Guidelines for Use of Tensile Data in the Calculation of the Hoekbrown Constant mi

"The intact rock constant, mi, is one of the three basic input parameters required for the widelyused Hoek-Brown failure criterion. Values for the constant should ideally be determined by statistical analysis of laboratory test data. In this paper the influence of the inclusion or exclusion of tensile test data on derived values is discussed. We firstly investigate the relationship between direct tensile strength (DTS) test values and Brazilian tensile strength (BTS) test values based on a literature search, finding that the correlation between the two test types is DTS = 0.9*BTS, independent of the lithology being tested. Secondly, using the shear strength database of Douglas (2002), we examine the influence of confining pressure range and the inclusion or exclusion of DTS and BTS test values on calculated values of mi.Chief amongst the influences was the exclusion of tensile data in the calculation procedure, particularly when the Hoek triaxial confining stress range requirement (?3?max = 0.5?c) was not satisfied. Based on a reasonable 1:1 relationship between R (the ratio of unconfined compressive to tensile strength), use of actual tensile data in the calculations for the Hoek brown failure criterion is considered to be a simpler and more robust approach to the use of a tensile cutoff based on empirical criteria such as that of Fairhurst (1964). We advocate the fitting a failure curve using compressive and tensile test data as the most reliable means of calculating mi, while at the same time satisfying requirements as originally advocated by Hoek relating to the appropriate triaxial confining stress range (?3?max = 0.5?c) and sufficient (>5) evenlyspread data points. In the absence of triaxial data the use of R provides a practical means of obtaining estimates of mi that are more robust than guideline values."