from the DOT for each applied pressure under which the
samples show swelling. The stress path independence is one of the big
advantages of DOT.
Akbay (1972), studied volume change characteristics of Ankara clay
with changes in moisture content under different surcharge loads.
Double Oedometer Tests were used,
The findings were: Void ratio versus applied pressure curve of a given
soil differs depending upon the initial water content and degree of
saturation. Changes in moisture content causes changes in volume, but
the direction of this change in volume during change in moisture content
is dependent on the magnitude of the surcharge on the sample i.e. soil
swells under a low surcharge and then under a larger one it exhibits
collapse. Higher surcharges reduce swelling upon soaking and for each
soil there is a particular pressure which will just admit swelling but will
not cause any collapse. This pressure may be defined as the swelling
pressure.
This researcher states that the double oedometer test as outlined above
may be used to predict the total collapse settlement or swell in partly
saturated soils. By these tests the settlement at field moisture and also
the additional amount to be expected on saturation can be estimated.
However as the tests do not completely reproduce the field conditions,
and thus represent somewhat the most critical state, the most accurate
test would be the one conducted in the field with the actual load in place.
So, the sampling errors would be eliminated. In such a case, a suitable
test set up enabling the researcher to load the soil along the predetermined stress path is desirable. This should preferably be made in a
shed utilizing settlement platforms, possibly inundating the soil and
observing the results and correlating these with laboratory observations
to get a correct picture.
As the soils above the water table do not necessarily reach 100 %
saturation, the amount of volume change for the degree of saturation
they will attain should be determined, using as accurately as possible the
stress path that the soil is expected to follow in the field.
13
Damla(1976), applied the Double Oedometer Method, (DOM), to
undisturbed samples taken from Kinik region, along Kurtbogazi Dam Ankara water line, a location where highly swelling soils are situated.
The swelling of the clay was observed while studying the moisture
variation against volume change behavior of the clay under different
loads. The swelling amount and the swelling pressure were also
determined by a series of double oedometer tests. The studies described
in this thesis lead to the following conclusions:
-Any partly saturated soil will undergo swelling under a given applied
stress when the suction is decreased if the void ratio-applied pressure
point for the saturated condition plots above the same point of that of the
natural moisture content condition.
-Double Oedometer Test (DOT) can be used to define the swelling
amount and the swelling pressure. Prediction of swelling pressure
according to double oedometer method (DOM) is done by letting
unsaturated sample swell fully under applied load from its initial void
ratio, and then re-compressing the sample to its initial void ratio. The
load needed to do this is the Swell Pressure.
- It is observed that for DOT to be a meaningful, loading in the oedometer
must be carried up to at least 32 kg/cm2 for Ankara clay to facilitate the
coincidence of the virgin curves.
-The time effect on expansive soils is a very important factor. For each
stress increment, time duration is different to reach a certain amount of
swell. Observations show that samples under low stresses need more
time to realize a certain percent of total swell than those under higher
ones.
-The amount of swelling pressure and swelling amount is directly related
to the initial conditions, and indirectly related to the Atterberg Limits.
Soils having high initial water contents exhibit lower swelling pressure
and lower swelling amount.
-The DOT provides a convenient measure for estimating the swelling
amount and the swelling pressure.
Sürgel (1976) modified the boundaries of classification of the soils on
the Reconnaissance engineering Soils Map of Ankara area reported by
Lohnes (1974) and determined the general properties of the Ankara
soils. He grouped Ankara soils into three different types named as
"Alluvial Soils " , " Residual Soils " and " Terrace Deposits " based on
Lohnes’ Reconnaissance Soils Map.
A frequency distribution analysis of engineering properties was made.
These frequency analysis were the basis of modification of the
classification boundaries. Sürgel also studied general properties of each
class of soils by following an Engineering Parameters versus depth
study.
The following conclusions were reached:
-The parent material of Ankara area is composed of three different soils
named as "Alluvial Soils ", "Residual Soils" and "Terrace Deposits"
which is also known as "Ankara Clay".
14
-Almost 75-85 % of the samples of the all three types of Ankara soils fall
above the A-line and a large part of it classifies as CH on the Unified
System.
-Terrace deposits are a little more active than residual soils, and residual
soils are more active than alluvial soils All three types of soils have been
subjected to surface drying and desiccation.
-Considering the swelling potentials of different soils it can be said that
terrace deposits may swell a little more than residual soils and residual
soils swell more than alluvial soils.
