Application of hybrid numerical and analytical solutions for the simulation of coupled thermal, hydraulic, mechanical and chemical processes during fluid flow through a fractured rock

Abstract

In many subsurface engineering geoscience applications the impact of thermal, hydraulic, mechanical and chemical (THMC) processes needs to be evaluated. Coupled process models require solution of the partial differential equations describing energy or mass balance. Ignoring the coupling of these processes can lead to a significant oversimplification which may not adequately represent the systems being modelled. Incorporation of coupled processes and associated phenomena inevitably leads to numerical stability issues due to very different scales in terms of spatial distribution, time and parametrical heterogeneity. One approach to simplify the computational demands is to integrate analytical and physical models into standard numerical modelling techniques (in this case finite elements), effectively adding sub-grid scale and sub-time scale information to the model. We present such an approach for the simulation of fluid flow through a fracture validated against experimental data and cross comparison with results of other modelling teams within the DECOVALEX 2015 (development of coupled models and their validation against experiments) project (http://www.decovalex.org). By replacing the mechanical behaviour and chemical transport processes with physical models, and by utilising the static nature of the temperature changes, only the hydraulic system required numerical solution in a highly coupled problem. Physical models for fracture closure due to pressure solution, fracture opening due to chemical dissolution, the development of channel flow and a change in the reactive transport characteristics with time were implemented and are described here. The main features of the experimental data could be replicated, although lying outside of the parameter range suggested by the literature. Comparison with other teams using different modelling approaches indicated internal consistency.