Automated HAZOP revisited

What is it about?

Automated HAZOP was developed in the 1970's, but required the user to model process plants with special notation which took nearly as long as the analysis itself. Taylor, in 1982 published a "An algorithm for fault tree construction" and reports om the RIKKE system which described a way of developing a library of component models which required much effort for modelling, but could then be used universally, in any logical configuration of components. This allowed process flow diagrams, piping and instrumentation diagrams and electrical circuit diagrams produced by plant designers to be read directly into a computer program and a HAZOP to be constructed fully automatically, or interactively. The interactive version allowed new types of failure causes and new disturbance consequence physics to be included into analyses, and for these to be fed directly back to the generic model libraries, extending the knowledge base. The physics based modelling allows new knowledge to be generalized so that it can be used outside the original context.. The modelling allowed extensive lists of causes and consequences to be included, and for disturbance propagation to be modelled on the basis of physical laws rather than functional descriptions. This allowed accidents involving reverse flows, sneak paths and emergent hazards due to subsystem interferences to be identified. The method allowed operator error analyses to be included using the action error analysis method, as described in Risk Analysis for Process Plants, Pipelines and Transport (1994) and Human Error in Process Plant Design and Operation (2016) Automated HAZOP was used to only a limited extent for to replace HAZOP workshops because this kind of use is based on an erroneous assumption. One of the main advantages of HAZOP workshops is to allow multi-disciplinary exchange of ideas and knowledge and to achieve consensus on risk reduction measures. Any automated method which bypasses the workshop eliminates most of the advantages of the methods. Automated HAZOP was used by the author's company, Taylor Associates, from 1987 when the HAZEX program was developed, through to 1914, for supplementary analyses, follow up analyses and for quality control of HAZOPs. They were also used to support post incident investigations and follow up studies as more knowledge of accident causes could be gathered. In all analyses were made for 102 plants including six refineries, four ammonia plants, a large number of pipelines, some mines, four petrochemical complexes, and several oil field both onshore and offshore. During the period the general quality of HAZOPs performed in industry were found to increase, with scenario completeness factors of between 85 and 95% in the early 1990's (as determined from HAZOPs for which third party review was carried out by our company) to between 95 and 98% for analyses carried out in highly competent engineering contractor and operating companies in the 2000's. (see 40 years of HAZOP in Loss Prevention Bulletin October 2012). Follow up using automated HAZOP allowed the completeness to be increased to 99.8% "historical completeness" for analyses completed after 2008. Each year, however there are one or two new accident types published in good sources such as those from US Chemical Safety Board.

Featured Image

Why is it important?

The paper Automated HAZOP revisited is probably unique in providing a detailed evaluation of both automated and manual HAZOP over a period of some 40 years, with automated HAZOP providing a precision tool for evaluations. This includes comparisons between the analyses made in the 1980's and early 1990's with repeat analyses made after 2010. (Repeat analyses a and comparisons are made with relatively little effort dure to the automation)

Perspectives