EFFECT OF ASPECT RATIO ON THE FREEZING-THAWING OF A CH CLAY

The geotechnical properties of fine grained soils which depends on the variation of aspect ratio (ξ=length/diameter) exposed to freezing-thawing is not known clearly. The effect of aspect ratio on unconfined compression strength and freezing-thawing resistance of fine grained soils after 0, 5, 10, and 20 freezing-thawing cycles have been investigated in this study. The freezing-thawing test was conducted in an automatic freezing-thawing cabinet, the freezing and thawing temperatures of which were selected as -20°C and +25°C, respectively. The unconfined compression test was conducted in an unconfined compression testing apparatus, the loading ratio of which was 0.8 mm/min. These values were held constant throughout the test. Tests were conducted on cylindrical samples with different ξ (length/diameter) ratios prepared at standard Proctor energy. The test results indicated that, as ξ ratio decrease and the number of freezing-thawing cycles increase, unconfined compression strength decrease and freezing-thawing grain loss increase.


INTRODUCTION
Foundation soils are exposed to freezing-thawing at least once a year in cold-climate regions.This phenomenon affects some engineering features of fine-grained soils, such as water content, bearing capacity, and permeability adversely.It is mentioned in the following references that freeze-thawing causes structural and volumetric changes in clayey soils (Wang et al., 2017;Wang et al., 2018;Xu et al., 2018;Lu et al., 2018).Andersland and Ladanyi (2004) indicated that freezing, thawing, and redistribution of water in the ground during seasonal temperature changes are responsible for variations in soil properties and the behavior of foundation materials.
The aspect ratios used in these studies vary depending on either the preferred standards or particular preferences (Yazıcı and Sezer, 2007;Yarbaşı et al., 2007;An et al., 2008;Del Viso et al., 2008;Zaimoğlu, 2010;Hazırbaba and Güllü, 2010;Ghazavi and Roustaie, 2010).The aspect ratio is supposed to affect some of the post-experimental determined geotechnical properties of soils.It has been stated in literature that, some of the determined engineering properties of concrete samples, such as unconfined compression resistance and durability, were significantly affected by aspect ratio (Yazıcı and Sezer, 2007;An et al., 2008;Del Viso et al., 2008).Additionally, Yalçın (2013) mentioned that the strength of sample decrease with increasing specimen size for the geometrically similar tested concrete samples.Yavuz et al., (2001) investigated the influence of sample size on the uniaxial compressive strength on the cubic shaped samples of different size for this purpose they used white marble and red travertine.Researchers reported that increasing specimen size is to cause significantly reduction of uniaxial compressive strength of rock and coal specimens.They are attributed this situation some studies (Hoek and Brown, 1980;Bieniawski, 1968;Swain, 1968;Kaczynski, 1986).Koç and Şener (2004) pointed out that stress decreases as the size increases on elements have a geometrically same but different size.The another study is conducted on concrete specimens of different sizes in Felekoğlu and Türkel (2005).Researchers stated "Due to the size effect, the relative strength of specimens varies at different dimensions" in their studies.Gök (2010) implied that the compressive strength increases while the sample size decreases.The researcher explained this increment in two theme: (1) Concrete surface in contact with the heads of the test press in small sample size is smaller than big ones.(2) Small-sized concrete samples' cross sectional areas and volumes are also naturally small.The extension of the sample size, statistically, increases micro-cracks or the amount of other faulty parts of the samples.
The variation pattern of freezing-thawing behavior under different soil and test conditions depending on the aspect ratio is not known clearly.There are scarcely any studies in literature researching the effects of aspect ratio used in soil freezing-thawing tests on the engineering properties of soils.Since decisions to be made at the field applications depends on the outcomes of laboratory tests, utmost attention should be given to determining aspect ratio to be tested in laboratories.The effect of aspect ratio on the unconfined compression strength and freezing-thawing resistance of fine grained soils after 0, 5, 10, and 20 freezing-thawing cycles have been investigated in this study.Tests were conducted on cylindrical samples with different ξ ratios prepared at standard Proctor energy.Stressstrain, freezing-thawing grain loss, and unconfined compression strength of samples with different ξ ratios were determined at the end of each freezing-thawing cycle.Test results were discussed comparatively.

