A New Model for Estimating the Resilient Modulus of Silt-Clay Subgrade Soils

A major concern in design of pavements is to determine subgrade soil in terms of Resilient Modulus (MR). The main objective of this paper is to utilize a novel and powerful branch of the Computational Intelligence (CI) techniques, namely Linear Genetic Programming (LGP), to develop a nonlinear and efficient model for estimating resilient modulus (MR). The LGP model is formulated in terms of soil index properties including percentage of soil particles passing through #200 sieve, liquid limit, plasticity index, percentage of optimum moisture content, percentage of moisture content, degree of saturation, unconfined compressive strength, confining stress and deviator stress. The database used for developing the LGP model is taken from test results conducted on cohesive Ohio soils including silt-clay (A-4) type of soils based on AASHTO classification, in the literature. The prediction ability of the model has been assessed using a part of laboratory data that hasn’t been used in calibration process (test data). The LGP model generates a practical equation with good prediction ability. study is performed for all variables

Keywords:

Pavement , Subgrade Soils , Resilient Modulus , Prediction , Linear Genetic Programming