Sampling strategy and statistical analysis for radioactive waste characterization

TL;DR

This paper presents a tailored sampling and statistical analysis methodology for efficiently characterizing radioactive waste drums' dihydrogen flow rate, ensuring relevant measurements with minimal samples and reliable predictions.

Contribution

Developed a sampling strategy and statistical framework for accurate characterization of radioactive waste drums with limited measurements, validated by factorial analysis and regression modeling.

Findings

Sample of 38 drums sufficed for reliable estimation.

Statistical analysis provided accurate mean and upper bound estimates.

Regression model effectively predicted dihydrogen flow rate.

Abstract

This paper describes the methodology we have developed to define a sampling strategy adapted to operational constraints in order to characterize the dihydrogen flow rate of 2714 nuclear waste drums produced by radiolysis reaction of organic mixed with \α-emitters. The objective was to perform few but relevant measurements. Thus, a sample of only 38 drums has been selected to be measured. Statistical analysis of drum measurement data of dihydrogen rate provided an estimation of the mean and the upper bound of the physical quantity of interest which gave a good convergence with global measurements from the ventilation system of the facility. Thereafter, performing a factorial data analysis has demonstrated the representativeness of the measurement data set and the sampling strategy assumption validity. Moreover, it provided information that has been used for a regression analysis to develop a linear prediction model of dihydrogen flow rate production for the waste drum characterization.