International Journal of Science, Engineering and Technology Research (IJSETR) Volume 1, Issue 1, July 2012 Soil Compression Index Prediction Model for Clayey Soils Seinn Moh Moh Dway, Daw Aye Aye Thant Abstract— This study presents soil compression index prediction model for clayey soils. As compressibility is the most significant parameter while evaluating the settlement of soil under the load of an infrastructure constructed on that soil mass, compressibility of a soil mass is indicated by soil characteristics like coefficient of compressibility, compression index and coefficient of consolidation. Compressibility index is one of the soil parameters that are required for calculation of foundation settlements, however , the determination of compression index is expensive, cumbersome and time consuming. Therefore, in this study, an attempt has been made to estimate compression index as a function of soil index properties. Soil samples are collected from three locations at 3 feet and 6 feet depths in Mandalay. The soil data used in this investigation is collected to develop the predictive models for estimation of compression index. Linear regression analysis is performed to establish empirical models relating compression index and index properties. Correlation between compression index as dependent variable and various index properties based on test results as independent variables is presented. Coefficient of determination (R2) is used as evaluation criteria to check for the empirical correlation that best fits studied soils. The root mean square error (RMSE) is used as statistical index for comparison of reliability of the different empirical equations. Keywords— Soil Compressibility, Index Properties, Compression Index , Linear Regression Analysis and Coefficient of determination I. INTRODUCTION Soil is one of the most important materials used in construction. The primary reason that civil engineers study the properties of soil is foundations of all structures have to be placed on soil. All structures will be designed depending on soil results. Compressibility is a measure of the relative volume of a fluid or a soil as a response to a pressure. It is the most significant parameter while evaluating the settlement of soil under the load of an infrastructure constructed on that soil mass. Compressibility of a soil mass is its susceptibility to decrease in volume under pressure and is indicated by soil characteristics like coefficient of compressibility, compression index and coefficient of consolidation. Although coefficient of volume compressibility is the most suitable, and most popular, of the compressibility coefficients for the direct calculation of settlement of structures, its variability with Manuscript received Oct 15, 2011. Seinn Moh Moh Dway, Department of Civil Engineering, Mandalay Technological University, (e-mail: seinnmohmohdway@gmail.com). Mandalay, Myanmar, 09-402534706 Aye Aye Thant, Department of Civil Engineering, Mandalay Technological University, Mandalay, Myanmar, 09-2010603 (e-mail: yelinthant 05@gmail.com). confining pressure makes it less useful when quoting typical compressibilities or when correlating compressibility with some other property. For this reason, the compression index of soils is generally preferred as its value does not change with the change in confining pressure for normally consolidated clays. However, the determination of compression index in the labs is a cumbersome and time consuming process. Hence, several attempts have been made in the past to predict the compression index using index properties which are relatively easier to determine and take lesser time to obtain in the laboratory. Therefore, this study reveals establishing empirical correlations in terms of index properties by linear regression analysis. This correlation can also be useful for checking of quality control which needs easy and faster ways. II. LINEAR REGRESSION Linear regression is a statistical tool for the investigation of relationships between dependent variable and independent variables. The methods of linear regression analysis are to describe the relationship between two variables by identifying an equation. The constants