Experimental Study
The clay soil utilized in the study was obtained from the Oltu-Erzurum vicinity in the Eastern Anatolia Region of Turkey.In this region, there is a long winter and snow remains on the ground from November until the end of April.From the data obtained at a station in Erzurum between 1988 and 2005, the long term mean temperature is about 5.1 o C, daily temperature average is 15.0 o C, the highest temperature measured so far is 35.6 o C and the lowest temperature is -37.2 o C (Toy et al., 2007).This soil resembles an area exposed to freezing-thawing and is used a great deal in engineering work in Erzurum.The fine-grained soil used in tests was sieved from 200 No (0,074mm) and kept in a stove at 105 ± 5°C for 24 hours.The soil was classified as high plasticity clay (CH) based on the Unified Soil Classification System (USCS).Some of the engineering properties of the clay soil are given in Table 1.The standard compaction test was conducted in accordance with ASTM D 698-78.The samples with specified optimum water content and maximum dry unit weight were prepared in different compaction molds under constant standard compaction energy in the following aspect ratios: 38mm x 76mm (ξ1= 2.0), 50mm x 100mm (ξ2= 2.0), 101.5mm x 116mm (ξ3 ≈ 1.14, standard compaction mold), and 151.7mm x 127.4mm (ξ4 ≈ 0.84, CBR test mold).Aspect ratios and some properties are given in Table 2.For CH clay the relationship between the water content and the dry unit weight is illustrated in Figure 1 (a).Also, the samples prepared at different aspect ratios are shown in Figure 1

69
samples were wrapped by aluminum foil sheets throughout the freezing-thawing test as illustrated in Figure 3 to prevent any changes in their water content (Kvǽrnǿ and Ǿygarden, 2006).Furthermore, small amounts of Vaseline were smeared on foil sheets to prevent aluminum foil from adhering to samples (Qi et al., 2008;Güllü and Hazırbaba, 2010).In the literature, there are numerous studies of freeze-thaw tests on fine-grained soils with various numbers of cycles, temperatures, and time intervals (Liu et al., 2010;Ghazavi and Roustaie, 2010;Hazırbaba et al., 2011).The number of cycles were selected as 0, 5, 10, and 20 in this study.The freezing and thawing temperatures were selected as -20 o C and 25 o C, respectively, for each cycle and the waiting time for each temperature was selected as 6 hours (Ghazavi and Roustaie, 2010).The samples on which a freezing-thawing cycle completed were taken out of the freezing-thawing cabinet so as to determine their freezing-thawing grain loss.Residuals on samples were removed by an appropriate brush and then their weights were measured.
The unconfined compression test in accordance with ASTM D 2166 was conducted end of each prescribed number of cycles on samples with different ξ ratios.Stress and strain values of samples were determined by means of a deformation-controlled unconfined compression testing apparatus, the loading ratio of which was 0.8 mm/min.Images of some samples with ξ₁, ξ₂, ξ₃, and ξ₄ ratios after the unconfined compression test are shown in Figure 2, while images of excessively deformed samples are provided in Figure 3.The grain losses of the samples after freezing and thawing (FTG) are calculated in percentages as in the following (Taşpolat et al., 2006): FTG (%) = (IW -FTAW) / IW * 100 (1) where: IW = initial weight of samples, and FTAW = weights of samples after freezing and thawing.The fact that the grain losses derived from Formula 1 is minimal indicates that the freezing-thawing resistances of these samples are high.In other words, the lower FTG values the better freezing-thawing resistance.
All tests were repeated on three samples for each number of freezing-thawing cycles for the reliability of the results and the averages of which are used in evaluations.

