Trust versus confidence: Microprocessors and personnel monitoring

Due to recent technological advances, substantial improvements have been made in personnel contamination monitoring. In all likelihood, these advances will close out the days of manually frisking personnel for radioactive contamination. Unfortunately, as microprocessor-based monitors become more widely used, not only at commercial power reactors but also at government facilities, questions concerning their trustworthiness arise. Algorithms make decisions that were previously made by technicians. Trust is placed not in technicians but in machines. In doing this it is assumed that the machine never misses. Inevitably, this trust drops, due largely to ''false alarms''. This is especially true when monitoring for alpha contamination. What is a ''false alarm''? Do these machines and their algorithms that we put our trust in make mistakes? An analysis was performed on half-body and hand-and-foot monitors at Oak Ridge National Laboratory (ORNL) in order to justify the suggested confidence level used for alarm point determination. Sources used in this analysis had activities approximating ORNL's contamination limits