Bowtie Methods for Analyzing Risk
New paper out in the Journal of Chemical and Safety describes “Bowtie methodology” for risk assessment. I took a look at it and it looks quite interesting.
Broadly speaking, what makes it good is an explicit statement of the stuff on both sides of something going badly wrong. Most ways of analysing risk tend to be convoluted, hard to read (think: Safety Data Sheets), and focus only on the existence of risks, with only a smattering of the ways in which the problem can be addressed. What’s missing is a good structure.
Here’s a bit from the hazard Identification of an acetic acid SDS:
Classification of the substance or mixture PHYSICAL HAZARDS Flammable liquids Category 3 Corrosive to metals Category 1 HEALTH HAZARDS Acute toxicity (oral) Category 5 Acute Toxicity (dermal) Category 4
…and so on. It then goes on to list “First Aid Measures” in section 4 (two sections later, please note). So not only do you need to know what exactly an “acute toxicity (oral) category 3) is, you need to look a whole new section to understand how to respond. Yes, you might be willing to do this, but how many lab workers are going to do that, at least until it all hits the fan?
These things really need to be linked, in the form:
Store in tight container BECAUSE Acute skin contact poison THEREFORE rinse for 15 min case of skin contact AND THEN contact medical personnel.
This is why this method looks so interesting. It is a neatly structured way of working out how to contain such issues.
How to tie a bow-tie
The way this works is you start with noting down the main hazard:
(cheesy ASCII diagram heading your way)
Next, you add the main “event”. In this case, we are talking about a spill and human exposure:
Spill and Human Exposure
Here’s where it get’s interesting. On the left side we list all the “threats”, meaning the ways in which the spill could happen. On the right we list all the “consequences”. To keep running with this example, we’d put “dropped beaker”, “jug knocked over in chemical hood”, and “Splash in face from bubbling materials” on the left, and “Fatality / injury”, “Legal Costs”, and “Lost reputation” on the right.
This is beginning to look like a bowtie already! Now the neat bit: we want to put as many barriers between each “threat” and each “consequence” and the central “Top event”. These are the “controls”. For example, when it comes to the “dropped beaker” threat we can put “Acetic acid only dispensed at site of use”, “acetic acid only transported in sealed containers”, “clean bench policy so that nothing is accidentally pushed over”, and “instruction on two handed carrying of beakers”. On the right side on the other hand, between this and “injury/fatality” we can put “instruction in washing acid exposure”, “presence of first aiders”, “working chemical cleaning shower”, “Medical response number posted prominently in lab, and stored on all user phones”.
The point of these controls should be that all need to fall to get from the threat to the top event, and from the top event to the consequence (first someone needs to be transporting acetic acid from the storage to the bench, in an open beaker, and not knowing how to accurately carry such a dangerous beaker, and then, following the accident, you need people to be unaware of how to treat an acid burn, first aiders to be absent etc. before it turns into an injury or fatality.
One quibble that’s immediately obvious is that certain controls only work together (for example, safety showers only work if the workers know where they are). That’s a detail, and I’m sure it’s easy enough to include a graphic representation of these . Overall, this method is is very neat, and I like it already.
Is the software worth it?
The nifty diagrams in the paper are all prepared using software that’s only available from GCE risk. The diagrams are good, but I am not convinced. First of all, you could make these diagrams using any image editor (Illustrator, Inkscape etc.), or even mindmapping software like FreeMind. Second, whenever possible, I prefer to work with my hands when it comes to stuff like this. Whenever you are doing analysis or trying to really get into something, using your hands creates a direct connection with the subject. I’ve drawn up this analysis by hand and it’s been faster and cheaper than any software.
The only exception I can think of is when you want to distribute your bowties to other lab users, so they have some idea of what is going on. Even so,g a regular illustration program will work well enough.