A REANALYSIS OF RESULTS FROM CONSTANT EXTENSION RATE TESTS ON SENSITIZED ALSL-304 STAINLESS-STEEL

One set of constant extension rate tests (CERTs) intended to define the environmental effects of boiling water reactor (BWR) chemistry on stress corrosion cracking (SCC) of AISI 304 stainless steel is re-analyzed by statistical methods. It is shown that a time to failure criterion has doubtful validity, since a fundamental concept of multiple regression correlation is violated. This is partly because of the dominance of the stress variable on the results and partly because the results comprise two different populations. Two alternate models are examined. It is postulated that improved statistical significance can be achieved by truncating the CERT results at stresses of 480 MPa and analyzing the two data populations separately. To implement these suggestions, the experimental matrices should be balanced.