Estimation of rock mass deformation modulus using a Bayesian approach

The deformation modulus (E-rm) is one of the most important rock mass parameters; however, measuring it directly by in situ tests could be costly, time-consuming and sometimes impossible. For that reason, many models have been proposed to estimate the deformation modulus of rock masses based on aspects such as RMR (or GSI). In this study, we present a Bayesian approach to estimate the deformation modulus based on RMR (or GSI) for rock masses. Unlike traditional methods, our approach can account for model uncertainty. We work with Serafim-Pereira's model, i.e., E-rm = 10(((RMR-10)/40)) and develop a "general" model for rock masses. Such general model is then updated using published and reliable data from in situ tests, hence allowing us to obtain updated models that are more accurate and with less uncertainty. We also show an example of how the predictive distributions of deformation modulus for rock masses based on the general model and the updated model can be employed to estimate the probability of failure of a tunnel project, and how reducing the model uncertainty affects the reliability results.