Geomechanical response due to nonisothermal fluid injection into a reservoir

Th geomechanical response of a porous reservoir due to injection of fluid can result from a complex interplay between the changes in porepressure and temperature near the wellbore. As a result, predictions are usually made using either simplified analytical models, which may apply unrealistic assumptions in order to produce a tractable model, or detailed numerical simulations that can be computationally expensive. LaForce et al. (2014a, 2014b) developed a semi-analytical model for the geomechanical response of a reservoir to nonisothermal, multi-phase fluid injection, which has been used in studies of CO2 sequestration. We demonstrate that a numerical solution using the MOOSE software precisely matches the analytical formulae. We then include various effects in the numerical model that relax the simplifying assumptions made in the analytical derivation. We find the analytic and numerical solutions for the fluid and temperature fronts still agree reasonably, while only qualitative agreement is observed for other quantities such as stress and displacement. We conclude the LaForce et al. (2014a,b) solutions are useful for rapid investigation of projects involving injection of cold fluid into warm aquifers. However, the enhancements afforded by MOOSE, such as high-precision fluid equations of state and the ability to more accurately capture geological complexity, along with its computational scalability which greatly reduces runtimes, means that MOOSE should be preferred for more sophisticated analyses. Because validating complex coupled codes is challenging, we propose that the model contained herein can be used as a benchmark for other coupled codes.

Automated summary

The geomechanical response of a porous reservoir to fluid injection is complex and can be predicted using simplified analytical models or detailed numerical simulations. The study compares analytical and numerical solutions for fluid and temperature fronts, finding qualitative agreement for some quantities but not others. While the LaForce et al. (2014a,b) solutions are useful for rapid investigation, MOOSE software offers enhancements for more sophisticated analyses and should be preferred.