Thermo-hydrologic analyses of the potential repository at Yucca Mountain, Nevada, have been performed to address issues that are important to performance assessment calculations. One- and two- dimensional simulations of the unsaturated zone (UZ) have been performed in conjunction with three-dimensional simulations of the saturated zone (SZ) to gain insight into the thermo-hydrologic response following emplacement of heat-generating waste packages with an assumed heat load of 83 kW/acre. Several unique features are incorporated into the current analyses that aim to complement previous thermo-hydrologic analyses of Yucca Mountain. First, recent material data have been used as a basis to develop geostatistically simulated property distributions in the models (Altman et al., 1996). Current heterogeneous property distributions are believed to yield more realistic representations of the modeled domain. Second, alternative conceptual models of flow through fractures are used in these analyses. Two equivalent continuum models (ECM) and a dual permeability model (DKM) are used to determine similarities and differences in the models that may impact performance assessment calculations. Finally, a suite of sensitivity analyses that investigate the effects of different boundary conditions are presented. Different infiltration rates are used and their effects on re-wetting and re-fluxing times are investigated. The sensitivity of the UZ model to heat transfer at the water table is determined by comparing constant temperature and heat-conduction boundaries, and simulations of the saturated zone are also used to assess the validity of different heat transfer boundaries at the water table. The sensitivity of the UZ model to lateral boundary conditions is also investigated by comparing adiabatic conditions with constant temperature conditions along the lateral boundaries.

The following results and conclusions address the four objectives of this study: 1) describe the different thermal regimes, 2) assess the thermally driven alterations to the hydrologic conditions, 3) quantify the time to return to ambient conditions, and 4) identify the most likely and significant scenarios that may lead to radionuclide releases.

1. The thermal regimes observed in the UZ are similar for both the ECM and DKM simulations. Heating of the repository occurs from approximately 0 to 1000 years, followed by cooling for tens of thousands of years. A large boiling zone develops above the repository because of buoyant convection, extending nearly 200 m above the repository between 500-1000 years. The boiling zone collapses entirely after several thousand years, but the region continues to cool for tens of thousands of years. Conduction, convection, and latent heat exchange due to evaporation and condensation all play an important role in distributing the heat around the repository. During heating, a dry-out zone develops around the repository. In most of the realizations, the dry-out zone extends approximately 20 m above and below the repository near the center and decreases in extent near the ends of the repository. The duration of the dry-out 5 m above the center of the repository is on the order of 1000 years in the ECM simulations. In the SZ, higher temperatures are predicted in a thermal plume that extends down-gradient from the potential repository at least 5 km. The simulated thermal plume persists well beyond 10,000 years. Higher than ambient temperatures at the water table are also predicted by the SZ simulations because of slow groundwater velocities.

2. Significant alterations to the ambient flow in the UZ develop as a result of the thermal perturbation. Gas flow is directed away from the repository due to buoyancy-driven convection and pressure gradients. Liquid flow just outside of the dried region is directed towards the repository due to large capillary pressure gradients. In a region above the repository, condensate forms and, depending on the conceptual model, either accumulates in the matrix or drains through the fractures. In the ECM, the condensate remains in the matrix until the matrix is satiated, forming large, highly saturated regions above the repository. In the DKM, the condensate is free to drain through the fracture continuum, allowing the matrix to remain at lower saturations. The liquid flow through the fractures in the DKM can be on the order of 100 times greater than the ambient infiltration due to condensate drainage. Increased flow through the fractures in the DKM persists above and below the repository for 1000 years. In the two-dimensional DKM simulations, the regions of increased fracture flow are distributed and locally focused because of heterogeneities. Histograms of the duration of dry-out and water flux directly above the repository for 10 realizations using the ECM showed that when re-wetting occurs, the flux can be several times greater than the ambient infiltration. The increased fluxes are caused by large hydraulic gradients that exist because of the accumulated condensate in the matrix above the repository. Simulations also predict altered conditions in the SZ for at least the first 10,000 years following waste emplacement. Groundwater velocities in the shallow SZ are somewhat higher than ambient conditions because of the higher temperatures and lower liquid viscosities. In addition, the simulated temperature changes in the SZ suggest the possibility of mineralogic alteration.

