At the end of July, FELiCS was represented with two presentations at the 41st International Symposium on Combustion in Kyoto, Japan. The symposium offered a fantastic opportunity to discuss recent advances in combustion science with colleagues from around the world.
Anant presented our work on resolvent analysis of turbulent hydrogen flames, carried out together with Marina from our research group and with Michael Gauding and Heinz Pitsch from RWTH Aachen. In this study, we apply FELiCS to a turbulent lean hydrogen–air slot flame and compare the predicted coherent structures with Spectral Proper Orthogonal Decomposition (SPOD) of high-fidelity DNS data. The results show that resolvent analysis successfully captures the dominant Kelvin–Helmholtz wave packets and their characteristic frequency range. Importantly, the study demonstrates that the approach remains applicable even for turbulent hydrogen flames affected by thermodiffusive effects.
A central part of Anant’s work is the development of a new data-informed active-flame closure. By calibrating a simple algebraic reaction model to the mean state obtained from DNS, we can substantially improve the prediction of progress-variable and heat-release fluctuations while retaining the computational efficiency of the linear framework. The results open an exciting route toward applying active-flame resolvent analysis to increasingly complex turbulent reacting flows.
Thomas presented our new approach (link) for calculating the dispersion relations of intrinsic flame instabilities using eigenvalue-based linear stability analysis, developed together with Sandra May and Peter Münch from TU Berlin and Thorsten Zirwes from the University of Stuttgart. Instead of perturbing a flame in costly two- or three-dimensional DNS and measuring the subsequent growth of individual wavelengths, the stability problem is formulated directly from the linearized governing equations around a one-dimensional flame. The resulting generalized eigenvalue problem provides both instability growth rates and the corresponding spatial eigenmodes.
After validation against the classical Darrieus–Landau instability, the method was applied to a finite-thickness reacting flame. We demonstrated that the calculated dispersion relations and eigenmodes achieve DNS-level accuracy while reducing the computational cost by up to eight orders of magnitude. This reduction brings calculations that previously required HPC systems onto a laptop, where they can now be performed within a matter of mili-seconds. This dramatic reduction makes it possible to investigate parameter spaces that would be prohibitively expensive using DNS alone.
The presentation in Kyoto went already beyond the results published in the symposium paper. Building on work carried out during the recent CTR Summer Program at Stanford, we showed first results extending the framework toward thermodiffusively unstable flames. This is an important next step toward our longer-term goal: using efficient linear stability calculations to systematically characterize intrinsic flame instabilities and ultimately incorporate their effects into LES sub-grid-scale models.
Both contributions highlight complementary directions in which FELiCS is developing: from identifying and modelling coherent dynamics in turbulent hydrogen flames to predicting intrinsic instabilities of laminar flames at a fraction of the cost of conventional simulations. We came back from Kyoto with many stimulating discussions, new ideas, and plenty of motivation for the next developments of FELiCS.