In a reprocessing facility where nuclear fuel solutions are processed, one could observe a series of power peaks, with the highest peak right after a criticality accident. The criticality alarm system (CAS) is designed to detect the first power peak and warn workers near the reacting material by sounding alarms immediately. Consequently, exposure of the workers would be minimized by an immediate and effective evacuation. Therefore, in the design and installation of a CAS, it is necessary to estimate the magnitude of the first power peak and to set up the threshold point where the CAS initiates the alarm. Furthermore, it is necessary to estimate the level of potential exposure of workers in the case of accidents so as to decide the appropriateness of installing a CAS for a given compartment.

A simplified evaluation model to estimate the minimum scale of the first power peak during a criticality accident is derived by theoretical considerations only for use in the design of a CAS to set up the threshold point triggering the alarm signal. Another simplified evaluation model is derived in the same way to estimate the maximum scale of the first power peak for use in judging the appropriateness for installing a CAS. Both models are shown to have adequate margin in predicting the minimum and maximum scale of criticality accidents by comparing their results with French CRiticality occurring ACcidentally (CRAC) experimental data.