The Hazard and Operability Study (HAZOP) is probably the most commonly applied process hazard analysis method. As such, HAZOPs are a very important tool for improvement of process safety. But HAZOPs are not as effective as they should be. Duhon and Sutton (SPE 120735, 2010) identified many reasons why we don’t learn all we should from HAZOPs. These insights suggest a path towards a more effective HAZOP process.

The process described here differs from the standard HAZOP process in several important ways. The most important difference is the definition of stream-based nodes rather than equipment-based nodes. In a stream-based node, a stream is followed from its inception to its logical conclusion. This is especially useful when considering flow deviations, because a flow disruption in any part of the stream affects all parts or at least all downstream parts of the stream. These stream-based nodes are much larger than typical equipment-based nodes and hence overcome the tendency of HAZOPs to create tunnel vision.

HAZOPs are supposed to evaluate operability, but that can’t be done effectively without reviewing the operating procedures or at least discussing how the system will be operated. There is no point in a typical HAZOP, because small nodes are selected, in which the procedures can be effectively introduced. Stream-based nodes provide a natural bridge to the procedures. High level operating procedures can be introduced during the stream-based node discussion providing an opportunity to do a Process HAZOP and a Procedure HAZOP simultaneously. A Procedure HAZOP often provides more insight than the Process HAZOP.

Another important difference is the use and ordering of guidewords. HAZOPs typically consider ‘flow’, ‘pressure’, ‘temperature’ and ‘level’ guidewords. But most pressure, temperature and level deviations are caused by flow deviations. Considering all four guidewords results in duplicated effort and tedium. The process described here avoids the duplication.

You can access this article if you purchase or spend a download.