Assessment of the Flamelet Generated Manifold method with preferential diffusion modeling for partially premixed hydrogen flames

This study presents a systematic analysis of the capabilities of a flamelet model based on Flamelet Generated Manifolds (FGM) to reproduce preferential diffusion effects in partially premixed hydrogen flames. Detailed transport effects are accounted for by including a mixture-averaged transport model when building the flamelet database. This approach adds new terms into the diffusive fluxes of the transport equations of the controlling variables coming from a set of coefficients computed from the data contained in the manifold. The manifold is constructed from the solution of a set of unstretched adiabatic one-dimensional premixed flames within the flammability range using mixture-averaged transport. Special attention is given to the numerical aspects related to the construction of the chemical manifold and how to reduce the numerical errors when evaluating the gradients in composition space required for the fluxes. Finally, a systematic application of the method to simulate laminar hydrogen flames in various canonical configurations is presented from premixed to stratified flames, including the case of a triple flame with different mixing lengths. The results demonstrate that the method describes accurately the flame structure and propagation velocities at a low cost, showing a remarkable agreement with the detailed chemistry solutions for flame structure and propagation speed.

Novelty and significance statement

The novelty of the paper lies on the presentation of an extended, robust and comprehensive model for tabulated chemistry incorporating mixture-averaged diffusion. The paper demonstrates the suitability of the model to describe stratified flames when preferential diffusion effects are important by simulating a relevant set of canonical flame configurations. The significance of the paper is that it allows to accurately reproduce a complex phenomenon as it is differential diffusion by the application of a tabulated method, without introducing overheads, and allowing to drastically reduce the computational cost of the simulations.