Projects

Media from a few projects I’ve worked on.

Hydrogym

Coming soon (but see project site). A framework for reinforcement learning in flow control problems.

Reduced-order modeling for fluid dynamics

The main topic of my PhD thesis was model reduction for unsteady fluid flow. This involved a fair amount of CFD simulation (mostly in Nek5000), math modeling, and some machine learning. For instance, we showed that you can use a deep autoencoder to learn a four-dimensional embedding of shear-driven cavity flow that tracks the entire breakdown of the laminar flow to chaos/turbulence as a function of Reynolds number:

Selected publications:

• Symmetry-breaking in a turbulent wake
• Manifold reduction with nonlinear correlations
• Multiscale model reduction

Topology optimization

I worked on a project where we developed 3D multiphysics level-set topology optimization using the Firedrake finite element analysis library and mesh adaptation using MMG. We applied this to a number of different industrial heat exchanger optimizations in automotive and aerospace - I can’t share those designs, but here are a few classic “benchmark” problems similar to what you’d see in the academic literature.