(135f) Flare Efficiency Characterization Using Neural Network and Sigmoid Models

Current EPA regulations have mandated a minimum combustion zone heating value of 270 BTU/scf and NHV_dil >=22 Btu/ft^2 for all steam/air/non-assisted flares and 380 Btu/scf for H2/olefin interaction cases while maintaining a high Combustion Efficiency(CE). To achieve the target performance along with satisfying the EPA regulations, it is necessary to understand the influence of various operating parameters and determine their set points. Studying the effect of operating parameters through experiments is both expensive and time intense. It would be more cost-effective to use validated models to guide flare operations. In this study flare data obtained from tests conducted from 1983 to 2014 has been analyzed. The data used in the study has a wide range of exit velocities, heating values and fuel compositions. The purpose of the study is to develop a model that can robustly be used in the industry to achieve the desired CE/Destruction Efficiency(DE) and reduce smoke. Its easier to control assist rates, exit velocities and the vent gas flow rates in an actual flare and so variables like these are used as independent variables in the models built. Multi-variable parameterized sigmoid models and a much complex neural network models were developed for the flare data using various types of fuels like propylene, propane, natural gas, methane and ethylene in this study. Contour plots were developed to determine the desired range of parameters for the independent variables. Both steam and air assist flares were modelled using curve fitting tool box and neural network toolbox in MATLAB. The models show high goodness of fit (R²)above 0.90. Desirable operating inputs can be set for the incipient smoke point (ISP) and for smokeless flaring (Soot<=Soot _ISP) with a high CE (>=96.5).