A parallel computing tool for large-scale simulation of massive fluid injection in thermo-poro-mechanical systems

Abstract

Massive fluid injections into the earth’s upper crust are commonly used to stimulate permeability in geothermal reservoirs, enhance recovery in oil reservoirs, store carbon dioxide and so forth. Currently used models for reservoir simulation are limited to small perturbations and/or hydraulic aspects that are insufficient to describe the complex thermal-hydraulic-mechanical behaviour of natural geomaterials. Comprehensive approaches, which take into account the non-linear mechanical deformations of rock masses, fluid flow in percolating pore spaces, and changes of temperature due to heat transfer, are necessary to predict the behaviour of deep geo-materials subjected to high pressure and temperature changes. In this paper, we introduce a thermodynamically consistent poromechanics formulation which includes coupled thermal, hydraulic and mechanical processes. Moreover, we propose a numerical integration strategy based on massively parallel computing. The proposed formulations and numerical integration are validated using analytical solutions of simple multi-physics problems. As a representative application, we investigate the massive injection of fluids within deep formation to mimic the conditions of reservoir stimulation. The model showed, for instance, the effects of initial pre-existing stress fields on the orientations of stimulation-induced failures.

Keywords:

• poromechanics
• geo-materials
• finite strain
• high performance computing

Acknowledgements

In particular, the author thanks R. Hartman-Baker (iVEC), C. Bording (iVEC), P. Ryan (CSIRO), and T. Poulet (CSIRO) for discussions.

Notes

1 Recall that the Legendre transform of a function , can be written as , where solves .

Additional information

Funding

This research was supported under Australian Research Council’s Discovery Projects funding scheme [project number DP140103015]. The authors gratefully acknowledge this support. The first author acknowledges the support of iVEC (http:www.ivec.org) through the use of advanced supercomputing resources that they provided [Project Number: geosciences711].