Development of an Intelligent and Statistical Model for Prediction of Rock Mass Deformation Modulus

Abstract

The rock mass deformation modulus (Em) takes into account the plastic and the elastic deformation of the rock mass. It has been widely used for designing structures such as dams, tunnels, caverns, and mines, etc. Since the tests available for ascertaining Em are expensive and time consuming, several equations were suggested in the past. However, it has been found that the existing models are limited to specific type of rock mass rendering constraints for its application to general use. Therefore, to cater to this need a comprehensive data set of 147 series, comprising of elastic modulus of intact rock (Ei), rock mass rating (RMR) and Em have been collected from the literature covering various types of rocks from across the globe. This data set has been used in constructing a non- linear regression (NLR) based equation to render ease to the design engineers for calculating Em. In addition to it, an intelligent model is proposed having a basis on artificial neural network (ANN). The credibility of the empirical and intelligent model has been ascertained using the R2 and RMSE. Based on these performance evaluation indices, both models have been found to predict the Em with considerable accuracy vis-a-vis the existing models. Subsequently, a chart has been designed to ascertain Em using Ei and RMR for several classes of rock mass.