The U.S. Chemical Safety and Hazard Investigation Board investigates serious chemical accidents at fixed facilities. It publishes what happened and why, and it sends recommendations to companies, trade groups and government agencies. It has no power to fine a company or to write a rule that a plant must follow. Congress designed it that way in the 1990 Clean Air Act amendments, using the National Transportation Safety Board as the model. This guide covers where the board came from, how it differs from the NTSB, and how an emergency planner can use a closed report. The example is the board’s long-settled report on the 2005 BP Texas City refinery explosion.

1984 to 1990: Bhopal, Institute, Pasadena and a board on paper

The board exists because a run of accidents in the 1980s convinced Congress of a gap. The federal government had no standing body whose job was to work out why chemical releases happened. Most accounts name the December 1984 release of methyl isocyanate at Bhopal, India, as the first of these events, and the site covers it in a separate article. The next year a release at Union Carbide’s plant in Institute, West Virginia, sent people from the surrounding area to hospitals. That plant was the company’s American methyl isocyanate facility, so the release brought the question home to the United States. In October 1989 an explosion at the Phillips Petroleum chemical complex near Pasadena, Texas, killed 23 workers, which is the figure given in the published accounts of that disaster.

Congress rewrote the Clean Air Act in 1990 and added accidental release prevention provisions in section 112(r). Those provisions led to the EPA’s Risk Management Program rule, which covers facilities holding listed toxic and flammable substances above threshold quantities. The same amendments directed OSHA to issue the rule that became its process safety management standard. Within section 112(r), Congress also created an independent Chemical Safety and Hazard Investigation Board. The thinking was that the agencies writing and enforcing the rules should not be the only ones explaining the accidents. An agency investigating a failure that its own regulation might have missed has an obvious conflict of interest.

The board existed in law for years before it existed in practice. Congress did not fund it until the late 1990s, and most accounts date the start of its operations to 1998. When you see 1990 given as the founding date, read it as the date of the legal authority, because investigators did not go to work until most of the decade had passed.

The mandate: a five-member board that investigates and recommends

As the statute is generally summarized, the board investigates accidental releases from stationary sources that cause a death, a serious injury or substantial property damage. For each one it reports to the public the facts, conditions and circumstances, along with the cause or probable cause. It also recommends measures to prevent a repeat, and it may study chemical accident hazards more broadly. I am paraphrasing here. Anyone who needs the exact wording should read section 112(r) of the Clean Air Act itself, which is codified in Title 42 of the United States Code.

The law provides for a board of five members, appointed by the President with the advice and consent of the Senate. A staff of investigators, engineers and analysts supports them. I am not giving a staff count or a budget figure in this guide. Both have changed over the years, and neither is needed to understand what the board does. Readers who want the current picture should check the board’s own website and its most recent budget documents, not a summary like this one.

The board does not investigate every chemical accident in the country, and the statute does not require it to. Thousands of releases are reported every year to the National Response Center and to state agencies. The board picks a small number for full investigation, generally based on how severe the incident was, whether its lessons apply beyond one site, and what resources the board has.

The board publishes several kinds of products:

• full investigation reports, which can run to hundreds of pages
• shorter case studies and safety bulletins
• animated safety videos, which many hazmat and industrial training programs show to their students

For a working agency, the videos are often the product that actually reaches the crews.

No fines and no rules: the separation and why Congress built it

The board cannot issue a fine, write a citation, order a facility to shut down, or adopt a safety standard that a plant must meet. Its authority covers two things. It can investigate, which includes gathering evidence at the site and from documents and witnesses, and it can publish what it finds. Any change that follows a CSB report has to be made by someone else. That might be OSHA or the EPA revising a rule or how they enforce it, a state legislature or fire marshal, an industry body revising a standard, or the company that had the accident. When a news story says the Chemical Safety Board fined a company, the reporter has confused the board with OSHA or the EPA. Those agencies often investigate the same event at the same time for enforcement purposes.

The reasoning behind the separation matches the reasoning behind the NTSB. A regulator investigating an accident at a plant it regulates is partly investigating itself, because one possible finding is that its rule was weak or its inspectors missed something. A body with no rules to defend and no penalties to collect can say that a standard was inadequate or that an agency’s oversight failed. The CSB has made recommendations to OSHA and the EPA that amount to exactly that. Defenders of the model also argue that witnesses and companies may cooperate more fully with an investigator whose findings will not become a penalty. That protection is partial, because enforcement agencies may be investigating the same event in parallel.

The statute reinforces the separation with a provision similar to one in the NTSB’s law. It keeps the board’s conclusions, findings and recommendations about an accidental release from being admitted as evidence in a lawsuit for damages. Whether the underlying factual material is treated the same way is a separate question, and how courts have handled that distinction is a matter for an attorney. For planners, the practical point is that a CSB report is written to prevent the next accident, and its authors are not deciding who owes money for this one.

