A novel projection strategy for manifold-based chemistry reduction models

The standard Flamelet Generated Manifold (FGM) approach employs a reaction progress variable which is a linear combination of species mass fractions. To find a proper combination for igniting hydrogen diffusion flames, however, is very difficult or even not possible, especially when preferential diffusion is involved. In this regard, this paper presents an FGM method that adopts a subset of the full set of species to describe the movement along the manifold rather than using a reaction progress variable. Moreover, the FGM method is based on an ideal parameterization that spans the local tangential space of the manifold by projecting the solution onto the manifold. This method is referred to as the Projected-FGM method. The method is first tested in a counterflow diffusion flame configuration, and further applied to two reacting mixing layers with and without preferential diffusion effects, respectively. The evaluation of the method’s performance is achieved by comparing the FGM predictions with the results obtained using detailed chemistry. It is shown that the predictions of the Projected-FGM method are in accordance with the results of detailed chemistry simulation for all cases. It is found that with a unity Lewis number approximation, the results are insensitive to the choice of species representing the movement along the manifold, a minimum of two species is adequate. When preferential diffusion is incorporated, more species need to be included. This method is also applicable to other cases where a suitable progress variable is hard to define.