265311 Parametric Study of Industrial Ethylene/Propylene Flares Using CFD Simulations

Thursday, November 1, 2012: 8:50 AM
316 (Convention Center )
Kanwar Devesh Singh1, Tanaji Dabade2, Preeti Gangadharan1, Daniel H. Chen1, Xianchang Li2, Helen H. Lou1, Christopher Martin3 and Kuyen Li1, (1)Dan F. Smith Department of Chemical Engineering, Lamar University, Beaumont, TX, (2)Mechanical Engineering, Lamar University, Beaumont, TX, (3)Chemistry & Biochemistry, Lamar University, Beaumont, TX

Air quality models often significantly under-predict observed peak ozone with the current VOC emission inventories in Houston-Galveston-Brazoria (HGB) area. There is a high likelihood that the emission inventories in the HGB area is significantly under-reported and one of the important under-reported sources is flare operations. Under the EPA guidelines, these open air combustion systems are assumed to have a high destruction and removal efficiency (DRE) and combustion efficiency (CE) if the flares are operated in compliance with 40 CFR § 60.18. As a result, VOC emissions are very low and no incomplete combustion products are assumed during the flaring operations.  These assumptions are currently under scrutiny by TCEQ in order to develop a viable state implementation plan (SIP). Even though strong correlations between air-to-fuel ratio, steam-to-fuel ratio, flare-tip velocity, vent gas heat content, and crosswind on the flare efficiencies have been observed, the effect of these parameters is known to regulatory agencies and the industries only qualitatively at best.  In this work, CFD simulations of industrial ethylene/propylene flares are performed to predict DRE and CE and VOC emissions over a wide range of operating conditions (from Standby mode to Startup mode). In addition, free radicals such as perhydroxyl and radical-producing species such as formaldehyde are also predicted. The CFD simulations employ a validated reaction mechanism suitable for the combustion of C1 to C3 light hydrocarbons coupled with Eddy Dissipation Concept (EDC) and Probability Density Function (PDF) turbulent models. The simulation results together with available controlled flare test data are further used to develop easy-to-use statistical correlations for future applications in air quality modeling and flare operations.

Extended Abstract: File Not Uploaded