From mathematical methods to industrial simulators
Simulating flow in porous media requires an understanding of fluids, geological structures and physical processes, combined with efficient numerical methods. We develop the full computational chain: from mathematical models and discretisation to solution algorithms, simulators and tools for analysis and optimisation.
Through sustained research and industrial collaboration, we have developed methods and open-source software used both to explore new ideas and to solve demanding reservoir problems. This connection between mathematics, software and practical applications is central to our work.
We help partners develop new models, improve accuracy and computational efficiency, extend existing simulators and establish workflows for simulation, sensitivity analysis and optimisation. Contact our Applied Computational Science research group to discuss the possibilities.
Models and methods for demanding flow problems
Our expertise covers single-phase and multiphase flow, compositional modelling, and the transport of heat and chemical components. We work with heterogeneous and fractured formations, complex computational grids, and coupling between flow and other physical processes.
A recurring objective is to make it possible to simulate more detailed models and investigate more alternatives within the available time and computing resources. Our work includes:
- Discretisation methods for structured and unstructured grids.
- Robust linear and nonlinear solvers and adaptive time stepping.
- Upscaling, multiscale methods and reduced-order models.
- Parallel algorithms and GPU acceleration.
- Automatic differentiation and adjoint methods for sensitivity analysis and optimisation.
- Flow diagnostics for rapid analysis of flow patterns and well connectivity.
The MRST website includes descriptions and examples of our multiscale methods and flow diagnostics tools.
Open-source software with international reach
Making research available through open-source software has long been a strategic priority. It enables others to examine methods, reproduce results and build on our work. The software also provides a shared platform for collaboration between research and industry.
MRST: from new ideas to reproducible calculations
The MATLAB Reservoir Simulation Toolbox (MRST), developed at SINTEF, is used internationally for research, teaching and the development of new simulation methods. It combines a broad range of physical models with numerical methods and examples that users can adapt and extend.
MRST provides established reservoir simulation models alongside specialised methods for applications such as CO₂ storage, geothermal energy and flow in fractured formations. Automatic differentiation and adjoint methods support sensitivity calculations, optimisation and model calibration.
Hundreds of scientific papers from other research organisations document how MRST is used and extended. Examples, documentation and software are available on the MRST website.
OPM Flow: open-source reservoir simulation in industrial use
We co-develop OPM Flow, an open-source reservoir simulator used for industrial field studies, including by Equinor. Our work covers physical models, numerical solvers and computational performance, together with professional support and maintenance delivered with our partners.
OPM connects numerical research with the demands of practical reservoir studies: large models, complex production histories, established data formats, and reliable and efficient simulation.
JutulDarcy and Fimbul: differentiable simulation and new workflows
JutulDarcy.jl is our Julia-based simulation engine for multiphase flow. It combines flexible models with automatic differentiation and efficient numerical methods. Differentiability allows simulations to be used directly in gradient-based optimisation, parameter estimation and workflows that combine physics-based models with machine learning.
The same platform underpins Fimbul.jl for geothermal reservoir simulation. Open and extensible simulators also provide a foundation for agent-assisted workflows in which users can construct, run and investigate models through dialogue with an AI agent.
Knowledge that others can build on
Our two textbooks on MRST and reservoir simulation bring together mathematical models, numerical methods and practical implementation. Both are available through open access and accompanied by code and examples that readers can run and extend.
The books reflect the breadth of our work, from fundamental flow models and discretisation to advanced reservoir physics, efficient solution methods and coupled processes.
Books
- K.-A. Lie (2019). An Introduction to Reservoir Simulation Using MATLAB/GNU Octave: User Guide to the MATLAB Reservoir Simulation Toolbox (MRST). Cambridge University Press. Open access.
- K.-A. Lie and O. Møyner, eds. (2021). Advanced Modeling with the MATLAB Reservoir Simulation Toolbox. Cambridge University Press. Open access.
