Establishing credibility in the environmental models used for safety and licensing calculations in the nuclear industry

Models that simulate the transport and behaviour of radionuclides in the environment are used extensively in the nuclear industry for safety and licensing purposes. They are needed to calculate derived release limits for new and operating facilities, to estimate consequences following hypothetical accidents and to help manage a real emergency. But predictions generated for these purposes are essentially meaningless unless they are accompanied by a quantitative estimate of the confidence that can be placed in them. For example, in an emergency where there has been an accidental release of radioactivity to the atmosphere, decisions based on a validated model with small uncertainties would likely be very different from those based on an untested model, or on one with large uncertainties. This paper begins with a discussion of some general methods for establishing the credibility of model predictions. The focus will be on environmental transport models but the principles apply to models of all kinds. Establishing the credibility of a model is not a trivial task, It involves a number of tasks including face validation, verification, experimental validation and sensitivity and uncertainty analyses. The remainder of the paper will present quantitative results relating to the credibility of environmental transport models. Model formation, choice of parameter values and the influence of the user will all be discussed as sources of uncertainty in predictions. The magnitude of uncertainties that must be expected in various applications of the models will be presented. The examples used throughout the paper are drawn largely from recent work carried out in BIOMOVS and VAMP. (DM)