*6.4. FMEA Results for Each System*

This study precisely identified the hybrid power system failure types and applied the reestablished RPN criteria to analyze the potential effects of failure. This study sought to derive consistent results between evaluators through newly applied evaluation criteria, obtaining results that could confirm safety and reliability when applied to the hybrid power system of a ship. Before applying Kendall's concordance coefficient, the hypothesis "The evaluation scores by item of the evaluators will be similar" was established according to the research objective. The significance probability between the existing and reestablished evaluation items was compared, confirming the validity of the research hypothesis. Kendall's concordance coefficient was applied using SPSS to confirm the concordance rate of the evaluation results between the existing and reestablished evaluation items as shown in Tables 10–12.

**Table 10.** Comparison of severity evaluation results between existing and reestablished evaluation items.


**Table 11.** Comparison of occurrence evaluation results between existing and reestablished evaluation items.


**Table 12.** Comparison of detection evaluation results between existing and reestablished evaluation items.


In this study, external evaluators assessed the same samples; based on the significance probability for the evaluation results of each item, the research hypothesis was supported. In addition, among the RPN items, the Kendall's concordance coefficient was 0.906 for S, 0.844 for O, and 0.861 for D. Compared to the existing evaluation items, the results for the reestablished evaluation items indicated that each evaluator applied essentially the same criteria when assessing the samples. The reliability of the evaluation results was therefore verified, and criteria for providing countermeasures for each failure mode were established based on the detected results.

To establish the criteria for countermeasures according to the RPN results of the fuel cell-based hybrid power system, it must be decided whether the absolute or relative RPN values will be used as the standard. To establish countermeasures via relative RPN values, the conditions of the targets for comparison must be similar (e.g., the number of items and the content of each item). However, as the internal device configurations and characteristics differ for each system, the number of evaluation items and the type and contents of each item also differ, making it difficult to apply relative criteria.

Therefore, this study defined the criteria of the reestablished evaluation items for countermeasures using absolute RPN values; specifically, the RPN evaluation class was defined as 1–10, and 1 ≤ RPN ≤ 1000. The following were set as the criteria for establishing countermeasures assuming a reliability of 90%: RPN of 100 or more, and either S, O, or D was 8 or more. Table 13 shows the number of items that should be set for each system according to the criteria.


**Table 13.** The ration of items required to establish countermeasures for a system based on these criteria.
