This paper describes how to apply fuzzy inference system (FIS) to estimate the strength of field scale rock masses by judgment and experience of practicing engineers. Three important parameters believed to influence the rock mass behavior namely intact rock strength, block size, and joint surface condition are defined as fuzzy variables. The strength of jointed rock masses is then estimated by incorporating different combinations of three fuzzy inputs into Mamdani rule-based FIS model. To validate the accuracy of the model results, a comprehensive rock mass data is collected from the literature and the strength of rock masses estimated using different empirical equations is compared to the strength of rock masses estimated from the FIS model. It is concluded that the newly developed model compares well with the estimated results and it can be recommended as an alternative method for the rock mass strength predictions instead of empirical approaches in practice.
Due to the presence of discontinuities preparation of core samples to perform laboratory tests on the rock mass into which the engineering structures are excavated is very difficult and not feasible. Laboratory tests, on the other hand, are inadequate because the strength of a small specimen is only indirectly related to the strength of the rock mass. The strength parameters required for design purposes must be those of the rock mass and not of the intact rock. Researchers, therefore, have focused on the studies about developing empirical equations for estimating the strength of a rock mass. As a result of these studies, numerous empirical equations have been proposed in the literature (Hoek et al. 2002, Palmstrom 1995, Kalmaris & Bieniawski 1995, Sheorey 1997, Yudbir et al. 1983, Aydan & Dalgic 1998, Ramamurthy 1986, Barton 2000, Bhasin & Grimstad 1996). Most of these equations consider uniaxial compressive strength of intact rock material (σci) as a scale parameter. The UCS value of a rock mass (σcm) can be estimated by reducing the UCS of intact rock based on the classification indexes showing quality of rock mass such as the RMR (Bieniawski 1989), GSI (Hoek & Brown 1997), RMi (Palmstrom 1995), and Q (Barton et al. 1974). However, the proposed empirical equations are derived from very limited strength tests conducted on large jointed rock samples and this questions how reliable to use strength values estimated by those equations for engineering applications. Researchers have extensively studied the results obtained from field tests for deformability of jointed rock masses and different empirical methods have been proposed for estimating the deformation modulus of jointed rock masses. For the strength of jointed rock masses, however, "further work is required to develop more precise, practical, and easy-to-use methods for determining the rock mass strength" (Edelbro et al. 2006).