Geotextiles and Geomembranes 16 (1998) 207—219

Geomembrane—sand interface frictional

properties as determined by inclined
board and shear box tests
Mizyal Izgin1, Yildiz Wasti*
Department of Civil Engineering, Middle East Technical University, 06531, Ankara, Turkey

Abstract

Geomembrane—sand interface shear strength parameters were measured using inclined board
and standard size direct shear box tests. Ottawa sand with well-rounded particles and a crushed
stone with very angular particles but of very similar size range and gradation along with
smooth and rough HDPE geomembranes were used in the experiments. Inclined board tests
were conducted under a normal stress range of about 5—50 kPa on variable size interface areas.
In direct shear tests the applied normal stresses were 14—200 kPa.
It has been found that inclined board test results produce plots of interface shear strength
versus normal stress that can be approximated by straight lines through the origin. Direct shear
tests on the other hand gave envelopes with adhesion intercept and larger interface friction
angles. Interface friction angles for sand with rounded particles were more sensitive to surface
roughness of geomembrane and were different for smooth geomembranes of different brands.
( 1998 Elsevier Science Ltd. All rights reserved

Keywords: Geomembrane; Sand; Interface shear

1. Introduction

In geotechnical applications such as the lining of canals or waste containment

systems, where geosynthetics are placed on inclined surfaces, the correct assessment of
the interface shear properties between the geosynthetics and soils or between different
types of geosynthetics becomes an important issue considering the fact that interfaces

1Present address: Kolin Ins, aat Imalat ve Ticaret A.S, ., Horasan sok. No.24 G.O.P., 06700 Ankara,
Turkey.
*Corresponding author. Tel: 90312-2102410; fax: 90312-2101262; e-mail: ywasti @rorqual.cc.metu.edu.tr

S0266—1144/98/$ — see front matter ( 1998 Elsevier Science Ltd. All rights reserved
PII S 0 2 66 — 11 4 4( 9 8 ) 00 0 10 — 7
208 M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219

with low in-plane shear resistance act as potential failure planes. Direct shear testing
has been almost exclusively used to determine interfacial shear strength parameters.
Tests are often conducted under high normal stresses and it is believed that test results
at low normal stresses may not be accurate due to mechanical difficulties. More
recently, the use of ‘inclined board’ or ‘tilting table’ tests for interface shear strength
measurements has been reported (Giroud et al., 1990, Girard et al., 1990, Koutsourais
et al., 1991). The apparatus consists essentially of a board hinged on one side and
raised on the other side till sliding occurs along an interface. To measure geosynthetic
to geosynthetic interface friction a rigid block with one specimen attached to its base
and providing the normal stress is placed on the other specimen attached to the upper
face of the board. The inclination at which the block just begins to slide is recorded.
A replica of a composite lining system having several interfaces can be constructed on
the board/table and sliding along the weakest or a predetermined interface can be
initiated. Inclined board tests are conducted with small normal stresses, because to
apply a large normal stress, a tall sliding block is needed and this would apply
a significant overturning moment. However, this is not considered as a disadvantage
in the sense that the test reproduces the real conditions more accurately in the lining
systems where the normal stresses are generally low.
Very limited data have been reported involving the use of inclined board as well as
direct shear tests for the same interfaces. Giroud et al. (1990) used 300 mm]300 mm
direct shear box tests with normal stresses between 25 and 160 kPa and inclined board
tests at a much lower normal stress of 0.7 kPa to test rough geomembrane—hard
geonet and rough geomembrane—nonwoven geotextile interfaces. For the hard
geonet—rough geomembrane interface direct shear tests gave an interface friction
angle d of 10° and no adhesion. In the inclined board test the same interface friction
angle of 10° was measured. On the other hand, for the rough geomembrane—needle-
punched nonwoven geotextile interface the results of these two methods of testing
yielded quite dissimilar results: The direct shear tests gave d of 15° and adhesion value
of 5.5 kPa; and in the other test the rigid block slid when the inclination of the board
was 48°. The authors used this single data point from the inclined board test to modify
the straight line envelope obtained from the direct shear tests, for the stress range less
than 25 kPa. This resulted in a curved envelope, the initial part of which lies below the
direct shear box test envelope as shown in Fig. 1.
Koutsourais et al. (1991) presented extensive data on interface friction shear
strength for a total of 71 interfaces involving soils and various geosynthetics. They
used a direct shear device with dimensions of 305]305]51 mm deep and normal
pressures ranged from 20—62 kPa. Geonet—geotextile or geomembrane, and geotextile
— geomembrane interfaces were also tested using an inclined board test device with an
interface area of 510 mm]102 m and a normal stress of approximately 2.7 kPa. The
data have been scrutinized by the present authors to compare the results of direct
shear and inclined board tests for the same interfaces. It has been observed that for
those interfaces for which the direct shear tests gave a linear failure envelope through
the origin, geonet—geosynthetic interface friction angles obtained from the inclined
board tests were generally a few degrees higher. The geonet—rough HDPE geomem-
brane interface friction angle was about the same as found from direct shear. This
 M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219 209

