Stochastic fracture propagation modelling for enhanced geothermal systems

Abstract

Fractures and fracture networks are the fundamental components of enhanced geothermal systems and determine their technical and economic viability. A realistic fracture model that can adequately describe a fracture-stimulated reservoir is critical for subsequent flow and heat transfer analyses of the system. Fractures in these systems are essentially the product of hydraulic stimulations of the reservoir that, together with ground conditions and the local stress regime, determine how fractures are formed and propagated. This paper describes three methods for generating realistic fracture models for enhanced geothermal systems; two of them incorporate the fracture propagation process in the modelling and hence provide a stochastic fracture propagation model. The methods are: a Bayesian framework in the form of Markov ChainMonte Carlo simulation, an extended Random Sampling Consensus model and a Point and Surface Association Consensus model. The conditioning data used in these methods are seismic events recorded during fracture stimulation. Geodynamics’ Habanero reservoir in the Cooper Basin of South Australia is used as a case study to test these methods.