Analysis of Dry Storage Temperature Limits for Zircaloy-Clad Spent Nuclear Fuel
- PDF / 406,932 Bytes
- 6 Pages / 418.68 x 637.2 pts Page_size
- 77 Downloads / 252 Views
LLNL used the DCCG analysis by Raj and Ashby [7] to predict the time to failure of zircaloy under dry storage conditions based on a 'limited damage' approach. The equations presented by LLNL can be used to calculate initial temperature limits for SNF for specified temperature decay profiles and stresses. PNNL predicted temperature limits using a fracture map to account for various fracture mechanisms predicted to be active over a relevant range of stresses and temperatures. For dry storage temperatures, the fracture map indicates that either DCCG (-0-110 MPa) or power law creep (-110-160 MPa) controls failure. The most relevant stresses for dry storage are less than 110 MPa, so both models predict DCCG controls failure for the bulk of SNF in dry storage. Instead of DCCG, some have suggested (e.g., [5,8]) using a creep-strain limit approach where creep-strain is limited to 1%. This approach is based on the proposition that cavity growth and fracture only occur after significant plastic strain (much more than 1%). DCCG, however, is not strain dependent and it has been shown (e.g. [9, 10]) that cavities can nucleate and grow in various metals after very low plastic strains (much less than the strain to fracture). Inconsistency between LLNL and PNNL models The two models predict nearly consistent temperature limits for stresses most relevant to dry storage (from about 40-100 MPa). However, it turns out that the near coincidence of the model predictions results from a fortuitous combination of differing assumptions by PNNL and LLNL. First, PNNL (but not LLNL) used a 'recovery factor' in all failure equations. This factor is used to account for the recovery of some of the reduction in ductility of zirconium alloys from irradiation damage. Because the DCCG fracture model is not a function of strain, a reduced ductility would not directly affect the fracture time. It is not clear, therefore, whether a "recovery factor" should be used with DCCG. Additionally, LLNL chose a value of 0.15 (15%) "area fraction of decohesion" to correspond to failure based on estimated post-dry-storage SNF handling forces [11]. The PNNL model, however, incorporates the original form of the Raj and Ashby [7] failure equation which inherently assumes that failure occurs at an area fraction of decohesion of 0.50 (50%). Finally, PNNL and LLNL chose different values for the material constants used in the DCCG failure equation. These differences include atomic volume, Q2, grain boundary thickness, 8, cavity spacing, X,grain boundary diffusion coefficient, DoJgb and Qgb, as well as other unspecified constants. Temperature sensitivity of current models Both models are very sensitive to changes in the temperature decay of the SNF. A calculated initial temperature thought to cause failure in 40 years essentially predicts failure in 5 years for the same initial temperature when using Equation 1,which PNNL suggests is reasonable. The temperature profile used by PNNL when comparing to the LLNL model is T(K)= C x [t(months)]"°'28 2
(1)
where C can be adjust
Data Loading...
