We present a method for estimating fracture porosity and permeability from measurements of fractures in small samples of a fracture system. A fracture system is divided into domains from which samples are collected. Lengths, widths, and numbers of fractures present in samples are measured and then summarized in density and distribution functions. These raw data, in turn, are combined using Monte Carlo techniques to estimate fracture porosities and permeabilities for domains and to develop their respective density and distribution functions. Finally, estimates from domains can be displayed in maps and cross sections to characterize the spatial variation of fracture porosity and permeability throughout the fracture system.

We have evaluated this procedure through comparison with estimates of fracture porosity and permeability reported in the literature and with data from three fractured reservoirs. Comparison with values reported in the literature indicates that the method yields the proper range, form of distribution function, and representative values for fracture porosity. Application of the method to specific reservoirs indicates the difficulties of estimating fracture porosity by any means. These applications also indicate, however, that very poor quality fractured reservoirs (i.e., with fracture porosities much less than 0.1%) can be recognized from core and that estimates of fracture porosity for viable fractured reservoirs through the use of core may be somewhat low. In contrast, estimates of fr cture permeabilities of samples, although not unreasonable as estimates for individual samples, are larger than those typically obtained by reservoir engineering methods for fractured reservoirs. In part, these differences may be diminished by fully including in the method reductions due to in-situ stresses and fracture fill. In part, however, the differences must be due to the conceptual difference between permeability of samples and gross permeability of fractured reservoirs.