Surficial layers of all three types of soils have been weathered and
affected by fissures.
-Consistencies of these soils types are as follows;
-i.Alluvial soils are in a semi-solid state and have medium to high
plasticity.
-ii. Residual soils are highly plastic. The soil within upper 3 meters is in
a semi - solid or solid state, and after 3 meters the soil is in plastic state.
-iii. Terrace deposits within the depths of 0.0-4.0 meters are in semi-solid
or solid state and have stiff consistency, the soils within the depths 4.019.0 meters are in plastic state and have stiff consistency, the soils
within the depths 19.0-25.0 meters are in semi-solid or solid state and
have high plasticity, and soils after 25.0 maters are in states varying
from semi-solid to plastic and have high plasticity.
iv. METU campus clay has a plasticity varying from high to very high
within upper 2.0 meters. Below this depth the soil has equal chance to be
in a semi-solid or plastic state.
-All three types of soils appear to be preconsolidated, mainly due to
desiccation.
-The parts of the all three types of soils remaining on the south side of
Ankara river are more active than the soils remaining on the north side.
-Plot of samples tested in this study on Birand(1978)’s Classification
chart for Ankara soils shows that Terrace deposits samples fall in the
terrace deposit region defined by Birand(1978). Nevertheless, the range
given for alluvial samples is not entirely compatible with
Birand(1978)’s range.
The properties of soils in Ankara region was treated by Ordemir et al
(1977) with special emphasis on their swelling properties. The
predominant types of clay formations in the region were introduced and
classified with respect to their expansiveness. Data on field volume
changes were presented and it is shown that the so called "Ankara Clay"
exhibits a potentially expansive character.
The researchers reached the following conclusions:
-Residual soils which cover a relatively smaller area whose significance
is not as great when city’s development is considered.
-Deposits of Pliocene - Pleistocene age which are mostly terrace
formations in flood plains. They refer to it as preconsolidated, stiff,
fissured "Ankara Clay". These formations contain solution deposits and
concretions of lime which decrease in a north-north west direction. It has
been noted in general that swelling potential decreases with increasing
lime content.
15
-Recent Alluvium deposits are encountered along local rivers and
streams.
-As related above, Ankara has a semi-arid climate according to the
Thorntwaite system. This is known to be a prerequisite for swelling
problems which are further accentuated by geological setting and soil
type. Problems due to swelling are to be expected mostly in the case of
"Ankara Clay ",Though recent alluvium may also deserve special
attention at certain places.Double Oedometer testing have been found
very useful and effective in the determination of swell pressures and
amount for this soil.
Birand (1978) studied the geotechnical properties of Ankara soils and
proposed a classification method.
-He suggested that. Preconsolidation of Ankara Clay was due to
capillary forces.
-He states that Ankara clay is active for swelling and shrinkage and this
may create problems depending on the position of the water table. If the
ground water table is at a shallow depth, Ankara clay and alluvial clay
creates settlement and stability problems, otherwise complications due to
volume change may arise.
-On the other hand, Swelling problems should be considered when using
Ankara clay for embankment material. Otherwise the structure on such
embankments may be damaged due to swelling. The cracking of the core
of Çubuk dam upon inundation is an example.
-Birand (1978) states that the Double Oedometer Test is a valuable tool
in geotechnical engineering for the solution of problem of expansive
soils. Its value derives not only from its ready applicability to
engineering problems but also from the fact that its results can be closely
related to the fundamental behavior of partly saturated soils in terms of
effective stresses. He has estimated swell pressure of Kinik clay samples
by three different methods:
a)Making use of the results of the DOT and reading off the value of
swell pressure,
b) Letting the sample to swell and then compressing it by application of
total stress increments until the initial thickness is obtained,
c)Increasing the total pressure continuously during inundation so as to
prevent any swelling.
In this paper he observed that the values of swell pressures determined
by the three methods are close, suggesting the existence of a stress path
independence in Double Oedometer testing. This is a very useful
concept.
-He plotted the swell (%) versus total pressure on several samples of
Kinik clay, all samples had been taken very near to one another. He
observed that the duration of swell measurements are important (i.e. 40
day swell amount value is higher than 24 hour swell amount value and
the DOT curves plotted using the 40 day values would yield higher swell
pressures than those would be obtained if 24 hour values were utilized.