Fig. 1. Geotextile—rough geomembrane interface shear strength: (a) results of direct shear box tests (m) and
inclined board test, point B (d); and (b) close-up in the vicinity of the origin of the axes. (The curve was
conservatively drawn as if it would meet the straight line for p"50 kPa.) (After, Giroud et al., 1990).

confirms the findings of Giroud et al. (1990). For interfaces for which the direct shear
tests gave an adhesion value, the data points from the inclined board tests were below
the direct shear envelopes as illustrated in Fig. 1. Various geomembrane — geotextile
interfaces tested gave results which make generalization impossible. For instance,
PVC—CSPE geomembrane—geotextile interfaces did not yield adhesion values in
direct shear tests. In the case of PVC geomembrane inclined board d values were 4—6°
higher than direct shear d values for woven geotextiles but 2 and 4° smaller for heat
bonded and needle-punched nonwovens, respectively. d values from both tests were
about the same for CSPE geomembrane-woven geotextile interfaces, but inclined
board d values were again 3 and 7° smaller for heat-bonded and needle-punched
nonwovens, respectively. On the other hand, for HDPE/VLDPE geomembrane—
geotextile interfaces, direct shear test envelopes had adhesion values generally ranging
between 1.7 and 2.8 kPa. d values from inclined board tests were 4—14° higher and the
single inclined test data point was below the envelope. For rough HDPE — needlepun-
ched geotextile however, the adhesion value was as high as 17.2 kPa and d from direct
shear and inclined board tests were 15 and 34°, respectively. Note that this result is in
210 M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219

agreement, at least qualitatively, with the findings of Giroud et al. (1990) for the same
interface as illustrated in Fig. 1.
An incident of partial slip of the geomembrane dam facing along the nonwoven
geotextile — PVC geomembrane interface and the data reported by Girard et al. (1990)
support the findings of Giroud et al. (1990) and Koutsourais et al. (1991) that the
inclined board tests carried out at low normal stresses (due the very nature of the
method) give lower interface shear strength values for some interfaces, which are
possibly more realistic. Girard et al. (1990) used a 375 mm]375 mm direct shear box
with the normal stress of 100, 200 and 400 kPa and d"34° and apparently no
adhesion was found, giving an appreciable safety margin when compared with the
21.8° angle of the dam slope. The measurement of the interface friction angle from
a 1.0 m]1.0 m inclined board test gave 25°. The authors concluded that the measured
d value from the inclined board test and, in addition, the dynamic stresses resulting
from vehicle traffic on the slope explained why the slip occurred.
As can be seen from the review of past work the reported inclined board tests have
been conducted at a single normal stress on one size interface area and interpreted as
though the envelope passes through the origin. The correlations with the direct shear
box results for different types of interfaces do not follow a general trend. The present
study focuses on the behaviour of geomembrane—sand interfaces as measured by the
inclined board tests on which there appear to be no previous data in the literature.
Normal stresses and interface area sizes have been varied and different brands of
smooth and rough (textured) HDPE geomembranes and two sands of comparable
gradation and mean size but having distinctly different particle roundness were used
in the experiments. Standard size (60 mm]60 mm) direct shear box tests were also
conducted to enable comparisons.

