Modelling Uncertainties of Non-stationary Fractured Reservoirs

This work is aimed to produce geomechanic realism of non-stationary fracture variations in naturally fractured reservoir (NFR) modelling, and quantify the flow response uncertainties by history matching. This project focus on fracture sets variation with respect to structural position. Outcrop analysis is used to realistically represent the prior range of the fracture properties, namely the intensities, length orientation and the region of distribution. The method consists of generating non-stationary fracture property prior models using Multipoint Statistics (MPS) and zonation concept to distribute the fracture trend.

The uncertainties in flow response are represented with multiple selection of history matched models through sampling variables combinations of the fracture properties. While the uncertainties of fracture sets distribution in the models is represented in Multipoints Statistics (MPS) realisations based on training images.

The results show that by using prior range of fracture properties identified through geomechanic realism will not only help to improve the efficiency in history matching, but also able to easily quantify the posterior range of the fracture properties for better understanding of the reservoir parameters. Also, producing realistic fracture variation is important to produce reliable reservoir forecast compare to unrealistic models.