In the previous section it was assumed that the defect would instantaneously snap open to the full COD associated with its length at the moment the pressure boundary was breached. In reality this will probably not happen. Instead, the very large defects, which are those of interest, will probably grow to different through wall depths at different points around the length. Thus, much smaller surface defects would begin to breach the boundary at different points around the defect. The COD of these small defects would then remain elastic until the whole defect progressed to the surface. In this scenario, the leak from the defect would start very small and grow, slowly at first and then probably very quickly before snapping open to the fully plastic COD.

During this time of surface crack combination, the leak rate may exceed the value at which the operators shut the reactor down to a safe state in order to investigate the leak. Provided this occurs before the crack reaches a critical size, i. e., before the leak rate moves very quickly to the final leak state. Whilst the high leak rate may still occur, the plant would be in a safe condition. This can be seen as leak detection.

This probability of leak detection is almost certainly associated with the length of the defect that is itself related to the rate of leakage in the previous section. Thus, expert judgement was again used to introduce a factor, based on the leak rate, which would represent this probability of leak detection. Figure G.6 shows this plot as a function of leak rate.

From this plot it can be seen that the reduction factor for Category 1 (380 lpm [100 gpm]) is about five, rising to a factor of about fifty at Category 6 (1,900,000 lpm [500,000 gpm]).