Результати порівняльної оцінки алгоритмів розрахунку коефіцієнтів масштабування (scaling factors) для радіонуклідів, що важко вимірюються у відходах ЧАЕС, на прикладі 241Am

Abstract

The criteria for radwaste acceptance valid in the Exclusion Zone of SSE “Chornobyl NPP” require that each batch (package) of solid radioactive waste (SRW) sent for burial be certified. For the radionuclides that are difficult to measure (DTM) with using standard control systems directly in a package, the IAEA recommends using the Scaling Factors (SF). In the course of special studies to determine their quantitative values, some difficulties were encountered when analyzing laboratory data on the DTM nuclide in SRW, which largely consisted of so-called nondetectable (ND) results declared in the reports as <MDA (less than the minimum detectable activity). The work was aimed to evaluate the known algorithms for SF determination used in the world practice of radwaste management, in terms of acceptability of their application to laboratory data sampling with different proportions of ND results (<MDA) on the example of 241Am content in ChNPP historical waste. Three data sampling were formed with the share of ND results equaling to 18, 42 and 55 percent. This work addresses several methods for SF calculation, which are used in radwaste management systems of the countries with developed nuclear fuel cycle. Among the selected algorithms for data process testing, the most powerful one of methods for ND results censoring is included — the method of maximum likelihood estimation (MLE), which allows by restoring the normal law of distribution of random data to most accurately adjust the value of mean contaminant content according to the probability of appearance of each of measurement results, with taking into account the added ND results after their censoring by a numerical value multiple of MDA. The possibilities of selected algorithms, from the viewpoint of accuracy of statistical indicators’ reproduction in the initial arrays of experimental data sampling with SF application, were investigated on “problematic” data sampling related to 241Am content in ChNPP historical waste. The studied algorithms are ranked according to the quantitative scale of acceptability (reliability) assessment for each of them for the use in radwaste management system of ChNPP for certification of DTM activity contained in the SRW packages. The data obtained allowed drawing conclusions on the most acceptable algorithms that can be recommended for SF calculation, depending on the content of experimental data collected after laboratory control. The influence of data censoring on the accuracy of reproduction of the original spectrum of experimental data for different algorithms is estimated. The validity of use of Mean Activity Method recommended by the IAEA for data sampling, which contains a significant proportion of ND results, was confirmed.