The experimental design of tests to measure the thermal stress resistance of spherical ceramic fuel elements for nuclear reactors.

Abstract

The statistical and thermodynamic problems introduced in quench testing the thermal stress resistance of spherical ceramic fuel elements for nuclear reactors are analysed. The statistical problem is one of deciding how many specimens to quench in order to achieve a required confidence interval in the working stress deduced from the test. Two methods are considered; quantal Response and the Up—and—Down Methods.. Estimates of confidence interval associated with a range of sample sizes are obtained by a computer simulation technique. It is shown that sample sizes of several hundred specimens would be required in order to obtain confidence intervals compatible with demands of reactor design. The thermodynamic problem is one of predicting the specimen temperature distribution from which the stress may be calculated. A solution for the case of the single specimen is already available. Multiple specimens, which offer the advantage of shortening the time scale of the experiment, introduce the problem of a time dependent quenchant temperature.; a solution, in terms of heat regenerator theory, is presented.