Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.370962
Title: A new treatment of low probability events, with particular application to nuclear power plant incidents
Author: Critchley, Octavius Hunt
ISNI:       0000 0001 3396 7654
Awarding Body: Queen Mary University of London
Current Institution: Queen Mary, University of London
Date of Award: 1984
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Technological innovation is inescapable if civilisation is to continue in the face of population growth, rising expectations and resource exhaustion. Unfortunately, major innovations, confidently thought to be safe, occasionally fail catastrophically. The fears so engendered are impeding technical progress generally and that of nuclear power in particular. Attempts to allay disquiet about these disastrous Low Probability Events (LPEs) by exhaustive studies of nuclear power plant designs have, so far, been less than successful. The New Treatment adopts instead an approach that, after examination of the LPE in its historical and societal settings, combines theoretical design analysis with construction site and operational realities in pragmatic engineering, the quality of which can be assured by accountable inspection. The LPE is envisaged as a singularity in a stream of largely mundane, but untoward incidents, described as 'Event-noise'. Predictions of the likelihood of plant LPEs by frequency-theory probability are illusory because the LPE is unique and not part of a stable distribution. Again, noise analysis seems to lead to intractable mathematical expressions. While theoretical LPE prognostications depend on the identification of fault sequences in design that can either be designed-out or reduced to plausibly negligible probabilities, the reality of LPE prevention lies with the plant in operation. As absolute safety is unattainable, the approach aims at ensuring that the perceived residual nuclear risk is societally tolerable. An adaption of elementary Catastrophe theory to model the prospective Event-noise field to be experienced by the plant is proposed whereby potential, credible LPEs could be more readily discerned and avoided. In this milieu of increasing sophistication in technology when management in the traditional administrative mold is proving inadequate, the engineer emerges as the proper central decision-maker. The special intellectual capability needed is acquired during his training and experience, a claim that can draw support from new studies in neuropsychology. The Nuclear Installation Inspectorate is cited as an exemplar of a body practising the kind of engineering inspection needed to apprehend those human fallibilities to which most catastrophic failures of technology are due. Nevertheless, such regulatory systems lack accountability and, as Goedel's theorem suggests, cannot assess their own efficiency. Independent appraisal by Signal Detection Theory is suggested as a remedy.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.370962  DOI: Not available
Keywords: Nuclear Engineering ; Low Probability Events ; Technological hazards
Share: