I am looking at implementing combustion modelling within a building fire scenario to provide a more accurate heat and smoke distribution (not just providing a heat and mass source). Have had very limited success. In general over prediction of temperature occurs with either eddy-dissipation models or non-premixed combustion models leading to unrealistic heat release rates and plume geometry.

Has anyone had any luck with implementing some form of combustion chemistry into such fire cases?

make sure you turn off species inlet-difussion using the TUI, otherwise diffusion of fuel vapor at inlet can overwhelm a small prescribed convective inlet flow.