PREDICTIVE ROCK MASS CLASSIFICATION PARAMETER MODELS FROM ROCK MATERIAL ENGINEERING TESTS

Abstract

Rock mass classification is one of the most widely used method for assessment of rock slope which is developed from empirical case studies. One of the most widely used are the Rock Mass Rating (RMR). Most of the rock mass classifications were developed through discrete rating of parameters, meaning that a high subjectivity from the practitioner is involved in assigned rating of the parameters. Thus, continuous functions were developed in order to reduce the subjectivity when assigning values to the parameters. This study attempts to develop predictive model of selected parameters from several available continuous function in the RMR rock mass classification, using input of laboratory test of engineering properties of weathered granite. Multiple regression analysis indicated several of the physical properties of weathered granite are statistically significant in predicting the parameters of strength of intact rock material and intact rock alterability. These prediction models are validated through graphical plotting of calculated continuous function RMR using measured field and laboratory test value for the parameter against calculated continuous function RMR using prediction model for parameter value. Several of the continuous function RMR show no statistical difference in mean value between the two values of continuous function RMR, indicating that calculation using prediction models for the parameter are closely similar to calculation using actual field and laboratory measurement of the parameter. Regression analysis are carried out for the rest of the continuous function RMR to provide correction factor to use these prediction models as parameters in the continuous function RMR.