As ever, there is a lot in the proceedings, and I found this conference to one of the harder ones I've attended to pick which session to attend, as there were a lot of times where at least 2 papers were on in parallel that I wanted to see!
KEYNOTES:
The keynote speakers included:
Sir Charles Haddon Cave: - Nimrod XV230
Disaster.
He had a number of key lessons (as a lawyer, he is very strong on lists), including:
"Complexity is the enemy of good safety".
The cause of the incident included:
1. Poor design: the incident was caused by fuel coming into contact with a hot surface (above 450°C). The modification to the aircraft that made these hot surfaces available to come into contact happened in the 1960's
2. History of Fuel Leaks throughout operation - these became seen as "normal".
3. Increase in operational tempo - heavy use of planes from Kosovo onwards.
4. Maintenance of aging aircraft - based on a 1950's design, with end of life being extended and extended.
5. Major organisational change and lack of funding.
6. Outsourcing of the safety case - they had a piece of paper that gave them a "warm feeling", but 40% of the risks ameliorated on paper were still live.
7. Dilution in responsibility, accountability and the associated processes.
He was particularly worried about a culture of "Paper Safety", where powerpoints slides were used as an alternative to thinking, rather than prompting thinking.
He encouraged us to keep asking questions, as "Questions are the antidotes to assumptions that so often result in mistakes".
Cheryl Grounds - BP VP Process Safety
Cheryl spoke to her history in Mobil - BP, and illustrated some of the advances and challenges she saw based on her development in safety. She clearly embodied a drive to make safety a central part of BP, and having a seat at the upper areas in BP to make the case for process safety being good business.
Alan Chesterman - Apache
Alan spoke to the approach within Step Change in Safety to manage best practice and make learning better across incidents. He particularly spoke about the "Joined Up Thinking" initiative within Step-change and the way that they are changing the incidents databases to allow sharing, rather than inhibit it.
He had a powerful anecdote that his personal commitment to process safety came from having survived a 30 tonne gas release and explosion early in his career, and his hope that others don't have to have quite as personal an incident to remember.
Ken Rivers - Chair COMAH Strategic
Forum
A really engaging speaker, he showed a document that was going to be entitled "Process Safety Management......" but is now more like "Managing the NON-FINANCIAL risks that could destroy your company". He is on a journey to try to make best practice common practice among their members and beyond.
Dr Paul Logan - HSE
He gave a good overview of the position of the regulator, and their strategies towards finding the places where the control measures that are critical and most vulnerable to failure exist.
Other Key Papers:
Vent Release, Maersk GPIII.
There were two papers which summarised experiments carried out by HSL to deal with 4 vent issues on GPIII. These looked at potential ignition sources and experiments to determine if a burning liquid droplet from the flare could ignite a vent, and also looked at the areas around a cold-vent where ignition could occur. This concluded that some ignitions happened even when the predicted LEL from PHAST was a low as 10%. They recommended that to account for eddy flow and uncertainty, a sphere of radius = the distance to 50%LEL along the centreline of the release would account for the ignition (an approach similar to IP-15 for area classification).
Hindsight Bias
The IChemE Safety Centre have released a set of training material focused on giving people experience of making some of the key decisions that were in the run-in to major incidents. The issue in general is that people will focus on the "obvious" cause of the incident once known, rather than if they would have made the same decision. They've put 200+people through the courses so far, and a large number (she said 90%) of people made the same decisions that ended up leading to the accident. Looks like a powerful tool in the training armoury. See http://www.icheme.org/media_centre/news/2016/tackling-memory-distortion-to-improve-process-safety.aspx.
Riser ESDVs
A paper was presented on the performance of Riser ESDVs based on the RIDDOR database. This concluded a few interesting points:
- Majority of failures were of valves 20-25 years old (although there would be a large population here due to post-Piper improvements).
- Most valves that had failed had failed more than once.
- A number of valves had not been tested before they failed for multiple years.
- On average the PFD was in the 1-2% range, which is in line with the kind of figures you would get using OREDA-type valve failure rates.
And finally
Revalidating HEART
HEART (or the Human Error Assessment and Reduction Technique) is one of the simpler ways of considering human error probability for use in fault trees/event trees. It was originally developed in the 1980's based on general tasks. The HSE have worked with the original author to check the underlying probabilities and error producing factors, and found that the model still holds true. A revised handbook for the method is planned for publication this summer.
In discussion, it was pointed out that the typical operator is probably not more likely to make a mistake in general, we just give them less warning and harder things to do now with fully automated control and shutdown systems than before.
