Hybrid Modelling using Neuro Fractal for Fractured Reservoirs

Due to geological reasons, fractured reservoirs are extremely heterogeneous. Modelling of these reservoirs has so far been considered complex and progress is still inadequate. This paper presents a novel and hybrid method to model discrete fracture networks in naturally fractured reservoirs. It involves investigation and systematic integration of tasks spanning cross-disciplinary areas: geological, statistical and artificial intelligence characterisation of natural fractures, rock and fracture mechanics. This paper also evaluates applications of fractal mathematics on characterising natural fracture distributions, especially discrete multifractal dimensions. A case study illustrates that discrete multifractal dimensions are greatly more suitable for such complex systems as natural fractures, compared to the commonly used single-fractal and statistical distributions. Key-Words: Petroleum Engineering, Naturally fractured reservoir, Discrete fracture network, Stochastic simulation, Fractal, Discrete Multifractal, Neural network.