Industrial Flare Performance at Low Flow Conditions. 1. Study Overview

by Vincent M. Torres, Scott Herndon, Zach Kodesh, and David T. Allen

Industrial & Engineering Chemistry Research 2012 51 (39), 12559-12568

A series of full-scale industrial flare tests were conducted at low flow and low BTU content of flared gases at an industrial test facility. Both a 24” diameter air-assisted flare with a flow capacity of 144,000 lb/h and a 36” steam-assisted flare with a flow capacity of 937,000 lb/h were employed in the testing. Flared gases were mixtures of natural gas, propylene, and nitrogen or natural gas, propane, and nitrogen. Natural gas to propane or propylene ratio was 1:4 by volume for all experiments. Nitrogen was used as a diluent to achieve the desired lower heating values (LHV) for the vent gas. The range of flared gas flow rates was 0.1% to 0.65% of the flare’s design capacity. Flare operation was characterized by measurements of flow rates to the flare, extractive measurements made of the vent gases fed to the flare, extractive measurements made at the end of the flare plume, and remote sensing measurements of the flare plume made by a variety of spectroscopic instruments. Destruction/removal efficiencies (DRE, fraction of vent gas reacted) of flared species were calculated based on the observed composition of the species in the plume. The tests demonstrated that destruction efficiencies for steam-assisted flares drop dramatically when combustion zone heating values fall below 250 BTU/scf. Air-assisted flares showed a linear drop in DRE as a function of air flow. While the primary focus of the measurements was on DRE, products of incomplete combustion were also measured. Dominant products of incomplete combustion were CO, ethylene, formaldehyde, acetylene, and acetaldehyde. CO represented approximately 24% to 80% (carbon basis) of the total products of incomplete combustion for DRE > 90%. While DREs of 98–99% were observed in some experiments, many operating conditions produced DREs of substantially less than 99%. Since prescribed methods for estimating emissions would have predicted 98–99% DRE for all the tests, some test conditions resulted in the production of flare emissions multiple times the value that would be calculated using the prescribed estimation methods. In practice, total emissions from flares will depend on both operating conditions and the duration of operation at the various operating conditions.

Read the article