The Phase 3 Total System Performance Assessment (TSPA) sponsored by the Electric Power Research Institute (EPRI) has led to new insights into critical models and parameters affecting estimated doses to humans from a potential repository of high-level radioactive wastes at Yucca Mountain, Nevada. The Phase 3 model has been extended to encompass time-varying climate and infiltration, detailed modeling of the source term and hydrology, and detailed specification of possible interaction between percolating ground water and waste containers. The model estimates doses to a time of one million years.

The three key radionuclides contributing to estimated total doses are Tc-99,1–129, and Np-237. Five other nuclides contributing to dose in lesser (but significant) amounts are U-233, Th-229, Pa-231, Ac-227, and Se-79. These results are consistent with other TSPAs.

From sensitivity studies, the most critical models and parameters are as follows. Infiltration and percolation assumptions, including the amount of lateral diversion of infiltration water, are important and need verification with site data and/or more detailed modeling. Parameters of the unsaturated zone (UZ) and saturated zone (SZ) determine dilution and delay of concentrations and peak doses downstream. The fraction of containers that become wet are not critical in our model, but this lack of sensitivity reflects our coupling of the fraction with a model of focused flow past the containers; an different model might indicate higher sensitivity. Also, the degree of coupling between fracture and matrix flow is important in affecting the times of peak doses but not their magnitudes.

Other critical design assumptions that could lead to reduced and/or delayed doses are a more robust container design, a capillary barrier around each container, the dilution during hydrologie transport from the repository to a potential agricultural community downstream, and the characteristics of an “average” individual in that community who might receive a dose.