A Stress-Test of Alternative Formulations and Algorithmic Configurations for the Binary Combinatorial Optimization of Bridges Rehabilitation Selection

Abstract

Optimal surface transport asset management is a major concern with multiple economic and operational implications developed in various infrastructure areas. Although relevant ‘mature’ analytical frameworks have been proposed and developed, the problem setup and the algorithmic choices are still issues requiring thorough and detailed investigation. In this chapter, an optimal budget allocation framework is developed and stress-tested for the optimal scheduling of a bridges upgrading program. A suitable test case is developed for performing in-depth analysis that takes into consideration the most important features involved in such scheduling problems, while alternative formulations are also presented and discussed. The proposed frameworks are applied on a real large-scale dataset from the highway system of US, able to provide an adequate test-bed for investigating the optimal upgrade problem. The paper aims in the investigation of the effects that alterations of the problem setup, but also the effects that algorithmic configurations are introducing, when addressing real-world applications. The binary/selection problem is handled with a suitably coded Branch-and-Bound (BaB) algorithm, which is regarded as a robust and fast heuristic for such optimization problems. BaB is tested in alternative standard and extreme configurations, offering insights on its performance. Interestingly enough, although the continuous relaxation introduced by the BaB enables fast convergence, the NP-hard problem’s nature should be cautiously taken into consideration. The results are discussed in order to provide insights of applying the proposed framework in realistic infrastructure upgrading schemes.