The CSB's Bio-Lab Conyers final report: when a warehouse is the process

On 21 July 2026 the U.S. Chemical Safety Board released its final investigation report into the 29 September 2024 fire and toxic-gas release at the KIK Consumer Products (Bio-Lab) Plant 12 warehouse in Conyers, Georgia. No one was killed, but the consequence footprint was community-scale: roughly 17,000 nearby residents were evacuated, about 90,000 people across metropolitan Atlanta were advised to shelter in place, and nightly shelter-in-place warnings continued within a two-mile radius for 17 days. The warehouse was destroyed, and KIK ceased operations at the site in May 2025. The mechanism is worth reading closely, because it is built out of ordinary storage conditions that a process hazard analysis is meant to surface but often does not, precisely because it is “just” a warehouse.

What happened

The full report describes a self-propagating reactive event rather than a conventional fire. Plant 12 held roughly 5,000 “super sacks” — each between about 2,200 and 2,800 pounds — of chlorinated isocyanurates, oxidizing pool-sanitizer chemicals that decompose exothermically on contact with small amounts of water or moisture, releasing toxic gases and generating enough heat to start fires. The woven polypropylene sacks were not waterproof, the warehouse was not air-conditioned, and its ventilation could not manage humidity given the mass of oxidizer stored.

Around 5:00 am, a corroded sprinkler-system component failed and water reached the chlorinated isocyanurates, initiating decomposition. The heat that reaction produced activated more sprinkler heads, wetting more oxidizer and other reactive stock stored nearby — cyanuric acid, aluminum sulfate, sodium bromide — some of which also reacted. The combined reactions drove further decomposition, producing plumes of smoke and a toxic vapor mix that included chlorine, hydrogen chloride and bromine. The fire-suppression system, in other words, was part of the initiating chain.

Findings that matter to a study team

Two threads in the CSB’s account translate directly into PHA practice.

The first is the corrosion pathway and the maintenance response to it. The CSB found the sprinkler system had been corroding and leaking since shortly after storage began in 2019: off-gassing chlorine combined with ambient humidity and condensed as hydrochloric acid on metal surfaces, attacking the very sprinkler heads meant to protect the building. At the December 2023 annual inspection — nine months before the incident — more than 1,100 corroded sprinkler heads were visually observed. The Board frames one of its five key issues as “run to failure.” For a study team this is the familiar gap between an inspection finding and an executed control: the degradation was known, documented, and slow enough to act on, and the credited safeguard was allowed to become the trigger.

The second is inventory. Documentation submitted in 2019 anticipated an average of about 6.2 million pounds of raw materials in the warehouse. At the time of the incident, nearly 14 million pounds of reactive chemicals were present — more than double. Quantity is a process-safety parameter: it changes the credible consequence, the segregation and ventilation the storage demands, and arguably the facility’s regulatory status. A change of that magnitude with no evident re-review is a management-of-change failure of the plainest kind, and it is exactly the sort of drift a periodic PHA revalidation exists to catch.

The regulatory thread

The CSB again noted that chlorinated isocyanurates — reactive and capable of producing toxic gas — are covered by neither OSHA’s Process Safety Management standard nor EPA’s Risk Management Plan rule, and reiterated recommendations, which it has made since 2002, that both agencies expand coverage of reactive hazards. It issued new recommendations to KIK and to the NFPA. The reactive-hazards coverage question sits alongside the ongoing debate over the scope of the RMP rule; teams working reactive-chemical storage should not assume that being outside PSM or RMP means being outside the hazard.

The wider point for facilitators is where the study boundary gets drawn. Bulk storage of reactive materials — water ingress, humidity, incompatible-material co-location, and the case where a protective system initiates rather than mitigates — belongs inside the nodes, not in the margin of the P&ID. The Bio-Lab report is a detailed, primary-source reminder of what that margin can hold.