An approach for determining reasonable sample sizes of MARSSIM final status surveys in decommissioning of nuclear facilities

Highlights: • This paper presents a practical approach to determining reasonable sample sizes of MARSSIM final status surveys for decommissioning of nuclear facilities. • The MARSSIM, which is a guide for final status surveys of a decommissioning site, recommends increasing the sample sizes 20% larger than necessary for satisfying statistical tests to provide a designed power of tests. • This study investigated the appropriateness of such MARSSIM recommendation by checking to what degree the sample sizes of Sign test and WRS test can be reduced still allowing a designed power of the tests for multiple cases of survey unit conditions. • The results show that the sample sizes smaller than recommended by the MARSSIM still are enough to provide deisgned power, suggesting the possibility of over-conservatism of the MARSSIM's recommendation. • Therefore, it is recommended to explore such possibility in designing the final status surveys by performing a detailed investigation of the prospective power curves of tests to ensure a reasonable conservatism in determination of the sample sizes. - Abstract: In many countries which adopted a radiological criterion for decommissioning or site release, a Multi-Agencies Radiation Survey and Site Investigation Manual methodology has been extensively used as a guide for designing decommissioning surveys, especially a final status survey, in order to demonstrate a decommissioned site's compliance with the radiological criteria. The methodology recommends the use of nonparametric hypothesis testing in a final status survey as a statistical tool in making a decision on whether or not the mean contamination level in a survey unit meets a concentration limit which is derived from the radiological criterion. When designing a final status survey, it is of a great importance to ensure that the hypothesis test of the survey has a sufficient power to release a survey unit in question when it is clean by determining the sample size necessary for achieving the power. For this purpose, the MARSSIM methodology recommends using sample sizes which are 20% greater than statistically necessary for the test. However, this study shows that, the MARSSIM-recommended sample size generally provides a conservatism much larger than necessary. Therefore, it is recommended to determine a reasonable sample size of the final status survey for a survey unit not by just following the recommendation of the MARSSIM but by performing a detailed investigation of the power of the test. This kind of effort is valuable as it could result in saving lots of costs when the determined reasonable sample size is smaller than recommended by the MARSSIM methodology.