Making decisions for incident management in nuclear power plants using probabilistic safety assessment

Abstract

Decisions by management during nuclear power plant operations may have safety-related and economic consequences, and may affect the relationship with external stakeholders such as regulatory authorities. Attributes such as worker exposure to radiation, potential radiation release, negative publicity, and regulatory intervention may conflict with concerns such as loss of income due to plant shutdown or reduced power. This article shows that the combination of probabilistic safety assessment and multi-attribute decision theory offers the potential for a structured decision-making methodology that could take into account risk-related aspects (plant and worker safety, for instance), as well as important factors like economics and regulatory requirements. The difficulties we encountered during the elicitation of utilities, and our solutions to those issues, are discussed. We demonstrate our decision analysis methodology using a case study in incident management—the decision making that follows events emanating from non-safety-related component failures. Particular attention has been paid to the assessment of tradeoffs among different objectives, especially when some of the consequences of decision options can be catastrophic.

Keywords:

• nuclear power plants
• probabilistic safety assessment
• decision analysis

Notes

*Permanent address: Idaho National Engineering and Environmental Laboratory, Idaho Falls, ID 83415-3850, USA. E-mail: [email protected]

¹The way the deliberation process should be done is beyond the scope of this article. We will focus only on Step 3 in Figure 1, i.e., the development of the formal process employed to rank and analyze the decision alternatives. Examples of the issues related to deliberation are given in Apostolakis et al. (1998).

²We call Performance Measures the fundamental objectives to highlight the fact that they have to be quantifiable.

³Initiating events are disturbances to normal operations that, if unmitigated, may lead to accidents. Examples are the loss of offsite power and various loss-of-coolant accidents. Minor incidents do not involve such events. However, a minor incident could lead to an initiating event or could increase the probability of failure in case an initiating event happens.

LOCA, Loss Of Coolant Accident; ATWS, Anticipated Transient Without Scram; LH, Loss of Heat Sink; LOOP, Loss Of Offsite Power; MWLB, Main Water Line Break; SGTR, Steam Generator Tube Rupture; MSL, Main Steam Line break; Sec-Tran, Secondary circuit transient; Pri-Tran, Primary circuit transient.

⁴These values come from estimates of the utility company.

⁵While the absolute difference between the PIs is only 5.2 × 10⁻³, the relative difference is what determines the relative importance of the alternatives, and this difference is more than 300%. It should be noted that this happens because even if the range of disutility values is between zero and unity, only small values will be expected because of the very low probability of undesired events.