Research

Generating design-relevant predictions with acceleration strategies and solution confidence

Energy conversion processes spans extreme disparities in spatial and temporal scales and couples shocks, chemical reactions, turbulence, interfaces, and geometric complexity. Our work develops the methods and models needed to understand these interactions and use them for predictive design.

High-Speed Propulsion

For airbreathing propulsion devices, the challenge is not only generating thrust but maintaining stability and performance across strongly coupled flow, mixing, ignition, and geometric effects. We study ramjets, scramjets, rotating detonation engines, oblique detonation engines, high-speed fuel injection, inlet and combustor interactions, shock-boundary-layer interactions, and propulsion-subsystem coupling.

The central problems are limited residence time, fuel mixing, ignition, shocks, unsteadiness, stability, geometry sensitivity, and coupling across engine components. We use resolved simulations to identify how these mechanisms interact and how they influence device performance.

Related projects Selected publications

Reacting & Multiphase Flows

Reacting-flow simulation is a coupled problem involving shocks, finite-rate chemistry, turbulent mixing, thermodynamics, and material interfaces. We study detonation structure, shock-reaction coupling, chemically reacting flows, solid-fuel pyrolysis and combustion, supercritical injection, multicomponent thermodynamics, and transport-dominated mixing phenomena.

High-fidelity computation provides access to detailed physics that are difficult to obtain from sparse experimental measurements alone. Our objective is to understand fundamental mechanisms with enough fidelity to support design and prediction.

Related projects Selected publications

Computational workflow: ChemGen chemistry generation, tetrahedral finite elements, and unstructured mesh discretization

Computational Methods

Complicated physics often reveals deficiencies in the numerical methods themselves. Our work develops the underlying algorithms needed for robust, high-fidelity simulation of chemically reacting compressible flow. This includes high-order CFD, finite-volume and discontinuous Galerkin methods, unstructured grids, robust conservative schemes, positivity preservation, entropy stability, and scalable CPU/GPU algorithms.

We do not simply run CFD; we develop the numerics that make it possible to maintain stability, conservation, and physical fidelity when strong shocks, turbulence, and chemistry interact.

Related projects Selected publications

Accelerated modeling workflow: neural networks, tensor networks, and flamelet progress variable manifolds

Accelerated Predictive Modeling

High-fidelity simulation is essential for insight and calibration, but it cannot be used at every scale or in every engineering calculation. We therefore investigate methods that preserve important physics while reducing computational cost using GPU computing, reduced and mixed precision, computational chemistry acceleration, machine learning, tensor-network compression, multifidelity modeling, reduced-order models, and adjoint optimization.

The longer-term goal is predictive simulation that links scale-resolving computation with robust numerics, modern hardware, and learned reduced models for efficient design and optimization.

Related projects Selected publications