A Voronoi-based Algorithm for Hydraulic Fracturing Simulation in Deep Geothermal Wells

Potential Hot Dry Rock (HDR) reservoirs underlie the entire world, and are characterized by hot rocks with insufficient or little natural permeability. To improve this aspect, Enhanced Geothermal Systems (EGS) are artificially created by injecting fluid into the subsurface. Fluid injection causes pre-existing fractures to re-open, thus creating permeability. This process is known as hydraulic fracturing. HDR field experiments are expensive to perform and cannot be repeated indefinitely from the same borehole due to progressive irreversible changes in the system under study. Numerical simulation provides a tool to demonstrate understanding of the available experimental results and to extend that understanding to other conditions. This leads to a predictive capability, lending support to design and operational planning. The modelling of a reservoir, including all interactions between fracture/channel flow paths and the host rock, is a challenging problem and has been dealt with in a number of studies. This paper extends and enhances a previous approach formulated by the Authors to simulate the hydraulic fracturing process and predict the fracture network between an injection and a production well. The Voronoi tessellation was used within the framework of the Distinct Element Method (DEM). To model fractures opening and propagation more realistically, an improved algorithm called HYDRA (HYDRofracking Algorithm) was implemented in the FISH programming language in application to the simple case study of a man-made geothermal field. HYDRA attempted to identify and discern shear and tensile fractures, neglecting those arising albeit not communicating with any well, and subsequently reproduced gradual fracture propagation, fluid inlet in fractures and pressure application on the fracture sides. At each stage, pressure inside fractures was the same as the one imposed over the wells boundary and was increased step by step when no more fractures developed. Based on the results of the computation, it was possible to capture the fracture pattern originated from the algorithm execution and the fluid pressure necessary to create a communicating channel through the wells.