2. Experimental program

2.1. Inclined board test apparatus

The specially designed apparatus consists of a tilting board which is actually

a shallow 600 mm]490 mm]25 mm rectangular tray hinged to a 0.9 m high steel
frame along one of the shorter sides (Fig. 2). For the lifting system on the opposite side,
the combination of an air compressor and a hydraulic jack rather than one of the two
was found necessary to ensure smooth raising of the inclined board, without any speed
fluctuations.
Tests were carried out at 2—9 normal stress levels on interface areas of 60 mm]
60 mm (for comparison with the direct shear box test of the same size),
200 mm]200 mm and 300 mm]300 mm. Dead weights of lead were also specially
cast each weighing 196.2 N to produce normal stresses in the range of about 5—50 kPa
except for the 300 mm]300 mm size for which the stress range of up to 25 kPa was
possible due to the maximum load capacity of the apparatus.
The geomembrane sample of the required size was glued to the bottom face of the
dead weight blocks with size adapters when necessary. Weights were secured to each
 M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219 211

Fig. 2. Inclined board test apparatus.

other tightly to prevent sliding and precautions were taken to minimize the overturn-
ing moment on the block by installing an extension strip of plexiglass to the base plate
under the block of weights in the direction of possible toppling over of the block and
supporting this strip by means of frictionless ball bearings.
The layer of sand placed on the tray had to be a thin layer of about 5 mm thickness
(which is approximately 10 times the average particle size) since for thicker layers the
block tended to sink and tilt to one side and topple over as the board was lifted.
Shibuya et al. (1997), who have conducted a very thorough and comprehensive
laboratory investigation recommended that the size of the opening between the two
halves of the direct shear box apparatus be maintained at a fixed value of approxim-
ately 10—20 times the mean particle diameter of the tested sand in order to produce
shear bands of thicknesses observed in plane strain compression tests and to subject
the thin soil element at the mid-height to a simple shear mode of deformation. Since
the present tests involve sliding of geomembrane over the top of the sand layer, the
thickness of the soil layer used (&5 mm) was regarded as acceptable. To prevent the
sliding of the sand grains on the metal tray surface rather than on the geomembrane —
sand interface, a thick paper with sand particles glued to it was stuck on to the tray.
The friction angle between sand-glued surface and sand (loose or dense) is about the
same as the peak friction angle of the sand at the dense state (Wasti, 1985) which is
much larger than the expected geomembrane—sand interface angle. Similarly sand—
sand interface friction is higher than of geomembrane—sand interface. Direct shear
tests were performed to verify and quantify this result (Izgin, 1997). The normal stress
distribution on the soil—geomembrane interface is assumed to be uniform.

2.2. Direct shear box test apparatus

The standard 60 mm]60 mm direct shear box test apparatus was used to test the
granular soils used as well as the geomembrane—sand interfaces. A wooden block with
212 M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219

the geomembrane sample glued on top is placed in the lower half of the box and the
upper half is filled with the soil. The use of the routine size testing apparatus was
preferred partly on the grounds that comparison with the inclined board tests should
refer to a type of equipment that is available in most laboratories.

2.3. Properties of the soils used

Ottawa sand with well-rounded particles and crushed stone with very angular
particles but having very similar size range and gradation were used in the experi-
ments (Table 1). In the present study it was considered adequate to assess the
loosest and densest states as follows: The minimum density is obtained by pour-
ing oven- dried samples from a small height ((20 mm) into a cylindrical mold
from a funnel having about 25 mm diameter spout. The maximum density is ob-
tained as the sand is ‘rained’ in air from a perforated funnel with low intensity
from large enough height found by trial (1.5 m). The angles of shearing resistance
for these densities were found to be 33 and 38° for the Ottawa sand; 42 and 46°
for the crushed stone in the direct shear box tests for the normal stress range of
25—200 kPa.

2.4. Geomembranes used

Three different brands of HDPE geomembrane were utilized in the experiments and
these will be designated as brand A (Werra Plastic; smooth and rough), B (Faser-
technik-Carbofol; smooth only) and C (GSE rough type only). The symbols S-GM(A)
or R-GM(A) refer to smooth and rough geomembrane of brand (A),(B) and (C) in the
text. The rough brand (C) geomembrane is visibly less rough than the rough brand (A)
geomembrane. The properties of the geomembranes supplied by the manufacturers
are given in Table 2.