RESULTS AND DISCUSSION
The unconfined compression strength (UCS) of the tested samples are given in Table 3.The stressstrain curves of the samples having ξ₁, ξ₂, ξ₃, and ξ₄ ratios after each freezing-thawing cycle, where the unconfined compression strength values specified in Table 3 were used, are shown in Figures 4.
Since , some large cracks were observed on samples with ξ₃ and ξ₄ ratios at the end of 10 and 20 freezing-thawing cycles, the stress -strain curves of these samples could be plotted (Figure 3) and also UCS values could not be determined.3 that, the decrease of the ξ ratio generally leads to a decrease in the unconfined compression strength in all freezing-thawing cycles.This decrement is due to arise from pore water pressures in a sample because of its size (Sakamoto and Shogaki, 2003).Likewise, it might be attributed that there is a terminal density (or terminal void ratio) for every soil and every process (Narsilio and Santamarina, 2008).Besides, this decrement is further anticipated to be affected by the change of contact area between the steel plate of the testing apparatus and the surface of samples with different ξ ratios.Yazıcı and Sezer (2007), who conducted compression strength tests on cylindrical concrete samples with different sizes, emphasized that a greater length/diameter ratio leads to less compression strength and stated that this was due to small samples have small surfaces contacting to the steel plate of the testing apparatus.In addition, samples with smaller volume of specimens (i.e., ξ₁, ξ₂) have less micro cracks in comparison with samples with larger volume of specimens (i.e., ξ₃, ξ₄), so that samples with larger ξ ratios may have higher unconfined compression strength.The study conducted by Yazıcı and Sezer (2007) promotes this conclusion.Furthermore, An et al., (2008) and Del Viso et al., (2008) highlighted in their studies on reactive powder concretes with different sizes and high-strength concretes, respectively that, samples with larger sizes have less strength.
The freezing-thawing grain loss-cycle number graphic plotted for samples with ξ₁, ξ₂, ξ₃, and ξ₄ ratios at the end of each freezing-thawing cycle is shown in Figure 5.It is also shown from Figure 5 that freezing-thawing resistances as a percentages of freezing-thawing grain loss of the samples.If there are grain losses of more the 15% in freezing-thawing tests, resistance of soil against freezingthawing are not calculated (Hassini, 1992).Therefore, freezing-thawing grain loss (>15%) in samples with ξ₃ and ξ₄ ratios as a result of 10 and 20 freezing-thawing cycles were not considered.
The fact that decreasing ξ ratio leads to an increase on freezing-thawing grain loss at the end of each freezing-thawing cycle numbers (5, 10, and 20) are shown in Figure 6 (Wang et al., 2007;Ghazavi and Roustaie, 2010).It has been anticipated that, this fact arises from the force of ice created by water within pores throughout the freezing-thawing process, which separates soil grains from each other and consequently increases the pore volume (Wang et al., 2007).
Pmax value, which is the stress value corresponding to 20% of total axial strain, was calculated by multiplying of the cross-sectional areas of the samples for only go to residual curves.The ξ -Pmax graph plotted using unconfined compression strength values are illustrated in Figure 6. Figure 6 indicates that a decrease in ξ ratio generally leads to an increase in Pmax values.This fact is anticipated as arising from the increase in surface areas (Sakamoto and Shogaki, 2003;Yazıcı and Sezer 2007;An et al., 2008;Del Viso et al., 2008).Similarly, despite ξ₁ = ξ₂ = 2.0, since the surface area of the sample with ξ₂ ratio is greater than the sample with ξ₁ ratio, the Pmax value appears to be greater in Figure 6.This result indicates that the surface areas of the soil samples should also be considered in freezing-thawing tests.
It has been observed based on all test results that aspect ratio is effective on some engineering properties of fine-grained soils exposed to freezing-thawing.Therefore, the aspect ratio used for evaluating test results should be taken into consideration.

CONCLUSION
In this study, a series of freezing-thawing tests were performed in a laboratory on clay soils with different aspect ratios.At the end of each freezing-thawing cycle number, the unconfined compression tests were performed on the samples.The following general conclusions were obtained from the tests: • The initial elasticity modulus decreased as ξ ratio decreased.
• A decrease in the ξ ratio led to a decrease in unconfined compression strength in each freezingthawing cycle in generally.• It has been determined that freezing-thawing grain loss increases as the ξ ratio decreases.
• It has been identified that surface area and volume of samples should also be considered in freezing-thawing tests.The variation patterns of freezing-thawing behavior depending on aspect ratios under different soil and test conditions are not identified clearly.Therefore, similar future studies on different soil classes and different ξ ratios would be beneficial for providing solutions to various engineering problems. (b).

Figure 1 .
Figure 1.(a) Relationship between the maximum dry unit weight -water content, (b) Samples with different ξ ratio

Figure 5 .
Figure 5.The variation of freezing-thawing grain losses

Table 1 .
Some properties of the clay used in tests *Obtained from standard Proctor tests.

Table 2 .
Aspects ratios and some properties

Table 3 .
Unconfined compressive strengths of the samples