2.2. Root cause analysis
This section introduces only the outline of the root causes in Fig. 2. For details of the analysis, refer to the following studies. Failure probability (Moon and Choi, 2021; Choi and Moon, 2022), Maintenance policy (Choi et al., 2020; 2021; Park et al., 2021), Maintenance technology (Choi and Moon, 2021), Schedule pressure (Oliva and Sterman, 2001). Failure probability is a root cause that affects many other factors. ROK Naval ships are maintained at regular intervals based on MTBF (Mean time between failure) according to the recommendations of the equipment manufacturers. Nevertheless, failures continue to occur. Recently, many efforts have been made to find the failure pattern over the total life using actual failure data (Zammori et al., 2020; Wang and Yin., 2019; Dikis and Lazakis, 2019). The failure pattern over the total lifetime is called the failure function. Ship failures are divided into two categories. One is that even if the performance of the equipment is partially degraded, it is possible to perform the mission. The other is a failure that immediately fails and requires repair by returning to the home port. In this study, the first is called normal failure and the second is called critical failure. The two failure functions were estimated by hierarchical Bayesian inference (Moon and Choi, 2021), combining trend and probability (Choi and Moon, 2022), respectively. The failure function is not in the form of a constant straight line. In some cases, failures during the total life cycle occur frequently and vice versa. Therefore, the MTBF-based policy of maintenance at regular intervals is inefficient. Choi et al. (2020; 2021) derived an appropriate maintenance interval based on the failure function. Park et al., 2021 analyzed maintenance intervals based on the naval fleet’s ship arrangement. A system dynamics model was implemented and analyzed for the phenomenon of queues occurring due to mission delay and maintenance work delay. The maintenance policy to which the failure function was applied was advantageous in terms of operability of ships and repair shops. Since maintenance is performed by humans, the efficiency may vary depending on the technology of the repair shop. Choi and Moon (2021) made Bayesian estimation of the change in equipment condition before and after maintenance. Results ROK Navy’s repair shop was able to complete an average of about 74% of repairs. In other words, if there are 100 faulty locations, it means that 74 locations are completely repaired after planned maintenance and 26 locations are insufficiently maintained. On the other hand, Oliva and Sterman (2001) argued that the speed of work varies by 75~125% depending on the workload. In this study, based on this, the variability of the maintenance amount performed by mechanics was given.