MeetingsRoundtable Meeting — June 2-3, 1999 History of Federal Process Safety
and Risk Management Regulation Briefing Paper No. 2 Jonathan Averback In the aftermath of the 1984 Bhopal, India chemical accident and subsequent incidents in the United States, the Federal government assumed a larger role in chemical safety issues affecting workers and people in communities around facilities that handle chemicals. Initially, both the Environmental Protection Agency (EPA) and the Occupational Safety and Health Administration (OSHA) used a combination of existing statutory authorities and voluntary programs to improve response planning and to promote accident prevention. Congress’s first legislative action in response to Bhopal was to pass the Emergency Planning and Community Right-to-Know Act (EPCRA) in 1986. EPCRA primarily addressed emergency planning and public information, but also included a requirement for a study on accident prevention. Subsequently, Congress included mandates for both OSHA and EPA to develop accident prevention regulations as part of the Clean Air Act Amendments of 1990 (CAAA). The focus of this paper is on the development of Federal regulatory programs implementing the CAAA accident prevention provisions. Special emphasis will be on the development of EPA’s risk management regulatory program, which implements the mandates of Clean Air Act (CAA) section 112(r). To a lesser extent, important efforts of other stakeholders will be referenced but will not be thoroughly addressed. This "history" will identify key decisions and policy choices that may be measurable to demonstrate how well the regulatory structure is working toward the declared purpose of section 112(r), "to prevent the accidental release and to minimize the consequences [of] any such release of any [listed regulated substance] or any other extremely hazardous substance." A general measurement issue reflected throughout this paper will be whether the combination of flexibility and accountability in both agencies’ programs is balanced in a way that meets the goal of preventing catastrophic accidents. After a brief overview of the rulemaking process, this paper generally tracks the organization of EPA’s regulations. Thus, the discussion of OSHA’s Process Safety Management (PSM) rule is split between the Program 3 and the Applicability portions of this paper, and follows comments on the development of Program Level (Tiers) and Hazard Assessment requirements. In each section, after explaining the basis for key policy choices, the paper identifies possible ways to measure whether the rule provision works as intended.
In the CAAA, Congress instructed both OSHA and EPA to develop rules on accident prevention. Congress instructed the agencies on its intent for the contents of these rules primarily in the language of the statute and secondarily in legislative materials such as committee reports. In developing the rules, OSHA and EPA each followed a "notice and comment" procedure in which each agency issued a draft "proposed rule" in the Federal Register, invited comments from the public in writing and at hearings, responded to these comments in writing, and published a final rule in the Federal Register. Each proposed and final rule has an accompanying explanatory "preamble," technical support information, economic analyses, comments, and responses contained in an official "record" for the rule. Federal administrative procedure allows parties to challenge agency rules in the US court system to test the legal sufficiency of the record as a justification for the final rules. OSHA’s rule principally appears in the Code of Federal Regulations compilation at title 29, section 1910.119. EPA’s rule appears in title 40, part 68. The key rulemaking notices for OSHA and EPA were as follows: 1) 1990 - OSHA proposes Process Safety Management
(PSM) rule; Both agencies chose to adopt what they call "flexible, performance-based" regulatory approaches to fulfill their mandates. This approach affords covered "facilities" (the PSM term) or "sources" (the section 112(r) term) substantial discretion in how to meet the rule requirements. Other programs at the state level, such as those in California and New Jersey, impose more specific requirements by regulation or by permit. Neither OSHA nor EPA expressed their performance standards as quantitative risk targets, as some European programs have. Due to the number of covered sources in both programs, both agencies lack the ability to provide the extensive source-specific oversight that states like Delaware have provided. As an alternative method of oversight, EPA’s program contains a "Right-to-Know" component involving publicly available information about source compliance activities. EPA has traditionally stated that it supports public involvement in emergency planning and accident prevention, as have some industry groups (at least at the local level). Theoretically, Right-to-Know can enhance the accountability of source compliance with performance-based programs.
