Hjem
Center for Modeling of Coupled Subsurface Dynamics

Varselmelding

There has not been added a translated version of this content. You can either try searching or go to the "area" home page to see if you can find the information there
Research project

Simulation technology for injection-related fault and fracture reactivation and induced seismicity

The project aims to develop simulation technology for injection-induced fault reactivation, accounting for interaction between rock and fluid mechanics.

Hovedinnhold

The approach is based on the open-source simulation tool PorePy, which is built on explicit representation of faults, and includes thermo-hydro-mechanical effects. The main research tasks are to adapt the framework to account for advanced fracture behavior including permeability increases and friction laws, and to adapt discretizations and linear and non-linear solvers to facilitate efficient simulations.


A main part of the project considers the further development and application of the simulation software PorePy. At the core, this activity involves extending PorePy so that it can accommodate an increasing range of physical phenomena and exploit the simulation capacity to study processes of relevance for subsurface applications. Examples of such extensions are new constitutive laws for fracture deformation and inclusion of process couplings. As the models give rise to equations that are difficult to solve, research into efficient and robust numerical methods is also an important part of this activity. 


A main component of the computational cost of such multiphysics simulation is spent on solving large linear systems of equations. This is a highly complex task which demands both availability of efficient linear solvers and tuning these solvers to the problem at hand. One goal of this project is to automatize selection and tuning of linear solvers for multiphysics problems, and thereby relieve users of simulation software of a task which is of paramount importance for computational efficiency, but which also can be tedious and challenging. The approach taken is to consider the solver selection as an optimization problem which can be treated with machine learning techniques.
 

Funded by the VISTA program
Foto/ill.:
VISTA Program (vista.no)