The Application of Metaheuristic Optimization Algorithms of Gravitational Search, Particle Swarm, and their Hybrid in Fracture Network Modeling

Fractured network modeling is the main prerequisite for fluid flow simulation in many applications such as groundwater resource management, oil and gas reservoir simulation, geothermal energy resource modeling and etc. The aim of this study is to develop an iterative object-based algorithm for fractured network modeling that considers both statistical parameters and spatial connectivity of fractures. The presented algorithm uses the functions of fracture neighborhood matrices in image processing to make clear the fracture continuity and to determine their spatial distribution using the matrix of interconnected fracture cell detection, Sobel matrix, Prewitt matrix and Laplacian matrix. The objective function defines the differences between the features of the reference fracture network and the stochastic generated fracture network using a norm of two. To solve this objective function, optimization algorithms of Gravitational search, particle swarm and hybrid of these two metaheuristic algorithms have been used. In this paper, the metaheuristic algorithm of particle swarm optimization enjoys more validity in fracture network modelling. Therefore, among the metaheuristic algorithms, PSO algorithm regenerates the reference fracture network with the accuracy of 98.89%.