16
Öner and Birand (1978) made a mathematical analysis of moisture
equilibrium under a pavement. It has been recognized for a long time
that the volume change behavior in the field can be characterized by
moisture profile (or suction profile) changes that occur when the surface
is covered.
Following these concepts, the researchers solved numerically the
diffusion equation for the soil under the Esenboğa highway pavement,
obtained the soil suction values and compared the results with the
tensiometer readings obtained earlier on by Elias(1967), which
compared favorably.
Furtun (1989) performed swelling tests in the oedometer on undisturbed
samples taken from Pliocene - Pleistocene age fluvial-lacustrine deposits
(terrace deposits) and recent alluvium deposits in various parts of
Ankara region to assess their relative expandability. He obtained the
swelling pressure and percent swell values with initial water content and
dry density. He compared the swelling pressures measured in constant
volume swell and swell under overburden tests. Additionally, data given
by previous researchers for Ankara soils was used to evaluate the
swelling potential.
An empirical equation giving swelling pressure in terms of LL, initial
water content and initial dry density were developed based on the test
results. The following main conclusions can be drawn on the basis of his
study:
Terrace deposits have larger values of Atterberg Limits and Clay content
than alluvial soils and the range of index properties of terrace deposits
are very wide because of their heterogeneous structure which contains
various sizes of silt, sand and gravel particles in the forms of bands and
lenses. For example PI values change between 20 and 55 % while clay
content values are between 10 % and 70 % . Another reason for this may
be the wideness of the area and the differences in the parent materials of
the terrace deposits formed in different locations. In order to gain an
idea about the swelling properties of terrace and alluvial deposits, data
was also evaluated by four different classification criteria, namely; Da
Nilow, Modified Casagrande, Seed et al and Van Der Merve (Chen,
1968). For all classifications between 70 - 86 % of the terrace deposits
are classified as high to very high expansive clay while only 26 - 37 %
of the alluvial deposits are classified as such.
According to the Da Nilow’s chart 68 % of the terrace deposits were
found to have appreciable swell potential while the value is only 12 %
for alluvial soils.
Swell potential for terrace deposits increase as initial water content
decreases and as dry density increases.
Liquid Limit (LL) appears to be a good indicator of swelling (i.e. swell
potential increases with increasing LL increases)
Damage due to swelling has been observed in Ankara basin where rapid
expansion of the city led to the construction of various kinds of
structures on the Ankara clay formation having a potentially expansive
character.
17
Swell pressure values obtained by swell under overburden test is
generally higher than the ones obtained from constant volume test.
The use of relationships present in the literature based on different soils
and locations for swell parameters in terms of index properties can be
misleading. Such correlation have to be derived locally.
Analysis has shown that swell pressure is strongly correlated with Liquid
Limit (LL), Initial Water Content (wi) , and Dry Density (d) of the soil.
In the analysis MINITAB Program was used and the analysis of 30
Constant Volume Swell test results of the study on soils from 6 different
locations in Ankara revealed the following empirical relationship to
predict the swell pressure (Ps) in kg/cm2.
logPs=-4.161-0.059wi+2.368 d+0.036LL
(Herein the coefficient of correlation is 0.90)
Erol (1990) has found that the swell amounts calculated from
oedometer test results are higher than the in-situ heave for an expansive
shale from Saudi Arabia. This may be due to the inadequate wetting of
the in-situ soil with respect to oedometer specimen and due to cracks
and fissures in the in-situ soil which causes lateral deformations, these
lateral deformations may decrease vertical swelling amounts. He defined
a "lateral restraint factor Lr" to reflect the effect of lateral deformations.
Where Lr=(In-situ heave amount/Swell amount found from oedometer
test) This appears to be a value between 1.0 (under the oedometer
conditions, no lateral deformation) and 0.33 (lateral deformation is
allowed). In the field, depending on the soil conditions, it is suggested
that the Lr value is in between these limits.
Yanikomeroglu (1990) studied this effect of lateral confinement on swell
behavior. He used Ankara clay and bentonite - sand mixture samples in
his studies. Due to the fissured structure of Ankara clay, the results of
his study are important in that the fissures may permit some lateral
swelling, hence relieving the vertical swell pressures. In conventional
oedometer tests the influence of the cracks can hardly be simulated. This
is because laboratory samples are usually prepared from smooth intact
cores which include minor discontinuities only. Therefore the swell
parameters obtained from oedometer tests tend to overestimate in-situ
heave in such soils. A lateral restraint factor introduced is intended to
account for such structural effects.
