Anode baking process has gained significant attention since the 1980s due to its high contribution to costs. The process involves various physics such as turbulent flow, combustion process, radiation and conjugate heat transfer which are highly dependent on each other. The process needs optimization for reducing NOx, saving energy and achieving a higher quality of anodes. A mathematical model of the anode baking furnace can provide significant information for improving the process. The goal of the present work is to find solutions for reducing NOx formation. The research method is chosen based on the simple models for each physics so as to have initial results. The complexity of the model is gradually increased. In this paper, a 2D reactive turbulent flow model of the heating section of the furnace is developed for initial analysis. The results are qualitatively analysed and the distribution of velocity, mass fractions of chemical species and temperature are presented. The distribution aligns with the expected physical behaviour of the system. Radiation in participating media is considered by modelling planck mean absorption coefficient. The temperature seems to be affected significantly by radiation. At the post-processing stage, Zeldovich mechanism is applied to calculate the source term for NOx and thereby, a distribution of mole fraction of NOx in the heating section is estimated. The NOx generation is observed in the high-temperature zone of the furnace as would be expected in reality. Furthermore, a 3D non-reactive flow model is developed and the results of velocity are compared with 2D. Analysis of the velocity in the Z direction suggests a significant difference in the velocity field near the fuel inlet. Therefore, for studying the mixing dominated combustion models, a 3D model is essential.