Run Toward the Leak to Close the Valve

Most U.S. chemical and refinery facilities lack remotely-operated emergency isolation valves, meaning workers cannot stop a hazardous release from a safe location during a loss-of-containment event. When cascading failures generate thousands of alarms in short periods ("alarm floods"), operators cannot effectively triage the information, leading to delayed or incorrect responses. OSHA's Process Safety Management (PSM) standard — the primary federal regulation for chemical facility safety — has undergone little reform since 1992, does not cover atmospheric storage tanks, does not require evaluation of remote isolation needs, and does not address alarm management. The CSB has declared OSHA's response to 7 open recommendations "unacceptable."

The CSB documented 81 serious chemical incidents between April 2020 and January 2025, resulting in 14 fatalities, 65+ serious injuries, and $3.5+ billion in property damage across 29 states. The ONEOK explosion (July 2022) alone caused $930 million in damages and evacuated 1,000 residents. The Bio-Lab Conyers, Georgia fire (September 2024) required evacuation of 17,000 people and shelter-in-place for 90,000+ across metro Atlanta. The BP-Husky Toledo refinery incident killed 2 and caused $597 million in damages.

Some large refineries voluntarily deploy remote isolation valves and advanced alarm management, but there is no regulatory floor — creating a patchwork where the least safety-conscious operators define the industry's risk exposure. At BP-Husky Toledo (2022), board operators were overwhelmed by 3,712 alarms in 12 hours during a cascading failure. ISA-18.2 provides alarm management guidance but lacks short-term alarm flood performance targets — the standard was not designed for the tsunami-like surges that accompany real cascading failures. At Philadelphia Energy Solutions (2019), HF release mitigation water spray pumps had remote-activation elements located in the fire zone — they failed when fire damaged them, and 40 minutes elapsed before manual activation. At ITC Deer Park (2019), a tank farm fire burned 3 days because the facility lacked remotely-operated emergency isolation valves, flammable gas detection, and a formal PSM program — atmospheric storage tanks fall outside PSM coverage. At KMCO Crosby (2019), the plant alarm system was never activated to alert 200+ people to evacuate. PSM was written in 1992, before modern alarm management, remote isolation, and inherently safer design concepts were mature.

PSM modernization — expanding coverage to atmospheric storage tanks, requiring inherently safer design analysis, mandating evaluation of remote isolation needs, and incorporating alarm management standards — would establish the regulatory floor that the CSB has been requesting for over a decade. Short-term alarm flood performance standards (beyond what ISA-18.2 currently addresses) would ensure operators can act during the precise conditions when action matters most. The CSB is an investigative body with no enforcement authority — OSHA must act for any of these changes to take effect.

A team could develop an alarm flood prioritization algorithm that dynamically ranks alarms during cascading failures based on process state, consequence severity, and operator action capacity — testing it against real incident alarm logs (several CSB reports include detailed alarm timelines). Another approach: design a decision framework for evaluating where remote isolation valves provide the greatest risk reduction relative to installation cost, applicable to small and mid-size chemical facilities. Relevant skills: process safety engineering, human factors, control systems, or alarm management.