Reliability growth by failure mode removal

• Failure mode discovery by testing and removal is a dynamic and uncertain process.
• The evolution of a failure mode subject to discovery by testing is a stochastic process.
• Different test types discover different types of failure modes.
• More than one statistical model can well summarize failure mode activation times.

Modern systems, civilian (e.g. automotive), and military (manned and unmanned aircraft, surface vehicles, submerged vessels), suffer initial design faults or failure modes (FMs), including software bugs, which detrimentally affect the system׳s reliability and availability. FMs must be removed or mitigated in impact during initial testing, including accelerated testing, in order for the system to meet its reliability requirements and operate satisfactorily in the field. This paper concerns models for reliability growth in which the behaviors of FMs are assumed independent, but of different types. Test effort is guided by prior information, expressed probabilistically, on the random number and tenacities of such FMs that are of various origins in the designs. Estimation of the numbers of FMs that will ultimately activate while in the field is considered here.