Électricité de France (EDF) has developed the Investment Portfolio Optimal Planning (IPOP) software tool [1] to be released with the Integrated Life Cycle Management (ILCM) software tool developed by the Electric Power Research Institute (EPRI) [2]. IPOP is an extremely powerful tool that uses genetic algorithms to provide an optimal strategy for investment in spare components and preventive replacements of multiple components at multiple power plant stations across an entire fleet. A drawback of IPOP is that it requires an extensive amount of user information to run even a single component. In response, Component Optimization Analysis Tools (COATs) was developed to simplify the process of deriving an optimal strategy for purchasing spares and replacements for a single component. This paper describes a two-layer algorithm used in the replacement strategy optimization in COATs. The inner layer consists of a Monte Carlo simulation that estimates the Expected Net Present Value (ENPV) of a given replacement strategy. A strategy consists of: the age of a component at which it needs to be replaced, the age of a component at which a spare should be purchased, years left in the plant at which to skip a scheduled replacement, and the end of life at which the scheduled replacement is skipped; and the years left in the plant at which no more spares are purchased. The Monte Carlo analysis uses these four strategy inputs with component costs, acquisition times, and reliability curves with plant downtime costs to calculate an ENPV for that strategy. The outer layer of the algorithm is an optimization layer that can use either Bayesian optimization or genetic algorithms to maximize the ENPV. These optimization algorithms are routinely available in various software packages and effectively treat the ENPV Monte Carlo as a black box function. An efficiency comparison is given between the two optimization algorithms to demonstrate under which conditions each algorithm out performs the other.

This content is only available via PDF.
You do not currently have access to this content.