| The geometry of the fracture system must be incorporated in the hydrogeological study of a fractured rock mass, particularly if flow porosity and fluid velocity are parameters of interest. For that purpose we present a method of analysis that integrates the statistical characterization of fracture geometry (fracture orientation, size, spacing and density) and the stochastic discrete fracture simulation of groundwater flow through fracture networks. The methodology developed for this study was applied to a granitic rock mass at Stripa, Sweden.;The stochastic discrete fracture numerical model developed for this study used a Monte Carlo method and field fracture and hydrology data to simulate groundwater flow in two-dimensional sections containing the Ventilation drift at Stripa. A number of synthetic networks of randomly located fractures were generated, using distributed orientations, lengths and apertures for each fracture set. The steady-state fluid flow parameters were computed for each generated network using boundary conditions determined from field hydraulic pressure measurements. The mean value of fracture aperture was progressively adjusted from one network realization to the other in order to match as closely as possible the total water flow measured during the macropermeability experiment. These simulations yielded estimates of mean fracture aperture that were comparable to those calculated from field injection tests. The modelling exercises also provided valuable insight into the effective fracture porosity and the fluid velocity, two determining factors for the transport properties of a fractured rock.;Four fracture sets were defined for the rock mass surrounding the Ventilation drift at Stripa, on the basis of fracture orientation data obtained from drill cores and fracture maps. For each fracture set the distributions of trace lengths and spacings were analyzed and an average value of fracture density (L('-1)) was computed. A number of sampling bias, including orientation bias, trace length censoring and trace length truncation, were considered in the statistical analysis, and appreciable differences were found between fracture sets for fracture densities as well as for trace length distributions. |
