Proceedings of the Eighth International Conference on Probabilistic Safety Assessment & Management (PSAM)
Download citation file:
- Ris (Zotero)
- Reference Manager
The Piper Alpha catastrophe in 1988 revealed the potential of operational failures with respect to dramatically destroying an entire offshore installation. When a severe hydrocarbon leak occurred due to operational oversights, other safety barriers collapsed mainly due to lack of a prudent safety culture.
Statistics for the Norwegian offshore industry show that each year there are about 5–6 hydrocarbon leaks at least the size of that kicking off the Piper Alpha sequence of events, the majority of these leaks have operational causes. The challenge with operational failures is that when leaks occur in such circumstances, other barriers may also fail due to common cause factors, as in the Piper Alpha case.
Accident investigations are often weak when it comes to analysis of root causes and common mode failures. In order to provide improved understanding of the operational failures that cause leaks, a set of incident reports have been analyzed in order to identify the common patterns in hydrocarbon leaks due to operational failures. Other scenarios where failures of operational barriers are dominating contributions are falling objects from cranes, as well as failure of station-keeping by large tankers in attendant mode. The former has been the dominating scenario for fatal accidents on Norwegian offshore installations for several years, whereas the latter is a potential scenario for extreme collision cases, where so far only near-misses have occurred. There are usually a number of redundant barriers against major accidents with technical causes on offshore installations, giving low risk values. With operational causes, there are significantly fewer redundant barriers, thus increasing the risk levels.
Of 40 incident reports and investigations, only six allowed a detailed analysis of barrier performance. It was shown that what was supposed to be four barriers for each scenario was reduced to only one barrier (the execution of the work itself) in five out of six cases, and two barriers in one case. The work execution failed in all cases, and this was the only effective barrier in five out of the six cases.
Also an analysis of a near-miss is presented, where it was shown that six operational barrier elements in sequence failed, and the seventh barrier element performed well in identifying the failure to conduct work in accordance with procedures (the initial barrier failure).
The analysis has demonstrated clearly that the simple barrier block diagram is useful in order to demonstrate performance of operational barriers in incidents and near-misses. The analysis has also demonstrated how hydrocarbon leaks are caused due to single failures in many cases, where it was supposed to be three or four barrier elements.