Predictions of creep closure for periods ranging up to more than 1,000 years are needed for designing a radioactive waste repository in a salt formation. Such long-term predictions must be based on the use of laboratory and field test results extrapolated by numerical modeling. Laboratory test results are presented which show highly variable creep behavior for salts from different sites as well as salts from the same site. Comparisons of measured creep closures against predicted closures are presented for two deep boreholes in Louisiana salt domes and for an exploratory shaft and drift in southeastern New Mexico bedded salt (WIPP). These data illustrate the significant uncertainties involved in long-term closure predictions and indicate the critical need for site-specific, in situ closure measurements over extended periods of time for repository design.
The design of a radioactive waste repository in salt must consider the effects of creep closure on waste emplacement operations and on postclosure behavior. Prediction of creep closure during the repository operations period is needed to plan for overexcavation or reexcavation to maintain openings at the minimum required dimensions for waste emplacement activities and to provide for possible retrieval. The time period of interest for repository operations ranges from a few years to as much as 80 years for certain parts of the repository. Prediction of creep closure rates for the period after sealing of the repository is needed to assess the degree to which the salt mass will converge on seal components such as bulkheads and backfill, thus enhancing the effectiveness of these man-made barriers against waste migration. Long-term closure rates are particularly important to the use of crushed salt for backfill material because of the ultimate consolidation of the crushed salt backfill into an essentially impermeable salt monolith (Kelsall et al., 1984). The stresses placed on waste packages due to creep closure are also important to waste package design. The period of interest for these aspects of postclosure behavior is hundreds of year s. These long time periods of interest, together with the complexity and variability of creep behavior, make it impossible to perform full scale tests that model the full time period and simulate the full range of anticipated repository conditions. Current creep closure predictions are thus based on analytical models and laboratory-derived materials properties. It is the purpose of this paper to suggest that high confidence in such predictions is obtained only when 1) laboratory test durations are sufficiently long to characterize steady-state creep, 2) sufficient laboratory tests are run to characterize the in situ variability of the site, and 3) field tests are conducted to validate the models.
LABORATORY CREEP MEASUREMENTS
Many different analytical expressions have been used by various researchers to describe the creep behavior of salt, frequently indicating very different types and magnitudes of creep strain. It appears that a creep equation of the following form is gaining acceptance as a reasonably accurate expression to describe primary and secondary creep (Hansen and Carter, 1982): (mathematical equation) (available in full paper)