ENV-11-27 Statistical Approach for MACT Floor Setting - Does One Size Fit All?

Description:

In setting emission standards under the Boiler MACT, EPA applied a statistical t-test across existing and new sources, all pollutants, subcategories, and even 112 listed categories without providing justification as to why it is appropriate in each and every case to define a limit that the best controlled similar source will "achieve under the worst foreseeable conditions." EPA’s widespread use of the t-test may be largely due to the fact that it is one of the few approaches that has not been over-turned in court, and because of its mathematical simplicity. It is easy to apply and often difficult to argue against - unless you consider the realities of unit sub-categories being evaluated. Although the t-test may be appropriate when defining limits with long-term averaging periods from random test data, it should not be used blindly in all cases. By using it solely, without attempting to evaluate how each HAP is formed and controlled, the analysis fails to address variability from the following: fuel variability, the status of the unit within a maintenance cycle (when tested), control device variability, impacts of external fluctuations (like weather), load changes, and fuel mixes. This paper addresses the limitations in EPA’s statistical approach to setting Boiler MACT emission limits using data submitted under the ICR.