The Value of LOPA (and other quantitative methods) versus HAZOP (and other qualitative methods)
I like LOPA as well as the next person; I’m one of the two co-originators as you likely know. But all of the quantitative methods use “statistical” values (but not as statistical as most think) that have in fact been “voted” upon by “experts”, so you can either let LOPA or QRA analyst use factors voted on by others or you can take the vote of your experts around the HAZOP table. Don’t get me wrong, I think LOPA is a great teaching tool, and it occasionally gives an answer that seems superior to a purely qualitative judgment. But, +95% of the time we have found HAZOP (or the other qualitative methods) gives an adequate understanding of risk, purely qualitatively. I agree that it appears on the surface that scoring with risk categories in a HAZOP will improve the risk judgment, or improve the consistency, but after many thousands of PHAs (using HAZOP, etc.) and a great many thousands of LOPA, and many QRA as well, we have found this is not the case and we have found the scoring interferes with the brainstorming process (this is decidedly bad, of course). And nearly half of the time the team members (when forced to use scoring) make the score match what they already believe is the right conclusion (and they are likely right).
Note that we continually do many PHAs (HAZOPs, etc.) and many LOPA and for 20% of the PHAs we are forced to use scoring still (the client overrules our advice on this) and for one client we still have to do LOPA on every scenario above a first aid or above $10,000 ultimate consequence. This is ridiculous, of course. But in those situations we continually try to fix the deficiencies with using scoring during meetings, etc. After 22+ years we do not believe the deficiencies are fixable, compared to doing the HAZOP meetings purely qualitative.
As mentioned, we do not recommend using risk scoring at all; it is too poor of a method. And we recommend using quantitative methods (starting with LOPA) only on scenarios that are too complex for the PHA/HAZOP team to make a judgment on tolerability of risk (i.e., only when they recommend further analysis). Of course, this approach depends on great HAZOPs that are staffed by, and led by, experts in their jobs…. but that is nothing new…that dependency is there regardless of risk judgment method used.
One last comment: Note that the values used in risk judgments have an order of magnitude variation on either side of the median… so the significant digit is at the exponent level … this is huge variance…. so for a PFD = 0.01 or 10-2 (+/-1), we are really dealing with an exponent of -2 (+/-1). The judgment of the qualitative teams though seem to be in a “tighter” range because they usually (95% of the time) have a great feel for what the residual risk is at a local plant/unit.
Anyway, that is our data and conclusions from the thousands of such analysis we have led and documented. If you want to find more data, please download our papers on the topic at www.piii.com