Species correlation measurements in turbulent flare plumes: considerations for field measurements.

Field measurement of flare emissions in turbulent flare plumes is an important and complex challenge. Incomplete combustion from these processes results in emissions of black carbon, unburnt fuels (methane), CO2 , and other pollutants. Many field measurement approaches necessarily assume that combustion species are spatially and/or temporally correlated in the plume, such that simple species ratios can be used to close a carbon balance to calculate species emission factors and flare conversion efficiency. This study examines the veracity of this assumption and the associated implications for measurement uncertainty. A novel tunable diode laser absorption spectroscopy (TDLAS) system is used to measure the correlation between H2O and black carbon (BC) volume fractions in the plumes of a vertical, turbulent, non-premixed, buoyancy-driven lab-scale gas flare. Experiments reveal that instantaneous, path-averaged concentrations of BC and H2O can vary independently and are not necessarily well correlated over short time intervals. The scatter in the BC/H2O ratio along a path through the plume was well beyond that which could be attributed to measurement uncertainty and was asymmetrically distributed about the mean. Consistent with previous field observations, this positive skewness toward higher BC/H2O ratios implies short, localized, and infrequent bursts of high BC production that are not well correlated with H2O. This demonstrates that the common assumption of fixed species ratios is not universally valid, and measurements based on limited samples, short sampling times, and/or limited spatial coverage of the plume could be subject to potentially large added uncertainty. For BC emission measurements, the positive skewness of the BC/H2O ratio also suggests that results from small numbers of samples are more likely to be biased low. However, a bootstrap analysis of the results shows how these issues can be avoided with sufficient sample size and provides initial guidance for creating sampling protocols for future field measurements using analogous path-averaged techniques. [ABSTRACT FROM AUTHOR]