3. The vast majority of the domain in the UZ base-case models returns to ambient conditions by 100,000 years, as evidenced by temperature, saturation, and mass flow profiles. In the SZ, temperatures are predicted to return to ambient conditions within about 65,000 years. However, re-wetting and re-fluxing times, defined as the time required for the saturation or the flux to first exceed the ambient values in the UZ, can occur within a few thousand years as a result of the large accumulation of condensate directly above the repository. The time required to return to a true ambient can be affected by a number of parameters addressed in the sensitivity analyses. Simulations with semi-analytical heat- conduction boundaries at the water table show higher temperatures near the lower boundary when compared to simulations using a constant temperature at the water table (base-case). Therefore, the time to return to ambient is slightly increased in the heat-conduction boundary simulations, but the majority of the model domain still returns to ambient by 100,000 years. Simulations that included both the UZ and SZ show similar increases in the temperature at the water table because of limited advective lateral heat transport in the SZ, but the overall impact on the behavior of the cooling in the UZ is expected to be minimal over the span of 100,000 years. Changing the infiltration rate by an order of magnitude after the start of heating does not show significant differences in the time to return to ambient, but changing the ambient steady-state infiltration rate in the DKM shows that re-wetting and re-fluxing times are very sensitive to the magnitude of the steady-state infiltration. For ambient infiltrations over 10 mm/year, no dry-out zone develops, and the flux of water directly above the repository never decreases below ambient levels.

4. The most significant aspect of the thermo-hydrologic simulations that could lead to radionuclide release in the UZ is the generation of large amounts of condensate and the subsequent behavior of the condensate. Depending on the conceptual model used in the simulations, different behavior is exhibited once the condensate is generated. In the ECM, the condensate remains primarily in the matrix above the repository. Once the repository begins to cool (~1000 years), the water moves slowly in the matrix as a large pulse through the repository, which creates the potential for significant amounts of water to contact the waste packages. In the DKM, large fluxes of condensate drainage in the fractures persist above the repository for the first 1000 years of heating and could allow focused flow to drip onto the waste packages. Large fluxes of condensate drainage in the fractures below the repository could also provide a fast pathway for radionuclides to reach the water table. It should be noted, however, that the base-case numerical simulations (0.3 mm/year) in this study did not show any liquid flow at the repository horizon while the repository elements were dry. The generated heat was sufficient to dry the repository elements, which prevented any through-flowing liquid for a period between ~100 years and several thousand years. In addition, elements adjacent to the dried out zone exhibited flow towards the repository because of capillary suction. Therefore, a strict interpretation of the simulations in this study would indicate that the release of radionuclides from within the repository is unlikely while the repository remains dry. From a performance assessment standpoint, the two conceptual models indicate possible time frames during which water can possibly contact the waste packages. The ECM predicts that water would most likely contact the waste packages as the accumulation of condensate moves through the repository via the matrix during cooling between 1000 and 5000 years. The DKM predicts that in addition to this pulse of condensate in the matrix, significant condensate drainage in the fractures can occur during the first 1000 years of heating. Water flux through the fractures is predicted to be over 100 times greater than the ambient infiltration at early times as a result of condensate drainage, yielding a significant possibility that focused flow through fractures could allow water to contact the waste packages during the heating phase as well. In the SZ, predicted temperature changes suggest the possibility of mineralogic alteration, including dissolution and precipitation reactions that could permanently alter the permeability structure of the aquifer. The resulting structure could impact the groundwater velocities and the transport of radionuclides.

The accuracy of the different conceptual models of the thermo- hydrologic behavior can be partially assessed in the upcoming thermal tests in the Exploratory Studies Facility (ESF) at Yucca Mountain. Predictions and extensions from this study can be applied to the thermal tests being established in the ESF. The DKM would predict significant condensate drainage through the fractures around the heater, and efforts should be made to monitor such behavior. If condensate drainage cannot be measured or detected in the thermal test in the ESF, then matrix saturations should be monitored. The ECM would predict a significant increase in the matrix saturations above the heater as a result of condensate accumulation. The DKM should predict lower matrix saturations due to drainage through the fractures. If the overall average water content around the heater in the ESF appears to be decreasing, then condensate drainage through the fractures is probably occurring as predicted by the DKM. If the average water content around the heater remains the same, then condensate drainage is not occurring, and the condensate simply accumulates in nearby matrix regions as predicted by the ECM. Another feature that can be investigated in the thermal test is the temperature distribution above and below the heater. Both the ECM and DKM predict that significant gas-phase convection will occur due to buoyancy effects and high fracture permeabilities. This causes asymmetrical temperature distributions with higher temperatures existing above the heater. Similar observations in the heater test would provide stronger evidence for the existence of significant natural convection and high fracture permeabilities.