“Do You Think We Should Consider This Risk”: The Need for Making Credible Scenarios in Risk Assessment

I am writing this at 30,000 ft on a plane. There have been several recent crashes attributed to a particular plane. Should I consider switching to a hand written note to my wife and family pinned to my soon to be created corpse instead of a new article? While I am a white knuckle flier despite almost two million airline miles I have decided to finish this article instead. (OK, I admit to a momentary mental review of my insurance policies.) Why? Because I don’t think there is a credible scenario that leads me to a high enough potential frequency to worry or, in risk assessment reviews, to consider mitigative measures being necessary.

As the participant, leader, or consultant on hundreds or risk assessments I have experienced incredible amounts of time wasted (and I chose that word carefully) on scenarios that are less likely to occur than being struck by lightning in clear weather inside your basement. (OK perhaps a slight exaggeration …) And the reason is that that the participants did not spend the time and effort to create a detailed scenario that they then could realistically evaluate. Instead they discuss, argue, and meander around a vague concern that is so vague no one can evaluate its potential. If I ask you could you be robbed at gunpoint, how can you answer without knowing a basis? If you are walking through a less desirable urban area at midnight all alone, ostentatiously counting money recently taken from an ATM I suspect the probability is higher than when I am grilling on my deck in an affluent suburban area that last saw an armed robbery in the 1930’s.

Let’s take a practical example. Your operation requires you to take a glass bottle of toxic material out of a cabinet and place it in a hood. “What happens if the bottle breaks and you are exposed?” A not unreasonable question to ask at a HAZOP or similar risk assessment. All too often, however, the participants begin to debate the frequency without any attempt to define the scenario. I would suggest that this definition is critical to the analysis. So, I would suggest that they construct a logic path for their analysis.

The technician is wearing the proper safety equipment (gloves, glasses, apron, lab coat, or whatever you require). The technician could either move the bottle without an incident or they could drop the bottle. They could lose control of the bottle. If they lose control of the bottle they might regain control without dropping it or they could drop it. It dropped it might break or not. If it breaks the fluid could splash on them or not. If it splashed on them it could be in an amount to cause a problem or be minor enough to treat. None of these options is 100% certain although all are open to discussion. If one applied the simplistic view that each one is a 50% probability, then the overall probability of the technician being injured beyond first aid is

Losing control 50%

Dropping the bottle 50%

The bottle breaking 50%

The fluid contacting the technician 50%

There being enough fluid contact to be serious 50%

The sum of these probabilities is (.5)(.5)(.5)(.5)(.5) or 3.125%. Yet many risk assessments treat it as almost 100%. And I would argue that 50% is a very, very high probability for some of these events. (The author, a notoriously fumble fingered individual usually manages to move bottles, intact at least 99% of the time.)

So, I contend that unless one creates a step by step scenario of how the accident could happen, no group can realistically assess the risk. It may be necessary to create multiple scenarios for evaluation. What if the technician in the above example was interrupted by someone walking into the laboratory? What if the technician was not wearing any safety equipment? These and other alternatives might have to be included in other, modified scenarios and assessed. The key is the detail.

Many risk assessment participants struggle with trying to assess the probabilities for each step. “How do I know the chances that the technician will drop the bottle?” Well I admit to having no hard data on this point either, but I do have 40 years of experience and a reasonable amount of common sense. So, I normally start off with questions. Based on you experience do you think dropping a bottle is a common, routine problem. Most people will say no. In that case would it be reasonable to suggest it is fairly rare say 1 time in ten or perhaps one time in a hundred. Usually some discussion ensures and participants think through their own experiences. “I did that operation at least 5-10 times a week and never dropped one.” That might suggest that 5 times a week x 40 weeks (assuming one took vacation and holidays) times two years translates to 1 chance in 2,000. Other participants might use similar logic to develop other figures. Then one could use the lowest, the average, or whatever seems most appropriate in the calculation. Where no hard data is available as, perhaps with the chance of the bottle breaking, the participants might develop all possible outcomes (e.g. it breaks, or it does not) and assign them equal probabilities. Lest you think this mathematical wizardry is not realistic, I suggest you look at what the frequency of many risk assessments suggest. In the above example, it might lead to you asking something like “well if we feel that the bottle drops and breaks 1 time in 3 why are we not having more accidents with dropped bottles?”. Validating the results against common sense is always a useful tool.

For more information on this topic and risk assessments in general, you may want to consider the following posts:

o  "But It’s Only an Instrument!”: Issues with Classifying Research Equipment (https://www.dhirubhai.net/pulse/its-only-instrument-issues-classifying-research-richard-palluzi)

o  Periodic Overall Cold Eyes Safety Inspections: An Overlooked Tool in Pilot Plant and Laboratory Safety (https://www.dhirubhai.net/pulse/periodic-overall-cold-eyes-safety-inspections-tool-pilot-palluzi/)

o  Preventing Fires and Explosions in Pilot Plants and Laboratory Units (https://www.dhirubhai.net/pulse/preventing-fires-explosions-pilot-plants-laboratory-units-palluzi/)

o  Pilot Plant Safety Systems Parts I and II (https://www.dhirubhai.net/pulse/pilot-plant-safety-systems-part-ii-richard-palluzi/ and https://www.dhirubhai.net/pulse/pilot-plant-laboratory-unit-safety-systems-part-i-richard-palluzi/)

o  Laboratory Accidents: A Continuing Problem (https://www.dhirubhai.net/pulse/laboratory-accidents-continuing-problem-richard-palluzi/)