Continuing from the previous post …
The formal AS 2885 definition of ALARP (Clause 1.5.3) says
“ALARP means the cost of further risk reduction measures is grossly disproportionate to the benefit gained from the reduced risk that would result.”
That means a cost-benefit analysis, comparing the cost of improving safety against the value (however that is determined) of the risk reduction.
The UK Health and Safety Executive have a method of doing this that takes some effort to interpret and apply. At a POG seminar a few years ago I presented an application of this method to the AS 2885 process, but it still looked a little complex. Since then I’ve realised that it can be simplified further by using the concept of Maximum Justifiable Spend (MJS):
MJS = cost of failure x probability of failure x proportionality factor
MJS is the amount of money it is worth spending, based on the magnitude of the risk, in order to reduce the risk to zero. It gives you a criterion for identifying what additional risk reduction measures are worth considering. Those that are obviously more costly than the MJS can be discarded immediately. Those that are obviously less than the MJS should be implemented in order to reduce the risk and contribute to achieving ALARP. And in between there might be a grey area where you need to go away and do some design and cost estimating to compare with the MJS, but so far I’ve found that to be very rare.
But what is the cost of failure? It includes the obvious costs of repair and supply interruption, and also some measure of the value of lives lost if that is an outcome of the failure. The UK HSE talk coyly about “the value of preventing a fatality”, which to my mind is just a euphemism for the dollar value of a life. They reckon it’s worth a million pounds, or somewhat less than $A2 million. In workshops I run we mostly use $4 million.
More on the probability of failure and proportionality in a future post. But meanwhile a caution: If you are using this method, its whole purpose is as a guideline for judgement. It’s an order-of-magnitude estimate. There is no point in obsessing over the detail and accuracy. Informed guesses are good enough for input values.