About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Materials Processing Fundamentals: Towards Sustainable Process Modeling, Design, and Operation
|
| Presentation Title |
HPC/AI Based Methodology for Flashback Detection in Hydrogen Blended Natural Gas Combustors |
| Author(s) |
Archit Bapat, Gregory Vogel, Veeraraghava Raju Hasti, Shashi Aithal |
| On-Site Speaker (Planned) |
Archit Bapat |
| Abstract Scope |
Methane fueled gas-turbines contribute to greenhouse gases. Using methane-hydrogen blends in gas turbines promises to reduce greenhouse gas emissions, since combustion of hydrogen produces only water vapor. As hydrogen concentration in the methane-H2 blend increases beyond 60%, higher reactivity mixture increases flame speed that can drive the flame back into the pre-mixer hardware, instantaneously melting the injector hardware. This unstable phenomenon known as flashback, is a serious durability concern and has significantly slowed down the design and the development cycle of H2-blended combustors and thus substantially increased the cost of developing such technology.
Innovative flashback-predicting computational approaches are needed to accelerate the development of such combustors. ANL and PSM is developing a design tool integrating Artificial Intelligence (AI) models trained on Reduced-Order Models (ROMs) with high-fidelity multidimensional CFD simulations to predict the flashback propensity in CH4/H2 fueled combustors. This technique promises to significantly reduce the design/development time of high-H2 combustors. |
| Proceedings Inclusion? |
Planned: |