Probabilistic risk assessment techniques

Complexity refers to the difficulty of identifying and quantifying causal links between a multitude of potential candidates and specific adverse effects (Renn and Walker 2007 Renn 2008). A crucial in this respect concerns the applicability of probabilistic risk assessment techniques. If the chain of events between a cause and an effect follows a linear relationship (as for example in car accidents, or in an... 

After determining what functions must be accomplished for each goal, it is necessary to determine how well each function must be accomplished to meet the top-level criteria and requirements. For this purpose, reliability evaluations and probabilistic risk assessment techniques (Ref. 3) have been used to supplement standard engineering techniques and to provide an integrated allocation of the top-level criteria and requirements to specific plant systems. PRA techniques have also been used to identify the relative importance of events and plant structures, systems, and components in responding to such events, as described in Section 3.2. 

The Safety Case produced for the Windscale Vitrification Plant in 1994 included a detailed and comprehensive assessment of fault conditions in the plant using HAZOP and Probabilistic Risk Assessment techniques. The Safety Case identified a number of major hazards. These major hazards, along with the protective measures, Operating Rules, and Safety Mechanisms designed to prevent these hazards or to mitigate them are briefly described below. 

Rasmussen, N.C. 1981 The application of probabilistic risk assessment techniques to energy technologies. Annual Review of Energy, 6, 123—138. 

Burmaster and Anderson (1994) have proposed 14 principles of good practice for using Monte Carlo techniques. They suggest that before an analyst undertakes a Monte Carlo risk assessment, the growing literature on probabilistic risk assessment should be thoroughly examined. Principles for a properly conducted Monte Carlo analysis have also been proposed by the USEPA (1997). 

HAZAN, on the other hand, is a process to assess the probability of occurrence of such accidents and to evaluate quantitatively the consequences of such happenings, together with value judgments, in order to decide the level of acceptable risk. HAZAN is also sometimes referred to as Probabilistic Risk Assessment (PRA) and its study uses the well-established techniques of Fault Tree Analysis and/or Event Tree Analysis ... 

Toward the end of the Second World War, systems techniques such as fault tree analysis were introduced in order to predict the reliability and performance of military airplanes and missiles. The use of such techniques led to the formalization of the concept of probabilistic risk assessment (PRA). The publication of the Reactor Safety Study (NRC, 1975)—often referred to as the Rasmussen Report after the name of principal author, or by its subtitle WASH 1400—demonstrated the use of such techniques in the fledgling nuclear power business. Although WASH 1400 has since been supplanted by more advanced analysis techniques, the report was groundbreaking in its approach to system safety. 

One method for analyzing human reUability is a straightforward extension of probabilistic risk assessment (PRA)—in the same way that equipment can fail, so can a human make mistakes and slips. One technique for predicting human error rates is the THERP, which was developed in the 1950s. As with other PRA techniques, THERP models can use either point. 

The goal of human error quantification is to produce error probabilities, building on task analysis and error identification techniques to provide a probabilistic risk assessment (PRA). This provides numerical estimates of error likelihood and of the probability of overall likelihood of system breakdown. Quantification of error is the most difficult aspect of HRA, often heavily reliant on expert judgement, rather than the more rigorous approach of actual observation and recording of error frequencies. Such techniques are little used in healthcare but have been successfully applied to anaesthesia (Pate-Cornell and Bea, 1992). Nevertheless, some hospital tasks, such as blood transfusion, are highly structured and the quantification of errors probabilities would seem to be eminently feasible (Lyons et al, 2004). 

Probabilistic Risk Assessment (PRA) builds on such techniques as FMEA and HAZOP, by adding modelling of fault and event trees and assignment of probabilities to events and outcomes. 

Mohaghegh, Z, Kazemi, R, Mosleh, A, 2008, Incorporating organizational factors into Probabilistic Risk Assessment (PRA) of complex socio-technical systems A hybrid technique formalization. Reliability engineering and system safety, 94, 5 1000-1018. 

You don t need to be reminded of the most recent nuclear accidents, principally Fukushima Daiichi in Japan in 2011. After the Three Mile Island accident in the late 1970s, the U.S. Atomic Energy Commission developed WASH 1400, The Reactor Safety Study. The WASH 1400 report laid the foundation for the use of probabilistic risk assessments (called probabilistic safety assessments in Europe). According to Henley and Kumamoto (1991), probabilistic risk assessment involves studying accident scenarios and numerically rank[ing] them in order of their probability of occurrence, and then assess[ing] their potential consequence to the public. Event trees, fault trees, and other risk-consequence tools are applied in developing and studying these scenarios. These techniques are extremely useful for the engineer but very expensive. The nuclear industry has been the leader in probabilistic safety analyses. 

As a reaction to a string of commercial nuclear reactor accidents, the U.S. Nuclear Regulatory Commission published a reactor safety study, popularly known as WASH-1400 (U.S. Nuclear Regulatory Commission, 1975). The study laid formal groundwork for conducting probabilistic risk assessments in the commercial nnclear industry. The assessment technique has been used in nnmerons indnstries inclnding automobiles, food safety, environmental, petrochemicals, and aerospace. 

Risk assessment is the formal process of calculating the risk of an event and making a decision on how to react to that risk. This risk assessment process (Bahr, 1993) is a useful technique for conducting a probabilistic risk assessment consistent with the U.S. Nuclear Regulatory Commission WASH-1400 reactor stndy. 

However, the application of probabilistic (or quantitative) risk assessment techniques means that there is a need to make judgments about the reliabilities of redundant systems, even when their failure rates are dominated by CMF. Two main approaches have been developed for modeling the effects of CMF in probabilistic (or quantitative) risk assessment ... 

Human health risk assessment has often been dominated by the use of default assumptions and worst case analyses, based on the use of upper bounds on the dose from exposure instead of distributional characterizations of that dose. There are severe limitations associated with the use of default assumptions and upper bounds instead of distributions when detailed exposure and/or dose-response data are available. The US National Academy of Sciences, the USEPA, and many others have recognized the need for new risk assessment methodology (NRC, 1983, 1993, 1994 USEPA, 1992 CRARM, 1997). This need has promoted the development of new quantitative risk assessment methods that use probabilistic techniques, especially Monte Carlo simulation and distributional characterizations of dose-response, exposure, and risk. For these reasons, this paper uses a probabilistic approach. An indication of some of these new methods and the type of results they produce are given below. 

If default constants are used for each of several different parameters in the risk assessment, then the conservative aspect of the individual components is compounded when they are combined in the risk characterization. Furthermore, the extent of the overestimation cannot be readily quantified, and so the magnitude of the overestimation of the risk is not identified. However, distributional techniques make it possible to combine exposures more realistically - whether from multiple years, subpopulations, exposure pathways, or chemicals - without having to assume the worst case for each component. By carrying all the information for each component of the risk assessment through to the end of the entire risk characterization, instead of requiring interim single-number characterizations, probabilistic techniques help avoid the compounding of the conservative aspect of multiple parameters. 

---