The October 1993 Risk Management Program proposed rule established uniform prevention program requirements for all sources that had more than a threshold quantity of a part 68 regulated substance in a process. Essentially, the proposal extended the OSHA PSM requirements to additional types of facilities, including facilities that were explicitly exempt from OSHA PSM. The proposal reflected the basic EPA philosophy that any source that handles chemicals capable of injuring people should systematically analyze its potential for a catastrophic release and take reasonable steps to prevent such a release. In EPCRA and section 112(r) notices as well as in outreach, EPA expressed the view that this responsibility extended not only to substances on specific lists but also to unlisted substances. This view is not unlike the view expressed in voluntary programs sponsored by EPA and by industry. PSM embodies the recommended approach to prevention, and EPA believed PSM’s flexibility allowed EPA to extend PSM to all types of sources that handled chemicals that presented a serious hazard if released. Early industry experience in implementing PSM convinced many commenters that EPA’s initial approach to prevention programs would be very costly. Industry commenters could be broken into two groups. Those accustomed to using PSM approaches to chemical safety, such as refiners and chemical companies, recognized that the details of PSM were costly, especially if EPA’s adjustments required completely revising their PSM implementation. Those not accustomed to PSM did not recognize any need for PSM-like approaches to safety either because they perceived other codes assured safety (for example, explosives manufacturers) or because they did not think of themselves as handlers of hazardous chemicals (for example, the ammonia refrigeration industry). Comments from states, environmental organizations, and academics were far fewer than from industry at the proposed rule stage. The latter two groups expressed some skepticism about whether PSM was too flexible an approach to actually prevent accidents. Environmentalists in particular asked for EPA to consider "inherent safety" options such as a mandated, top-down, "technology options analysis" (TOA), to force greater rigor in prevention assessment.
In the Supplemental Proposal, EPA moved from the "PSM for all sources" approach of the initial proposal to a "tiered" approach to prevention. A source that could establish though modeling and accident history that it would have no off-site impacts in the event of an accidental release would file a short notice and certification to comply with Tier 1. A source in an industry that historically had a high frequency of significant releases would still be subject to at least PSM in prevention Tier 3. All other sources covered by part 68 would be subject to a generally defined Tier 2, under which a source would describe its existing prevention activities under the General Duty of 112(r) and other programs and continue such activities. EPA’s Economic Impact Analysis (EIA) accompanying the Supplemental Proposal reveals that EPA significantly revised its estimates of the costs and benefits of the program between the first proposed rule and the supplemental proposal, with EPA deriving a large reduction in projected costs from the adoption of "tiering" in the prevention program. The statutory mandate that EPA’s regulations be "reasonable" made lowering the costs prudent because costs are legally relevant when reasonableness is an explicit criterion in a statute. The only benefit increment foregone was the difference between what an explicit PSM mandate would provide to sources not subject to PSM over what the General Duty of 112(r) and other existing programs supplied. While there was recognition by industry and many states that tiering represented an approach that could make their level of effort more proportionate to risk, there were two types of criticism about how tiers were assigned. First, some believed that using accident history data by industry class was inappropriate because the same process in two different industries could be treated differently (for example, chemical storage). Second, there was recognition that a source already subject to PSM would not get relief from Tier 2. In fact, by making a source subject Tier 2 describe its existing prevention programs, Tier 2 had the potential to increase a source’s enforceable prevention requirements. The Supplemental Proposal drew substantially more comment from environmental and labor organizations than the initial proposal. Generally, these groups questioned the legality of tiering and objected to the failure to impose TOA and public participation requirements.
The Final Risk Management Program Rule retained three tiers, renamed "Programs." Two significant changes were made to the assignment of Program levels. First, the rule assigns Program level by process rather than by source. Second, processes subject to OSHA PSM were assigned to Program 3. Both of these changes were in response to comments. Program 1 retained both a modeling component (no public receptors within the worst case zone) as well as an accident history component (no accident with off-site death, injury, or environmental response / restoration). This latter component serves as a "reality check" on using modeling to get into Program 1.
