Risk-informed Performance-based Regulation for Advanced Nuclear Power Reactors: Role of Artificial Intelligence as an Enabling Technology

Fisher, Riley; Sakurahara, Tatsuya; Bui, Ha; Afzali, Ami; Rowell, Arden; Mohaghegh, Zahra

Permalink

https://hdl.handle.net/2142/121808 Copy

Description

Title

Risk-informed Performance-based Regulation for Advanced Nuclear Power Reactors: Role of Artificial Intelligence as an Enabling Technology

Author(s)

• Fisher, Riley
• Sakurahara, Tatsuya
• Bui, Ha
• Afzali, Ami
• Rowell, Arden
• Mohaghegh, Zahra

Issue Date

2023

Keyword(s)

• Risk-informed performance-based regulation
• Advanced nuclear power reactors
• AI and machine learning
• Probabilistic risk assessment
• Probalistic Risk Assessment (PRA)

Abstract

As a promising solution to climate change and energy sustainability, advanced nuclear power reactors have gained rapidly growing attention. One of the crucial areas for the successful commercialization of advanced reactors is the establishment of a regulatory framework for advanced reactors that can ensure adequate protection of public health, safety, and the environment while enhancing the predictability and efficiency of regulatory processes [1]. Recent regulatory developments have indicated increasing support for risk- informed, performance-based (RIPB) approaches to addressing the risks and safety of advanced reactors; in particular, the U.S. NRC, in engagement with stakeholders, is working on rulemaking of 10 CFR Part 53, "Risk-Informed, Technology-Inclusive Regulatory Framework for Commercial Nuclear Plants," to provide a RIPB option for licensing future nuclear power reactors, and the Licensing Modernization Project (LMP)’s industry guidance document “Risk-Informed Performance-Based Guidance for Non-Light Water Reactor Licensing Basis Development” [2] has been endorsed by the NRC for licenses, certifications, and approvals of advanced reactors. Nevertheless, key puzzles in implementing RIPB approaches have yet to be resolved. This is because RIPB safety regulation presents unique challenges and opportunities in high- hazard industries, especially where regulators must choose amongst regulatory designs for preventing low- frequency, high-consequence events. [3] Academic research suggests that RIPB approaches can serve socially valuable purposes by improving macro-level societal safety. How well regulations serve this purpose, however, depends on contextual factors, including regulatory and legal structure and goals and the capabilities of firms and regulators, as well as the engineering realities on the ground. Although the potential for using RIPB regulatory strategies has been examined in other high-hazard industries (such as offshore oil and gas development), the safety case for using such approaches to regulate new nuclear reactor risk while realizing the benefits of innovative technologies to the full extent remains to be investigated. This paper reports the progress of the authors’ ongoing research to evaluate the recommended design of RIPB regulation for advanced nuclear reactors, focusing on the engineering-informed legal and regulatory context in which such regulation occurs, and the likely role of artificial intelligence (AI) in achieving regulatory and safety requirements. The preliminary insights from a comparative analysis of the current constructs of regulatory approaches (both the NRC’s regulation and those in other high-consequence industries), such as prescriptive, deterministic, versus RIPB approaches, will be discussed. This paper will conclude by discussing the role that artificial intelligence (AI) can play in automating probabilistic risk assessment (PRA) processes, which are used as one of the key pillars in the RIPB regulatory framework by the U.S. NRC. The potential benefits and issues of using AI in the RIPB regulation will be discussed from engineering and legal construction viewpoints. In particular, the authors will discuss their views that automating the development and updates of PRA is the key to ensuring the effectiveness and practical feasibility of the RIPB approach for advanced reactors.

Type of Resource

text

Language

eng

Handle URL

https://hdl.handle.net/2142/121808

Owning Collections