Checkout my CONFERENCE POSTER turML Model Poster.pdf for more details.

Deep-Learning-Turbulence-model

Use deep learning to learn a turbulence model from high fedelity data. The model can reasonably predict other turbulent flows.

We combine OpenFOAM C++ code with deep neural network python code to build a deep learning turbulence model.

Procedures:

1. Extract high fidelity turbulence flow information in OpenFOAM using c++ code.
2. Preprocessing the dataset and training a mode using python code.
3. Emmbeding the DNN weights to OpenFOAM using c++ code.
4. Combine the deep learning turbulence model with OpenFOAM turbulent flow calculation.

The model makes reasonable predicitons and outperforms conventional turbulence model for coures-mash condition. Below is some features map obtained using our deep learning turbulence model:

Enjoy!