Industrial chemical safety is often described in terms of regulations, audits, and procedures, but at its core it is a data problem. OSHA Process Safety Management (PSM) and EPA Risk Management Program (RMP) requirements exist because catastrophic releases of toxic, reactive, flammable, and explosive chemicals have repeatedly shown that incomplete hazard knowledge can kill workers, injure communities, destroy assets, and damage public trust. OSHA states that its PSM standard is intended to prevent or minimize the consequences of catastrophic releases of toxic, reactive, flammable, or explosive chemicals, while EPA’s RMP rule requires covered facilities to analyze potential accident consequences, implement prevention programs, and prepare emergency response measures.
PSM and RMP are complementary safeguards
OSHA PSM, codified at 29 CFR 1910.119, focuses primarily on protecting employees from catastrophic chemical releases inside the workplace. It applies to processes involving listed highly hazardous chemicals at or above threshold quantities and to certain quantities of flammable gases and liquids. EPA RMP, codified at 40 CFR Part 68 under Clean Air Act Section 112(r), extends the prevention and planning framework to the surrounding community by requiring covered facilities to submit a Risk Management Plan, perform hazard assessments, document accident history, implement prevention programs, and establish emergency response programs.
The two programs overlap because the same physical event can injure both workers and neighbors. A runaway reaction, vapor cloud explosion, dust explosion/fire, toxic gas release, or relief system discharge does not respect the fence line. PSM asks, “How do we prevent the process from failing?” RMP asks, “If it fails, who offsite could be affected, how far could impacts extend, and how will the facility and community respond?” EPA’s RMP overview explicitly states that the plan identifies potential effects of a chemical accident, prevention steps, and emergency response procedures, and that the information helps local fire, police, and emergency responders prepare for chemical emergencies.
The foundation is Process Safety Information
Every high-quality PHA, LOPA, consequence analysis, relief evaluation, emergency response plan, and siting study depends on Process Safety Information (PSI). OSHA and EPA both require PSI before process hazard analysis. EPA’s Program 3 PSI rule requires written information on the hazards of regulated substances, process technology, and process equipment, including toxicity, physical data, reactivity data, corrosivity data, thermal and chemical stability data, hazardous effects of inadvertent mixing, process chemistry, maximum intended inventory, safe operating limits, consequences of deviations, P&IDs, materials of construction, relief system design basis, ventilation design, and safety systems.
This is where laboratory testing becomes indispensable. Safety Data Sheets are useful starting points, but they are often generic, may not reflect actual process mixtures, may not cover impurities or recycle streams, and may not describe behavior under credible abnormal conditions. PSM and RMP analyses require facility-specific data: decomposition onset temperature, heat of reaction, adiabatic temperature rise, gas generation rate, vapor pressure of mixtures, flammability limits, minimum ignition energy, dust explosibility, corrosion rate, compatibility with elastomers and metallurgy, reaction kinetics, vent sizing data, toxicity data for mixtures, and physical properties across the expected operating envelope. EPA’s PSI requirements explicitly include reactivity, corrosivity, thermal and chemical stability, safe upper and lower limits, and consequences of deviations, which are precisely the types of values often generated or validated through laboratory and pilot-scale testing.
Laboratory testing turns “unknown unknowns” into design requirements
A disciplined laboratory test program does more than populate a spreadsheet. It changes engineering decisions. Reaction calorimetry can show whether a batch operation has sufficient cooling capacity or whether a credible cooling failure will lead to thermal runaway. Accelerating rate calorimetry, differential scanning calorimetry, and vent sizing package (VSP2) tests can measure onset temperatures, pressure rise rates, gas generation rates, self-heating rates, and emergency relief requirements. Flash point, autoignition temperature, vapor pressure, and electrostatic ignition sensitivity data influence area classification, ventilation, inerting, bonding and grounding, and hot work controls. Corrosion coupon testing and electrochemical testing can validate materials of construction and inspection intervals. Compatibility testing can reveal whether an inadvertent mixing scenario produces heat, pressure, toxic gas, polymerization, or unstable intermediates.
These data directly support both PSM and RMP. PHA teams need credible causes, consequences, safeguards, and severity estimates. Relief system engineers need runaway pressure and gas generation data. Mechanical integrity teams need corrosion and degradation rates. Operating procedures need safe temperature, pressure, composition, and feed-rate limits. Management of change reviews need evidence that a new raw material, catalyst, solvent, impurity profile, or operating condition does not invalidate prior assumptions. RMP offsite consequence analysis requires release quantity, release rate, release duration, physical state, scenario type, endpoint distance, and mitigation assumptions, all of which depend on accurate physical and chemical data.