If a substantial number of processes qualify for Program 1 and subsequently have disqualifying accidents, then that would draw into question the use of the current worst case to assign program levels. The worst case now has substantial flexibility for different models, and the compliance cost differences for Program 1 and the other tiers are great. Therefore, the rule provides incentives to use models that project very short dispersion distances. The rates of accidental releases in Program 2 relative to Program 3 processes may demonstrate whether exempting Program 2 processes from PSM is appropriate. If Program 1 and 2 processes experience a relatively greater proportion of accidents than projected based on prior history, then that calls into question not subjecting such processes to PSM. III. Hazard Assessment - Off-site Consequence Analysis and Accident History Requirements
The requirement to conduct and report a worst case release analysis and, for most sources, alternative release scenarios, was probably the most commented-upon issue during rule development. EPA’s initial proposal for the "Offsite Consequence Analysis" (OCA) was simple and sweeping - model a total release of each regulated substance in a process under absolute worst case conditions, and additional "more likely scenarios" under other conditions. EPA had previously issued guidance for release modeling under EPCRA and had worked on other public-domain modeling such as ALOHA, which could serve as a basis for dispersion analysis. EPA’s requirements for the OCA tracked the legislation and those aspects of the CAAA House-Senate Conference Report that were not otherwise reflected in the prevention program. The proposed release scenario analysis issue may have sensitized many in industry and the states as to the potential impact of part 68. Industry ridiculed these provisions, questioning the conditions for modeling, the quantity to be modeled, the types of receptors to be noted in impact calculations, and the number of potential releases that needed to be modeled. Through the Supplemental Proposal and the Risk Management Program Final Rule, EPA established more specific release conditions and modeling parameters. Among the critical changes were modifying the worst case quantity to be modeled to the greatest quantity in a single vessel, specifying what impacts were to be noted, allowing a source to substitute site-specific parameters for default worst case conditions if it could document that the default conditions did not exist at its site, and clarifying how many releases were to be modeled. Also, EPA provided few restrictions on the type of models that could be used. Impacts in dispersion zones generally could be identified off of census data and street maps. The issuance of EPA modeling guidance, the reliance on publicly available data for receptors and modeling conditions, and the clarity in what needed to be reported greatly simplified the worst case and alternative case modeling. EPA claims that sources do not need a consultant to complete the required modeling. Thus, as was the case for tiering / program levels, the EIA indicates that EPA assumed substantial cost reductions over the initial proposal. EPA did not attribute quantifiable benefits to the OCA in any of the versions of the rule. EPA claimed a qualitative benefit in giving the public access to information about the hazards in their communities and believed that the information would be useful for local zoning and other purposes.
One particular measurement of the effectiveness of EPA’s OCA requirements would be to see if actual releases travel farther than projected worst case scenarios. It will always be difficult to prove a negative (that is, no release traveling as far as the worst case does not prove such a release is impossible), but a worse than worst case release may indicate a shortcoming in the rule. Two possible ways such a release could occur would be if more than the contents of one vessel is released or if the model selected by the source substantially understates dispersion distances. If many sources report substantially different dispersion distances when they model releases involving the same chemical and quantity, then that may indicate the rule is structured to favor sources that have the resources to use on modeling or those that are willing to make less conservative assumptions about potential releases. If actual releases travel significantly shorter distances than EPA guidances and other commonly used public domain models predict, then this may confirm the view that these models are unnecessarily conservative. If this is the case, then the OCA may impair debate about hazards and risks rather than promote such debate. A key measure of any Right-to-Know program is to see whether the data is actually used by citizens, governments, and industry. How the OCA information is used, positively and negatively, also should be measured, although this may tend to devolve into anecdotes.
