Regression Analysis of Index Properties of Soil as Strength Determinant for California Bearing Ratio (CBR)

Abstract

Investigation of the variation in different soil types and origins is an essential task for Geotechnical engineers. To overcome the effects of this change, the Geotechnical engineers, as well as other professionals, attempted to develop empirical equations unique to a region and soil type to use the soil for its intended purpose. However, these empirical equations are more reliable for the kind of soil where there is the strong relationship of parameters is developed. Hence, it is good practice to develop empirical equations that best fit in the soil available in the location that we can access. On the flexible pavement, subgrade is considered to be an ideal layer to resist wheel load, and its CBR value is considered as the strength measuring a parameter. Conducting CBR test is an expensive and time-consuming procedure. Also, it is challenging to mold the sample at a desired in-situ density in the laboratory. Furthermore, if the available soil is of poor quality, suitable additives are mixed with soil, and resulting strength of soil is assessed by the CBR value which is cumbersome. To overcome such problem, statistical approach such as regression-based models are used. This can be used for quick and easily determined parameters. Relative to this, the study has been conducted to develop an equation to show the relationship between the index properties and CBR values precisely located along WelkiteArekit –Hosanna Road of about 121km stretches. It was carried out using thirty samples retrieved from this road and tested in a laboratory. The test result of the regression based statistical analysis was used to develop the predetermined relationship. The relationship development was performed in the form of an equation of CBR as a function of grain size parameters, Atterberg’s liquid limits and compaction parameters by considering the effect of individual soil properties as well as the effect of a combination of soil properties on the CBR value. Based on the results of the study for both linear and multiple linear regression analyses, it was revealed that there was relatively fair relationship obtained by combining plasticity index, the percentage of fine content and maximum dry density which are strength determinant of fine-grained soils. Also, the results showed that the coefficient of determination for multiple linear regression is R2=0. 731, while for single linear regression is R2=0. 682. Therefore, it is concluded that the index properties of soils are sufficiently accurate in determining the CBR values, of which it can be utilized for preliminary characterization.

Full document