DEVELOPMENT OF A PROBABILISTIC PERFORMANCE ASSESSMENT MODEL TO SUPPORT ENVIRONMENTAL DECISION MAKING

Regulators and other decision makers are faced with the problem of making decisions in light of uncertainty, and typically do not have an adequate appreciation for the uncertainty inherent in environmental analysis. Until recently, radiological performance assessments and other environmental models have been restricted to a deterministic approach, producing a single set of values (concentrations, receptor risks) to represent the performance of a contaminated site. The decision maker had only to compare these values to those presented in relevant regulations in order to determine if a site is in compliance. These single values, however, are inherently uncertain, and this uncertainty is not generally appreciated by the decision makers or the public.

Through probabilistic modeling, a more thorough and honest estimate of risks can be offered to the decision maker and the public. This probabilistic result represents the state of knowledge about the site and its behavior. The decision maker’s job may seem more difficult (more difficult than comparing two values) but in fact he is empowered by a greater appreciation for the uncertainties present in the conceptualization of the site and in its analysis. The decision maker is also provided with information related to the chance of making the wrong decision.

When probabilistic modeling is coupled with subsequent multivariate sensitivity analysis a more powerful tool emerges. This analysis can identify which parts of the model are most significant to the results in question. This information can help to guide further investigation to be the most effective at reducing uncertainties should they be uncomfortably large.

This presentation demonstrates a generic radiological performance assessment model (e.g. for shallow land burial of low-level radioactive waste) built using the GoldSim probabilistic systems analysis programming platform. The model considers typical stochastically-defined performance assessment input parameters and processes, and produces probabilistic results ready for a sensitivity analysis. Similar models are in use at existing low-level waste sites, and not only assess regulatory compliance, but are used to inform operational decisions such as the acceptance of candidate waste streams.