The accident history provisions of the final rule require specific data about actual release incidents over the 5 years prior to the submission of a source’s "risk management plan" (RMP). This accident history is different from the accident history used to qualify for Program 1. For instance, releases that led to onsite deaths, injuries, and property damage, or known off-site evacuations and sheltering in place must be reported but do not disqualify a source from Program 1. EPA followed comments that suggested that keeping a source out of Program 1 because it requested the community to evacuate or seek shelter would discourage such precautions when impacts are possible but are not projected to rise to levels of concern in part 68. EPA does not require reporting of near miss events nor does it require reporting incidents until an RMP update is due. Preliminary drafts of the Supplemental Proposal that were subject to interagency review contained reporting requirements for accidents but the Supplemental Proposal did not propose such reporting. The Supplemental Proposal contains EPA’s official explanation for this change. Researchers in particular believe that near miss events offer important opportunities for learning about potential safety failures and have been critical of this omission. IV. Program 3 Requirements - Adoption of PSM
Prior to promulgation of the PSM Standard, OSHA had addressed accident prevention failures through its General Duty Clause under the OSH Act. For example, OSHA used the General Duty Clause to assess Phillips Petroleum what at the time was its largest penalty for the Pasadena, Texas explosion. The General Duty was useful to OSHA as an after-the-fact enforcement tool but was difficult to use up-front to prevent an incident. Proving a General Duty case under OSHA policy and subsequent case law required OSHA to show that there was no other applicable standard addressing the incident. Industry finds the General Duty to be a "gotcha" provision because it is used after-the-fact when there is no violation of a specific government rule. After Bhopal, the chemical industry began developing voluntary process safety initiatives while environmentalists sought to have EPA take on an additional regulatory role beyond its voluntary prevention programs and data gathering efforts. When the Senate vigorously pursued giving EPA this additional authority, OSHA’s ability to address process safety faced possible preemption under the structure of the OSH Act. Provisions were added to the CAAA to avoid the EPA authority preempting OSHA from acting. Organization Resource Counselors, Inc. (ORC) prepared a widely-circulated draft of the PSM standard for OSHA. The Senate Report on the CAAA recognized that this draft contained much that was worthwhile but criticized it for, among other things, preempting state rules in the accident prevention area, protecting hazard evaluation material from public disclosure, and imposing "vague performance standards . . . which would be essentially unenforceable" (Sen. Rep. at 245). OSHA proposed the PSM Standard in the summer 1990 and later that year received an explicit mandate to promulgate a final rule in the CAAA. The final rule, promulgated in 1992, tracks the proposal fairly closely, although it contains a revision to the definition to the "process" that provides more specificity to that term. Proponents of the rule praise its flexibility and its "performance" basis. Critics continue to characterize it as not being substantively enforceable and some have criticized it as unconstitutionally "void for vagueness."
As previously discussed, EPA’s October 1993 Proposed Rule used the PSM rule as a starting point for its prevention program. To address some of the enforceability concerns, EPA added documentation requirements to several provisions. EPA also adjusted some of the portions of PSM that targeted effects on workers to target effects on the public and the environment. Industry criticized the proposal for any inconsistency with PSM, not only for the prevention aspects but also for the lack of compatible exemptions (exemptions are discussed below in the applicability / list / threshold section). For example, proposed requirements for Process Hazard Analyses (PHAs) to consider release impacts on the public and the environment were thought to require revising Hazard Evaluations performed under PSM. Even provisions EPA copied from PSM without modification raised concern because the general nature of PSM might lend itself to different interpretations of the same language. Industry argued for OSHA primacy in the area of prevention throughout the rulemaking process. Also, the ammonia refrigeration industry and other sectors that OSHA had not initially identified as impacted by PSM criticized the EPA proposal when it became clear that they were going to be subject to both OSHA and EPA regulation. Environmentalists, who may have thought the separate mandates to EPA and OSHA, ensured that EPA would not adopt PSM, generally criticized the prevention program and, as noted above, pushed for TOA.
In the final rule, EPA eliminated many of its modifications to PSM for Program 3 sources so that a company subject to PSM for a process would not have to revise its PSM compliance. EPA committed to coordinated interpretations for provisions that overlap with PSM. EPA retained an enforcement, implementation, and oversight role in the prevention area. Complete deferral to OSHA would be contrary to Congressional intent expressed both on the face of the statute and in the Senate Report. EPA did not adopt a TOA or inherent safety mandate for Program 3 processes. At least some commenters questioned the cost/benefit relationship of an inherent safety mandate. EPA expressed the view that good PHAs would identify opportunities for new technologies and that companies had market incentives for inherent safety when designing new sources.
