ABSTRACT
Traditional system reliability optimization problems, such as the redundancy allocation problem (RAP), are proven to be NP-hard (Chern 1992). Many solution methodologies have been proposed to solve them (Gen and Yun 2006, Kuo et al. 2001, Kuo and Wan 2007). Research findings on solving the traditional reliability optimization problems mostly pertain to series-parallel systems with either 1-out-of-n:G or k-out-of-n:G subsystem configuration where at least 1 or k components should be working for the sub-system to function, respectively. In addition, each parallel/standby sub-system has been limited to either hot or cold standby redundancy. A hot standby unit operates in synchrony with the online primary unit and provides immediate replacement once the online unit fails; while a cold-standby unit is initially unpowered and is switched into the power mode when it is needed to replace the failed online unit. Therefore, the hot standby redundancy is generally used for applications where the recovery time is critical; the cold standby redundancy is commonly used for applications where the energy consumption is critical.
