Multivariate Regression Analysis in Modelling Geotecthnical Properties of Soils Along Lambata-Minna-Bida Highway

Abstract

In pavement design, three important geotechnical properties- CBR, OMC and MDD are often used to determine the strength of a subgrade layer. To determine these properties in the laboratory is time consuming, laborious, very costly and sometimes infrequently performed due to lack of equipment. The aim of this study is therefore to develop regression models to estimate the strength of properties using relatively easier index properties. Thirty-four soil samples were collected from various locations along Bida-Minna highway between 0.6-1.5 m depths for index, consistency, compaction and CBR tests. Based on the laboratory results, the CBR significantly related with sand, %fines, LL, PL, PI, OMC and MDD parameters. Satisfactory correlations (R2 > 0.59) were found between the three strength properties and other index properties of the experimented soils. Seven best predictive models were developed to estimate the strength properties based on multiple linear regression analysis.