Accurately simulating how fuel burns in turbulent flows is essential for designing and running jet engines, rocket motors, and new hypersonic propulsion systems, which are important for national security. However, fully capturing the complex chemical reactions and fluid transport in these flames can require solving hundreds to thousands of coupled equations, which can take supercomputers days or even weeks to complete.
To overcome this challenge, researchers created a new method called Time-Dependent Bases with Local CUR decomposition (TDB-L-CUR). Instead of looking at the entire engine flow as one big system, the method automatically breaks the flow field into smaller regions where the chemical reactions behave in similar ways. Within each region, it builds a low-dimensional “manifold” that captures the essential species transport and reaction dynamics. At the same time, an on-the-fly scheme identifies unusually fast chemical pathways and penalizes them to avoid numerical stiffness without any prior training or offline data.
When tested on a series of benchmark flames, TDB-L-CUR matched the accuracy of high-fidelity models while reducing computational cost by orders of magnitude. Because it runs entirely “on- the-fly,” the method can be embedded in real-time design tools and digital twins for defense applications, from engine-in-the-loop testing to adaptive control of hypersonic cruise missiles.
By delivering fast, reliable predictions of how complex combustion processes respond to changing operating conditions, TDB-L-CUR paves the way for safer, more efficient, and more agile propulsion systems critical to maintaining technological advantage.
Sandia researchers linked to work
- Lacey, C.E.
Sponsored by
The Department of Energy, Office of Science
Associated Publications
Kim, J.J., Lacey, C.E., Lee, J.H., Jung, K.S., Chen, J.H., and Yoo, C.S., (2026). Time-dependent-bases with local CUR decomposition method for accelerating turbulent combustion simulations” Combustion and Flame, 286: 114848. https://www.sciencedirect.com/science/article/pii/S0010218026000842
August 24, 2026