The NTSB model: the same method and a different standing at the scene

Congress modeled the CSB on the National Transportation Safety Board, and the resemblance is real. The NTSB was created in 1967 within the new Department of Transportation. The Independent Safety Board Act of 1974 made it independent of that department. It has no enforcement power either. The FAA, the Federal Railroad Administration, the Pipeline and Hazardous Materials Safety Administration and the other modal regulators write and enforce the rules. The NTSB investigates, makes recommendations, and publicly tracks whether they are adopted. Both boards rely on the regulators and on public attention to turn a recommendation into a change, and the press routinely credits each of them with more power than it actually has.

For a working responder, the differences matter more than the similarities. In transportation accidents, the NTSB’s statute generally gives its investigation priority over those of other federal agencies. The main exception is a case the Attorney General determines involves an intentional criminal act. The NTSB also runs a party system, in which the operator, manufacturers and others assist under its direction. The CSB has no equivalent priority at a fixed facility. OSHA and the EPA may be running their own investigations at the same site, under their own authority and for their own purposes, and the board has to coordinate its evidence collection with theirs.

The two boards also divide the ground between them. Releases during transportation, such as a derailed tank car or an overturned cargo tank, generally fall to the NTSB. The CSB focuses on fixed facilities such as refineries, chemical plants and storage sites. The NTSB opens a case on every civil aviation accident in the United States, while the CSB takes up only a selected few of the serious chemical incidents that occur.

The reports differ too. An NTSB final report closes with a formal probable cause statement, which the site’s article on reading a closed NTSB case covers in detail. CSB reports generally present key findings, causes and recommendations without a single one-sentence statement of that kind.

BP Texas City 2005: an overfilled tower and trailers near the unit

The example here is the board’s investigation of the explosion at the BP refinery in Texas City, Texas, on March 23, 2005. Do not confuse it with the 1947 ammonium nitrate ship explosion in the same port city, which the site covers separately. The board released its final report on the 2005 explosion in March 2007, which closed the investigation. The related enforcement actions concluded years ago, and the report has been used in process safety training ever since, so it is a settled record to work from. Everything I say about causes in this section and the next is the board’s own finding from that final report, attributed to the board, and not my own judgment about what went wrong.

The CSB’s final report describes the sequence this way:

• Operators were restarting the refinery’s isomerization unit after a maintenance period.
• During startup, a distillation tower called the raffinate splitter filled with flammable liquid far above its normal level. The board found that the level instruments gave operators a misleading picture of how full the tower was.
• The liquid overflowed into the relief system and went to a blowdown drum. That drum vented to the atmosphere through a stack instead of to a flare.
• Flammable liquid and vapor discharged from the top of the stack and formed a vapor cloud at ground level, which then ignited.

The final report also addresses how the cloud came to ignite. Readers who need that finding should take it from the report itself, in the board’s own wording, rather than from the condensed versions that circulate in training slides and news coverage.

The CSB’s final report gives the toll as 15 killed and 180 injured. I use those figures because they come from the investigation of record. If you see a different injury count elsewhere, go back to the report rather than average the two. The people who died were contract workers, and summaries of the report place them in or around temporary trailers set up near the isomerization unit to house workers for maintenance elsewhere in the refinery. That detail is what makes the report useful to emergency planners, because the deaths were concentrated among people who had no role in the startup and who were near the unit only because of where their temporary workspace had been put.

The 2007 report: organizational causes and where the recommendations went

The report went well beyond the tower and the release. As the board’s closed final report is generally summarized, the board attributed the accident to organizational factors, which it set out as its own conclusions:

• a startup procedure that was not followed and had not been kept current
• instruments and alarms that were not working as needed
• a long-standing decision to keep an atmospheric blowdown system in service when a flare system would have been safer
• cost-cutting and staffing decisions
• a safety management system that tracked personal injury rates and paid too little attention to process safety

The finding about injury metrics has had the longest life outside the refinery industry. The board argued that a low rate of slips, trips and recordable injuries said very little about whether a refinery was heading toward a catastrophic release. The distinction between personal safety and process safety is now standard vocabulary in the industry.

Before the final report came out, the board issued an urgent recommendation in 2005. It asked BP to commission an independent panel to examine safety culture and corporate oversight at its North American refineries. BP did so, and the panel, chaired by former Secretary of State James A. Baker III, published its report in January 2007. The board also recommended that the American Petroleum Institute develop guidance on keeping occupied trailers and other portable buildings away from hazardous process areas. API later issued a recommended practice on portable buildings at process plants, known as RP 753. If you apply that guidance, use the current edition from API, not the description here.

The recommendations to OSHA concerned its oversight of refineries and its process safety management standard. OSHA started a national emphasis program for petroleum refineries in 2007. To see how the board judged the response to each recommendation, look up its status in the board’s recommendation records, which show whether a recipient acted or declined.

The enforcement consequences came from other agencies entirely. OSHA cited and penalized BP, and the Department of Justice brought a criminal case under the Clean Air Act that ended in a guilty plea. None of that was the CSB’s doing, and it shows the division of labor clearly.