Table 1
Properties of granular soils used in the tests

Ottawa sand Crushed stone

Effective size, D (mm) 0.40 0.40

10
D (mm) 0.43 0.43
15
D (mm) 0.48 0.52
30
D (mm) 0.55 0.77
60
D (mm) 0.67 0.88
85
Uniformity coefficient, C 1.375 1.75
6
Coefficient of curvature, C 1.05 0.97
#
Unified Soil Classification SP SP
Specific gravity of solids, G 2.65 2.94
4
Minimum void ratio, e 0.594 1.1
.*/
Maximum void ratio, e 0.820 1.226
.!9
 M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219 213

Table 2
Properties of geomembranes used in the tests

Property Unit ASTM DIN GM GM GM

test method test method (A) (B) (C)

Thickness mm D 1593 53353 1.5 2.5 1.5

Density g/cm3 D 792 53479 0.946! 0.942 0.944
Tensile properties D638 53455
(each direction)
Strength at break N/mm2 31.4 27 35
Strength at yield N/mm2 17.6 *15 17
Elongation at Break % l "50 mm *750
0
l "64 mm 511" 700" *600
0
Elongation at yield % l "50 mm 11
0
l "33 mm 6.3" *10" 15
0
Tear resistance N D 1004 53515 147 *125 200
Puncture resistance N 53373 1658 — 400#
Carbon black content % D 1603 — 2—3
Carbon black dispersion D 3015 — A1/A2
Dimensional stability % D 1204 53377,
(each direction) 120°C*, 1 h )1 $2
Environmental stress h D 1693 '1000 — '1500
cracking
Tensile impact strength mJ/mm2 53448 — 800
Low temperature °C D 746, — — (!84
brittleness (by impact) Cond.B
Low temperature 53361 — — No break
brittleness at !20°C
Oxidative induction min ASTM D — — 100
time 3895, 200°C
Pure O , 1atm
2
Ozone resistance ASTM D — — No cracks
1149,7 days
100 ppm
Water absorption wt %. 53521 — — )0.01
change
Moisture vapor g/m2 d ASTM E 96 — — )0.001
transmission
Melt flow index g/10 min D 1238 190/2 53735 190/2,16 — — )1.0
Melt flow index g/10 min D 1238 190/5 53735 190/5 — 2.4 )3.0

!ASTM D 1505.
"ASTM D 638.
#FTMS 101, Method 2065.

2.5. Some notes on the tests

Preliminary inclined board tests performed at rising speeds of 1, 1.65 and 6.5° per
second indicated no significant effect of speed and therefore, the value of 6.5° per
second was adopted. Similarly, the effect of the density of the very thin (about 5 mm
214 M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219

thick) granular soil layer was found to be very small in the inclined board tests with
60 mm]60 mm interface area. For this reason tests with larger interface areas were
carried out on dense soil layer only, and the presented results are for the dense state
which was achieved by tamping the loosely placed soil.
Tests were repeated two or three times and the average is reported. Very similar
results were obtained in the repeated tests. A fresh geomembrane sample was used in
each test. Out of a total of 520 tests the results of 296 inclined board tests have been
considered as relevant to the present study.
The details of the experimental work can be found in Izgin (1997).

3. Experimental results

3.1. Inclined board test results

A typical result of the inclined board tests is given in Fig. 3. The lines are drawn
after linear regression analysis. The adhesion intercept which appears on some the
envelopes is very small (&0.5 kPa). In Table 3, the inclination of the best fitting line
has been reported as the interface friction angle and the small adhesion value is
neglected. The actual measured sliding angles (i.e. the slope of the inclined board at
which the block slides) at the smallest and the largest normal stresses used in the
respective test series are also given in the table in parentheses for comparison. The first
value which represents the result for the smallest normal stress is the same or
a maximum of 3° higher than that found for the highest normal stress except in one
case.
The examination of the data given in Table 3 shows that d values are higher for
rough or textured geomembrane surfaces and for angular particles as expected. The