A model study has been performed to simulate the cracks and fissures
(macropores) in expansive soils with artificial holes drilled in two
different samples. Free swell, swell pressure and swell under surcharge
tests have been performed for three restraining conditions
i. No lateral restraint,
ii.Partial lateral restraint,
iii.Full lateral restraint.
An expression for lateral restraint factor in terms of percent macropore
content and normal pressure has been derived.
On the basis of his study, the following conclusions can be drawn:
18
There is a linear relationship between percent swell and logarithm of
normal pressure (percent swell decreases as logarithm of normal
pressure increases ).
The amount of swell as well as swell pressures increase with increasing
values of dry density.
Both the swell amounts and swell pressures are reduced with increasing
values of macropore content. This reduction when it occures in situ,
would be similarly due to crack closure and a possible reduction in lateral
swell pressures.
The rate of swell is significantly increased by the presence of macropores
which increases the permeability and consequently the rate of water
intake.
The predicted heave based on experimental swell data obtained from
intact samples in the oedometer should be corrected by multiplying the
results with a lateral restraint factor, if a reasonable estimate of this
factor can be made for a real soil profile. The relationship obtained for
the soils tested was :
Lr=-0.12*10-3 (Pm) - 0.19P+0.89
Lr= Lateral restraint factor
Pm= Percent macropore content by volume, %
P = Surcharge pressure, kg/cm2
This relationship for Lr is valid for Pm >0, and Lr is unity for Pm=0
Pm= (1-(dm/d))*100
dm = bulk dry density (with macropores)
d = intact dry density
The crack closure process is more efficient in the vertically oriented pores.
It is also indicated that the magnitude of the "Index of Propensity to Water
Intake (Pw)" (Birand, 1976) is reduced significantly due to the presence of
macropores.
This Index is defined as :
Pw=de/dw
de= change in void ratio during swell
dw= change in water content that cause swell
Conclusions
The parent material of Ankara area is composed of three different soils
named as Alluvial Soils, Residual soils and Terrace deposits, which is
also known as Ankara clay.
19
Recent alluvium deposits are encountered along local rivers and streams.
Ranges of index properties of terrace deposits are very wide because of
their heterogeneous structure, which contains various sizes of silt, sand
and gravel particles in the forms of bands and lenses. Another reason for
this may be the wideness of the area and the differences in the parent
materials of the terrace deposits formed in different locations. Because
of this, the engineering problems change from one location to another on
terrace deposits. Therefore, terrace deposits are divided into sub groups.
Residual soils and terrace deposits are mostly classified as CH.
Residual soils and terrace deposits appear to be preconsolidated mainly
due to desiccation. Alluvial soils are normally consolidated.
All three types of soils have been subjected to surface drying and
desiccation to different extents.
Surface materials of residual soils and terrace deposits have been
weathered and affected by fissures.
Skempton’s original definition of activity as A=Plasticity Index/Clay
Content for natural London clays is applicable to natural Ankara clay.
The parts of the all three types of soils remaining on the south side of
the Ankara River are more active than the soils remaining on the north
side.
Terrace deposits are a little more active than residual soils, and residual
soils are more active than alluvial soils. Since swelling potential depends
on activity, it can be said that terrace deposits may swell a little more
than residual soils and residual soils swell more than alluvial soils.
The mineralogy is a subject of controversy. Some clay mineralogists
claim that the mineralogical composition of Ankara clay is Kaolinite Illite-Montmorillonite with predominant Kaolinite presence and lower
montmorillonite content. Based on sparse data, it is stated that the
exchangeable ion is calcium.
The most abundant clay mineral in the samples examined by Soylemez
(1972) is illite. Expanding minerals (montmorillonite and vermiculate )
are present in the samples in minor amounts.
However, according to Kasapoglu (1980) the dominant mineral in clay
fraction of Ankara soils is montmorillonite.
Kiper (1983) states that in Ankara clay montmorillonite and illite are the
dominant minerals and chlorite and kaolinite are present in minor
amounts.
Terrace deposits contain solution deposits and concretions of lime
constituency, which decrease in a north - northwest direction. It has been
noted in general that swelling potential decreases with increasing lime
20
content. According to Kasapoglu (1980), the main source of the Calcium
is limestone formations present in the region.
The effect of precompression of the clays on swelling is to reduce the
swelling potential.
