Combustion Efficiency and Methane Emission Rate of Flares Subjected to Crosswind

An estimated 139 billion m3 of gas are burned globally each year in flares at upstream oil and gas production sites.  However, data and models to reliably predict emissions from flares are notably lacking.  Especially concerning are the impacts of strong crosswinds, which can reduce the overall combustion efficiency and strip unburned methane from the flare, although the specific mechanisms behind these emissions are not well understood. Building on a methodology developed by Burt et. al, this study reports quantitative measurements of carbon conversion efficiencies and methane emission rates of 25-100 mm diameter flares subjected to turbulent crosswind in a large closed-loop wind tunnel.  Experiments considered multicomponent flare gas mixtures and operating conditions representative of flares at upstream oil and gas production sites which generally operate as simple non-premixed flames without supplementary air or steam injection.  Emission rates and combustion efficiencies were found to be influenced by wind speed, burner diameter, exit velocity, and particularly composition of the flared gas.  Comprehensive Monte Carlo uncertainty and sensitivity analysis was used to assess the uncertainties associated with such experiments. Attempting to propose a practically implementable model capable of encapsulating known affecting factors, good correlation was observed between carbon conversion inefficiency and both burner diameter and exit velocity for a single fuel mixture over a range of wind speeds.  However, in different flare gas mixtures, minor changes in the composition dramatically affected the overall efficiency and emission rates.  To capture the effect of fuel chemistry in a general model, different correlating factors (such as stoichiometric molar air-fuel ratio, volumetric hating value, carbon number, etc.) were considered and introduced to the model.  The final semi empirical model was shown to be capable of predicting combustion efficiency with reasonable accuracy (less than 1% absolute error for conversion inefficiency at a 95% confidence interval) over a wide range of flaring conditions.