Research Interests
The overall goal of our research program is to combine fundamental studies of combustion, scientific machine learning, and novel materials synthesis technology to meet energy and environmental challenges. This subject is interdisciplinary in nature, comprising thermodynamics, fluid dynamics, chemical kinetics, and materials sciences, and our pursuit of both fundamental understanding and application-oriented engineering aims to advance the frontier of the research.
Scientific machine learning and uncertainty quantification in combustion and energy applications
Our research aims to develop and leverage machine learning techniques in physically meaningful applications. Instead of pure black box machine learning, we investigate machine learning and neural network applications that can extract physical insights and uncertain model representations from experimental data in an accelerated or augmented capacity.
One signature work is Chemical Reaction Neural Network (CRNN). By embedding inductive bases such as the law of mass action and the Arrhenius law, CRNN is able to construct physically interpretable chemical reaction models from data, in a wide range of reaction systems such as catalytic reactions, biomass pyrolysis, and battery cathode thermal runaway. Current research topics include:
- Optimal design of experiments
- Uncertainty quantification in reacting flows and other energy applications
- Flame dynamics: ignition, extinction, and stability
- Battery thermal runaway model development
- Machine learning methods for stiff ordinary differential equations
Flame-based methods for synthesis of micro and nanoparticles used in energy storage and sensing
Our work also focuses on developing flame-based synthesis methods that are cost-effective and environmentally friendly. Our efforts include advancing flame-based methods with new strategies that lead to significant reduction in energy consumption and carbon emissions. We are also conducting transdisciplinary research on relating the operating parameters to particle morphology and material performance. The specific topics we are working on are:
- Minutes-level cathode material synthesis for Li-ion batteries
- Low-cost synthesis of solid-state electrolytes
- Low-carbon material synthesis using hydrogen
- Synthesis of nanomaterials for high performance gas sensors
Combustion of energetic materials and carbon-free energy conversion with metal fuels
The goal of this research portfolio is to enable programmable performance and tailored material designs for energy conversion of energetic materials and metal fuels. We discover new physics using advanced experimental techniques and develop novel frameworks that autonomously construct transport and chemical reaction models from lab-scale experiments. We further utilize these approaches to optimize material and structural properties in a large design space, aiming to accelerate the design process. Below are some selected research topics:
- Combustion physics with space and time resolved thermal imaging of flame propagation of energetic materials
- Regression models that account for multi-phase physics of hybrid rocket engines
- Green hydrogen production from metal-water reactions for carbon-free energy conversion
- Inverse modeling of transport and chemical reaction models towards tailored material designs
Project Portfolios
Chemical Reaction Neural Network (Link)
Kinetic uncertainty quantification in combustion simulations (Link)