Towards reduction of computational cost for large-scale combustion modelling with a multi-regional concept

The objective of the work is to validate the feasibility and the performance gain of a multi-regional approach, which has the potential to improve computing performance significantly for large-scale modelling of combustion processes. The basic idea is to solve the non-reactive, less CPU-intensive domain within the burner and the much more CPU-intensive domain with the flame downstream, separately. For the fresh gas flow within the nozzle, only the fundamental Navier-Stokes equations are solved, whereas complex combustion models accounting for the combustion reactions are switched on after the fresh mixture has left the burner exit. The methodology has been implemented into the OpenFOAM code and applied to a large eddy simulation of a turbulent, premixed methane/air flame. The multi-zonal simulation has shown a very good agreement with results obtained from the conventional single-regional. The multi-regional modelling, however, has been proved to be considerably faster than the single-zonal computation.