Confidence in the Long-term Safety of Deep Geological Repositories

Confidence in the long-term safety of deep geological disposal, and the ways in which this
confidence can be obtained and communicated, are topics of great importance to the radioactive waste
management community.1
The technical aspects of confidence have been the subject of considerable debate, especially
the concept of model validation. It has, for example, been pointed out that it is impossible to describe
fully the evolution of an open system, such as a repository and its environment, that cannot be
completely characterised and may be influenced by natural and human-induced factors outside the
system boundaries.2 A complete description is not, however, a requirement of decision making in
repository development. Repository development proceeds in stages, and the depth of understanding
and technical information available to support decisions will vary from stage to stage. Decision
making requires only that a description of the possible evolutions of the system has been compiled that
gives adequate confidence in safety to support the decision at hand, and that an efficient strategy exists
to deal at future stages with any uncertainties in the description which have the potential to
compromise safety. Furthermore, flexibility should be built into the process of repository
development, allowing account to be taken of new understanding and technical information, as well as
the demands of societal review.
This report is aimed at practitioners of safety assessment and at technical specialists wishing
to become versed in the subject. In its current form, it is intended to improve communication among
these specialists by clarifying the concepts related to the development of confidence, and by placing
the various measures that are employed to evaluate, enhance and communicate confidence in the
technical aspects of safety in a clear, logical framework. These measures are increasingly embodied in
actual procedures applied in today’s safety assessments, and can be incorporated in a common
framework, despite differences in approaches, practices and constraints both within and between
repository projects.