Short-term (less than 1 hour) emissions such as upset flaring represent a challenge in dispersion modelling, which typically assumes constant and continuous release. This assumption precludes an accurate prediction of the progressive decrease in ambient concentrations that results when the release stops. Since the emissions from such events are usually high, assuming continuous emissions will result in an exceedance of Alberta’s ambient objectives that are based on the 9th highest 1-hour receptor concentration. Risk-based criteria based on lower percentiles of predicted concentrations provide an alternative for compliance, but sour-gas facilities can have significant emission rates that fail even these criteria.
The CALPUFF model can simulate a short-term release during a specific time and date, but regulatory applications and risk assessments require that the impact of a release under all meteorological conditions must be evaluated. In Alberta where a five-year meteorological modelling database must be used, this means that nearly 44,000 possible conditions must be tested.
This paper presents the results of using the CALPUFF modelling system to achieve this requirement. Two hypothetical sources, one in simple terrain and another in a valley, were tested. To efficiently model each possible condition, the intermittent-release method described in the Alberta Non-Routine Flaring Management model guideline was applied together with a custom post-processor for extracting receptor concentrations at various hours after each release.
Results of the intermittent-release approach showed three advantages over conventional modelling. First, it can demonstrate compliance with ambient objectives even in a large release that fails the risk-based criteria. Second, it can provide a realistic estimate of maximum concentrations at distant receptors which are affected only after the release has ceased. Finally, it can identify potential locations of maximum concentrations resulting from stagnant conditions that cannot be inferred from conventional dispersion modelling. Such features make the approach a useful tool in assessing risk from any short-term release event.
