Comparison of Reduced-Order Manifold Approaches for Simulating a Turbulent Lifted Jet Flame

Reduced-order manifold approaches use assumptions about combustion mode to constrain the thermochemical state space to a low-dimensional manifold generated from one-dimensional component problems relevant for the assumed combustion mode. These models significantly reduce the computational cost associated with the simulation of turbulent combustion systems. In this work, reduced-order manifold models assuming the nonpremixed, premixed, and autoignition modes are all applied in Large Eddy Simulations of the Cabra flame, a lifted flame formed by a rich methane/air jet with a vitiated coflow generated by upstream lean premixed combustion of a hydrogen/air mixture. Due to the elevated temperature in the coflow, autoignition contributes the stabilization of the globally nonpremixed lifted flame, which challenges mode-dependent modeling approaches. Comparison of the predictions using these different approaches to the experimental data indicates that no single mode is able to adequately capture the multi-modal turbulent flame structure. The analysis indicates that a more general modeling framework that retains the benefits of low-dimensionality while relaxing the assumptions on combustion mode would lead to significantly improved predictions.