Parametric sensitivity analysis to select the relevant natural fractures in reservoir rock cells based on the equivalent permeability response

Over the last few years, several works have been conducted to improve the productivity of unconventional highly fractured reservoirs, in which the fracture network strongly affects the hydraulic behavior of the porous medium. However, fracture networks can present a complex geometry making difficult its spatial discretization. This work proposes to use a statistical method to select a group of natural fractures that reproduces the equivalent permeability response. An adapted Box-Behnken experimental design together with steady state fluid flow analyses is carried out in order to determine the relevance of each fracture, which is instrumental to select the dominant fractures based on the statistical significance. The results of the model with only few dominant fractures were validated and verified against the ones from the original model, which contains all natural fractures. The time evolution of pressure and the accumulated fluid curve obtained by means of transient analyses showed that a model with the selected fractures is able to properly represent the hydraulic behavior and main trends associated with the original cell.