(b) The minimal cut sets of the system define directly the failure modes of the system. The system will fail in the above example if components 1 and 5 are failed. In addition the method quantifies the impact of each failure event on the system in terms of λ, r and U. Therefore, it is possible to determine from this analysis not only the system reliability indices but also the contributions made to the system indices by the various failure modes. This information is very important in any properly structured reliability assessment since it identifies critical areas of system weakness and suggests where reinforcement and investment should be made. None of this information is readily obtained from the network reduction method.
(c) From an assessment of this type, it may be decided that reinforcement and investments should be made to improve the system reliability. This investment could be in the quality of the components or redundancy of the system thus decreasing the failure rate or in the availability of spares and repair facilities thus decreasing the average repair time. There are many possible alternatives and each can be studied in turn to identify its effect on the system performance. The evaluation of λ, r and U in a failure modes and effect analysis however if of enormous benefit in deciding which components and area needs due consideration, and which of the alternatives may give the most desirable result.
11.6 Inclusion of scheduled maintenance
The previous section of this chapter have assumed that the system is operated with all components in service until a forced outage or failure occurs. In practice, system frequently change their state because various system components are removed at intervals in order to perform preventive maintenance.
This effect can be included in the assessment by dividing the total period of interest into intervals for which each interval represents one particular operating state of the system. The reliability indices of the system can then be evaluated for each time interval. Finally the system indices for the whole period of interest can be evaluated by weighting the indices for each interval by the probability of that interval and summating the weighted results together.
This method is both tedious and frequently unnecessarily complicated.
An alternative and practical method is to consider the maintenance outages in a similar manner to the forced outages but with a particular constraint attached, this being: a component will not be taken out for maintenance if this will cause failure of the system.