Modelling Instantaneous Risk for Major Accident Prevention

Quantitative risk analyses have traditionally been used to make decisions about long term risk, e.g. when we are designing a new offshore oil and gas installation and want to know how to achieve the lowest possible risk within the lifetime of the installation. Experience has however shown that these analyses are not equally well suited to support short term decisions related to operation of the installation. The project has looked at this problem. Different explanations have been given for why these analyses can not provide the answers we need. It has been assumed that the analyses are too complicated to use for non-experts in risk analysis. Decision-makers have therefore not been able to retrieve relevant information. Further, the analyses are normally updated with an interval of 5-10 years, implying that they do not provide updated information. In MIRMAP, we have looked into this problem in more detail and tried to understand why current risk analyses do not provide the answers we need. We have looked at the types of decisions that typically are made in operation compared to design. This has formed the basis for evaluating the risk models that we used today and if they are sufficient to provide the right answers. The results have revealed several weaknesses in the way we approach the problem today. First of all, there are different aspects of risk that are important for short term decision making compared to long term. In the long term. It is relevant to minimize statistically expected losses (how we normally express loss), but in the short term, avoidance of accidents is more important, i.e. the probability of an accident. Further, we also need to distinguish between risk during the performance of an activity and risk resulting from completing an activity (after the activity). An example is performance of maintenance on safety critical equipment. We perform this work to reduce risk to the plant after the work has been completed, but the personnel involved in performing the maintenance may be exposed to high risk during the performance of the work. Further, the risk models that we apply do not meet our requirements. All models are simplifications of reality and we prioritize including those factors and aspects that influence the results that we want. The project has highlighted that the aspects that influence risk during operation are not explicitly modelled in today?s quantitative risk analyses. Typically, relevant technical systems are modelled in detail, but activities and operations are not modelled explicitly. During operation, the largest variations in risk from day to day are caused by operations, but since these are not modelled, we are not able to calculate these variations. The basis for making decisions is therefore not good enough. From this, the project has developed a generic methodology for how we can develop improved risk models for operation, that to a greater degree reflect the factors that change on a daily basis and that influence risk. All activities that are planned can be modelled explicitly in the risk analysis and other factors that influence the risk level, such as status on technical systems, can also be modelled. The method has been tested on a case study where a limited risk analysis has been developed for an offshore installation. The analysis was designed to support decision-making during planning of activities. All activities that were planned for a two-week period were entered into the model and it was possible to see the effect of adding or removing activities. Similarly, it was also possible to see how activities could influence each other and thus help to avoid combinations that could increase risk. The personnel from the company involved in the case study expressed that this was useful information that could provide improved risk management during operation. The project has shown that there is a need to think differently about risk in operation compared to design and that we need other models and results than we apply today. This may also be an indication that the way we manage risk in operation should be changed and that we should focus on other parameters then we use today. We plan to investigate this aspect further.

The first task will be to analyse situations where information about risk is required as input to decision-making, and to determine situations where average risk is sufficient as opposed to where instantaneous risk needs to be considered. A range of situa tions will be identified in cooperation with the industrial partner. The situations will be considered and a classification scheme will be developed, to enable a structured evaluation of the type of decision support required for different situations. The work will be based on general experience and knowledge about operations, combined with discussions with relevant operating personnel. From the classification of decisions, a set of cases that will be used as basis for the analyses will be identified. Th e introduction of instantaneous risk also implies that present ways of expressing risk not necessarily are suitable. Through a literature review, and a seminar/workshop, alternative ways will be identified and discussed. This will form the basis for sugge stions for alternative ways of expressing and thus also communicating risk in different situations. This may also open for a discussion about acceptance criteria and the applicability of the ALARP principle. The main part of the project is related to exp loring and developing methods for analysing instantaneous risk. Alternative approaches (traditional QRA supplemented with influence modelling, STAMP, FRAM and other approaches) will be investigated and evaluted with respect to what is the most promising a pproach. Most likely, the project will pursue a method based on work in the RISK_OMT project, but supplemented with knowledge from other methods and approaches. The theoretical basis will be developed and the method will be applied to several case studies at the Kårstø processing plant.