The operating temperatures typical of water reactors are not high in terms of homologous temperatures (<0.4 Tm) so that thermal creep is relatively slow. Furthermore, the creep strength is considerably en­hanced by the presence of dissolved oxygen and hy­drogen present as hydride. In practice, however, the in-reactor deformation seems to be primarily radia­tion, induced. In-pile creep tests show that the creep rate (e) is almost athermal and approximately linearly dependent on the fast neutron flux (Ф) and on the — stress (or) (Wnod and Watkins, 1971 [18]). The me­chanisms werg assessed by Hesketh (1968) [19] who attributed the enhanced in-pile creep to yielding creep [4]. Diameter measurements on PWR fuel rods by Franklin (1982) [20] produced an equation of the form t = к Ф0-6 a1 5.

In ‘fair agreement’ with the results of in-pile creep experiments (k is a constant). In this study, ovalisa­tion of the clad and ridge formation due to the clad taking on the (hour-glassed) shape of the fuel pellets occurred during the second reactor cycle; subsequent (diametral) clad shape changes were controlled by the pellet stack.

In the early days of PWR, problems were encoun­tered with collapse of the clad into large interpellet gaps. These had formed by accumulation of smaller gaps caused by fuel densification (Roberts et al, 1977 [21]). Whilst few failures were produced, such an effect is obviously undesirable and steps were taken to prevent its re-occurrence. These were the use of a higher starting density for the fuel and internal pre — pressurisation of the pin (to about 10 atmospheres, which roughly doubles at operating temperature) to

reduce the driving force for collapse (Frost, 1982 [22]). Experience has shown that these can prevent clad col­lapse into large inter-pellet gaps and, as shown by Franklin’s results, clad/fuel contact is delayed until some way into the second cycle. At the ends of the fuel rod, of course, the plenums always consist of lengths of unsupported tube. In this case clad col­lapse does not occur because both the neutron flux

and the temperature are low, hence there is little scope for either irradiation or thermal creep.

Gro w(h

Just as with uranium (Section 10.1.2 of this chapter), fast neutron irradiation of zirconium alloy cladding results in growth because the vacancies and intersti-

. js which are produced haw different preferences for^sink positions. If (be metal trains were perfectly randomly oriented, this would haw little effect other [han the” generation of high internal stresses. In prac — of course, any metallurgical treatment will result I j” s’ome kind of texture and, although this can be — patrolled to some extent by the fabrication proce­dure in Zircalos tubes г he ‘exture is usuallv sucii dial fast neutron irradiation tends ю make the tubes I oncer and imperceptibly thinner. Other factors which uTect erowth are prior cold work and temperature (Adamson, Tucker and Fileris, 1982 [23]) and, since the process produces rod length changes it is appa­rent that the presence of cross-rod flux and tempera­ture sradients will also be capable of producing dif­ferential length changes which will lead to rod bowing.

Other, less important, contributions to bowing are possible from non-uniform, radiation-induced stress re­laxation (Montgomery, Mayer and French, 1977 [24]). To date rod bowing seems to have caused few failures and this may be due to the very firm built-in support offered by most spacer designs and the fairly short unsupported rod lengths (typically 0.5 m).