Title: RODOS and Decision Conferencing on Early Phase Protective Actions in Finland

Authors: R. P. Hämäläinen, K. Sinkko, M. Lindstedt, M. Ammann, A. Salo

Status: Finnish Centre for Radiation and Nuclear Safety, STUK-A159, 1998.

Keywords: Nuclear emergency management; decision support; multiattribute risk analysis

This work was undertaken in order to study the utilisation of decision conferencing and of the rodos system when considering early phase protective actions in the case of a nuclear accident. Altogether four meetings with various people were organised. The meetings were attended by competent national safety authorities and technical level decision-makers, i.e., those who are responsible for preparing advice or making presentations of matters for decision-makers responsible for practical implementation of actions. In the first set of meetings the aim was to elicit the factors/attributes that have to be considered when making a decision on sheltering, evacuation and iodine tablets. No uncertainties nor a threat phase were considered but everything was assumed to happen as described in the given scenario. The theme in the second set of meetings was to study the implications of probabilities. All information was calculated with the support of the rodos system.

In the early phases of a nuclear accident time is limited. Prestructured generic value trees or a list of possible attributes can help to save time. A possible approach is to present a large generic value tree. Either the decision-makers select the attributes that are suitable for the case in hand or the facilitator offers a choice between more structured value trees. The decision-makers then just examine the suggested value trees, check the generic tree to make sure that no important factors have been omitted and choose the appropriate one.

As in previous rodos exercises, the participants felt that rodos could be used for providing information but found it more problematic to use decision analysis methods when deciding on countermeasures in the early phase of a nuclear accident. Furthermore, it was noted that understanding the actual meaning of ‘soft’ attributes, such as socio-psychological impacts or political cost, was not a straightforward issue. Consequently, the definition of attributes in advance would be beneficial. The incorporation of uncertainties also proved to be difficult. The participants felt uneasy about probabilities and they focused rather on the worst possible consequences of the accident.

All in all the results from this study are promising. Further meetings, however, have to be organised in order to deepen insight into the features of the decision-making process in the early phases of an accident and to familiarise decision-makers with decision analysis techniques. And more research is needed on how to implement decision conferencing in nuclear emergency management.