A new framework for electronic assembly/system reliability prediction

Abstract

The inaccuracies of existing reliability prediction methods necessitate a fresh look at the methodology. At least 3 papers appeared in 1995 proposing new approaches for reliability prediction. A new framework for the prediction of part failure rates was published in the March IEEE Trans. Reliability 1995. Still lacking is an approach to bridge the gap between the part failure rates and the system probability of success. This paper presents a reliability prediction framework for assembly/system levels using predicted part failure rates as the takeoff point. Multipliers and environmental stress screening (ESS) times are used to modify part failure rates to give assembly/system failure rates. These multipliers reflect the influences of parts/assemblies interactions, learning curve effects, user malfunction tolerance, repair effectiveness, and management priority on maintenance. The ESS times manifest themselves as time shifts in the equations. Avionic system failure data show that avionic failure rates tend to be higher at the early part of a mission. A mission-time related multiplier is used to modify the standard survival function to give the final probability of system success. If data are available for quantifying the multipliers, the reliability practitioner should use them within this framework immediately. If data are not available one must begin to accumulate the information so that reliability predictions at the assembly/system level can be improved.