Optimization of life-cycle preventative maintenance strategies using genetic algorithm and Bayesian Updating

Abstract

The authors have developed an optimization genetic algorithm (GA) methodology that enables the optimization of preventative maintenance (PM) strategies applied to reinforced concrete (RC) bridges. The PM strategies are used to delay/prevent the reinforcement corrosion of bridge beams due to contamination from chloride ions and maintain the reliability profile within acceptable limits and minimum whole life costing. A key element in predicting optimum PM strategies using the GA methodology is the accuracy of estimating the degree of deterioration of an element. The use of Bayesian Updating improves the reliability of this estimation by enabling the updating of the probability of failure based on data from inspection and the adjustment if necessary of the timing of subsequent PM interventions. The case studies presented demonstrate the application and the effectiveness of the proposed updated GA methodology and also examine the influence of applying updating at different time frames in reaching the optimum PM maintenance strategy.