Flare testing using the SOF method at Borealis Polyethylene in the summer of 2000

Report, 2001

A study of flaring efficiency at the Borealis low-pressure plant in Stenungsund was conducted using a new patented method called Solar Occultation Flux (SOF), in combination with a new type of direct measurement of flow rate and ethylene concentration in the flare stack. The SOF method is based on the measurement of hydrocarbon concentrations over a cross-section of the emission plume. Multiplied by wind speed, this gives the flux of gas through the cross-section, i.e., the source emission in kg·s-1. In the measurements, the sun was used as the light source. An infrared FTIR spectrometer linked to a sun tracker was placed on top of a van that was driven in such a way that the sunlight shone through a cross-section of the plume being measured. From the size of the molecular fingerprints in the infrared solar spectra, the concentration of ethylene and other constituents can be calculated. The results show that the flare we studied has a good combustion efficiency of about 98% at high loads (>1100 kg·h -1), but that at low loads, which are the normal operating conditions most of the time, it has a significantly lower efficiency (50-90%). Emissions thus appear to vary between 20 and 50 kg/h irrespective of load – a result that is consistent with other long-term FTIR measurements taken outside the plant. Similar emission results were also obtained during the course of the project from a newly installed flare, so that the problem does not appear to be specific to the flare that we studied. From the measurements in this study it can be concluded that a major contributing factor in the poor efficiency is an overdose of steam at low operating loads, as a result of trying to avoid soot formation by optimizing flare combustion at high loads. The problem can presumably be solved by introducing some form of dynamic steam metering linked to combustion load, or by eliminating the presence of ethylene at low operating loads. The direct flare stack measurements show that the heating value is generally low, and that theoretically we could therefore expect poorer efficiency levels. It was not possible to demonstrate this unambiguously in the present study, however, since the steam and ethylene levels were generally covariant. Only in a few cases did we find a low efficiency at low ethylene concentration, independent of steam quantity.