Why this matters
Urban road networks carry many origin–destination (O–D) trips at once, and even small routing changes can have a big impact on total system travel time. With connected and automated vehicles, a central controller could route traffic more intelligently and, with platooning, even increase link capacities. This project used an example network in “Rivendell, Middle-earth” to quantify how much system-wide benefit such technologies could provide.
What I did
I formulated the optimal routing problem as a nonlinear network flow model that minimizes total system travel time subject to flow conservation at each node and consistent link flows across all O–D pairs. To implement this in GAMS, I first built a node–link influence matrix to systematically encode which links start from and end at each node, which made writing the flow-balance equations straightforward. I then imported the network’s link attributes (free-flow travel time and capacity) and the O–D demand matrix from CSV files, defined decision variables for link flows by O–D pair, and implemented the Bureau of Public Roads–type travel time function used in the assignment.
Key results
I solved the model for two scenarios: one with conventional vehicles and one where all vehicles are automated and allowed to platoon, effectively tripling link capacities. In the no-platooning case, the total system travel time was 22364.723 hours; with platooning, it dropped to 20889.354 hours, a reduction of 1475 hours of travel time across the network. Plots of link flow versus capacity in each scenario show how higher capacities relieve congestion on previously overloaded links and improve overall network performance.