in the linear relationships were calculated by utilizing a least-squares approach. The form of the regression equation is commonly written as Y = MX + C (1) where, Y = the dependent variable X = the independent variable M = the slope of the regression line C = Y - axis intercept The strength of the evidence of a linear relationship between two variables is described by the coefficient of correlation, r. This measures has no units. The coefficient of correlation always assumes a value between -1.00 and +1.00. The sign establishes whether the relationship is direct (positive correlation) or inverse( negative correlation). The following table describes for interpreting the strength of either positive or negative correlation. TABLE I STRENGTH OF CORRELATION Value of r Interpretation 0.00-0.25 Little or no relationship 0.30-0.45 Fair relationship 0.50-0.75 Moderate to good relationship 0.80-1.00 Strong relationship to perfect correlation 1 All Rights Reserved © 2012 IJSETR International Journal of Science, Engineering and Technology Research (IJSETR) Volume 1, Issue 1, July 2012 The coefficient of determination (r2) represents the percent of the data that is the closest to the line of the best fit where the highest amount of r2 shows the best relationship between parameters. The root mean square error (RMSE), also called the standard deviation of the errors, is a measure of the typical spread of the data around the regression line. The accuracy of proposed model was also checked by calculating the statistical parameter root mean square error. III. PERFORMED TESTS The capability of soils to bear loading are differ depending on the soil types. Generally, fine-grained soils have a relatively smaller capacity in bearing of load than the coarser grained soil. Hence, fine grained soils therefore have a greater degree of compressibility. In order to get the correlation between compression index and index properties, following tests are performed. A. Water -Content Determination Water-content determination is a routine laboratory test to determine the amount of water present in a quantity of soil in terms of its dry weight. It can be calculated from the following equation. W ω w 100% Ws (2) where, Ww = Weight of water Ws = Weight of dry soil B. Specific Gravity Test Specific gravity is defined as the ratio of the unit weight of a given substance to the unit weight of water. G t Ws (3) G s can be divided into four basis states - solid, semi-solid, plastic and liquid. The moisture content , in percent, at which the transition from solid to semi-solid state takes place, is defined as the shrinkage limit. The moisture content at the point of transition from semi-solid to plastic state is the plastic limit, and from plastic to liquid state is liquid limit. The plastic index is the difference of liquid limit and plastic limit. The following equation is to find the plasticity index. PI = LL - PL (4) E .Free Swell Test Ten grams of oven-dried soil specimens passing No.40 sieve (0.425 mm openings) is placed in the graduated cylinders containing distilled water and kerosene. Sediment volumes are measured after complete sedimentation of specimens in respective fluid. It takes about 24 hours to 120 hours in distilled water. Kerosene is used instead of carbon tetrachloride since it is easily available in Myanmar. Table II shows classification of clays on the basis of their free swell ratio. Calculation; V (5) FSR w Vk where, FSR = Free Swell Ratio Vw = Sediment volume of soil in distilled water in cm3 Vk = Sediment volume of soil in kerosene in cm3 TABLE II CLASSIFICATION OF CLAYS ON THE BASIC OF THEIR FREE SWELL RATIO Free Swell Ratio Clay Type Soil Expansivity ≤ 1.0 Non- swelling Negligible 1.0 – 1.5 Mixture of swelling and non-swelling Low 1.5 – 2.0 2.0 – 4.0 > 4.0 Swelling Swelling Swelling Moderate High Very High Ws W2 - W1 where, Gs = Specific gravity of soil Gt = Specific gravity of water at t, temperature Ws = Weight of dry soil W1 = Weight of bottle plus water plus soil W2 = Weight of bottle plus water C. Grain-Size Analysis of Soil Grain size analysis is the determination of the size range of particles presented in a soil, and can be expressed as a percentage