Influence of dispersion length on volume-averaged simulations of ammonia/air combustion in porous media burners

Ammonia is a carbon-free alternative to fossil fuels and can potentially be integrated in the existing energy infrastructure. However, due to poor flame stability and high pollutant emissions, clean combustion of ammonia is a current topic of research. Porous media burners have shown potential to improve the combustion characteristics of ammonia and ammonia blends, which are otherwise difficult to stabilise in conventional burners. Combustion in porous media can be investigated in great detail by performing three-dimensional direct pore-level simulations (3D-DPLS). However, 3D-DPLS with complex ammonia chemistry are computationally expensive. Volume-averaged simulations (VAS) are an efficient alternative for numerical investigations of porous burners. In this work, a comprehensive VAS framework is proposed for 1D, 2D, and 3D transient VAS, taking variable porosity, detailed chemistry and diffusion into account. The numerical framework allows for on-the-fly definitions of constitutive models for effective properties, e.g. tortuosity, dispersion and permeability. After successful validation with other VAS cases from literature, the new code is used to analyse an experimentally investigated novel porous ammonia burner. The analysis is performed to study the effect of the characteristic dispersion length of the solid matrix, which is hard to measure for practical geometries, on pollutant formation and energy balance. All other effective properties are obtained directly from μ -CT scans. Both fuel-lean and fuel-rich conditions of ammonia/air combustion in porous media are investigated. As the characteristic dispersion length increases, local peak temperatures decrease. This significantly affects the predicted and NH 3 emissions. Higher dispersion lengths lead to a broadening of the flame zone that can lead to larger lift-off heights from the burner inlet and merging of neighbouring flames Therefore, reliable estimates of characteristic dispersion lengths are required to achieve good predictions from VAS.