Optimal Reduction of the C1-C3 Combustion Mechanism for the Simulation of Flaring Helen H. Lou,* ,† Daniel Chen, † Christopher B. Martin, ‡ Xianchang Li, § Kuyen Li, † Hitesh Vaid, † Kanwar Devesh Singh, † and Preeti Gangadharan † † Dan F. Smith Department of Chemical Engineering, ‡ Department of Chemistry and Biochemistry, and § Department of Mechanical Engineering, Lamar University, Beaumont, Texas 77710, United States ABSTRACT: Flaring is a combustion process designed to relieve pressures and safely dispose of vent gases from chemical and petrochemical plants. An industrial flaring activity typically involves various combustible waste gases and a large number of reactions and species. Because most of the detailed kinetic mechanisms for the speciation study of flaring events are too complicated to use in the computational fluid dynamics simulation of industrial-scale flares, several techniques for reduction of the detailed combustion mechanisms have been developed. In this paper, a new rigorous skeleton mechanism (RSM) based reduction technique, namely, the LU 2.0 algorithm, is proposed. It falls under the category of identification of redundancy. Other techniques in this category try to remove redundant species and reactions based on criteria such as sensitivity and quasi-steady- state analyses. These are highly dependent on the preanalysis of the mechanism and require species concentration sets for the conditions of interest. This algorithm tries to find out the skeleton mechanism with the lowest possible error. It works by rigorously testing all of the possible combinations of species sets. This RSM-based optimized mechanism was validated successfully against experimental data for various key performance indicators (laminar flame speeds, burner-stabilized flame, adiabatic flame temperature, and ignition delay) for methane, ethylene, and propylene flames. The efficacy of this algorithm was demonstrated by its improved predictability. ■ INTRODUCTION Flares are used to dispose of unwanted vent gases, by burning them in an open flame, generally at elevated heights. 1 According to the U.S. Environmental Protection Agency, the destructive com- bustion efficiency of the flares should be greater than 98%. 2 During the combustion of flares, besides CO 2 ,H 2 O vapor, and CO, many other intermediate products are generated, in addition to the unburned hydrocarbons. 2 The combustion efficiency depends on the amount of stoichiometric air supplied for com- bustion, fuel-air mixing, wind speed, fuel exit velocity, temper- ature of the flare, and heat value of the gas. 3,4 Recent studies have shown that the combustion efficiency of the industrial flares is underdetermined and uncertain. 5-8 The unburned hydrocarbons include volatile organic compounds (VOCs) such as formaldehyde and highly reactive VOCs (HRVOC). 1,3-Butadiene and all isomers of butene, propylene, and ethylene have been designated as HRVOCs in Texas. Industrial point sources in the Houston-Galveston area are possible sources of HRVOCs that result in the formation of ozone. 9 It is desirable to identify speciated emissions and the combustion efficiency using field measurements. 10 However, because of problems such as the minimum detection limit and accuracy of the system setup, the development of a practical and reliable system for the routine detection of VOCs through field measurement is still under study. Field measurements are also difficult and costly, which contributes to the current lack of field deployment. Therefore, a reliable and practical alternative, such as a computational model for flare speciation, is greatly needed. Computational methods employed to predict the amount of underestimated VOCs, like formaldehyde, require solution of the governing chemical transport equations with detailed kinetic mechanisms. Detailed chemical kinetic mechanisms for some of the fuels are available, 11-14 but they contain hundreds of species and up to thousands of reactions. Hence, solving equations of hundreds of species with such complicated mech- anisms coupled with continuity, momentum, energy, radiation, and gravity equations in a computational fluid dynamics (CFD) grid becomes computationally expensive and, in most cases, infeasible. To alleviate this computational difficulty, several techniques for the reduction of mechanisms have been proposed. These techniques can be broadly classified into three categories: 15 (1) Tabulation methods, where the reactive propensity and corres- ponding system status are stored in the form of tabulated entries. Typical tabulation methods include in situ adaptive tabulation (ISAT) 16 and intrinsic low-dimensional manifolds (ILDM). 17 (2) Identification of redundancy: species and reactions that can be removed from the mechanism without inducing significant errors are considered as redundant. Kuo and Wei 18 proposed a lumping approach, which lumps concentrations of chemical species into a reduced species set. Other approaches that aim at identifying redundant species and reactions include sen- sitivity analysis proposed by Turanyi 19 and Rabitz et al., 20 optimization-based approaches proposed by Androulakis, 21 Bhattacharjee et al., 22 and Petzold and Zhu, 23 the relation graph method method proposed by Lu and Law, 24 a dynamic adaptive Special Issue: Industrial Flares Received: November 28, 2011 Revised: February 11, 2012 Accepted: February 13, 2012 Published: February 13, 2012 Article pubs.acs.org/IECR © 2012 American Chemical Society 12697 dx.doi.org/10.1021/ie2027684 | Ind. Eng. Chem. Res. 2012, 51, 12697-12705