VOC EMISSIONS FROM REFINERY BIOTREATERS: MODEL VALIDATION STUDY

Wastewater collection and treatment systems can be significant sources of VOC emissions in refineries and petrochemical plants. The ability to estimate VOC emissions from wastewater treatment units is important in identifying the need for and developing technologies for control of VOC emissions. A variety of models have been developed to estimate emissions of specific compounds from various wastewater treatment units, but these models cannot give an estimate of total VOC emissions, primarily because not all of the VOCs in the wastewater can typically be identified. Even if all of the compounds could be identified, however, parameters required by the model (e.g. Henry's Law constant, biodegradation rate constants) are not available for every compound.Perlmutter et al. (1998), working at the Shell Global Solutions (US) Inc. laboratories, developed a set of three empirical models for total VOC emissions from diffused-air activated sludge systems, each model with a different level of complexity and requiring a different amount of input data. The tiered models were tested in a field validation test using measured VOC emissions from a biotreater at a Shell Oil Company refinery in the U.S. (Perlmutter et al., 1998). Model estimates were found to agree within 16% of the measured VOC emissions.To provide further support to the models, a second field validation test of the Tier 1 and Tier 2 models was conducted at another Shell Oil Company refinery in the U.S. Biotreater off-gas, influent, and effluent samples were collected and analyzed for total VOCs. Actual VOC emission rates were determined and compared to model estimates.The Tier 1 model estimates were in close agreement with the measured emissions. The estimated total VOC emissions were within 4% of the total measured VOC emissions. This result is similar to that from the earlier validation study with this model (Perlmutter et al., 1998) The Tier 2 model underestimated total VOC emissions from the biotreater by about 40% when compared with measured emissions. The C1-C5 and C11-C14 range emissions were significantly underestimated while the C6-C10 estimated emissions were slightly high compared to measured emissions. The differences between model estimates and measured emissions for the three carbon ranges showed a similar pattern with what was observed in the previous validation study (Perlmutter et al., 1998), although the error was somewhat greater in this study. Improvements to the Tier 2 model could be made by adjusting the biodegradation factors, βi, in the model, but specific recommendations for adjustment are beyond the scope of this work. Measuredemissions of compounds heavier than C14 were less than 3% of total VOC emissions, validating the assumption made by Perlmutter et al. (1998) that C1 to C14 range compounds comprised the bulk of VOC emissions.