Fig. 3. Inclined board test results: Ottawa sand—geomembrane interface 60 mm]60 mm area.
Table 3
Inclined board test results

Ottawa sand Crushed stone

Interface Smooth Smooth Rough Rough Smooth Smooth Rough Rough

area (mm2) GM (A) GM (B) GM (A) GM (C) GM (A) GM (B) GM (A) GM (C)

60]60 16.50(!) 18.5° 28° 24° 23.5° 24° 30° 27.5°

(18°, 17°)(") (20°, 19°) (29°, 28°) (26°, 24°) (26°, 23°) (25°, 24°) (33°, 30°) (31°, 28°)
200]200 18.5° — — 26° 26.5° — — 30.5°
(21°, 19°) (26°, 26°) (26°, 26°) — — (32°, 31°)
300]300 19° — — 26° 25° — — 31.5°
(20°, 19°) (26°, 26°) (25°, 25°) — — (31°, 32°)

!Slope of the straight line envelope.

"Values in parentheses are the inclination of the board at which sliding occurs at the lowest and the highest normal stresses used.
M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219
215
216 M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219

experimental values are quite consistent except one (i.e. the value correspond-
ing to crushed stone-smooth GM(A) and 300 mm]300 mm area) which appears
to have been measured a few degrees too low. The effect of the interface area
size can be seen by comparing d values for smooth GM(A) and rough GM(C)—soil
interfaces which have been tested using all three sizes. The average increase in d is
about 3° as the size is increased from 60 mm square to 300 mm square. The values
found for 200 mm and 300 mm square areas seemed to be closer to each other.
A minimum of 200 mm square is considered to be the appropriate area for similar
materials. In Fig. 4 the results from Table 3 for smooth GM(A) and rough GM(C)
for both soils and all three interface areas are plotted, showing the effect of interface
area size as well as increased surface roughness of the geomembrane and particle
roundness.
It is seen from Table 3 and Fig. 4 that d for smooth geomembrane—angular sand
interface is about the same as rough (even comparatively less rough in this case)
geomembrane—round sand interface. Also it is interesting to note that d angles of sand
with rounded particles are much more sensitive to changes in surface roughness and
even to the changes in the brand of smooth geomembrane compared to the angular
sand. For example d for smooth GM(A) and rough GM(C) differs by 7°—7.5° for the
Ottawa sand, whereas this is 40 for the crushed stone. Similarly, smooth GM(A) and
GM(B) gives 20 difference for Ottawa sand and only 0.5° for the crushed stone; rough
GM(A) and GM(C) resulted in 4° difference in d values for the Ottawa sand and 2.5°
for the crushed stone.

Fig. 4. Interface friction angle versus interface area relationship in inclined board tests.
Table 4
Direct shear box and Inclined board test results (both 60 mm]60 mm interface area)!

Interface Interface friction angle (d)! and adhesion (a)!

Difference
Direct shear box Inclined Board (*d°)

d"22° a"2.76 kPa d"16.5° a&0 kPa 5.5°

Ottawa sand/ p"14 kPa p"25 kPa p"50 kPa p"100 kPa p"200 kPa p&14 kPa p&25 kPa p&50 kPa
S-GM (A) 31°(2)" 26.6° 24.2° 23.7 23° 17° 17° 17°
d"32° a"5.0 kPa d"24°° a"0.5 kPa 8°
Ottawa sand/ p"14 kPa p"25 kPa p"50 kPa p"100 kPa p"200 kPa p&14 kPa p&25 kPa p&50 kPa
R-GM (C) 44.5° 36.9° 33° 31.5 30.2 26° 24° 24°
d"31° a"4.25 kPa d"23.5° a"0.5 kPa 7.5°
Ottawa stone/ p"14 kPa p"25 kPa p"50 kPa p"100 kPa p"200 kPa p&14 kPa p&25 kPa p&50 kPa
S-GM (A) 42.1° 38.7° 35.4° 34° 33.3° 25° 25° 23°
d"37° a"2.89 kPa d"27.5° a"0.5 kPa 9.5°
Ottawa stone/ p"14 kPa p"25 kPa p"50 kPa p"100 kPa p"200 kPa p&14 kPa p&25 kPa p&50 kPa
R-GM(C) 43.8° 40.7° 39.4° 38.7° 38.2° 30° 29° 28°

!From straight-line envelopes.

