ON THE USE OF EXPERT JUDGMENTS TO ESTIMATE THE PRESSURE INCREMENT IN THE SEQUOYAH CONTAINMENT AT VESSEL BREACH

The use of expert judgments in probabilistic risk assessments has beco me common. Simple aggregation methods have often been used with the re sult that expert biases and interexpert dependence are often neglected . Sophisticated theoretical models for the use of expert opinions have been proposed that offer ways of incorporating expert biases and depe ndence, but they have not found wide acceptance because of the difficu lty and rigor of these methods. Practical guidance on the use of the v ersatile Bayesian expert judgment aggregation model is provided. in pa rticular, the case study of pressure increment due to vessel breach in the Sequoyah nuclear powerplant is chosen to illustrate how phenomeno logical uncertainty can be addressed by using the Bayesian aggregation model. The results indicate that the Bayesian aggregation model is a suitable candidate model for aggregating expert judgments, especially if there is phenomenological uncertainty. Phenomenological uncertainty can be represented through the dependence parameter of the Bayesian m odel. This is because the sharing of assumptions by the experts tends to introduce dependence between the experts. The extent of commonality in the experts' beliefs can be characterized by assessing their inter dependence. The results indicate that uncertainty is possibly underest imated by ignoring dependence. Two Bayesian approaches are used. The f irst approach uses the experts' opinions as evidence to update the dec ision maker's state of knowledge. The second approach, in recognition of the fact that the experts are highly dependent on a common informat ion source, assumes that the common information source is the actual e xpert and the participants are assessing its biases and credibility. T he results lend validity to the use of weighted averaging schemes beca use the Bayesian aggregation method encompasses simple arithmetic and geometric averaging techniques.