Mathematics (Dec 2022)
Availability Optimization Decision Support Design System for Different Repairable <i>n</i>-Stage Mixed Systems
Abstract
This study attempts to propose an availability optimization decision support design system for repairable n-stage mixed systems, in which different combinations of subsystems, such as parallel, standby, and k-out-of-q, are connected in series configuration. The enumeration method, tabu search, simulated annealing, non-equilibrium simulated annealing, and the modified redundancy allocation heuristic combined with a modified genetic algorithm will be proposed to solve the system availability optimization problem and further determine the appropriate system configuration design. Several simulated cases are conducted by following the procedural flow of the proposed availability optimization decision support design system to reach the optimal allocations of the component redundancy amount, the optimal repair rates, and the optimal failure rates of all subsystems to minimize the total system cost under several configuration constraints for different repairable n-stage mixed systems. Simulated results display that the proposed availability optimization decision support design system can definitely take advantage of different component redundancy system designs, including the parallel-series system, n-stage standby system, n-stage k-out-of-q system, and n-stage mixed system, to save a lot of cost and meet the high level of the system availability requirement compared to the n-stage single component series system. Additionally, the results for all proposed combined methods also show that the parallel-series system can obviously reach the same level of system availability requirement with less system total cost, in contrast to the n-stage standby system, by presuming the identical deteriorating probability for both the operating components and the standby components. The performance comparisons of five proposed combined methods for four proposed system configurations are analyzed comprehensively. It can be concluded that the performances of the modified redundancy allocation heuristic method, combined with a modified genetic algorithm on the criteria of the optimal system costs for four proposed system configurations, are not only superior to the other four combined methods, but also to the performances on the criteria of CPU running time for four proposed system configurations.
Keywords