"d values at indicated normal stresses assuming adhesion is zero.
M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219
217
218 M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219

3.2. Direct shear box test results

The interface shear strength parameters ‘d’ and ‘a’ obtained by fitting a straight line
through the plots of interface shear strength q versus the applied normal stress p used
in the direct shear tests, and d values calculated at each normal stress assuming
adhesion to be zero, are presented in Table 4. The results from the inclined board tests
with the same size interface are also given in Table 4 for comparison.
It is seen that the direct shear tests yield about 5—10° higher d values as well as
higher adhesions. These differences are a few degrees less if the results of the inclined
board tests with larger interface areas are used as a basis. A greater degree of
improvement in interface friction as a result of increased surface roughness in the case
of sand with round particles as opposed to angular particles is seen in the direct shear
test results as well.
The interface friction angle values from the shear box tests decrease as the
normal stresses increase, suggesting a curved shape for the q vs p relationship. Plots
of d vs p (not given) show that the drop in d is more significant up to 50 kPa and
the difference in d determined from direct shear box and inclined board tests is
quite large for lower normal stresses. The examination of the relationships bet-
ween change in specimen thickness and shear displacement in the shear box tests
indicates some tendency for dilation of the sand at the interface under low normal
stresses. But dilation does not seem to fully account for the observed behaviour
since the direct shear test results were still higher in cases where dilation is not
detected. It is believed that intrinsic drawbacks of direct shear box testing as described
in Shibuya et al. (1997) and different mode of deformation in the inclined board tests
contribute to the disagreement between the d values obtained from these two types of
testing.

4. Conclusions

Based on the results of the inclined board tests in the normal stress range of about
5—50 kPa and shear box tests carried out in the standard (60 mm]60 mm) direct
shear apparatus with the applied normal stresses from 14—200 kPa, the following
conclusions can be drawn regarding the HDPE geomembrane—sand interface shear
strength:

1. The interface shear strength vs. normal stress relationships from inclined
board tests can be represented by a straight line relationship through the
origin. The direct shear box tests, on the other hand, produce envelopes with
adhesion intercept, as well as higher interface friction angles. Since the mechanism
of sliding in the inclined board tests appears to be more realistic, direct shear box
tests probably give unconservative assessment of interface shear strength. When
lower normal stresses are used with the standard small size direct shear box to
match those applied in the inclined board tests, the discrepancy is found to
increase.
 M. Izgin, Y. Wasti/Geotextiles and Geomembranes 16 (1998) 207—219 219

2. Interface friction of sand with rounded particles is more sensitive to the degree of
roughness of the geomembrane and even to the difference in the brands of the same
general type of geomembrane when compared to sand with angular grains.

References

Girard, H., Fisher, S., Alonso, E., 1990. Problems of friction posed by the use of geomembranes on dam
slopes — Examples and measurements. Geotextiles and Geomembranes 9 (2), 129—143.
Giroud, J.P., Swan, R.H., Richer, P.J., Spooner, P.R., 1990. Geosynthetic landfill cap: Laboratory and fill
tests, design and construction. Proceedings of the 4th International Conference on Geotextiles,
Geomembranes and Related Products, Balkema, Rotterdam, pp. 493—498.
Izgin,M., 1997. Geomembrane—sand interface friction. M.Sc. Thesis, Middle East Technical University,
Ankara, Turkey.
Koutsourais, M.M., Sprague, C.J., Pucetas, R.C., 1991. Interfacial friction study of cap and liner compo-
nents for landfill design. Geotextiles and Geomembranes, 10 (5—6), 531—548.
Shibuya, S., Mitachi, T., Tamate, S., 1997. Interpretation of direct shear box testing of sands as quasi-simple
shear. Geotechnique 47 (4), 767—790.
Wasti, Y., 1985. Determination of the soil-reinforcement friction coefficient. METU Journal of Pure and
Applied Sciences 18 (33), 405—417.