Probabilistic Integration of Geomechanical and Geostatistical Inferences for Mapping Natural Fracture Networks

Geomechanical modeling of the fracturing process accounts for the physical factors that inform the propagation and termination of the fractures. However, the resultant models may not honor the fracture statistics derived from auxiliary sources such as outcrop images. Stochastic algorithms, on the other hand, generate natural fracture maps based purely on statistical inferences from outcrop images excluding the effects of any physical processes guiding the propagation and termination of fractures. This paper, therefore, focuses on presenting a methodology for combining information from geomechanical and stochastic approaches necessary to obtain a fracture modeling approach that is geologically realistic as well as consistent with the geomechanical conditions for fracture propagation. As a prerequisite for this integration approach, a multi-point statistics-based stochastic simulation algorithm is implemented that yields the probability of fracture propagation along various paths. The application and effectiveness of this probability integration paradigm are demonstrated on a synthetic fracture set.