Much of the information to date on the success of PSM in reducing accidents is anecdotal. Some advocates of PSM claim that its implementation actually reduces costs by leading to fewer malfunctions and by identifying other efficiencies. It will be hard to use accident history data to identify whether PSM itself is working because one must determine whether particular benefits are attributable to PSM or other aspects of rule compliance. For example, are reduced accidents due to PSM or to increased management focus brought about by assembling data into one RMP? Does public involvement cause risk reduction independent of the prevention program? Does dialogue with the public contribute to greater sensitivity about public impacts? Conversely, if reported accidents seem to increase, is the increase due to heightened sensitivity to reporting obligations? Finally, if one can show PSM works or even produces cost savings, does it make sense to extend PSM to more chemicals? The success of the prevention program may also be measured indirectly over time by the information submitted in RMPs on the types of process controls, detection, and mitigation systems and on changes implemented. These data give some indication of whether industry is investing in safer systems.
Rather than adopting the "tell us your program and stay with it" approach of the Supplemental Proposal for the content of Tier (Program) 2, the final Risk Management Program rule specified a "streamlined" version of PSM for the elements of Program 2. By including specific elements, EPA could also design an electronic reporting format for Program 2 to simplify RMP preparation and management. Consistent with EPA’s expressed intent to integrate the rule with existing safety programs that adequately addressed prevention, Program 2 allows a source to rely on existing industry codes and regulatory programs to fulfill parallel part 68 requirements. EPA has developed guidances that explain how several business sectors can comply with Program 2 and part 68 in general. EPA also explored other ideas for a non-PSM tier, such as a special set of requirements for Merit or Star rated sources under OSHA’s Voluntary Protection Program. Concerns about capacity of that program to adequately handle a large influx of new participants did not allow EPA to proceed with that approach. Mobil offered the most specific comments on how such a program might be structured.
The Emergency Response provisions of the Risk Management Program rule may carry out integration of part 68 with preexisting requirements to a greater extent than any other portion of the EPA rules. In parallel with the rule development process for part 68, EPA and other agencies that participate in the National Response Team developed the "One Plan" guidance on emergency response planning. An emergency response plan adopting the structure of the One Plan would have a single core of actions to take in planning and responding to emergencies and annexes that would address additional mandates of particular statutes and regulations. Part 68 codifies the statutory minima of CAA section 112(r) and allows a source to meet its response planning requirements by preparing a One Plan that contains these minima and that is coordinated with the community emergency response plan developed under EPCRA. C. Measurement Issues The efficacy of EPA’s extensive reliance on guidance, outreach, and integration of existing programs in designing and implementing the Program 2 and Emergency Response requirements can be tested against accident data as much as other parts of the regulations. In addition, a measure of success for the regulations would be to see if compliance with the preexisting integrated programs improves through the outreach and visibility of EPA’s program. There is already some evidence that EPA’s coverage of ammonia refrigeration has led to better awareness within the industry of PSM applying to types of refrigeration.
The Extremely Hazardous Substance (EHS) List consists of a group of toxic substances initially identified after Bhopal based on acute lethality. There are 360 chemicals that meet the listing criteria. EPA developed Threshold Planning Quantities (TPQs) for these chemicals based on a volatility and toxicity ratio score ("ranking") and a dispersion analysis assuming a 100-meter distance from release to the fence line (with like ranking chemicals "binned" with like thresholds). There are no exemptions from the EHS list. Facilities that have on-site more than a TPQ must participate in local emergency planning. EPA has considered expanding this list to include other classes of chemicals, such as flammables, explosives, and reactives, but has not made any regulatory proposal beyond an Advanced Notice of Proposed Rule for flammables and explosives in the early 1990s.
