Why the Question You Ask Shapes What an Accident Investigation Finds
- Robert Williams
- 656 reads
At IChemE's Hazards 36 conference, speakers argued that blame-focused investigations hide root causes, and an insurer reported refineries running near full capacity.

The most useful idea from the second day of IChemE's Hazards 36 Process Safety Conference was about wording. Richard Scaife, a chartered occupational psychologist and director of the Keil Centre, argued that the question an investigation starts with largely determines what it can find. Ask "who made the error?" and you tend to get a name. Ask "why did conditions allow this?" and you start to get a system.
Starting the investigation with the wrong question
Scaife's frustration was with how major accidents are reported. Plane and train crashes, and other widely publicised incidents, are often summarised with a single "root cause" that is a human or technical error. That framing stops short of the more useful question: why did the behaviours that led to the error make sense to the people involved at the time?
The problem is practical as well as philosophical. Thorough investigations depend on reliable evidence, and in many cases that evidence comes from witness interviews. Scaife said corporate investigations often make people less willing to speak candidly, particularly when they feel their own conduct is the focus. If operators think the process is a search for someone to blame, they will protect themselves, and the investigators lose the detail they need.
His remedy was a "psychologically safe culture and just culture that really encourages people to speak up." That phrase is doing a lot of work. It isn't about going easy on individuals. It's about getting accurate testimony in the first place.
Scaife's example made the point sharply. In a case involving a patient who died after contracting septicemia during an operation, the investigation initially asked "why did the patient die?" Operating theatre staff felt managers were trying to "pin the blame on someone", and they said little. When the scope changed to "why was bacteria in the operating theatre?", witnesses started talking, and the investigation reached more systemic causes. The same facts were on the table both times. The change was the framing, and it changed what people were willing to say.
Practitioners who recognise the pattern
IChemE Fellow Michael Godfrey, who works as a part-time consultant and goes by "that process safety guy" in that role, called the presentation "absolutely spot on." After nearly four decades in process safety, he said the point about resisting the urge to blame individual operators was the one that landed hardest. "You can't blame the operator, you can't sack the pilot for the accident," he said. "You've absolutely got to get to the root causes."
That's a familiar tension for anyone who has sat on an incident review panel. Removing a person from a role can feel like closing the case, and it often does nothing to change the conditions that produced the error. Scaife's argument is that the conditions are where the useful engineering lies: procedures that don't match how work is actually done, pressure to keep a unit running, or alarm systems that train people to ignore warnings. None of these show up if the investigation stops at the operator.
An insurer's view of refining and gas plant risk
The second plenary came from Nigel Cairns, a chartered engineer at insurer Liberty Specialty Markets. His talk covered recent major accident trends across refining, petrochemicals and gas plants, as seen by the onshore energy insurance market. The headline was demand pressure. Cairns pointed to what he called the "long-term dominance" of the US in business downtime, and said that pressure had grown as operations expanded in response to geopolitics.
He tied the trend to specific market conditions. Before the Strait of Hormuz closure prompted a ramp-up in oil and gas production outside the Middle East, "global refining margins had already doubled" from 2015 to 2026, Cairns noted. He linked the conflict involving Iran to North American refining utilisation rates that have "crept over 95% as a number of oil majors have postponed planned maintenance activities." Alongside the war in Ukraine, he said, US refineries have been "operating near maximum capacity for the longest sustained period this century."
Anyone who has worked near a refinery turnaround schedule will see the risk here. Deferred maintenance and sustained high throughput both shrink the margin for error in equipment that was already running hard. Cairns didn't claim a direct causal line to particular incidents in his summary, and it would be wrong to add one. But the operating pattern he described is the kind that tends to show up in loss data over time.
Regulation, cost-benefit and the improvement gap
Cairns also raised a more uncomfortable point about incentives. Operators, he said, often won't make improvements unless they are legally required to. "We can have a very frustrating time, particularly working in North America, where we might suggest an improvement that might make good sense and would pass a cost-benefit analysis, but the site is just not interested because they're not required by OSHA [the US Occupational Safety and Health Administration] to do that."
That's a real gap between what a sound risk assessment recommends and what a site chooses to fund. For engineers in consulting, insurance or plant design, the practical lesson is to frame recommendations in terms of the site's own economics and risk exposure, not only in terms of compliance. Where a proposed change clearly passes a cost-benefit test, documenting that case is often the most persuasive step available.
A more diverse programme, and a practical one
Hazards technical committee chair Eamon Chandler reflected on his fourth and final year in charge, a year that saw record attendance. He said the conference "has been an order of magnitude better than four years ago." Chandler was particularly pleased with the range of speakers. "I've always wanted to have another human factors speaker, and we got Richard this morning," he said, adding that the perspective from an insurance broker was "something different, and we've always wanted to do that."
Committee member Ashley Hynds said the practical focus is what keeps him coming back. "The things that you learn at Hazards are really applicable. It's not just some concept that might work in 5-10 years' time. It's stuff that's happening now."
My read is that the two plenaries pointed at the same problem from different directions. Scaife showed how investigation technique can either open up or shut down the evidence that explains an accident. Cairns showed how commercial pressure and weak incentives can leave known risks in place. Both point toward the same next step: asking questions that lead to conditions and systems, and then making the case for changing them in terms a site will act on.