Abstract
This paper models the reliability of a performance-sharing k-out-of-n: G system with heterogeneous degrading components and a performance-redistributing common bus. Each component may behave at various performance levels to meet its random demand. If one component exhibits performance beyond its demand, the redundant performance is redistributed to components with deficit performance via the common bus with limited capacity. The system fails if the number of operating components is less than k after sharing the redundant performance. A new analytical method based on multi-valued decision diagrams (MDDs) is put forward, which comprises an efficient model generation algorithm leveraging top-down simplification rules and a new ordering heuristic for improving MDD generation efficiency. The MDD evaluation engages the continuous-time Markov Chains to compute the steady-state probabilities of system components considering the degradation effects. Case studies of a wind power generation system and a data processing system as well as benchmark studies are conducted to illustrate the applicability and efficiency of the proposed method. A comparative study with the universal generating function-based method is also provided to further demonstrate the efficiency of the proposed MDD-based method.