Redundant systems must include some means by which the system detects, isolates and reconfigures in a manner that maintains system performance in the event of failures among its redundant components; this process is called "redundancy management" (RM). The probability that the RM process is accomplished without error is called "fault coverage" or "coverage." Coverage for complex redundant systems is seldom, and perhaps never, perfect. A system with less than perfect coverage is said to be subject to "imperfect fault coverage" (IFC). Since coverage levels even slightly less than unity will have a significant impact on system reliability, it is critical to correctly model the effects of IFC on redundant system reliability. The correct reliability modeling for redundant systems depends on the nature of the system’s RM process: if each redundant component has an associated coverage value for each of its redundant components, the appropriate model is Element Level Coverage (ELC); if, however, system RM uses a voting process, such as mid-value-select, then the appropriate model is Fault Level Coverage (FLC). In this presentation, mathematical models for determining IFC reliability for both ELC and FLC systems will be discussed; firstly for k-out-of-n:G systems, and, secondly, for a complex multi-channel digital fly-by-wire system. These examples will clearly illustrate the significant effects of imperfect fault coverage on redundant system reliability.

Key Words: Imperfect Fault Coverage, k-out-of-n:G System Reliability, Reliability of Systems Subject to Imperfect Fault Coverage

Albert Myers
Myers Consulting
La Habra Heights, California