| OCR Text |
Show of formaldehyde emissions but no significant difference between units with N O x controls and those without was found. P A H results are relatively scarce for units with N O x controls. Although it m a y appear that P A H emissions are slightly lower for units with L N B or L N B + S CR compared to units with no controls, the difference is not practically significant because only three data points on a single unit in each subcategory are available; hence, the uncertainty associated with the two data sets is large and the differences are within the confidence intervals of the data. DISCUSSION Figure 4 showed that the data from the pilot-scale furnace compare well to actual field data from the W S P A - C A T E F database. The figure illustrates the range of emissions measured in the field from actual petroleum industry boilers and process heaters. Formaldehyde and total P A H data agree well, given the variability of the results. The field data indicate benzene emissions somewhat higher than measured in the pilot-scale tests. In most cases this is the result of including undetected results in the field data at the full minimum detection limit; the test methods used for the P E R F program were simply more sensitive than for most field tests. Nevertheless, the agreement between the PERF and field results demonstrates the excellent overall representativeness of the P E R F data. Analysis of organic H A P emission factors derived from field tests of gas-fired petroleum industry boilers and process heaters revealed no significant differences due to process design (including different N O x emission controls) or fuel gas type. It is reasonable to combine data to achieve more robust emission factors applicable to all gas-fired boilers and process heaters. Table 3 presents combined organic H A P emission factors for gas-fired boilers and process heaters. Emission factors for these organic H A P s range from approximately IO"4 to 10~9 - o n e ten thousandth to one billionth-lb/MMBtu. Note, undetected results generally have been included in the average emission factors at the full value of the minimum detection limit to provide a conservative estimate of emissions. Some undetected results were not included if the detection limits were significantly higher than other results detected at lower concentrations. The upper and lower 95 percent confidence intervals also are shown as an indication of the uncertainty in the emission factors. The range from the lower to the upper 95 percent confidence interval represents approximately 50 to 300 percent of the mean organic H A P emission factors. This is a consequence of the fact that organic H A P emissions are very low and he very close to the lower limits of detection. This is typical of other H A P emission factor data sets where emission levels are low, and merely highlights that factors of two or three in organic H A P emissions at this level should be considered insignificant Figure 9 shows calculated organic H A P mass emissions in tons per year for a hypothetical process heater or boiler with a fuel heat input value of 100 M M B t u per hour. BTX, formaldehyde, and P A H emissions for a source of this magnitude are below one tenth of a ton, one tenth of a ton and one thousandth of a ton per year, respectively The figure compares these emission levels to the equivalent gas concentrations, to emphasize that these mass emission rates correspond to very low concentrations. The figure also shows estimated total organic H A P 7 |
