Optimization of Pavement Rehabilitation Scheduling and Budget Under Uncertainty
Advisor: Professor Imad L. Al-Qadi
The significance of transportation infrastructure is increasing because of rising demand, limited resources, and potential benefits of emerging technologies. This reality and the complexities it introduces affect the future of transportation infrastructure and planning. A key aspect of building resilient transportation is the efficient and optimal use of resources. One vital resource of transportation infrastructure is the highway system, valued at more than $3 trillion in the United States. It plays a critical role in transporting a substantial amount of goods in the country (72% of the nation’s goods are carried by trucks). The highway wearing surface is a crucial component of this infrastructure, impacting user cost, safety, and greenhouse gases. Optimal maintenance and rehabilitation of pavement wearing surfaces would reduce agency and user costs.
Hence, the main objective of this dissertation was to provide a framework to assist agencies in highway planning while considering current and future challenges such as accommodating truck platoons. Hence, platoons’ lateral positions were optimized to control asphalt pavement damage. Metaheuristics were used to minimize pavement damage (i.e., rut depth, fatigue cracking, and international roughness index–IRI). Optimum platoon configurations could reduce rut depth and fatigue cracking by up to 18% and 50%, respectively. Compared to channelized truck traffic, they could reduce rut depth and fatigue cracking by up to 29% and 70%, respectively. A platoon factor could be incorporated in a mechanistic pavement design (e.g., Illinois full-depth pavement design method).
The life-cycle cost of a pavement network was minimized utilizing IRI progression. The life-cycle cost was linearized to arrive at an exact solution. Linearization was achieved by linearizing all nonlinear components. The linearization mainly focused on creating a generalized method to linearize IRI progression. Due to its impact on user cost, instead of simply linearizing it, IRI progression was piece-wise linearized. A novel application of circular shift in integer programming was applied to achieve that. To demonstrate the outputs of the model, a case study of a hypothetical city similar to Chicago was conducted. The study illustrated that increasing an agency cost by investing one dollar per lane-mile per year has a high return on investment until a certain threshold, beyond which allocating more budget does not lead to a reduction in life-cycle cost.
The uncertainty of the (near) optimal value resulting from the linear approximation was quantified. The uncertainty of the main and most volatile parameters (e.g., cost of maintenance and rehabilitation, fuel cost, agency budget, and discount rate) was considered. Monte Carlo simulation was used to sample the distributions of such uncertain parameters. It was found that the distribution of optimal agency could be visually characterized as lognormal distribution. Risk analysis of the probability of investment backlog was also conducted. The approximation of normality of both agency’s available resources (i.e., budget) and optimal spending could result in a probability that is different from what reality could be.