Abstract
In this paper, the authors argue an optimal inspection/replacement policy for a tunnel lighting system which consists of lamps, ballasts and luminaires on an expressway. Deterioration processes of lamps and ballasts can be estimated by Weibull deterioration hazard models, and deterioration process of luminaires can be estimated by a Markov deterioration hazard model. By utilizing the deterioration forecasting result, the inspection/replacement process of the tunnel lighting system can be expressed as a Markov process model. As policy variables, the authors establish a) a flow of non-stationary inspection/replacement intervals of lamps, b) a repair policy of luminaires, c) an interval of simultaneous replacement of ballasts, and d) an interval of simultaneous replacement of the tunnel lighting system. At that time, the number of candidates of the optimal inspection/replacement policy is enormous due to the non-stationarity of inspection intervals. Therefore, the authors develop the method employing genetic algorithm in order to determine the optimal inspection/replacement policy which satisfies given risk control levels and minimizes the life cycle cost. Finally the effectiveness of the proposed method can be discussed through the empirical study of an actual expressway tunnel lighting system.
