This compares with the total stress release of 42% for seal 1 at T = -50°C (see Fig. 8). Considering the approximate nature of the analysis and the limited creep data, this estimate is in accord with experimentally measured stresses. It does confirm that the major discrepancy between calculated and observed stresses is due to stress release.The foregoing analysis and experimental data point out that the glue-spall stresses may be minimized by a judicious choice of the organic sealant, such that it has a low modulus and a low expansion coefficient and remains compliant at low temperatures. Further reductions in glue-spa11 stresses can be effected by using a thin layer with feathered edges. This observation was verified by the finite element analysis with stress reductions of up to 40% being possible for the specific examples considered here.
V. Summary and ConclusionsWe have met the three basic objectives of the paper, viz. we have presented a simple elastic analysis of the glue-spa11 stresses and verified its accuracy via the finite element analysis. It has been shown that both the shape factor and the physical properties must be known accurately to obtain a good estimate of the glue-spall stresses. That these stresses can lead to fracture has been demonstrated, both theoretically and experimentally, notably at low
Solid-statetemperatures when stress release is ineffective and the glue-spa11 stresses approach fracture values.We have also provided a crude estimate of the stress release which agrees with experimental observations and accounts for most of the discrepancy between calculated and measured compressive stresses in glass. Thus, if the stress relaxation curve of the sealant is known, the elastic analysis can be modified to yield an accurate picture of the state of stress. Finally, the key requirements for the organic sealant to minimize glue-spa11 failure have been noted, i.e. it should have a low elastic modulus and a low thermal expansion coefficient, it should remain compliant at low temperatures, and its thickness should be kept to a minimum.A solid silica-based matrix containing 30 cm3 of Kr(STP)/cm3 of glass was prepared by sintering 96% Si02 with 28% porosity under 140 MPa krypton pressure. The glass was heated to 850" or 900°C and held at temperature until the glass density was ;=2 g/cm3. At 420"C, only 0.7% of the krypton would be released after one half-life of "Kr (10.7 years). At T>600"C, release of krypton is accompanied by crack development, comminution, and glass softening. Advantages and disadvantages of this technique for radioactive gas storage and diffusion data are presented.
