Empirical Estimation of Unconfined Compressive Strength and Modulus of Elasticity Using ANN

Abstract

The strength parameters such as unconfined compressive strength (UCS) and Modulus of Elasticity (E) of rocks are important for design of foundations. Both the parameters are determined in laboratory after rigorous and destructive testing. In this study Artificial Neural Network (ANN) models are developed for prediction of UCS and E from index test parameters such as Unit Weight (γ), porosity (n) and point load index Is(50). Multi variable regression models are also developed to compare the accuracy of prediction from different models. Coefficient of determination (R2 ), Root Mean Squared error (RMSE) and Standard Error of Estimate (SEE) has been used as the controlling factor to determine the prediction accuracy of both ANN and multivariable regression. The ANN models increased the R2 values from 0.53 to 0.72 and 0.51 to 0.75 for UCS and E respectively. The variation between experimental and predicted values of UCS and E for ANN model are ± 23% and ± 29% and for regression model are ± 40% and ± 31% respectively.