Abstract

Most of the empirical approaches used today were proposed 40 years ago, and their updates do not include most of the new understanding of mining geomechanics acquired since then. Even though numerical modeling is used today for designing purposes, rock mass characterization is still based on the input parameters required by traditional rock mass classification methods without considering their limitations. In recent years, data acquisition methods have improved the quality of geotechnical data and machine learning algorithms have become widely used. However, it is important to note that geomechanics is a data limited problem and that empirical methods are still widely used, which indicates that the relation between the amount of data and decision support has not changed due to limitations in geotechnical methodologies. This paper discusses the limitations related to the identification of critical parameters influencing the rock mass behavior, alternatives to characterize the geology and the options to define weightings for critical parameters. Additionally, a regression that reduces the multidimensionality of a problem is presented to have a better visualization of the problem. The use of traditional statistical techniques showed to be useful to highlight which parameters have a more significant effect on the final response and, therefore, where the efforts in rock mass characterization should be directed.

This content is only available via PDF.
You can access this article if you purchase or spend a download.