Author(s)

J. Vitázková & E. Cazzoli

Abstract

The principle of the Defence-in-Depth concept has been set forth by the IAEA as fundamental for the safety of Nuclear Power Plants in INSAG10 (1996). Within the time, essentially after severe accidents in Three Mile Island and Chernobyl, the concept evolved and the currently accepted definition is based on five successive layers of safety and four physical safety barriers, including organizational and administrative measures. Safety demonstrations are required to show that the safety layers and barriers provide sufficient margin to prevention or mitigation of releases of radioactivity to the environment. These safety demonstrations are well established for Design Basis Accidents. Probabilistic Safety Assessments (PSAs) assumed to be risks assessment tools also for Severe Accidents, however, enter the description of the Defence-in- Depth concept only as an afterthought and their role is marginalized to \“improvement of defence in depth” and \“optimization of efforts to implement defence in depth”. This article investigates aspects related to implementation of defence in depth in the wake of the Fukushima accidents and in view of the historical evidence of nuclear power plants’ risks considering also risk targets complying with IAEA safety objectives and principles. Analysis of the levels of currently accepted defence in depth is performed focusing on real accidents vs. PSA analyses results with objective to answer the question, if some of current defence-in-depth levels may not actually be yet a reason of concern instead of prevention or mitigation of risk, and, what should be the role of risk targets and risk assessment within the defence in depth concept. Keywords: risk, risk target, Defence-in-Depth, Fukushima, severe accident, real accident, administrative measure, safety barrier, safety level, safety margin.

Keywords

risk, risk target, Defence-in-Depth, Fukushima, severe accident, real accident, administrative measure, safety barrier, safety level, safety margin.