An improved method of fitting experimental data to the Hoek–Brown failure criterion

In this study, a special fitting technique is proposed to apply the Hoek–Brown failure criterion to available laboratory strength data collected on samples of Ankara andesites. The original method of fitting the Hoek–Brown failure criterion using the spreadsheet and the fitting method utilized in the RocLab software are statistically compared with this new modified nonlinear fitting technique. The Hoek–Brown failure parameters obtained by a simple multiple regression technique, utilizing the solver function within an Excel spreadsheet and the revised least squares procedure significantly improved the fitness of the Hoek–Brown envelope especially within the tensile and high confining stresses. The Brazilian tensile strength is found to be a useful parameter to fix a rock failure criterion at low confining stresses, particularly, in the tensile region. If tensile, uniaxial and triaxial test results are available, when compared to two existing fitting methods the modified multiple regression method is found to predict more precise tensile and compressive strengths, and material constant mi for the Hoek–Brown failure criterion.

► A special fitting technique is proposed to apply the Hoek–Brown failure criterion.
► Tensile data is an essential part of fitting process to provide improved estimates.
► Hoek–Brown envelope fitted by three methods will give dissimilar failure parameters.
► There are special requirements for the treatment of local variability.
► The modified least squares produce accurate estimates for the Hoek–Brown parameters.