Any sufficiently catastrophic research accident is indistinguishable from a PSM case study. Let's audit this unfortunate facility's paperwork, because the fictional disaster maps perfectly onto three real programs that real facilities really skip.
Before the test: Process Safety Management. Processes involving highly hazardous chemicals require, among fourteen elements, a process hazard analysis that systematically asks what can go wrong and what happens when it does (29 CFR 1910.119(e)), written operating procedures (1910.119(f)), and training (1910.119(g)). A PHA worth its salt would have flagged 'unexpected sample behavior in the test chamber' before anyone was told to push the cart. Skipping the analysis because the schedule was tight is the most realistic detail in the entire scenario.
During the event: the emergency action plan. Employers covered by a standard requiring an EAP must maintain one covering procedures for reporting the emergency, evacuation procedures with exit route assignments, and accounting for all employees after evacuation (29 CFR 1910.38(c)), backed by an employee alarm system (1910.38(d)). If your facility's evacuation plan can be defeated by blocked corridors and a monorail, it wasn't a plan; it was an aspiration. Employees responding to the uncontrolled release — rather than evacuating — puts you in HAZWOPER emergency-response territory, which demands its own written plan, trained responders, and an incident command system (29 CFR 1910.120(q)).
And the suit. A marvelous protective ensemble does not substitute for any of the above: PPE requires a workplace hazard assessment and selection of equipment matched to the hazards identified (29 CFR 1910.132(d)), and in safety practice it is the last line of defense after engineering and administrative controls, not a stand-in for them. If your safety program is 'we issued one employee an excellent suit,' you have not written a safety program — you've cast a protagonist.