Evaluating Methods and Algorithms for Multicriteria Bridge Management at the Network Level

In the current era of greater stakeholder participation, limited funding, and increased user awareness, bridge managers are using multiple-criteria techniques in their decision-making processes so that their decisions can be more accountable and transparent. Issues that arise in making such decisions include the multiplicity of the performance criteria and associated decision constraints and the large bridge inventories in most states. These issues translate into increased problem dimensionality, many variables, and excessive computational time. This paper presents alternative formulations and methods for decision problems, identifies alternative algorithms for implementing the solutions, and evaluates the performance of the algorithms. The algorithms include the incremental utility–cost ratio (IUC), Lagrangian relaxation, and pivot and complement methods. In investigating the suitability of these algorithms for agencywide bridge management, the study carried out computational experiments with field data. The criteria for evaluating the performance of the algorithms were computational speed, accuracy, simplicity, and robustness. The results were unequivocal: of the three heuristics, IUC consistently performed the best. IUC provided the simplest and quickest way to compute the optimal solution when small changes were made to the input parameters without the need to redo the entire optimization for each budget level.