CFD Modeling of Soot Emission and Flare Efficiencies for Controlled Flare Tests

Flare efficiencies, presented either as Combustion Efficiency (CE) or Destruction and Removal Efficiency (DRE), can drop below stipulated 96.5% and 98% for various reasons that do not sustain a stable flame. It is also well known that flare operators tend to apply excessive steam or air assists to suppress smoke in order to comply with regulations at the expense of combustion efficiency. In order to sort out various factors that influence CE and DRE, a sophisticated modeling tool like Computational Fluid Dynamics (CFD) can be used to conduct parametric studies and to fill out data gaps. The CFD models require combustion mechanisms that can handle both soot and the volatile organic compounds (VOCs) emissions, i.e., they need to include sufficient soot precursor and VOC species to be useful in ANSYS Fluent simulations. In this study, a reduced mechanism of 50 species (LU 3.0.1) is developed from the full version of USC II mechanism and is successfully validated against experimental data. This reduced mechanism has been used in CFD modeling of various controlled flares; Flare tests in 2010 TCEQ flare campaign in Tulsa, Oklahoma and lab scale flare tests in Carleton University. Two different turbulence-Chemistry models namely Eddy Dissipation Concept (EDC) and non-premixed Probability Density Function (PDF) models are being used to simulate various air-assisted, steam-assisted and non-assisted flares. To predict soot satisfactorily, the model parameters in the built-in Moss-Brookes soot model (along with the soot oxidation sub-model) are modified according to fuel species and the Reynolds number. The chief driver behind the study is to validate the CFD models against the experimental data so as to provide a useful tool in the absence of needed experimental data and to facilitate the understanding of key factors that influence opacity, DRE and CE.