AUTHOR=Göktolga M. Ugur , de Goey Philip , van Oijen Jeroen TITLE=Modeling Temperature Variations in MILD Combustion Using MuSt-FGM JOURNAL=Frontiers in Mechanical Engineering VOLUME=6 YEAR=2020 URL=https://www.frontiersin.org/journals/mechanical-engineering/articles/10.3389/fmech.2020.00006 DOI=10.3389/fmech.2020.00006 ISSN=2297-3079 ABSTRACT=

The energy demand in the world is ever increasing, and for some applications combustion is still the only reliable source, and will remain as such in the foreseeable future. To be able to mitigate the environmental effects of combustion, we need to move to cleaner technologies. Moderate or intense low oxygen dilution (MILD) combustion is one of these technologies, which offer less harmful emissions, especially nitric oxide and nitrogen dioxide (NOx). It is achieved by the recirculation of the flue gases into the fresh reactants, reducing the oxygen content, and thereby causing the oxidation reactions to occur at a milder pace, as the acronym suggests. This results in a flameless combustion process and reduces the harmful emissions to negligible amounts. To assist in the design and development of combustors that work in the MILD regime, reliable and efficient models are required. In this study, modeling of the effects of temperature variation in the oxidizer of a MILD combustion case is tackled. The turbulent scales are fully resolved by performing direct numerical simulations (DNS), and chemistry is modeled using multistage flamelet generated manifolds (MuSt-FGM). In order to model the temperature variations, a passive scalar which is created by normalizing the initial temperature in the oxidizer is defined as a new control variable. During flamelet creation, it was observed that not all the compositions are autoigniting. Several approaches are proposed to solve this issue. The results from these cases are compared against the ones performed using detailed chemistry. With the best performing approach, the ignition delay is predicted fairly well, but the average heat release rate is over-predicted. Some possible causes of this mismatch are also given in the discussion.