A 2D, two-phase, black-oil, immiscible, incompressible reservoir simulator using TPFA (two-point flux approximation). Both explicit and implicit time steppers are available. Documentation.

Based on Matlab codes (2007) from NTNU/Sintef by Jørg E. Aarnes, Tore Gimse, and Knut–Andreas Lie.
Translated to Python by Patrick N. Raanes.

The Python code produces the same output as the Matlab version (up to errors from the linear solvers and randomness). This is verified by the tests/, which also serve as examples, producing the following illustrations (with the original "jet" colour maps).

Still, some changes have been made. The main ones are:

• 83293bcb: Converted from 3D to 2D for simplicity.
• 27208d5d: Index ordering is C-major (numpy standard), not F-major.
• 7543f574: Vectors are "numpy-thonic", in using 1d arrays, not (2d) columns.
• cade3156: Several linear solvers suggested.
• f33c571a: OOP
• 55ce7325: Facilities for working on the grid.
• e0d12b06: Convenient well config (ensures total sink + source = 0).

Used in

Please let me know (or make a PR) if you use this in your work, and I will add it to this list.

• History matching tutorial

Installation

Prerequisites: Python >= 3.9 with a virtual environment from conda or venv or virtualenv, etc...

Either: As dependency

pip install git+https://github.com/patnr/TPFA-ResSim.git

NB: This will install it somewhere "hidden" among your other python packages. Thus, it will be easy to import, but hard to modify. If you want to play around with the model, install for development:

OR: For development

Clone (or download and unzip) this repo, cd into it, then do pip install -e .

Contributions

Get poetry and do poetry install, which will give you a new venv with very same dev-environment that I used, after which you can run the tests with pytest (no args), and linting with flakeheaven lint.

Suggestions (ToDo)

• Typing: I don't know anything about it, so contributions are most welcome. Must be compatible with Python 3.7 (Colab's version).
• Compressibility: Without compressibility the pressure field propagates information infinitely fast, which is a little boring. It must be carefully considered whether implementing compressibility would make the model too complex for its intended use.