Year
2006
Abstract
The objective of this research is to perform a statistical analysis on the fissile mass flow monitor (FMFM) associated with instrumentation that monitors the downblending of HEU to LEU, including equipment such as that used under the 1993 HEU Purchase agreement made with Russia. The purpose of the analysis is to identify expected statistical behavior of the measurements obtained and, furthermore, to identify a threshold for delineating usable from unusable data. The confidence measurement used in verifying the existence of flow, providing the probability of flow, is referred to as “quality.” From this research, we conclude that any quality measurement having a value greater than 20% provides high confidence of the presence of flow of fissile material in the pipe. Furthermore, any quality value less than this threshold may be regarded as statistical noise mistakenly identified as a measurement. The confidence measurement quality is mathematically defined as a stretched, inverted Snedecor Fisher F-test1 which tests the inequality of residuals from actual BDMS data to residuals of zero flow data. A calibration method is employed in determining the threshold value for usability in which a simulation is set to zero flow and a histogram of the quality values obtained produces a distribution surrounding zero (i.e. nonzero). At this point, a range of probability values exists between quality values of zero and approximately 30%. A significance level can be concluded by counting those below some threshold probability (0.5% in this case) out of the number of tests taken. If N tests are taken, and n tests out of these have less than 0.5% probability of having a value greater than 20%, then our significance level is n/N%. From this assessment, this research has concluded at a 95% significance level that less than 0.5% of 600,000 tests have a quality value greater than 20% at the zero flow assumption. The outcome of this work will greatly help the quality of the measurements obtained from the BDMS.