Development of a differentiable code for solving flamelet equations
Development of a differentiable code for solving flamelet equations
One possible modeling approach to simulate turbulent flames are so-called flamelet-based models. An accurate and yet computational efficient approach is to precompute the underlying flamelet equation and integrate the solution via look-up tables into CFD simulation environments.
Target:
In this master’s thesis, a code to solve flamelet equations and to generate look-up tables for turbulent combustion simulation is aimed to be developed. Based on an existing code base, an adaptive numerical discretization scheme is to be implemented. Through the use of a machine learning library in Python, differentiable simulation capabilities are guaranteed and GPU utilization is enabled.
Tasks:
- Literature study and familiarization with the unsteady flamelet/progress variable turbulent combustion model
- Discretization of the equations and application of appropriate numerical schemes
- Implementation of adaptive discretization
- Preparation of the code to be GPU-ready
- Testing and validation for different fuels against available data
- Writing the master’s thesis
Starting date: As soon as possible
Duration: Approximately 6 months
Contact:
Univ.-Prof. Dr.-Ing. Nicole Wermuth, +43 (316) 873-30087, nicole.wermuth@lec.tugraz.at
Dipl.-Ing. Dr. Stefan Posch, +43 (316) 873-30084, stefan.posch@lec.tugraz.at