Towards lean premixed hydrogen combustion modelling using a data-driven approach

Lean premixed hydrogen flames are characterized by sub-unity mixture Lewis number, which causes the formation of intrinsic instabilities, driven by hydrodynamic and thermo-diffusive mechanisms that interact leading to complex corrugations of the flame front. In this contribution, an approach for lean premixed hydrogen combustion modelling is presented, based on a two-scalar manifold representation of the thermochemistry. The framework is based on a data-driven filtered tabulated chemistry approach, where a fully resolved, two-dimensional, unstable, self-wrinkling simulation is used as a source for the filtering procedure. The method is validated a-priori and a-posteriori against a two-dimensional DNS multi-step chemistry hydrogen/air dataset, consisting of statistically planar, premixed, laminar flames in rectangular computational domains periodic in the cross-wise direction.