High Altitude Relight Performance of Hydrogen-Air Micromix Combustion Systems

Abstract

Abstract In order to reduce the environmental impact of commercial aviation, hydrogen is being considered as a replacement fuel for kerosene in commercial aircraft. Due to the significantly different physical properties as compared to kerosene, this change in fuel promotes a potential revolution in the design of the combustion chamber. One of the alternative combustion systems recently proposed for hydrogen is micromix. The micromix injection consists of replacing a single air and fuel injector with many hundreds or thousands of small-scale simplified jet in crossflow type injectors. These simple injectors consists of an air gate that generates a small axial jet and a hydrogen injection hole that introduces the fuel perpendicularly into the path of the injected air. This configuration then leads to a large number of relatively rich partially-premixed flamelets in an otherwise globally lean combustor. The high local mixture fraction enhances flame stability, while the size of the flamelets enhances mixing and reduces the residence time of hot products, leading to low production of nitrous oxides. This relatively revolutionary approach has still to be tested thoroughly over the full range of engine conditions and operability specifications. One such requirement is high altitude relight: the low temperature and pressure of the combustion chamber reduce the laminar fuel speed and turbulence levels up to conditions where ignition using conventional methods may not succeed. This study used high fidelity Large Eddy Simulation to investigate high altitude relight for a micromix system. In particular, two approaches for chemistry description are compared: tabulated chemistry based on premixed flamelets and transported chemistry. Flame kernels of different sizes are computed with both approaches to determine the minimum kernel size required for ignition and a probability of ignition. Results show that due to the small recirculation zones downstream of these injectors, ignition must take place close to the injector face. Tabulated chemistry methods fail to predict ignition and a detailed analysis is proposed to better understand the differences with transported chemistry. A simplified model is also presented that links the probability of ignition with the flame speed and radius of the initial flame kernel.