An explanation for the relatively stable efficiency under ABR and the decreasing efficiency under CBM when the
response time is increased is as follows. Under ABR, the chance that during a response time for the maintenance of a component a PM activity for one of the other components is scheduled is higher than under CBM. The failed component can then more often be repaired before the response time, at the scheduled PM of a component. The results for the average maintenance delay and the average group size confirm this. The average delay under ABR is smaller than under CBM. A larger response time increases the grouping of maintenance activities under ABR, while it does not increase the average group size under CBM. With a response time of 10 days, almost all maintenance events are grouped under ABR, while the average group size is 1.2 under CBM. Therefore, under CBM, it is most likely that a component has to wait the entire response time after which it is maintained individually. In a serial configuration, the waiting of a single component results in blocking and/or starvation of the other components,which explains the steeper decrease in efficiency compared to the parallel configuration.