Non-linear regression analysis of reactor containment leakage rate

Abstract

A nuclear reactor’s containment boundary provides a pressure-retaining barrier that limits the escape of any radioactive matter released from reactor processes. The requirements for leakage rate testing of a reactor’s containment boundary are specified in standards such as CSA N287.7 [1] and ANS 56.8 [2]. The recommended data analysis approach in these standards is the Absolute Method using the Mass Plot Analysis Technique, and both Annex C of CSA N287.7 and Section 5.4 of ANS 58.2 present analytical approaches that use the linear least squares fit (or linear regression) method for quantifying leakage rate. Section C.3.3 of CSA N287.7 also permits the usage of non-linear regression, but does not describe how. This paper presents a non-linear regression analysis method for containment leakage rate data analysis which can be readily integrated within the framework of an established standard such as CSA N287.7. The non-linear regression approach is shown to give similar results to the traditional linear regression approach when the total leakage is a small proportion of the initial mass. When the leakage is greater, the method affords significant improvements in accuracy compared to the traditional method.