Measured and estimated benzene and volatile organic carbon (VOC) emissions at a major U.S. refinery/chemical plant: Comparison and prioritization.

Estimates of emissions for processes and point sources at petroleum refineries and chemical plants provide the foundation for many other environmental evaluations and policy decisions. The most commonly used method, based on emission factors, results in unreliable estimates. More information regarding the actual emissions within a facility is necessary to provide a foundation for improving emission factors and prioritizing which emission factors most need improvement. Identification of which emission factors both perform poorly and introduce the largest error is needed to provide such a prioritization. To address this need, benzene and volatile organic compound (VOC) emissions within a major chemical plant/refinery were measured and compared with emission factor estimates. The results of this study indicate estimated emissions were never higher and commonly lower than the measured emissions. At one source location, VOC emissions were found to be largely representative of those measured (i.e., the catalytic reformer), but more often, emissions were significantly underestimated (e.g., up to 448 times greater than estimated at a floating roof tank). The sources with both the largest relative error between the estimate and the measurement and the largest magnitude of emissions in this study were a wastewater treatment process, an aromatics concentration unit and benzene extraction unit process area, and two sets of tanks (sets 7 and 8). Emission factors for these sources are priorities for further evaluation and improvement in this chemical plant/refinery. This study presents empirical data that demonstrate the need to validate and improve emission factors. Emission factors needing improvement are prioritized by identifying those that are weak models and introduce the largest error in magnitude of emissions. The results can also be used to prioritize evaluations of the emissions sources and controls, and any operational conditions or erroneous assumptions that may be contributing to the error.