The Highly Hazardous Chemical (HHC) list and thresholds goes beyond the current EHS list to address flammable and reactive chemicals as well as a different, smaller group of toxics. The selected toxic chemicals generally have high scores on an industry-developed "Substance Hazard Index," although this was not an explicitly adopted criterion. Thresholds tend to be substantially higher than EHS TPQs for the same chemicals. In addition, a facility determines whether it has a threshold quantity on a process-by-process basis. The determination of what is a process involves engineering judgment (any activity or combination of HHC activities), although OSHA added some objective criteria in the final rule (interconnected vessels or vessels located so that the contents could be released at once). The HHC list and thresholds focus on workplace safety. It is unstated in the OSHA rule, but dispersion analysis is less critical to OSHA thresholds than to EPA because OSHA can assume that a worker may always be next to an accident at a non-remote process. OSHA did not have a mandate to protect workers from all potential exposures to HHCs caused by an accident; thus OSHA had the flexibility to adjust the thresholds on other criteria. With respect to exemptions, OSHA has specific safety programs for many types of processes and therefore could exempt such processes from the less specific, performance-based PSM rule without exposing workers to unreasonable risks.
EPA did not repromulgate either the EHS list or the HHC list when it developed the part 68 regulated substance list. Congress specifically directed EPA to look at other programs like New Jersey’s Toxic Catastrophe Prevention Act and other classes of chemicals like highly flammable substances when promulgating the list. The proposed part 68 list identified several classes of chemicals that EPA considered regulating under section 112(r): toxics on the EHS list, other toxics with similar properties, flammable substances, explosives, radionuclides, and reactive chemicals. The latter two classes were not proposed for regulation. Because the statute suggested prioritizing chemicals, EPA cut down the number of toxics from the EHS list using physical state and volatility criteria, and selected NFPA 4 flammables for its final list. These criteria focused on inherent properties of the chemicals rather than judgmental formulas such as the Substance Hazard Index method, and thus were thought to be more "objective." While the proposed EPA list was criticized for its differences with the HHC list on the one hand and for not covering all EHSs and more on the other hand, the most significant change in the final list was just to raise the volatility cutoff to eliminate many chemicals that were not involved in historic releases. In developing thresholds for part 68, EPA moved away from the TPQs while retaining the EPCRA ranking and binning method. EPA had commented to OSHA that its HHC thresholds were inconsistent with and significantly higher than EHS thresholds, but this did not lead OSHA to lower the HHC thresholds. For part 68, EPA proposed thresholds that exceeded the EPCRA TPQs and then increased the thresholds again in the final rule. Ultimately, EPA ended up with thresholds that were greater than or equal to OSHA’s for all chemicals. EPA’s rationale for exceeding the OSHA thresholds was based on the notion that thresholds that were protective of workers would be protective of the public. To maintain consistency with PSM, EPA also determines applicability on a process-by-process basis. All exemptions that EPA has promulgated to date, except for the exemption for agricultural nutrients held by a farmer, have been based on the threshold provision. Generally, the thresholds are in the form of "when determining whether a threshold has been exceeded, do not consider . . . ." EPA has generally based its exemptions on a hazard or risk-based rationale. D. Measurement Issues Many interpretive questions to the agencies since promulgation of both PSM and part 68 indicate there is some confusion in determining what is a process. Some have recommended that a site-wide calculation of threshold quantities would be more objective and simpler to implement in practice. The process definition could be gamed by subdividing processes to avoid the applicability thresholds or to become subject to Program 1. Conversely, grouping together processes minimizes some compliance requirements. Uncertainty in applicability can be measured through compliance assistance, reports, inspections, audits, and enforcement. The dissimilarity between the three Federal lists discussed above, as well as state lists, would allow for a meaningful comparison of which lists capture most accidents. According to EPA’s technical support documents and response to comments, the adjustments made to the part 68 list and thresholds allowed EPA to retain most of the historical accidents involving air releases while substantially reducing the cost of the program over the original proposed list. If an analysis of release reporting databases like ERNS demonstrates that facilities / sources that do not have threshold quantities of chemicals in processes are involved in significant accidents, then it may suggest that thresholds are too high. The assumption that part 68 thresholds should be higher than HHC thresholds because workers are closer to accidents than the public could be tested by observation and by modeling. For example, it may be easy to measure how protective the various thresholds are by modeling different quantity releases on various models.
