Data is in from EPA’s “work practice” requirement that petroleum refineries monitor ambient air for benzene concentrations around their fence lines. The regulations set an “action level” of 9 µg/m3 benzene, using benzene as a “surrogate” for fugitive hazardous air pollutants. The purpose? To “protect the health of the populations surrounding the facility, including minority and low-income populations.” EPA set the action level at a concentration that no refinery would exceed as long as its fugitive emissions estimates were “consistent with the level of fugitive emissions actually emitted.” In other words, if operators reported their fugitive emissions accurately, the benzene action level would be entirely theoretical.
Surprise! Benzene concentrations in air around 10 oil refineries blew the limit. The offending refineries include operations by major players such as Chevron, Shell, Marathon, Valero and BPF Energy. Does this tell us something about using unverified industry estimates of emissions as a basis for protecting public health?
In theory, the regulatory structure that governs hazardous air pollutants—such as benzene from oil refineries—is brilliant. It includes elements to appeal to fans of both “technology-based” and “risk-based” regulation. Technology-based standards require that facilities reduce dangerous pollution as much as practical given the state of the art. These standards are relatively straightforward to set and enforce. There is no guarantee, however, that technology-based standards will protect people from all excessive risks. In contrast, risk-based standards are designed to eliminate unacceptable risks, ideally with a margin of safety. Confidence in risk-based regulation, however, requires a leap of faith that risk assessment techniques will generate accurate results. Risk assessments tend to rely on questionable estimates of the amounts of chemical pollutants that people breath, drink, or absorb, and on controversial assumptions about what a safe level of exposure would be. The fact that people are exposed to many chemicals leads to further uncertainty about cumulative and synergistic risks.
Originally, Congress designed the Clean Air Act’s hazardous air pollutant program to use risk-based standards. The Act required EPA to set emission standards that would protect public health with an ample margin of safety. For EPA, this mandate raised the prospect of banning some chemicals completely, at least when “the only level … which would appear to be absolutely protective of health is zero.” The agency essentially froze up. As of 1990, EPA had only promulgated eight hazardous air pollutant standards.
Congress responded in the 1990 Clean Air Act Amendments. At least initially, that law shifted the hazardous-air-pollutant program to rest on technology-based standards. The Act required EPA to determine maximum achievable control technology (MACT) for a list of 191 chemicals. Congress, however, did not stop there. To ensure that a MACT standard is actually protecting the public, the Act mandates an EPA “residual risk” analysis within six years of the promulgation of technology-based limits. This sounds like the best of both the technology-based and risk-based approaches—right?
But look at EPA’s historical approach to residual risk: In Natural Resource Defense Council v. EPA, the D.C. Circuit reviewed EPA’s 2006 analysis of risk from facilities that use or produce synthetic organic chemicals. EPA relied on the results of an American Chemistry Council questionnaire with a 44% response rate. Why? The agency explained, inter alia, that reliance on “industry sources is a well-established practice” and it would have been “very costly and time-consuming” for the agency to require collection and submission of data. EPA’s approach survived the appeal.
With respect to the 2015 petroleum refineries rule: Hats off to EPA for its innovative work-practice/fenceline-monitoring approach. Because the monitoring results illustrate the fallacy of continued reliance on industry estimates of fugitive emissions, the agency should now expand the fenceline-monitoring approach to other sectors.