| Probabilistic risk estimates are typically not obtained for time-dependent releases of radioactive contaminants to the geosphere when a series of sequentially coupled transport models are required for determining results. This is due, in part, to the geophysical complexity of the site, numerical complexity of the fate and transport models, and a lack of a practical tool for linking the transport components in a fashion that facilitates uncertainty analysis.;Using the theory of convolution integration, sequentially coupled submodels can be replaced with an independent system of impulse responses for each submodel. Uncertainties are then propagated independently through each of the submodels to significantly reduce the complexity of the calculations and computational time. The impulse responses of the submodels are then convolved to obtain a final result that is equivalent to the sequentially coupled estimates for each source distribution of interest.;In this research a multiple convolution integral (MCI) approach is developed and the decoupling of fate and transport processes into an independent system is described. A conceptual model, extracted from the Inactive Tanks project at the Oak Ridge National Laboratory (ORNL), is used to demonstrate the approach.;In this application, uncertainties in the final risk estimates resulting from the ingestion of surface water show that the range of variations of the right tail of the PDFs are over several order of magnitude. Also, sensitivity analysis shows that uncertainty in the final risk is mainly attributed to uncertainties inherent in the parameter values of the transport model and exposure duration. These results demonstrate that while the variation in the tail of time-dependent risk PDF (the region of interest to regulatory decisions) are large, the resulting confidence level that human health has been protected is only slightly increased. In terms of remediation cost, this slight increase yields huge costs, and might reflect poor management decisions.;Ultimately, the MCI approach made uncertainty analysis a highly automated process and restored it as beneficial tool for investigators and decision makers while maintaining an effective balance between cost and level of accuracy of the model predictions. |
