Statistical Assessment of Bias Significance in SNM Measurement Systems

Measurement bias is systematic error affecting the accuracy of measurement systems. Measurement bias is the difference between the measured value of a standard and the accepted value of the standard. All measurement systems have some inherent measurement bias. Determining how much bias is tolerable for Special Nuclear Material (SNM) Material Control and Accounting measurements is another matter. According to NRC regulations, SNM measurement bias must be estimated and ‘significant’ biases eliminated from inventory summaries. At Global Nuclear Fuel Americas (GNF-A), Wilmington, NC, a manufacturer of nuclear reactor fuel for LEU power plants, bias is calculated for all MC&A measurement systems during the annual SNM inventory. The systematic measurement error is calculated from repeated measurement data of standards traceable to nationally recognized standards. The systematic measurement error is compared to the accepted standard values to determine if a significant measurement bias exists for the measurement system. This paper discusses the statistical practices used at GNF-A for calculating and evaluating measurement biases in MC&A measurement systems.