Cloud-Based Scalable Software for Optimal Long-Range, Network-Level Bridge Improvement Programming

The current state of the practice in bridge management highlights a growing need to develop scalable optimization software tools to support the development of truly optimal bridge improvement programs and ensure that limited financial resources are optimally allocated. The heuristic project selection approaches employed in today’s bridge management systems are not capable of generating optimal programs. Agencies that rely on suboptimal programs may inadvertently direct a significant portion of their budget to the wrong projects, leading to an increase in maintenance backlogs and overall system risk levels. The subjective project selection criteria may also hinder the ability to quantify project benefits or justify projects to funding agencies and stakeholders. This paper presents a novel dynamic programming–based multiobjective optimization approach that is capable of generating global optimal network-level, long-range bridge improvement programs. The algorithm considers three objectives: the minimization of system-level risk, the maximization of system-level condition, and the minimization of life-cycle costs, subject to agency-defined constraints and planning scenarios. The algorithm efficiently explores the enormous search space to find optimal project lists for each year in the planning horizon under any given scenario. Alternative planning scenarios are defined to quantify the impact of different investment levels on system-level performance metrics and to determine the investment required to achieve the desired performance and risk targets.