Predicting NOx Emissions of a Lean Hydrogen Flame Using High and Low Order Computational Fluid Dynamics Models

Abstract

Recent efforts in numerical methods to study hydrogen combustion have allowed the development of affordable and reliable strategies that can reproduce the main structure of the flame. Although this objective represents a vital goal of the design process of a new combustor, properly estimating the emission remains an aspect that must be further investigated. In fact, due to the lack of experimental data, few numerical works addressed the evaluation of NOx emissions in hydrogen-fueled rigs. The present work aims to study turbulent combustion and NOx emission formation through different numerical approaches on a laboratory-scale atmospheric rig. The burner consists of a swirl-stabilized, technically premixed hydrogen-air flame, with detailed NOx emissions estimated via an experimental campaign at the Technische Universität Berlin (TUB). A first estimation is obtained through a high-fidelity simulation performed in order to assess the capability of a computationally expensive strategy to estimate NOx emissions. A species transport simulation adopting a thickened flame model in which NOx chemistry is included in the chemical mechanism is carried out. After that, a cost-efficient method is explored, allowing a quick assessment of the NOx. With this approach, named LES-to-RANS (L2R), time average fields are evaluated from an large eddy simulation (LES) species transport simulation with simplified chemistry. In particular, the NOx equations are performed on a frozen Reynolds-averaged Navier–Stokes (RANS) framework as a postprocessed stage. The capabilities of the model are then tested under two different scenarios: adiabatic and non-adiabatic wall temperature. The computational accuracy of each approach is compared and discussed, with emphasis on computational cost.