manufacturing · family: the rulebook was written for a world that no longer exists
run toward the leak to close the valve
Chemical facility remote isolation and alarm flood management
Problem statement
Many U.S. chemical facilities are not protected with remote isolation equipment, the CSB finds, 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 federal workplace rule for processes involving highly hazardous chemicals — has undergone little reform since 1992, does not cover atmospheric storage tanks, and does not require evaluation of remote isolation needs; managing abnormal situations such as alarm floods, the CSB's Toledo report says, "goes beyond PSM compliance alone." In 2013 the CSB declared OSHA's response to 7 open recommendations — on combustible dust, fuel gas, and PSM — "unacceptable."
Why this matters
The CSB's 2024 safety study reviews six incidents it investigated — four of them in 2019 alone — in which the lack of remote isolation made the outcome worse. The Philadelphia Energy Solutions refinery fire and explosions (June 2019) released over 5,000 pounds of hydrofluoric acid, caused an estimated $750 million in property damage, and led to the refinery's closing. The TPC Port Neches explosions (November 2019) caused $450 million in on-site and $153 million in off-site property damage. The BP-Husky Toledo refinery incident killed 2 and caused $597 million in damages.
What’s been tried and why it hasn’t worked
Voluntary industry guidance on remote isolation has been available for years, but the CSB finds it has not been consistently followed, and there is no regulatory floor. 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 — its metrics are based on at least 30 days of data, so a site can meet the monthly target while the operators on a single shift are overloaded. At Philadelphia Energy Solutions (2019), the water pumps meant to suppress an HF release could not be started remotely because the elements to remotely operate them were damaged by the fire and explosions, and 40 minutes passed before a worker turned a pump on by hand. At ITC Deer Park (2019), a tank farm fire burned 3 days; the CSB found that the absence of flammable gas detection at the tank where it began, and of remotely-operated emergency isolation valves on the tank farm's tanks, contributed to its severity, and that no formal PSM program was applied to that tank — atmospheric storage tanks fall outside PSM coverage. At KMCO Crosby (2019), workers could not isolate an isobutylene release from a safe location; operations staff entered the release area and closed valves by hand, and the vapor cloud exploded, killing one operator and seriously burning two other workers. PSM was written in 1992 and still does not require an evaluation of remote isolation, though the CSB study calls remote isolation "neither a technically complex issue for the chemical industry, nor a new one."
What would unlock progress
PSM modernization — expanding coverage to atmospheric storage tanks, requiring inherently safer design analysis, and mandating evaluation of remote isolation needs — would establish a regulatory floor; the CSB has been asking for parts of it for over a decade (tank coverage since 2002) and for the remote isolation evaluation since 2024. Short-term alarm flood performance standards (beyond what ISA-18.2 currently addresses) would help 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 the PSM changes to take effect, and the alarm flood targets were recommended to the International Society of Automation, which publishes ISA-18.2.
Entry points for student teams
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.
Genes
Tags marked “under review” were questioned by a later calibration pass; they stay visible here but are left out of filters until re-adjudicated.
Structural cousins: stuck the same way, in other fields
Sources
CSB Remote Isolation Safety Study (2024-01-H), CSB investigation reports (BP-Husky Toledo, PES Philadelphia, ITC Deer Park); accessed 2026-02-19 go to source ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
- CSB Remote Isolation Safety Study (2024-01-H) — https://www.csb.gov/assets/1/6/csb_ripe_study_final.pdf
- CSB BP-Husky Toledo Final Report — https://www.csb.gov/assets/1/6/final_report_-_20241.pdf
- CSB PES Philadelphia Final Report — https://www.csb.gov/csb-releases-final-report-into-2019-pes-fire-and-explosion-in-philadelphia/
- CSB ITC Deer Park Final Report — https://www.csb.gov/assets/1/6/itc_report_-_final_(july_6,_2023).pdf
- CSB OSHA "Unacceptable" Response — https://www.csb.gov/us-chemical-safety-board-determines-osha-response-to-seven-open-csb-recommendations-on-dust-fuel-gas-and-process-safety-management-to-be-unacceptable/
- CSB "Most Wanted" PSM Modernization — https://www.csb.gov/csb-board-members-identify-modernization-of-process-safety-management-regulations-as-the-agencys-second-most-wanted-safety-improvement-/
- PSM modernization is CSB's second "Most Wanted" safety improvement (announced December 2014).
