Numerical Modeling of NOx Formation in Turbulent Flames Using a Kinetic Post-processing Technique

The numerical prediction of formation of nitrogen oxides (NOx) in combustion devices (domestic and industrial burners and furnaces, internal combustion engines, etc.) is gaining rising importance because of strong limitations imposed by the legislation. Such predictions usually require the adoption of detailed kinetic mechanisms, because the numerical modeling of NOx can be accurate only considering all of the relevant chemical paths (thermal, prompt, N2O, fuel NOx, NNH, etc.). Despite the continuous increase of computational resources, the direct coupling of detailed kinetics and complex computational fluid dynamics (CFD) is very expensive in terms of central processing unit (CPU) time, and alternative approaches must be taken into account. In this work, the prediction of NOx in combustion systems is performed using a kinetic post-processing technique, which is based on the assumption that pollutants, such as NOx, only marginally affect the main combustion process (because they are usually present in very small amounts). In particular, the proposed numerical methodology, which can be applied to turbulent flames (premixed and diffusive) in arbitrarily complex geometries, transforms the original CFD simulation of the combustion device under investigation in a network of perfectly stirred reactors at a fixed temperature, which is solved by employing a very detailed kinetic scheme, accounting for the formation of NOx. The main novelty with respect to similar approaches documented in the literature is represented by the possibility to solve networks with a very large number of reactors (of the order of ∼100 000), also in the presence of very complex fluid dynamics (recirculations, swirled flows, multiphase flows, etc.). For this purpose, a robust and accurate numerical algorithm was developed, to reduce the CPU time, without sacrificing the reliability of the numerical predictions. The kinetic post-processing technique was validated on several turbulent jet flames (fed with syngas, hydrogen, and methane). The comparison to the experimental measurements is quite good, demonstrating the feasibility and reliability of the proposed approach. Then, the formation of NOx was successfully modeled in a small-scale combustor, operating in moderate to intensive low oxygen dilution (MILD) conditions.