Both the Commission of the European Communities (CEC) and the United States Nuclear Regulatory Commission (USNRC) have expressed their concerns with respect to the expected consequences in cases accidental releases of radionuclides might occur, in various ways. Both commissions have issued ACA codes (COSYMA by the CEC, MACCS by the USNRC), which enable users to make estimates of the risks presented by nuclear installations based on postulated frequencies and magnitudes of potential accidents.
Knowledge of the uncertainty associated with these risk estimates play a crucial role in the effective allocation of resources to be put in risk reduction measures. In the absence of such estimates, suboptimal priorities might be assigned to particular topics. fairly comprehensive assessments of the uncertainties regarding the code risk estimates have already been made. These uncertainty assessments were performed by feeding in probability distributions of the relevant parameters to both models.
These uncertainty assessments were largely done by the code developers as opposed to experts in each of the many different scientific disciplines which feature within an ACA code. The formal use of expert judgement has the potential to circumvent these criticisms. Although the use of expert judgement is common in the resolution of complex problems, it is most often used informally and rarely made explicit. The formal use of expert judgement elicitation and evaluation was used extensively in both Europe (by Delft University of Technology) and the US (NUREG-1150 study by Sandia National Laboratories).
These ACA codes consist of several modules, with numerous input parameters of which a relatively small number will determine the uncertainties of the code output. These key input parameters are selected based on previous sensitivity analyses of the ACA codes or modules. The important phenomenological areas for which the key parameter uncertainties are assessed with expert judgement are:
* atmospheric dispersion and deposition
* behaviour of deposited materials and the calculation of its related doses
* foodchain models
* internal dosimetry
* early or deterministic health effects, and
* late or somatic health effects.
The current joint project is undertaken with the aim of quantifying the uncertainty on key parameters in these areas.
The area of atmospheric dispersion and deposition (2 expert panels with 8 experts each = 16 experts) was the first subject of the study. Teams from the EC and USNRC were able to successfully work together to develop a unified process for the development of uncertainty distributions on consequence code input parameters. Furthermore, in this project, formal expert judgement elicitation has proven to be a valuable vehicle to synthesise the best available information by a most qualified group.
Three more panels of experts were elicited thereafter:
* foodchain processes, consisting of two separate expert panels:
** on soil/plant transfer and processes (7 experts), and
** on animal transfer and behaviour (9 experts)
* the behaviour of deposited material and related doses ( one panel of 10 experts).
Three additional expert panels were started (each will have approximately 10 to 12 experts = circa 36 experts in total):
* deterministic or early health effects
* internal dosimetry
* late or somatic health effects.
The expert judgement methodology applied in this project is a combination of methods from previous US and EC studies: the NUREG-1150 method and methods developed in Europe at Delft University of Technology.
The objectives of the proposed joint project are to further develop and apply expert judgement elicitation techniques in estimating the uncertainties associated with the predictions of probabilistic accident consequence assessment codes, and to investigate the use of the results of these studies as input to uncertainty analyses of such codes.
Within the project the following contributions will be made: (i) To further develop methodological backgrounds and
implementational aspects of expert judgement elicitation techniques to be applied in the joint project (TUD contribution).
(ii) To further develop mathematical techniques for handling modelling uncertainty (TUD contribution).
(iii) To perform modifications and extensions of the System for Uncertainty and Sensitivity Analysis (SUSA), that emerge from methodological improvements, for instance from (ii), as well as from the practical applications of the system to results from the COSYMA and possibly MACCS codes (GRS contribution).
(iv) To provide the necessary detailed information on the relevant parameters and models, in order to:
(a) assist in the selection of appropriate experts in specific subject areas of consequence assessment (NRPB-KfK-SRD-ECN
(b) provide experts both in the general area of accident
consequence assessments, and in particular modelling aspects, who will interact with and advise the chosen experts in other fields (NRPB-KfK-SRD-Ecn contributions),
(c) consider the implications of the results of the expert
elicitation with regard to an uncertainty analysis of the COSYMA code (NRPB-KfK contributions).
The following subject areas will be investigated: Atmospheric dispersion and deposition (NRPB-SRD), Internal and external
dosimetry (SRD-ECN), Countermeasures modelling (KfK-NRPB), Health effects modelling (KfK-ECN), Foodchain modelling (NRPB-KfK).
Funding SchemeCSC - Cost-sharing contracts
WA3 4NE Warrington
OX11 0RQ Didcot,harwell,chilton