Underground haulage drifts serve a crucial purpose of moving blasted ore material from the stopes. These drifts are driven in either footwall rock or hanging wall rock depending on the geology of the mine. As there is constant movement of men and material, the stability of these drifts is of critical importance and every measure is taken to make sure that these drifts are stable and functional. Significant effort is put into understanding the stability of these drifts and reinforcing them with suitable supports during the mine life. By making use of some user defined and built-in functions available in FLAC, a simple method to implement multiple simulations using random variables has been developed to determine the stability of underground haulage drift in the vicinity of mining activities, using numerical modelling. The probability of unsatisfactory drift performance is derived based on pre defined performance function of drift wall deformations.
Numerical modelling has become increasingly popular as a tool for geomechanics mine design and mine stability analysis. Methods like the finite elements, boundary elements and finite difference have proven to be extremely useful, and have been successfully used for the development of a wide range of commercially available numerical modelling software. In practice, commercial numerical modelling software is often used to perform deterministic analyses and the design values are selected accordingly. For example, the material is considered to be elastic or elastoplastic and the stability is assessed by examining the extent of yield zones, the deformations causing drift wall convergence or roof sag, etc. and, the design of support system for underground openings is based on a combination of past experience, empirical methods, and deterministic numerical models. Because of the deterministic nature of numerical models, model parametric studies or sensitivity analyses, in which model input data are varied within plausible range, are often carried out to allow for better understanding of the problem before hand, e.g. stability of mine openings, as a result of constant variation in estimated rock mass properties. Although, deterministic numerical modelling has been quite useful in estimating failure, it has been well recognized by many practitioners that numerical modelling alone cannot predict the probability of failure. Because of the inherent uncertainty associated with parameters like the in-situ stress fields, rock properties and geological features around the openings, there is also high uncertainty in the selection of support design based on such parameters. Thus, there is a need to perform stochastic analyses to derive a probability of failure to better understand the risk associated with choosing design parameters based on uncertain input data. Stochastic analysis techniques have been used with success in conjunction with limit equilibrium methods (LEM) to calculate the probability of failure in geotechnical applications like slope stability and isolated pillar collapse (Fenton and Griffiths, 2008). These problems are characterized by a single equation defining the safety factor as a function of one or two random variables, typically cohesion and angle of internal friction of the rock.