of the total dry weight. Two methods are generally used to find the particle size distribution of soil. Sieve Analysis Hydrometer Analysis Sieve analysis is used for particle sizes larger than 0.075mm in diameter and consists of shaking the soil sample through a set of sieve that has progressively smaller openings. Hydrometer analysis is used for particles sizes smaller than 0.075mm in diameter. Hydrometer analysis is based on the principle of sedimentation of soil grains in water, the particles settle at different velocities, depending in their shape, size, weight and viscosity of the water. D. Atterberg Limit Test Consistency varies with the water content of the soil. The consistency of a soil can range from (dry ) solid to semi-solid to plastic to liquid (wet ). The water contents at which the consistency changes from one state to the next are called consistency limits or Atterberg limits. Hence, on an arbitrary basis depending on the moisture content, the behavior of soil F. Modified Free Swell Index Test Sivapullaiah et al. (1987) suggested a new test method of obtaining a modified free swell index for clays, which appears to give a better indication for the potential of clayey soils. This test begins with an oven-dried soil with a mass of about ten grams. The soil mass is well pulverized and transferred into a 100 ml graduated jar containing distilled water. After 24 hours, the swollen sediment volume is measured. Table III shows the soil classification on modified free swell index. The modified free swell index is then calculated; Modified free swell index = V Vs Vs (6) where, V = Soil volume after swelling Vs = Volume of soil solid Ws Gs = Weight of oven-dried soil = Specific gravity of soil solid = Unit weight of water γw = Ws Gsγw 2 All Rights Reserved © 2012 IJSETR International Journal of Science, Engineering and Technology Research (IJSETR) Volume 1, Issue 1, July 2012 TABLE III SOIL CLASSIFICATION SCHEME BASED ON MODIFIED FREE SWELL INDEX Modified Free Swell Index < 2.5 2.5 to 10 10 to 20 >20 Swelling Index Negligible Moderate High Very High G. Standard Proctor Compaction Test In the Proctor compaction test, the soil is compacted in a mold that has a volume of 1/30 ft³. The diameter of the mold is 4 in. During the laboratory test, the mold is attached to a base plate at the bottom and to an extension at the top. The soil is mixed with varying amounts of water and then compacted in three equal layers. The hammer weighs 5.5 lb and has a drop of 12 in. For a given soil, the process is repeated at least five times, the water content of the samples being increased each time. This test is used for determining optimum moisture content and maximum dry density. H. Consolidation Test Consolidation is the process in which reduction in volume takes place by expulsion of water under long term static loads. It occurs when stress is applied to a soil that causes the soil particles to pack together more tightly, therefore reducing its bulk volume. When this occurs in a soil that is saturated with water, water will be squeezed out of the soil. In the Classical Method, developed by Terzaghi, soils are tested with an oedometer test to determine their compression index. I. Compression Index (Cc ) The compression index is useful for the determination of the settlement in the field. The compression index, Cc, is equal to the slope of the linear portion of the void ratio(e) versus log pressure (log p) plot or vertical strain ( v ) versus vertical applied stress (log p). Thus compression index can be calculated as follow: e e (7) Cc 2 1 P2 log P1 But, Δε Δe1 (8) 1 e (9) ε 2 ε1 Cc log P2 P1 (1 e 0 ) Mandalay. For the performed tests, disturbed samples collected from three locations at 3 feet and 6 feet depths in Mandalay are used to know the engineering properties of studied soil. These locations are denoted as location A, B and C. A. Specific Gravity Test Specific gravity is determined using equation 3 and results are described in Table IV. TABLE IV TEST RESULTS OF SPECIFIC GRAVITY TEST Location ɛ Compression index Void ratio at mid height = e2 - e1 Void ratio at applied load Initial void ratio Effective overburden pressure Vertical strain (%) IV. TEST