Engineering Services and Analysis – Reliability Analysis
Reliability analysis is the use of statistical models that provide information regarding how often a product experiences a failure. The model takes into account the expected operating environments, the area where the product will be stored, the quality of parts used in manufacturing, and the length of time the product will be operated.
After these basic questions are answered, the product can be analyzed. Typically, a reliability prediction is performed using a top-down structure, where all the individual components are analyzed, and their failure information is rolled up to the subassembly and assembly level. This structure allows the design team to substitute components to see how they impact the reliability of the product. It also allows them to determine what systems or assemblies may be high failure items that require a backup mode or another design change to increase reliability.
The component failure rate can be determined using a variety of means. Electronic parts and assemblies are usually analyzed using the methods contained in MIL-HDBK-217. Mechanical parts can be analyzed using the similar part method based on the practices contained in Non-Electronic Parts Reliability Data (NPRD), version 1995. They can also be analyzed using historical data based on established return and service information for similar items manufactured in the past.
Reliability predictions typically use a base time of one million hours. Other time bases can be used if desired. Once established, the failure rates are converted into a time basis (Mean Time Between Failures or MTBF) by dividing the time basis by the failure rate.
Reliability prediction reports typically include a front matter that provides the end user an idea of what assumptions and practices were established for use in the analysis.