“…The screening must be carried out in a well-documented manner, including formal expert elicitation and peer review panels (Hora and Iman, 1989;Bonano et al, 1990;Ortiz et al, 1991). Note that expert elicitation and peer review are not superficial activities, but carefully structured and documented procedures to obtain the best opinions corresponding to the state-of-knowledge.…”
“…The screening must be carried out in a well-documented manner, including formal expert elicitation and peer review panels (Hora and Iman, 1989;Bonano et al, 1990;Ortiz et al, 1991). Note that expert elicitation and peer review are not superficial activities, but carefully structured and documented procedures to obtain the best opinions corresponding to the state-of-knowledge.…”
“…Plus spécifiquement, le jugement représente l'état de la connaissance de l'expert au moment de sa réponse à la question (Ortiz et al, 1991). En général, on se sert du jugement expert dans les analyses de risques, lorsque les données de retours d'expérience sont insuffisantes (en termes de quantité ou de qualité) voire inexistantes.…”
RésuméEn France, la réglementation concernant l'évaluation du niveau de sécurité des linéaires de digues a été renforcée. Elle nécessite l'évaluation probabiliste du risque de défaillance des ouvrages. Or ceux-ci et leurs mécanismes de défaillances demeurent dans une large mesure mal connus, conduisant à une prédominance de l'exploitation du jugement expert dans les études de diagnostic. Ainsi une erreur de jugement peut avoir des conséquences catastrophiques et la production d'une information fiable induit que l'ingénieur est capable de décrire fidèlement ses évaluations ainsi que l'incertitude qui s'y rattache. Répondant à ce besoin, notre recherche, au sein du projet INCERTU, vise à produire des approches et des outils scientifiques pertinents pour la collecte et le traitement des déclarations d'experts fiables ou assorties d'un degré de vraisemblance. Nous présentons ici nos recherches concernant l'identification et le traitement des biais susceptibles d'entacher les évaluations expertes.
SummaryIn France, the regulation on the evaluation of linear levees security level has been strengthened. It requires the probabilistic assessment of failure risk of structures. However thereof and their failure mechanisms remain largely not well known, leading to a predominance of the exploitation of the expert judgment in diagnostic studies. Thus an error in judgment can have catastrophic consequences and the production of reliable information induces the engineer is able to accurately describe these evaluations and the uncertainty associated whit it. To meet this need, our research, within INCERTU project, aims to produce relevant scientific approaches and tools for the collection and treatment of reliable expert statements or accompanied by a degree of verisimilitude. We present our research on the identification and treatment of biases may taint the expert evaluations.
“…To create high-quality assessments of these prior probabilities, a more formal elicitation of expert probab~ty judgments would be wuI. "his type of process has been applied in the nuclear safety and related areas for several years (Merkhofer 1987;Wheeler et al 1989;Ortiz et al 1990; Keeney and von Winterfeldt 1991). the amount and activity factors levels of vitrified TRU…”
Executive SummaryThis report presents the rationale for adopting a recommended characterization strategy that uses a risk-based decision-making eamework for managing the Tank Waste Characterization Program at Hanford. The risk-managemendvalue-of-information (VOI) strategy that is illustrated explicitly links each information-gathering activity to its cost and provides a mechanism to ensure that characterization funds are spent where they can produce the largest reduction in risk. The approach was developed by tailoring well-known decision analysis techniques to specific tank waste characterization applications. 'This report illustrates how VOI calculations are performed and demonstrates that the VOI approach can definitely be used for real Tank Waste Remediation System (TWRS) characterization problems.The goal of a characterization strategy is to provide timely information to support decisions. Information has value if it leads to better decisions, which are those with better expected consequences. Thus the VOI should be gauged by the expected increase in the value of decisions. Frequently, information will reduce uncertainty about consequences. Idormation will also often lead to decisions that involve reduced risks.Qualitatively speaking, a source of information (e.g., a tank waste sample) is valuable if it has the potential for changing subsequent decisions, such as whether to keep transuranic (TRU) waste separate from high-level waste (HLW). For a source of information to be valuable, two conditions must be met:1. The decision alternatives must have uncertain consequences (otherwise, there is nothing to be learned from the information).2. Depending on the information obtained, different decisions may be best (otherwise, the information would not have any impact).VOI decision analysis techniques are used for determining risk-based characterization requirements.This approach provides an understandable technical basis that explicitly links sampling, analysis, physicalchemical modeling, and other "learning" activities to risk reduction. Furthermore, VOI analysis provides a clear measure of completion because additional characterization activities are not justified when the costs exceed the calculated risk reduction value.Risk-based requirements defined in terms of the VOI become a defensible basis for integration and prioritization of needs across tanks and program elements requiring information about waste composition, phenomenology, and performance. Knowing the value of obtaining specific information will provide an explicit basis for investments in research and/or technology development to reduce costs and technical uncertainty. This approach will also provide a defensible basis for budgeting and scheduling decisions by providing the means for selecting the most impactfid characterization work for funding each year. This integration, coupled with statistically and technically sound sampling and analysis, monitoring, and laboratow methods, will result in an implementation plan that provides the most cost-effect...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.