RESULTS FOR THE STUDIED SOIL All tests were made at Soil, Concrete Laboratory and Irrigation Technology Centre, Patheingyi Township, 2.68 6 2.69 3 2.73 6 2.73 3 2.54 6 2.73 C B. Grain Size Analysis Sieve analysis and hydrometer test are performed for grain size distribution. Results are shown in Table V. Sieve analysis gives the intermediate dimensions of a particle; hydrometer analysis gives the diameter of an equivalent sphere that would settle at the same rate as the soil particle. TABLE V RESULTS FOR GRAIN SIZE ANALYSIS Sample Location A Location B Location C 3 feet 6 feet 3 feet 6 feet 3 feet Gravel (%) 0 0 0 0 0 0 Sand (%) 13.6 10 10.3 7.7 22.2 18.2 6 feet Silt (%) 64.5 54 55.7 61.0 49.8 36.8 Clay (%) 21.9 36 34 31.3 28.0 45.0 F200 89.6 91.2 93.6 93 80.2 85.4 R 2.2 2.4 2 4.6 9.6 7.6 F 89.6 91.2 93.6 93 80.2 85.4 R 2.2 2.4 2 4.6 9.6 7.6 F 100 100 100 100 100 100 R (GF) 0 0 0 0 0 0 2.2 2.4 2 4.6 9.6 7.6 200 200 200 SF=R = = = = = = 3 B 4 Cc Δe e e0 p0 Specific Gravity (Gs) A 4 where, Sample Depth (ft) R 200 - 4 C. Atterberg Limit Test Results For describing the limit consistency of fine-grained soils on the basis of moisture content, three basic limits are determined. These limits are calculated from liquid limit test, plastic limit test and shrinkage limit test. The plasticity index indicates the range of consistency within which a soil exhibits plastic properties. Atterberg limit test results are described in Table VI. 3 All Rights Reserved © 2012 IJSETR International Journal of Science, Engineering and Technology Research (IJSETR) Volume 1, Issue 1, July 2012 TABLE VI ATTERBERG LIMIT TEST RESULTS Location A B F .Modified Free Swell Index Test This method is for obtaining a modified free swell index for clays, which appears to give a better indication for the potential of clayey soil. Results are shown in Table IX. Sample Depth (ft) 3 6 3 PL LL SL PI 16.8 21.3 21.5 54.7 70.1 58.2 13.1 17.8 17.4 37.9 48.8 36.8 6 16.0 41.0 - 25.0 3 18.0 43.7 - 25.7 6 17.6 50.0 12.7 32.4 C Atterberg Limits (%) D .Classification of Soil According to USCS The Unified Soil Classification System is based on the grain size distribution, plasticity index and liquid limit of soil. The studied soil can be classified as shown in Table VII according to USCS. TABLE VII CLASSIFICATION OF SOIL TYPES FOR DIFFERENT LOCATIONS Location Depth (ft) 3 6 3 6 3 6 A B C Group Symbol CH CH CH CL CL CH Group Name Fat Clay Fat Clay Fat Clay Lean Clay Lean Clay Fat Clay E. Free Swell Test Results The free swell test is one of the most commonly used simple tests in the field of geotechnical engineering for getting an estimate of soil swelling potential. Results for location A,B and C are shown in Table VIII. TABLE VIII SOIL CLASSIFICATION SCHEME BASE ON FREE SWELL RATIO Location 3 feet Free Swell Index 1.16 A 6 feet 3 feet 1.6 1.2 B 6 feet 3 feet 1.1 1.1 C 6 feet 1.2 Clay Type Mixture of swelling and non-swelling swelling Mixture of swelling and non-swelling Mixture of swelling and non-swelling Mixture of swelling and non-swelling Mixture of swelling and non-swelling Soil Expansivity TABLE IX SOIL CLASSIFICATION SCHEME BASED ON MODIFIED FREE SWELL INDEX Modified Free Swell Index Location Swelling Potential 3 feet 2.89 Moderate 6 feet 3.37 Moderate 3 feet 3.03 Moderate 6 feet 2.96 Moderate 3 feet 2.55 Moderate 6 feet 2.96 Moderate A B C G. Compaction Test Results The Standard Proctor test is used for determining optimum moisture content and maximum dry density. Table X shows compaction test results. TABLE X COMPACTION TEST RESULTS Samples Optimum Moisture Content (%) Dry Unit Weight (lb/ft3) Location A 3 ft 6 ft Location B 3 ft 6 ft Location C 3 ft 6 19.3 20.6 18.5 19.2 18.9 19.5 107.7 105 106.7 109.2 104.1 108.5 ft H. Consolidation Test Results Consolidation is the process whereby soil volume decreases under the application of a load. Consolidation is caused by loads being applied to soil and the grains of being packed together more closely as a result. From this test, compression indices are described in Table XI. TABLE XI COMPRESSIBILITY INDICES OF DIFFERENT LOCATIONS Low Moderate Low Location Depth (ft) Cc Cs eo 3 0.451 0.088 1.356 6 0.396 0.108 1.365 3 0.298 0.060 1.400 6 0.323 