The following is an excerpt from The Reliability Engineering Handbook by Bryan Dodson and Dennis Nolan, © QA Publishing, LLC.
The elements of a reliability program must be selected to meet reliability needs. Identifying and quantifying these needs should be accomplished prior to release of a proposal for the appropriate acquisition phase so that tasks and requirements commensurate with the needs may be included. The tasks and requirements, which are included, establish the framework for the continuing reliability dialogue between procurement and those who will ultimately be selected to develop the hardware. It is essential to make appropriate analyses and exercise mature judgment in determining reliability needs.
In making this determination, it is necessary to assemble program data concerning mission and performance requirements (preferably at the sub-system level), anticipated environments, and mission reliability and basic reliability requirements. This information is initially gathered in the conceptual phase and refined throughout development. It is the base upon which the reliability needs are determined and adjusted. The initial life profile defines, as a minimum, the boundaries of the performance envelope and provides the timeline, both environmental conditions and induced stresses versus time, typical of operations within that envelope. The quantitative requirements, basic reliability and mission reliability, are then determined for the defined life profile.
Using the information on equipment contemplated to provide the required performance, a separate apportionment or allocation of basic reliability and mission reliability can be made to the equipment level. This apportionment is usually based on available reliability data modified to reflect changes in performance requirements, duty cycles, and anticipated environments. If the hardware to be procured is either a sub-system or equipment, the allocations would apply down to the lowest assembly level in terms of Mean-Time-Between-Maintenance-Action (MTBMA), or Mean-Time-Between-Failure (MTBF), or failure rate. The required modifications are largely a matter of judgment, particularly when a new or considerably modified equipment concept must be synthesized to provide a specified function.
A reliability estimate should be made for each item of equipment independent of, and reasonably soon after, completing the initial apportionment. The equipment estimates should be combined to provide an initial estimate of basic reliability and mission reliability. During the conceptual phases and validation phases, design details will probably not be available. Therefore, estimates made during these phases and early in FSED will provide "ball park" numbers, which are nevertheless adequate for initial comparisons with, and for establishing the reasonableness of, the initial apportionment. Reapportionment based on a comparison with details of the estimate may be advisable at this time. The apportionment and the estimate procedures should be repeated until reasonable apportioned values are obtained. The apportionment should be frozen prior to awarding subcontracts that have firm reliability requirements.
Some reliability tasks should be accomplished for an entire system, e.g., development and use of a failure reporting system, periodic estimates of basic reliability and mission reliability. In most cases, needs are self-evident while others, which must be selected, may apply only to sub-systems. While experience plays a key role in the task selection, it should be supplemented by analysis and investigation.
A useful initial analytical procedure is to compare reliability estimates at the sub-system and equipment level, with the corresponding apportionments. If the estimate is less than the apportionment, the need for improvement is indicated. Where "considerable" improvement is required, the subsystem or equipment should be identified as "reliability critical." This identification should be done as early as possible in the program so as to impact the equipment through the proper selection of tasks.
Reasons for the disparity between the apportioned and the estimated values of the reliability critical items should be investigated. Discussions of these reasons and tentative ways to attain the apportioned values, (e.g., relaxed performance requirements, either more or less design redundancy, additional environmental protection), should be held with appropriate project personnel. The object of the investigations and discussions is viable recommendations for action to overcome the deficiencies. A significant benefit that can be gained from this process is a consensus on the specific equipment that is considered reliability critical. When systems or equipment performance requirements create a wide and irreconcilable disparity between apportioned and estimated values of what is required, reliability should be challenged. Elimination of less essential equipment functions can reduce equipment complexity and significantly enhance reliability.
Tasks and requirements can be prioritized and a "rough order of magnitude" estimate made of the time and effort required to complete each task once recommendations for task applications have been determined. This information will be of considerable value in selecting the tasks that can be accomplished within schedule and funding constraints.