Case 1: T2 Laboratories, Jacksonville, Florida
The 2007 T2 Laboratories reactive chemical explosion is one of the clearest examples of why laboratory-generated process safety data matters. The CSB reported that four people were killed and 13 were transported to the hospital when an explosion occurred during production of methylcyclopentadienyl manganese tricarbonyl, a gasoline additive. The NTIS abstract of the CSB report states that the explosion injured 32 people, damaged buildings within a quarter mile, sent debris up to one mile away, and resulted from a runaway exothermic reaction during the first process step after loss of sufficient cooling led to an uncontrollable pressure and temperature rise.
The technical lesson is not simply “cooling failed.” The deeper lesson is that process scale-up without adequate reactive hazard characterization leaves engineers blind to how fast a reaction can accelerate when control is lost. A PHA based on incomplete calorimetry can underestimate severity, safeguards can be mismatched to the hazard, and relief systems can be inadequate for the true pressure generation rate. The CSB identified key issues including reactive hazard recognition, emergency preparedness, process design, and scale-up, all of which depend on process-specific experimental data rather than generic literature values.
Case 2: BP Texas City Refinery, Texas
The 2005 BP Texas City refinery explosion shows how PSM failures can transform abnormal operation into mass casualties. CSB reported that explosions occurred during restart of a hydrocarbon isomerization unit, killing 15 workers and injuring 180 others. The release occurred after a distillation tower flooded with hydrocarbons and was overpressurized, causing a geyser-like release from a vent stack. OSHA later summarized that the incident involved 15 contractor fatalities and at least 170 injuries, and cited the CSB conclusion that the disaster was caused by organizational and safety deficiencies at all levels of BP.
Laboratory testing alone would not have prevented Texas City, because operating discipline, siting, alarms, management systems, and culture were central issues. But accurate process data is still part of the prevention chain. Safe operating limits, consequences of deviations, vapor generation potential, relief system behavior, human factors, and startup transient behavior should all be reflected in PSI, PHA, procedures, operator training, and emergency planning. If a PHA treats overfill, flooding, or relief discharge as a manageable nuisance rather than a credible catastrophic vapor cloud scenario, safeguards will be weak. The PSM lesson is that data must be translated into operational controls, and those controls must be maintained through training, audits, MOC, and leadership accountability.
Case 3: Chevron Richmond Refinery, California
The 2012 Chevron Richmond refinery fire illustrates why materials data and degradation testing are part of process safety, not merely inspection program details. CSB reported that an August 6, 2012 release of flammable vapor at the Chevron Richmond refinery led to a fire. The CSB recommendations included revisions to API guidance to address sulfidation corrosion in low-silicon carbon steel piping and to require users to identify susceptible piping circuits, inspect components, or replace at-risk carbon steel with more resistant alloys.
The broader lesson is that PSI must include accurate materials of construction, corrosion mechanisms, damage rates, and inspection data. Laboratory and field metallurgical testing can identify alloy composition, corrosion susceptibility, and degradation under actual service chemistry. In this case, the hazard was not a mysterious chemical reaction in a reactor; it was a predictable damage mechanism that required better recognition, better data, and stronger safeguards. PHAs that do not integrate damage mechanism reviews may miss scenarios where a small piping component becomes the initiating event for a major release.
Case 4: West Fertilizer, Texas
The 2013 West Fertilizer explosion demonstrates the RMP dimension of chemical safety: community consequence. CSB reported that a massive explosion at a fertilizer storage and distribution facility fatally injured 12 volunteer firefighters and two members of the public and caused hundreds of injuries. The CSB final report identified key issues including regulatory oversight, hazard awareness, emergency planning and response, fertilizer-grade ammonium nitrate storage practices, and land use planning and zoning.
West Fertilizer is important because it shows that a facility’s hazard analysis cannot stop at the process boundary. Fire exposure behavior, decomposition potential, storage configuration, contamination sensitivity, combustible construction, separation distances, emergency responder awareness, and offsite receptors all matter. RMP-style thinking asks how an incident develops, what populations or responders could be exposed, and what the facility has communicated to local emergency planners. Laboratory testing and published test data on thermal stability, contamination, burning behavior, and detonation sensitivity are essential to classify hazards correctly and design storage, separation, firefighting, and emergency response strategies.
Case 5: DuPont La Porte, Texas
The 2014 DuPont La Porte methyl mercaptan release shows the deadly interaction between toxic release hazards, building ventilation, troubleshooting, emergency response, and process safety culture. CSB reported that four workers were killed and one was injured when methyl mercaptan, a toxic chemical used in insecticide and fungicide manufacturing, was released. CSB’s news release stated that nearly 24,000 pounds of methyl mercaptan escaped through two valves in a poorly ventilated manufacturing building, and that the investigation identified flawed engineering design, lack of adequate safeguards, and safety management shortcomings.
