Michigan redesigned its campus bus network as hubs and spokes to run buses half full

When COVID hit, the University of Michigan's campus bus system faced a contradiction: state distancing rules capped buses at roughly half of seats, but the campus still had to move the same students with the same fleet. The obvious response was to cut frequency and let waits explode.

A team from UMich's industrial and operations engineering department instead re-architected the network. They built mixed-integer programming models that turned the route-and-schedule redesign into a hub-and-spoke problem, and used real data on student activities to decide where hubs should sit and which stops the new routes would serve.

The design rules encoded the safety constraints directly: every new route had to be under 15 minutes end to end, buses would run at no more than 50% seat capacity, and the total number of buses stayed the same as before the pandemic. A Monte Carlo simulation sampled peak-demand variation, distancing-requirement changes and bus breakdowns to stress-test the plan before it went live.

The university implemented the new hub-and-spoke routes and schedules for the 2020-21 academic year, running half-empty buses on shorter routes while keeping the fleet constant — trading a redesigned network for what would otherwise have been a service collapse.

Redesign the network topology instead of cutting service

Hub-and-spoke concentrates flows where capacity exists: under a hard 50%-seat cap, routing riders through hubs lets each bus fill its capped load instead of running half-full on direct routes.

Real student-activity data grounded the model: hub placement came from where students actually move, not from administrative guesses about campus demand.

Simulation before deployment: sampled scenarios over demand peaks, distancing rules and breakdowns made the plan resilient before a single route changed, which is what let the university commit to it.

The fleet stayed fixed: by keeping bus counts unchanged, the design cost reallocation effort only — critical when procurement and staffing were frozen mid-pandemic.

When a constraint makes the old network infeasible, the answer can be a new topology rather than less service: optimize which hubs concentrate flows, and frequency survives where routes cannot.

The new routes operated through the 2020-21 academic year at the University of Michigan. The team, led by professor Siqian Shen with Gongyu Chen, Xinyu Fei, Huiwen Jia and Xian Yu, published the models and simulation results in the INFORMS Journal on Applied Analytics' special issue on analytics remedies to COVID-19 (November 2022), with support from the National Science Foundation, and framed the hub-and-spoke approach as a template for any transit system forced to operate under distancing constraints.