The statutory imperative of 112(r)(7)(B) is that a source not only must prevent and respond to accidents, but also must disclose its actions to the public. At the same time, the very provision that mandates this disclosure also recognizes that businesses have a legitimate interest in confidential business information (CBI). Process information can be a business’s most sensitive information. From the proposal to the final rule, part 68 has struck a balance between these two conflicting mandates by distinguishing the substantive actions a source must undertake, called the "Risk Management Program," from the report on these activities, called the "Risk Management Plan" or "RMP." Some of the environmentalists most active in addressing chemical accidents and in Right-to-Know issues have been among the most vehement critics of EPA’s failure to mandate more disclosure in RMPs. They make a statutory argument that because the statute refers to the hazard assessment, prevention program, and response program as the components of the RMP, the RMP must contain much, if not all, of the information required to document compliance with these statutory requirements. Moreover, in light of the previously referenced Senate Report criticism of the ORC draft PSM rule, they can argue that the Senate intended a different balancing between CBI interests and disclosure. Finally, many sense that the emergency planning provisions of EPCRA do not effectively integrate public access into planning, especially in contrast to the accessibility of the Toxic Release Inventory and its effect on stimulating pollution prevention. Industry’s criticism of the RMP during rule development derived from the lack of clarity in what needed to be reported. Industry may have read the proposed RMP to say, "now that you’ve done the program, tell us what you did, and keep updating your summary when you do more." While EPA emphasized the summary nature of the RMP and that it did not intend companies to have to disclose CBI, commenters looked at the proposed rule and believed that careful lawyers would advise submitting cabinets of information to designated recipients to minimize liability. This image also tended to provoke hostile reactions among state officials whose programs lacked resources to manage the perceived volume of information. EPA always was sensitive to the data management impacts of the rule, not only for the burden it imposed on reporters and implementing agencies, but also because it did not want the RMP to become an unwieldy and inaccessible document that would be warehoused and not used. While the proposed RMP left great freedom for different presentation of information, the final rule adopted specific standardized data requirements. The standardized data are intended to capture enough information about the implementation of part 68 so that the public can understand basic facts about the source’s Risk Management Program and a knowledgeable reader will be able to make inferences about the quality of the source’s program. For example, if dates of certain documentation updates significantly precede major changes at the source, this RMP information may suggest a weakness in a source’s implementation of PSM and part 68. The EIA counts standardization of content as the most significant cost savings between the Supplemental Proposal and the Final Rule. Recent action by EPA to require up-front substantiation of CBI claims and limitations on what types of data may be claimed as CBI refines the balance between legitimate business secrecy concerns and the Right-to-Know. The Chemical Manufacturers Association expressed concern that the broad statements by EPA that the RMP did not require disclosure of CBI were meant to prohibit such claims. EPA did not intend to prohibit such claims when meritorious. However, EPA also recognized that the ordinary structure of handling CBI claims might encourage gamesmanship by parties seeking to frustrate public access even for a short time until the initial attention cycle for RMPs passes. The new provisions prohibit CBI claims for many data elements that do not meet the test for CBI.