For PSM and RMP practitioners, the data lesson is clear: toxicological properties, vapor density, odor warning limitations, ventilation effectiveness, detector placement, release rate, and dispersion behavior must be understood before emergency plans are needed. Laboratory and engineering data support detector selection, alarm setpoints, ventilation design, respiratory protection, evacuation zones, and alternative release scenarios. Under RMP, facilities must submit offsite consequence data including chemical name, physical state, scenario type, quantity released, release rate, duration, distance to endpoint, and receptors within that distance.
Case 6: Bhopal and the origin story of modern process safety
The 1984 Bhopal disaster remains the defining warning for toxic chemical risk. CSB’s 30th anniversary safety message stated that water entered a storage tank containing more than 80,000 pounds of methyl isocyanate, which reacted violently with water, overheated the tank, and produced a massive toxic gas release over the city. CSB also stated that the tragedy killed thousands immediately, injured tens of thousands, and helped drive new laws for chemical emergency preparedness, process safety, risk management programs, worst-case release reporting, and creation of the CSB. Britannica describes Bhopal as a 1984 chemical leak in which methyl isocyanate escaped from a Union Carbide subsidiary plant, with investigations later identifying substandard operating and safety procedures at an understaffed plant.
Bhopal remains relevant because MIC’s reactivity with water, toxicity, storage inventory, refrigeration, scrubber and flare availability, emergency planning, and community proximity were all central to the disaster narrative. These are not abstract compliance items. They are data-driven design and management decisions. A modern RMP worst-case and alternative release analysis exists because communities need to know whether a credible toxic release can travel beyond the fence line, and emergency responders need realistic planning assumptions before an event occurs.
What laboratory testing should feed into PSM and RMP
A strong testing strategy should be risk-based, documented, and connected to decisions. For reactive systems, test to determine heat of reaction, onset temperature, adiabatic temperature rise, pressure and temperature rise rates, gas generation, accumulation potential, contamination sensitivity, and emergency relief requirements. For flammable systems, test for flash point, boiling range, vapor pressure, flammable limits, autoignition temperature, minimum ignition energy, electrostatic properties, and combustible dust parameters where applicable. For toxic systems, validate volatility, vapor density, detection limits, decomposition products, and mixture hazards. For mechanical integrity, measure corrosion rate, compatibility, erosion, stress corrosion cracking susceptibility, elastomer compatibility, and metallurgy. For RMP consequence analysis, ensure accurate release quantity, liquid fraction, physical state, release rate, mitigation effectiveness, and endpoint inputs. EPA’s required RMP offsite consequence fields demonstrate why these values must be technically defensible.
The most effective companies treat testing as part of lifecycle process safety. During R&D, testing identifies hazards before scale-up. During design, it sets safe operating limits and relief requirements. During operations, it supports PHA revalidation, MOC, mechanical integrity, and procedure updates. During incident investigation, it tests hypotheses and prevents recurrence. During RMP updates, it ensures offsite scenarios are based on current chemistry, inventory, equipment, and mitigation assumptions. EPA requires RMP resubmission every five years, and OSHA requires PSI to support PHA and the broader PSM system, so the data must remain alive and maintained, not just archived after startup.
Conclusion: compliance is the floor, data-driven safety is the goal
OSHA PSM and EPA RMP are not paperwork exercises. They are structured ways to force organizations to understand major hazards, design safeguards, train people, maintain equipment, manage change, and prepare for emergencies. But the system is only as strong as the data beneath it. If reactivity data are missing, a runaway reaction can be underestimated. If corrosion data are incomplete, a pipe can fail before inspection plans catch it. If vapor pressure or toxicity data are wrong, an RMP endpoint can be misleading. If safe operating limits are copied from old documents rather than validated, operators may be asked to run inside an unsafe envelope.
The published accident record is painfully consistent. T2 Laboratories shows the cost of inadequate reactive hazard understanding. BP Texas City shows the cost of weak PSM execution during startup. Chevron Richmond shows the cost of incomplete damage mechanism control. West Fertilizer shows the cost of poor hazard awareness and community planning. DuPont La Porte shows the cost of toxic release scenarios that were not adequately controlled. Bhopal shows why modern PSM and RMP expectations exist at all. The lesson for every covered facility is simple: invest in laboratory testing, keep PSI current, challenge assumptions during PHA and RMP analysis, and convert data into safeguards that work on the worst day, not just the normal day.