The method proposed by Lambe and Whitman (1959) should be used in
estimating the potential heave of Ankara clays. Using this method, one
would be on the safe side and also take the climatic factors into account.
Swell pressure values obtained by swell under overburden test is
generally higher than the ones obtained from constant volume test.
The following empirical relationship to predict the swell pressure Ps in
kg/cm2 can be used for Ankara Clay as cited above.
logPs=-4.161-0.059wi+2.368 d+0.036LL
The value of swelling pressure decreases with increasing silt content.
The swell and swell pressures are reduced with increasing values of
macropore content. This reduction; based on the tests performed on
Ankara clay and bentonite - sand mixtures is due to crack closure and a
possible reduction in lateral swell pressures.
The rate of swell is significantly increased by the presence of
macropores, which increase the permeability, and consequently the rate
of water intake.
The predicted heave based on experimental swell data on bentonite sand mixture samples and on Ankara clay as from intact samples should
be corrected by multiplying with the lateral restraint factor Lr, if a
reasonable estimate of this factor can be made for a real soil profile.
Double oedometer test is useful and effective in the determination of
swell pressures and amounts. It seems that there is stress path
independence in swelling behavior. In this procedure, test duration is
found to affect the test results.
Under a given compactive method and energy, Ankara clay specimens
compacted at low initial water contents exhibit more vertical swelling
than specimens compacted at higher initial water contents.
Under a given initial water content and dry density, remolded clay
samples swell more than undisturbed samples.
Increase in the initial moisture content of the soil decreases swelling for
undisturbed and compacted samples.
Increasing the surcharge decreases the swelling.
21
Inorganic chemicals NaCl and CaCl2 produce a net reduction in the rate
and amount of vertical swelling.
Samples compacted on the dry side of optimum have higher initial rate
of swelling than those compacted on the wet side.
Liquid Limit appears to be a good indicator of swelling potential
Increasing Plasticity Index increases the Swelling Index value (Cs).
Ankara has a semi- arid climate according to Thorntwaite system. This is
known to be a prerequisite for swelling problems, which are further,
accentuated by geological setting and soil type.
Swelling problems should be considered when using Ankara clay for
embankments. Otherwise the structure that will be built on this
embankment may be damaged due to swelling.
References
Abbasoğlu, C., 1971, " Ion Exchange Process Affecting Swelling and
Other Properties of Ankara Clay ", M.S.Thesis, METU, Civil
Engineering Department, 78 pages.
Akbay, Ö.Ü., 1972, " The Influence of Saturation on Volume Change
Characteristics of Ankara Clay Under Various Surcharge Pressures",
M.S. Thesis, METU, Civil Engng. Dept., 53 pages.
Arda, Ş.,1966, " Preconsolidation of Ankara Clay", M.S. Thesis,
METU Civil Engng Dept., 58 pages.
Birand, A., 1963, " Study of the Characteristics of Ankara Clays
Showing Swelling Properties", M.S. Thesis, METU, Civil Engineering
Department, 40 pages.
Birand, A.A., 1976, " Presentation of a Case of Damage to an Airfield
Pavement", METU Journal of Pure and Applied Sciences, Vol.9, No.1,
pp.99-111.
Birand, A.A., 1977, " Ankara Yöresi Zeminlerde Ön Yükleme
İsotropisi",4. Tubitak Teknik Kongresi,
Altinyunus, Izmir, pp.277287.
Birand, A., 1978, "Ankara Yöresi Zeminleri ve Jeoteknik Sorunlar",
Yerbilimleri Açısından Ankara’nın SorunlarıSempozyumu, Türkiye
Jeoloji Kurumu, ss.55-60.
22
Chen,F.H.,1968. "Foundations on Expansive Soils", Elsevier Scientific
Publishing Company, Amsterdam-Oxford- New York, 280 pages
Damla, Ö.R., 1976, " Prediction of Swelling Potential
and Swelling Pressure From the Double Oedometer Test", M.S. Thesis,
METU, Civil Engng. Dept., 33 pages.
Doruk, M., 1968, " Swelling Properties of Clays on the METU
Campus", M.S. Thesis, METU, Civil Engng. Dept., 46 pages.