Local planning: a closed report beside Tier II and RMP data

In the United States, local planning for chemical hazards runs largely through the structure set up by the Emergency Planning and Community Right-to-Know Act of 1986. Local emergency planning committees write plans covering facilities that hold extremely hazardous substances. Facilities file annual hazardous chemical inventories, the Tier II reports, with the state, the LEPC and the local fire department. Facilities covered by the EPA’s Risk Management Program also prepare risk management plans, which include hazard assessments and worst-case release scenarios. Access to the off-site consequence analysis portions of those plans is restricted for security reasons. Ask your EPA regional office or your state agency how access works where you are.

Those filings tell a planner what is in the county and roughly how bad a release could be. A CSB report adds something they cannot: a detailed account of how a real facility with a similar process actually failed, and how people on and around the site were affected.

Read a report for mechanism. If your jurisdiction has a refinery, a chemical distributor or a large ammonia refrigeration system, search the board’s published investigations for the same process or chemical. Read the executive summary and the key findings first. Then read whatever the report says about emergency response, warnings and the population off site. Texas City is a refinery case, but its planning lesson about where people are placed relative to a process unit applies to any facility where occupied buildings, contractor trailers or a command post might end up inside the hazard area.

For response planning, Texas City points to several items for a hazmat annex or a refinery preplan:

• where the plant places contractor trailers during turnarounds, when the number of people on site can rise well above normal staffing
• how the facility accounts for those people after a release, and how that headcount reaches the incident commander
• where mutual aid companies stage and where the command post sets up, relative to the units that could release a vapor cloud

The board’s videos are a practical way to put the mechanism in front of crews. A short animation of a release sequence is more likely to reach a company at shift change than a report several hundred pages long.

Limits: what a closed CSB report cannot tell a planner

A closed report describes a facility as it was on the day of the accident and in the period before it. Since then the plant may have been rebuilt, sold or closed, and its procedures have very likely changed. A report is therefore no evidence about current conditions at any site, including the one investigated. BP sold the Texas City refinery in 2013. Anyone planning around that site today should work from the current owner’s plans and current regulatory filings, not from the 2007 report.

The board’s choices also shape the record. It investigates only a selected few incidents, so the absence of a CSB report on a type of facility means only that the board did not select one or has not finished one, and it carries no information about how safe that type of facility is. Some hazards that matter a great deal to local responders may be represented by very few investigations, and planners should not read that as a statement about risk.

The reports are written for prevention, and most recommendations go to companies, trade associations and regulators. Some investigations have dealt directly with emergency response and community planning, and those sections deserve close reading. Even so, tactics, protective action distances and isolation zones should come from sources written for that purpose, including the current Emergency Response Guidebook, your hazmat team’s procedures and the agency with jurisdiction, with the CSB report serving as background on how a facility failed. I have also kept this guide away from the board’s current budget, staffing and status, because those change over time. Check the board’s website for where it stands now.

What to do at your agency

• The LEPC coordinator or county emergency manager takes the current list of facilities in the LEPC plan and searches the CSB’s published investigations for each facility’s main process or chemical. The job is finished when every facility has either a matching report title or a note that none was found.
• The training officer adds one CSB safety video that matches a local hazard to an existing drill or in-service session this month. The drill sheet records which local facility the crews discussed afterward.
• The hazmat team leader or operations chief reviews the preplan for each refinery or large chemical plant in the response area. Each preplan gets one line stating where contractor trailers and temporary occupied buildings are placed during turnarounds, and if the answer is missing, the team leader gets it from the facility.
• The emergency manager asks the emergency response coordinator at each RMP-covered facility how the site accounts for contractors after a release and how that headcount reaches the incident commander. The answers go on the agenda of the next regular LEPC meeting.
• The emergency manager adds one paragraph to the existing hazardous materials annex. It states that CSB reports are independent safety investigations used for hazard analysis and training, and that enforcement findings are cited from OSHA, EPA or state records instead.
• The planner checks the status of each CSB recommendation that applies to a facility type in the jurisdiction, using the board’s recommendation records. The hazard analysis then notes which ones were closed without the recommended action.

Takeaways

• The Chemical Safety Board was authorized by the accidental release provisions of the 1990 Clean Air Act amendments, and most accounts date the start of its operations to 1998, after Congress funded it.
• The board investigates selected serious chemical accidents at fixed facilities and makes recommendations, and it has no power to fine, cite or regulate anyone.
• Enforcement after a chemical accident comes from OSHA, the EPA, state agencies and the Justice Department, which may investigate the same event in parallel with the board.
• Congress modeled the CSB on the NTSB, but the NTSB holds a statutory priority in transportation investigations that the CSB does not have at fixed sites.
• The CSB’s final report on the 2005 BP Texas City explosion gives 15 killed and 180 injured, and its findings on trailer siting and on process safety versus personal injury metrics are still widely used.
• A closed CSB report is most useful to a planner when read for mechanism, matched to a local facility with the same process or chemical, and used alongside Tier II and risk management plan data.
• The board’s recommendation records show what happened after a report, and a recommendation closed without action can describe a gap that is still open.
• A CSB report describes a facility as it was at the time of the accident, and it is no evidence of current conditions at any site.