Probability of Process Laboratory Errors Affecting Reprocessing Operations

摘要

Probabilistic methods are used to assess the risks of postulated accidents at the Savannah River Plant (SRP), operated for the United States Department of Energy by E.I. du Pont de Nemours and Company. The SRP includes two reprocessing areas, one for depleted uranium targets and one for enriched uranium fuels. In the preparation of the Safety Analysis Report for these reprocessing areas, rare-event frequencies are calculated by fault-tree analysis (FTA). In many SRP reprocessing operations, nuclear safety is provided by dilution of fissile materials to safe concentrations rather than by vessel dimensions and geometry. Thus the process laboratory at SRP has an especially important role in both process control and safety. Data on the reliability of results reported from process sample analyses are frequently needed in FTA. Further, these data must reflect both the accuracy of the laboratory analyses and their credibility with operating personnel, an important recovery factor in human reliability analysis. To meet this need, SRP experience for the eight-year period 1976 to 1983 was analyzed. The probability that an incorrect sample result was reported by the laboratory and that it was used as a basis for process operation was 2.6 per 100,000 analyses. The best fit to the time-between-errors data is a normal distribution. A least-squares straight-line plot of errors per year has a slightly negative slope. The consequences of the laboratory errors were actually small or nonexistent due to other safeguards. Most of them caused only short operating delays, and no hazardous materials were released from the plant as a result of the reported errors. (ERA citation 11:019147)