The statute itself mandates that sources register RMPs with EPA and also submit RMPs to the Chemical Safety and Hazard Investigation Board, states, and local emergency responders. It does not specifically mandate submission to State Emergency Response Commissions and Local Emergency Planning Committees created under EPCRA. Unless one reads the registration requirement to also require submission of an RMP, there is no mandate to submit a copy to EPA or whatever agency is implementing the program in an area. Even though the requirements of section 112(r) are applicable requirements of CAA operating permits, there is no mandate for sources to submit RMPs to the permitting agencies. Finally, while section 112(r) mandates that RMPs be made available to the public, the statute does not provide for who should fulfill this duty or how it should be met. Thus, the submission language barely begins to address how all these potential users of RMPs could have their information needs met. When EPA proposed the Risk Management Program rule, EPA invited comment on how to simplify data management for RMPs. Specifically, EPA sought comment on electronic reporting as a way of handling data. This concept of electronic reporting was not developed further or addressed at the Supplemental Proposal stage. Nevertheless, at the final rule stage EPA adopted centralized electronic reporting as the way to fulfill the multiple submission and data availability requirements of the statute, and deferred specifying the method and format for such submission to a stakeholder process after the rulemaking. EPA identified electronic reporting as a way to fulfill numerous executive orders and statutes requiring the agency to select the least burdensome methods of implementing statutory mandates, including the requirement to avoid unfunded mandates on state and local governments. Subsequent to the final rule, EPA determined that it would not post certain portions of the RMP, particularly OCA data, on the Internet while all other parts of the RMP would be available in that manner. This results in a bifurcated access scheme, where some data are easily accessible to the statutorily mandated recipients while other data require a different means of access with additional data management burdens.
EPA has not quantified the accident reduction benefit associated with the preparation, submission, and availability of RMPs. If one believes that what gets measured gets managed, then assembling a summary of disparate activities that have an effect on the likelihood of an accident and presenting this summary to responsible corporate officials will result in greater management focus on preventing accidents. This should be true even for sources that can comply with part 68 without any rule-mandated prevention or response program changes. Certification and submission of the RMP increase the significance of the information summarized and should help bring additional management focus on accident prevention. Of course, many attribute reductions in toxic releases to the public availability of TRI, and if one accepts that belief, then public access to RMPs should reduce accidents. Given EPA’s extensive integration of PSM and other programs in its part 68 regulations, the RMP may be the single most significant addition to accident prevention resulting from the EPA rules. During the oversight phase after June 21, 1999, it should be possible to see if RMPs provide summaries that are useful to various stakeholders. If one cannot begin to tell a good risk management program from a bad one by looking at an RMP, then EPA will need to consider again the balance it struck in deciding the appropriate types of disclosure. The raw number of website hits should tell something about the utility of the RMP to people who access the data, although a hit count will not tell much about how the RMP is used. What will be difficult to measure is the effect of not having complete RMPs accessible to the public and agencies via the Internet. How many terrorist incidents were avoided? How many accidents were not prevented? How many people failed to look at the data at all because the data of most interest to the public are not readily available? How is the balance between flexibility and accountability altered? Conclusion The general structure of both the PSM rule and the Risk Management regulations affords substantial flexibility to regulated entities to prevent accidents using good engineering judgment. Both rules affect more facilities than either or both agencies together could oversee. The discussion above and that which will occur over the next two days will suggest ways to measure whether the agencies’ programmatic approach of balancing flexibility and accountability minimizes catastrophic accidents and what aspects of this approach need improvement. Note About Sources The publicly-available record for the decisions and policy choices made during EPA’s program development is massive. The most important historical resources for research into EPA’s decision making would include the published rulemaking notices, the technical support documents and Economic Impact Analyses (EIAs) accompanying these notices, the 1700 page Response to Comments for the 1996 "Risk Management Program Rule," and, to a lesser extent, interagency preliminary drafts of the rulemaking notices available in EPA’s Air Docket. The Senate Report on its version of the bill that became the CAAA contains the most systematic discussion of most of the statutory provisions that became CAA section 112(r) due to the House Commerce and Energy Committee Bill being substantially rewritten in a floor amendment. Additional insight into EPA’s thinking can be gained from numerous public meetings it has had with stakeholders, subcommittee and work group meetings of the Accident Prevention Subcommittee of the Clean Air Act Advisory Committee, and publicly available guidance. Statements about the basis for EPA’s decisions are drawn from the public record. The author has mainly relied on his memory of these documents for statements about rule development because this paper is for discussion purposes and not intended for publication in the current format. Thus, there are no citations for most statements. Any inconsistencies with matters in the record for the part 68 rules or otherwise previously in the public domain are inaccurate misstatements for which the author is solely responsible.
|
||
|
About
NCSP | Reports | Data
| Tips | News
| Calendar | Education
| Contact us Texas
Engineering Experiment Station
|