2019 CSCE Annual Conference - Laval (Greater Montreal)

2019 CSCE Annual Conference - Laval (Greater Montreal) Conference

Trilevel Optimization Framework for Municipal Co-located Infrastructure: City of Montreal

Dr. Soliman Abu-Samra, KPMG Canada (Presenter)
Dr. Luis E Amador-Jimenez , Concordia University

Municipalities are experiencing high inefficiency and financial burden imposed by their under-performing infrastructure. One-third of Canada’s municipal infrastructure are in fair, poor and failing condition states, increasing the risk of service disruption and leaving the decision-makers with no choice but undertake immediate interventions. Furthermore, the massive number of infrastructure intervention activities occurring in cities leads to detrimental social, environmental, and economic impacts on the community. Thus, coordinating the interventions of the co-located assets (i.e. roads, water, and sewer) to reduce the duplicate activities, service disruptions, and corridor rehabilitation cost, is progressively becoming of paramount importance to cope with those tough challenges. Numerous attempts were made by previous scholars to enhance the infrastructure performance within the limited budget. Yet, most of their efforts were geared towards short-term intervention planning for a single asset, without accounting for the potential coordination benefits and savings. Thus, this paper develops a tri-level goal optimization framework to coordinate the intervention planning and efficiently allocate the funds among the co-located assets. The framework revolves through five models: (1) asset inventory, which comprises the asset characteristics of the corridor infrastructure; (2) deterioration and future condition models, which relies on probabilistic Weibull models to obtain the reliability of the co-located assets and account for the uncertainties across the assets’ life-cycle; (3) financial model, which computes the direct and indirect costs for interventions and service disruptions; (4) temporal model that computes the disruption time for different intervention scenarios; and (5) Tri-level optimization model that features an integrated non-pre-emptive goal optimization and genetic algorithm engine to maximize the financial, reliability, and temporal improvements, as opposed to the conventional infrastructure management approach. The tri-level hierarchy is similar to the management levels (i.e. strategic, tactical, operational) such that; the number of systems represents the strategic level; the systems represents the tactical level; and the interventions’ types represent the operational level. It revealed huge search space savings, estimated at 720*25, as opposed to conventional optimization techniques, assuming 20 corridors, with 2 intervention types, and 25 years planning horizon. The system was applied to a 9-km stretch on the city of Montreal network and the optimized intervention schedule exposed promising results as opposed to the heuristic-based with an overall improvement of 10% broken-down into 11%, 7%, 6%, for the reliability, life-cycle costs, and time. The developed framework facilitates the decision-making process for planning the corridor infrastructure interventions.