Scaling and Validation of Adaptive-Grid Turbulent Mixing Simulations

Project Description

This work lays the foundation for applied research in methane rocket injector design, simulating non-premixed turbulent combustion. While the mixing layer simulation is suitable for both weak- and strong-scaling development, it is also representative of the combustion region at rocket injector tips and is therefore a crucial step towards applied studies. This provides insight on the optimal use of adaptive-wavelet simulation technologies at very large scales. The computational complexity of multi-physics combustion, along with the size of applied engineering cases, demands this ability to effectively scale to clusters of this magnitude.The project will also be the first time SpaceX has published any of its CFD work and will hopefully pave the way to a more open interaction between the company and the research community at large. Furthermore, this work will allow SpaceX to gain experience running simulations at the petaflop scale and assist the company in evaluating investment into its own large-scale GPU cluster.Successful scaling of this code will allow further grants to focus on applied work, running problems specific to SpaceX’s engineering design and providing insight into the physics which underlies supercritical turbulent combustion. The scaling metrics acquired during this project will be used in submission of an ALCC grant application, which if successful would target a Gordon Bell submission in 2017.