0.036 0.632 3 0.320 0.058 0.565 6 0.296 0.066 0.713 A Low B Low Low C 4 All Rights Reserved © 2012 IJSETR International Journal of Science, Engineering and Technology Research (IJSETR) Volume 1, Issue 1, July 2012 V. CORRELATION BETWEEN COMPRESSION INDEX AND INDEX PROPERTIES It is well-known that compression index is related to different index properties of the soils. By using linear regression, analysis and validation for the studied soil are done and empirical equations for prediction model of compression index are developed and described in equation 10-14. The correlation between compression index and soil index properties are evaluated. These indices are liquid limit, plasticity index, natural water content and initial void ratio. A linear correlation using liquid limit is done with seven data sets to get higher correlation. Cc = 0.0027 LL + 0.1994 (10) The following proposed models are done with six data sets obtained from studied soil test results. Cc =0.0038 PI + 0.22 (11) Cc = 0.01 ωn +0.027 (12) Cc =0.196 eo + 0.207 (13) Cc = 0.52 - 0.03 LL + 0.03 PI (14) The statistical indices for the above equations are shown in Table XII. TABLE XII STATISTICAL INDICES Independent variables Correlation coefficient, r Coefficient of determination,r2 LL PI 0.50 0.550 0.701 0.75 0.923 0.250 0.303 0.491 0.563 0.852 ωn eo LL, PI permission and supervision. The author would like to thank Daw Khin Thida and her staff members of Soil, Concrete Laboratory and Irrigation Construction Quality Control, Chaung Woon, Patheingyi, for their kind help during the study of this paper. Finally, the author would like to thank to her parents and family members for their supports and encouragements. REFERENCES [1] [2] [3] [4] [5] Slamet Widodo, Abdelazim Ibrahim, Estimation of Compression Index(Cc) Using Physical Properties of Pontianak Soft Clay, Gustav Strasse 1,09599 Freiberg (SN), Germany Nader Abbas, Akbar A Javadi and Reza Bahramloo, Empirical Prediction of Compression Behaviour of Normally Consolidated Fine-Grained Soils, Iranian Agricultural Engineering Research Institute (IAERI), Geotechnical Engineering, School of Engineering, Computing and Mathematics, University of Exerter, Exerter EX4.40F, Hamadan Agricultural and Natural Resource Center Azzous, A.S.,Krizek, R.J, and Corotis,R.B., Regression analysis of Soil Compressibility, Soils and Foundations,16 (2),19-29,1976 Braja M. Das, : Principal of Geotechnical Engineering, Fourth Edition.Boston. U.S.A : PWS Publishing Company (1998) Braja M. Das, : Principal of Geotechnical Engineering,Fifth Edition.California State University, Sacramento, PWS Publishing Company (2006) Root mean square error, RMSE 0.048 0.047 0.084 0.038 0.030 VI. DISCUSSION AND CONCLUSIONS Compression index and index properties are used to conduct a statistical study to determine correlations. Regression analysis is carried out to develop the predictive models for estimation of compression index (Cc).In this study, single and multiple variable empirical equations are developed for estimation of the compression index. Linear regression analysis is performed using variables that obtained from the test results, individually to develop single variable equations. Also, two variable equation is developed utilizing combinations of a representative parameter, liquid limit and plasticity index. Equation-14 with the least RMSE and the largest r2 is a reliable model for prediction of the compression index. The values of RMSE (0.038) and r 2(0.563) show the equation-13 that it has good relationship. The other equations with LL, PI and ωn cannot be used to accurately predict the compression index in the study area. ACKNOWLEDGMENT Firstly, the author would like to thank Dr. Myint Thein, Rector of Mandalay Technological University for her motivation, supports and guidance. The author is particularly grateful to Dr. Kyaw Moe Aung, Associate Professor and Head of Civil Engineering Department of Mandalay Technological University, for his motivation, supports and guidance. The author would like to express her heartfelt gratitude to her supervisor Daw Aye Aye Thant, Lecturer of Mandalay Technological University, for her kind advice, 5 All Rights Reserved © 2012 IJSETR