This repository contains the code used in the article A case for machine learning in CFD published by FlowFusic.

We recommend using a separate virtual environmet. To create one, run

python3 -m venv <ENV_NAME>

Enter it by running

source <ENV_NAME>/bin/activate

To install the requirements run

cd flowfusic_cylinder_flow
pip install -r requirements.txt

There are three main steps in creatin a machine learning model, which could replace parts of the CFD pipeline. Before going further, make sure that you are in the repository's main directory.

1. Generating geometries

python make_geometry_dataset.py

2. Generating sample flows

3. Training the Convolutional Neural Network model

python train.py