Reconciliation 2026-09-30: Fact repair after reading the whole write-up against every source it names. All six were fetched on 2026-09-30 directly from csb.gov. Read in full: the Remote Isolation safety study (https://www.csb.gov/assets/1/6/csb_ripe_study_final.pdf, 24 pages) and the three news releases. Read in part: the BP-Husky Toledo investigation report (https://www.csb.gov/assets/1/6/final_report_-_20241.pdf, 169 pages; the full text was searched, and the passages on alarm flood, regulation and recommendations were read) and the ITC Deer Park investigation report (https://www.csb.gov/assets/1/6/itc_report_-_final_(july_6,_2023).pdf, 100 pages; the full text was searched, and the passages on cause, gas detection, isolation valves and PSM were read). The news releases are the CSB news release on the PES Philadelphia final report (the link above is the news release, not the report itself); the 2013 news release on OSHA's "unacceptable" responses; and the 2014 "Most Wanted" news release. (1) The text said "most" U.S. chemical and refinery facilities lack remote isolation valves. The study says "many chemical facilities are not protected with remote isolation equipment"; the sentence now says many. (2) The text called PSM "the primary federal regulation for chemical facility safety" and said it "does not address alarm management." Neither statement is in the sources. PSM is now described as the ITC report describes it, and the alarm clause now quotes the Toledo report's lesson that managing abnormal situations such as alarm floods "goes beyond PSM compliance alone" (the report treats alarm floods as one kind of abnormal situation). (3) The "unacceptable" vote was in July 2013 and covered seven recommendations on combustible dust, fuel gas and PSM together; the sentence now gives the year and the subjects. (4) Removed because none of the six sources contains them: 81 serious incidents between April 2020 and January 2025 with 14 deaths, 65+ serious injuries, $3.5+ billion in damage and 29 states; the ONEOK explosion figures ($930 million, 1,000 residents evacuated); and the Bio-Lab Conyers figures (17,000 evacuated, 90,000+ sheltering). The paragraph now uses figures from the study: six investigated incidents, four of them in 2019; the PES losses (over 5,000 pounds of hydrofluoric acid, an estimated $750 million, the refinery closed); and the TPC Port Neches losses ($450 million on site, $153 million off site). (5) The sentence that some large refineries voluntarily deploy remote isolation and advanced alarm management, leaving "the least safety-conscious operators" to define the industry's risk, is not in the sources; it now gives the study's finding that voluntary guidance has been available for years and has not been consistently followed. (6) The text said ISA-18.2 "was not designed for" surges. The Toledo report says the standard's metrics are based on at least 30 days of data and that an operator on one shift can be overloaded while the site meets the monthly target; the sentence now says that. (7) The text said the ITC fire burned three days "because the facility lacked" isolation valves, gas detection and a formal PSM program. The ITC report says the absence of gas detection at Tank 80-8 and of the valves on the tank farm's tanks contributed to the severity, and that no formal PSM program was applied to that tank; other parts of the terminal were covered. The sentence now follows the report. (8) The claim that the KMCO plant alarm was never activated to warn 200+ people is not in any source named here. The sentence now gives what the study says about KMCO: workers could not isolate the release from a safe location, staff went into the release area to close valves by hand, and the explosion killed one operator and seriously burned two other workers. (9) The text said PSM was written before remote isolation concepts "were mature." The study says the opposite about remote isolation, calling it "neither a technically complex issue for the chemical industry, nor a new one," and cites guidance from 1969 and 1975; the sentence now quotes the study. (10) "Incorporating alarm management standards" was listed among the PSM changes the CSB has requested for over a decade. No source shows the CSB asking OSHA for this; the Toledo report's alarm recommendation went to the International Society of Automation. The item was removed; the sentence now says the CSB has been asking for parts of the change for over a decade (tank coverage since 2002) and for the remote isolation evaluation since 2024; "would ensure" became "would help ensure," and the last sentence now says which body each change depends on. (11) The PES sentence said the water pumps' remote-activation elements were "located in the fire zone." The news release says the elements to remotely operate the pumps were damaged by the fire and explosions and that forty minutes passed before a worker turned a pump on by hand; the sentence now says that and no more. (12) "Currently 12 open recommendations to OSHA" in the notes above is not in any of the sources and was removed; the date of the "Most Wanted" announcement was added. Confirmed in the sources and left as written: PSM issued in 1992 and "little reform" since; the atmospheric storage tank exemption; the study's recommendation that OSHA require an evaluation of remote isolation; 3,712 alarms in 12 hours at Toledo; two deaths and about $597 million in property damage at Toledo; the lack of short-term alarm flood targets in ISA-18.2; the ITC fire burning three days; the CSB does not issue citations or fines; and the entry point's statement that several CSB reports include detailed alarm timelines (the Toledo report gives alarm counts in 10-minute blocks and event-log details in its Appendix A, and the PES report's Appendix A timeline lists individual alarm activations to the second). No title changed. Genome tags untouched. A second reader went over these repairs on the same day, 2026-09-30, fetching all six sources again plus the PES investigation report itself (https://www.csb.gov/assets/1/20/PES_Final_Report_Published_October_2022_r1.pdf, 107 pages, searched for the water-pump, isolation and timeline passages). The repairs held, three of them tightened to say no more than the sources do (the abnormal-situation quotation, the ITC cause sentence and the dating of the CSB's requests), with one exception: a first repair had replaced the entry point's "several CSB reports include detailed alarm timelines" with a pointer to the Toledo report alone, but the PES report's Appendix A timeline also lists individual alarm activations, so the original sentence is supported and was restored. The PES water-pump sentence was reworded as described in item (11).