If there’s one thing that has struck me here at the International Pipeline Conference it’s that there is a great diversity of approaches to pipeline risk management. If you’ve only been exposed to the AS 2885 approach you might not appreciate how many other ways there are of looking at and managing pipeline risk.
It seems that to many (most?) pipeline people in North America “risk assessment” means something like “evaluating the likelihood of corrosion leaks”. And that means interpreting ILI data and making assessments of corrosion rates and where leaks might occur. Very different to the situation in Australia where external interference is the overwhelmingly dominant cause of failures (86%), and our safety management studies are focussed accordingly. Third party damage is the major cause of failures in North America too, but by a much smaller margin over other causes (perhaps only 30% of the total).
Risk methods discussed at IPC range from a quite unimpressive variation on the Muhlbauer index method (if you don’t know it, don’t ask) to esoteric statistical analyses. There are paper titles like “On the use of generalised lambda distribution and parametric bootstrap method in the prediction of maximum pit depths: comparison with experimental pipeline pitting data”.
No-one seems to use anything like the Australian qualitative safety management study. And yet I still think it is by far the most appropriate method for the Australian industry.
The ultimate objective of any risk assessment is to make a decision – what to design or what to repair. Sometimes in complex situations a sophisticated statistical analysis can help with that. But the decisions facing Australian pipeline engineers can almost always be resolved very easily, without specialist risk expertise, using the qualitative AS 2885 risk matrix approach. The fact that any pipeline engineer with common sense can use the AS 2885 method is one of its greatest strengths.
The precision of numerical risk assessment is superficially seductive. I guess it’s because as engineers we are used to calculating everything, from wall thickness upwards. But in that case the calculation is the end result – the answer (a wall thickness value) is the decision (the pipe will be 7.8 mm thick). Risk decisions are different. If the probability of someone dying is 3.5 x 10E-5 that’s not the answer to anything. You still have to make a decision on whether that probability is acceptable, and what to do about it if not. One can usually reach a conclusion that a risk is acceptable (or otherwise) via much easier routes than complex statistical analysis.
None of which is to say that the corrosion risk assessments used in North America are wrong. In that part of the world they face a different suite of threats to their pipelines, and have chosen methods accordingly. And in Australia we have chosen a method that’s appropriate for us.