Catálogo de publicaciones - libros

The probabilistic risk assessment (PRA) is the most powerful approach to quantification of risk and safety. Risk is a combination of probability of harm and severity of that harm, while safety is freedom from unacceptable risk [1].

A plant consists of a variety of systems, structures, and components (SSCs) operated and maintained directly or indirectly by humans. Some SSCs and human activities (HAs) are more important than others from the point of view of risk. A risk-informed safety assurance utilizes risk information to 1) satisfy safety goals, 2) gain public trust, 3) increase safety assurance effectiveness, and 4) to remove unnecessary burden. The first step of the risk-informed safety assurance is the categorization of SSCs and HAs. The second step is the realization of requirements demanded for each category (Chapter 3)

Safety goals, quantitative health objectives, subsidiary numerical objectives, and tolerable risks are dealt with in Chapter 1. Risk-informed categorizations of safety systems, SSCs and human actions are described in Chapter 2 from the point of view of safety significance in satisfying tolerable or acceptable risk levels. This chapter considers how the requirements demanded for each category can actually be satisfied by uncertainty management, compliance with standards and regulations, dependent failure management, safety margins, human-factors review, early detection and treatment, defense-in-depth, and performance evaluation.

Risks cannot exist without hazards. A reasonably complete identification of hazards should be made. Initiating events as accident initiators are found, and risk-reduction measures are established. This chapter describes risk-reduction approaches based on hazard identification, hazard elimination, prevention and mitigation of initiating events and accident mitigation. Safety systems described in Chapters 2 and 3 are types of products from the risk-reduction framework given in this chapter.

This chapter overviews PRAs over three different levels. The PRA has been used most intensively in the nuclear field. The process industry is another intensive user of the PRA. Whenever there is a need for risk quantification, simpler versions of PRA are used in other fields. Risk quantification without the PRA is imperfect and in a very near future any industry with risks will use more and more complete versions of the PRA.

A plant can be decomposed into basic components including hardware and human. Event trees and fault trees contain events related to these basic components. The PRA integrates these events to quantify risks of the plant. The event-tree and fault-tree models facilitate the integration. This chapter describes basic event quantification prior to the integrations.

A top event is defined as an undesired state of a system (, a failure of the system to accomplish its function). The top event is the starting point (at the top) of the fault-tree model [5]. A basic event is defined as an event in a fault-tree model that requires no further development, because the appropriate limit of resolution has been reached [5]. This chapter focuses on the relationships between the top event and the basic events. The reliability parameters presented in Chapter 6 can be extended to the top event [53].

Risk reduction almost always would succeed if there were no dependent failures. The dependent failure is a source of a collapse of dependable risk reduction. Some dependencies are modeled explicitly in PRA, while others are dealt with by common-cause failure analysis. This chapter first describes a relatively recent methodology called the alpha-factor model for common-cause quantification. The well-known beta-factor model can be regarded as a variant of the alpha-factor model. The second dependency described in this chapter is a graceful degradation process where changes to inferior states occur in a gradual manner. The graceful or gradual degradation is a key approach to problems with risks because time is available to correct the current inferior situation.

To quote Alexander Pope (1688–1744), “to err is human”. Human errors in thinking and rote tasks occur, and these errors can destroy aircraft, chemical plants, and nuclear power plants. Our behavior is both beneficial and detrimental to modern engineering systems. The reliability and safety analyst must consider the human element; otherwise, the analysis is not creditable [53]. This chapter briefly discuss the human-error quantification. Refer to references [56],[69]–[73] for more detail. There is a software version called HRA Calculator [74]. The ASME PRA Standard recommends the use of THERP [56] and ASEP [69].