Elias, M., 1967, " An Investigation of Moisture Movement in Soils
and the Concept of Equilibrium Moisture Distribution with its Bearing
on Pavement Performance ,M.S. Thesis . METU Civil Engineering
Departmen.,99 pages
Erol, O., 1990, " Gercek Kabarmaların Şişme-Ödometre Metodu ile
Karşılaştırılması ", Zemin Mekaniği ve Temel Mühendisliği Üçüncü
Ulusal Kongresi, Boğaziçi Üniversitesi, Cilt 1, ss.1-8.
Gibbs, H., 1967, “Stability Problems of Collapsing Soils”, ASCE,
journal of Soil Mech. And Found. Eng.
Grim, RE, 1953. "Clay Mineralogy ", Mc Graw- Hill Series in the
Geological Sciences
Furtun, U., 1989, " An Investigation on Ankara Soils With Regard to
Swelling", M.S. Thesis, METU, Civil Engng. Dept., 151 pages.
Kasapoğlu, K.E., 1980, " Ankara Kenti Zeminlerinin Jeomühendislik
Özellikleri", Doçentlik Tezi, Hacettepe University, Geological
Engineering Department, Beytepe, Ankara.
Kiper, O.B., 1983, " Etimesgut-Batıkent Yöresindeki Üst Pliosen
Çökellerinin Jeomühendislik Özellikleri ve Konsolidasyonu", Ph.d.
Thesis, Hacettepe University, Ankara.
Kocabayoğlu, E., 1971, " Contribution of Desiccation to the
Preconsolidation of Ankara Clay ", M.S. Thesis, METU Civil Engng.
Dept., 18 pages.
Krynine, D.P., (1941), Soil Mechanics, Its Principles and Structural
Applications, Mc Graw Hill Co.
Lambe, T.W. and Whitman, R.V., 1959, " The Role of Effective Stress
in the Behavior of Expansive Soils", Quarterly of the Colorado School
of Mines, Vol.54, No.4, pp.33-61.
Lambe, T.W., 1960, " The Character and Identification of Expansive
Soils (Soil PVC Meter )", A Report Completed for The Technical
23
Studies Program of The Federal Housing Administration, MIT, 46
Pages.
Lohnes ,R.V. 1974 "A Report on the Geologic Characteristics of
Ankara Soil Formations" Unpublished Report Submitted to the Civil
Engineering Department, METU
Omay, B., 1970, " Swelling of Clays on METU Campus", M.S. Thesis,
METU, Civil Engng. Dept., 73 pages.
Ordemir, İ., Alyanak, I. and Birand, A.A., 1965, "Report on Ankara
Clay", METU, Publication No.12, Ankara, 30 pages.
Ordemir, I., Soydemir, C., Birand, A., 1977, " Swelling Problems of
Ankara Clays", 9th. Intentional Conference of Soil Mechanics and
Foundation Engineering, Tokyo, Vol.1, pp.243-247.
Öner,M. and Birand, A., 1978, " Effect of an Impervious Surface Cover
on the Soil Suction Equilibrium", METU Journal of Pure and Applied
Sciences, Vol.10, No.2, pp.207-222.
Özmelek , A.İ., 1984, " An Investigation of the Concept of Equilibrium
Moisture Distribution and the Effect of Climatic Factors on Subgrade
Moisture Conditions", M.S. Thesis, METU Civil Engng. Dept. , 197
pages.
Parcher, J.V., and Means, R.V., 1968, “Soil Mechanics and Foundation
Engineering, Ohio, Charles E. Merrill Pub. Co.
Seed, H.B., Woodward, R.J. and Lundgren, R., 1962, “Prediction of
Swelling potential for Compacted Clays”, Journal of the Soil Mechanics
and Foundations Division, Proc. Of the ASCE, pp.53-87.
Söylemez, N., 1972, " Mineralogical Analysis of Ankara Clay by XRay Diffraction ", M.S. Thesis, METU Civil Engng. Dept., 89 pages.
Spangler, G.M. and Handy, L.R., (1973), Soil Engineering, Third
Edition
Surgel, A., 1976, " A Survey of The Geotechnical Properties of Ankara
Soils", M.S. Thesis, METU Civil Engineering Department, 96 Pages.
Trask, P.D., (1959), Applied Sedimentation, John Wiley and Sons.
Yanikömeroğlu, K., 1990, " Effect of Lateral Confinement on Swell
Behavior", M.S. Thesis, METU, Civil Engng. Dept., 105 pages.
Yüncü, H., 1972, " An Investigation of Volume Change of Ankara
Clay", M.S. Thesis, METU, Civil Engng. Dept., 52 pages
24