Life-Cycle Cost Design Using Improved Multi-Objective Genetic Algorithm

Abstract

In order to establish a rational bridge management program, it is necessary to develop a cost-effective decision-support system for the maintenance of bridges. In this paper, an attempt is made to develop a bridge management system that can provide practical maintenance plans by using an improved multi-objective genetic algorithm. A group of bridges is analyzed to demonstrate the applicability and efficiency of the proposed method. INTRODUCTIONIn Japan, maintenance work on bridge structures is increasing in volume and importance. The number of structures requiring repair or replacement will drastically increase in the coming ten years. In order to establish a rational maintenance program, the concept of Life-Cycle Cost (LCC) has gained great attention, which minimizes the total cost of whole lives of structures [1]. So far, the authors have developed LCC based bridge maintenance systems for existing concrete bridge structures [2, 3]. Concrete bridges are deteriorating due to the corrosion of reinforcing bars and the neutralization of concrete. Then, it is necessary to achieve an optimal maintenance plan that can provide appropriate methods and times of repair or replacement. However, the optimal maintenance problem is very difficult to solve, because it is one of combinatorial problems with discrete design variables and discontinuous objective functions. Furthermore, the tougher problem becomes , the larger and more complex it is. Although low-cost maintenance plans are desirable for bridge owners it is necessary to consider lifetime constraints when choosing an appropriate maintenance program. For example, the minimization of maintenance cost needs the target safety level and the expected service lifetime to be both prescribed. The predetermination of requirements may affect the possible maintenance plans. Namely, it may be possible to find out a better solution that can largely extend the service life if the safety level can be slightly decreased using the same amount of maintenance cost [4, 5]. In this paper, an attempt is made to improve the bridge management system proposed in [6], by developing an improved multi-objective genetic algorithm. Many alternative 17 ANALYSIS AND COMPUTATION SPECIALTY CONFERENCE th Copyright ASCE 2006 17th Analysis and Computation Conference maintenance plans with different characteristics can be obtained by introducing the concept of multi-objective optimization. Single-objective optimization can provide various solutions by changing the constraints. However, this approach requires enormous computation time. When selecting a practical maintenance plan, it is essential to compare feasible solutions obtained under various conditions. This process is effective for the accountability through the disclosure of information. A new multi-objective genetic algorithm is developed for the bridge management problems that have a lot of constraints. Several numerical examples are presented to demonstrate the applicability and efficiency of the proposed method.