When you manage a campus that spans 657 hectares and serves over 180,000 students, transportation is not a convenience problem. It is an infrastructure problem. At King Abdulaziz University in Jeddah, we have been grappling with this reality for years, operating a fleet of 33 buses across 9 routes and 38 stops. Recently, our team undertook a comprehensive, data-driven review of the entire system, and what we found challenged many of the assumptions that had guided campus transit planning for the past decade.

The Scale of the Problem

KAU generates an estimated 350,000 daily trips across its campus. That figure alone should give any transportation planner pause. The campus road network, mapped in detail using OpenStreetMap data covering 499 buildings, reveals a complex web of corridors, pedestrian zones, and vehicle routes that were never designed with current enrollment figures in mind.

The existing bus network evolved organically. Routes were added as new buildings went up, schedules were adjusted semester by semester, and stop locations were chosen based on administrative convenience rather than ridership data. The result was a system that moved buses efficiently along fixed paths but did not necessarily move students efficiently to where they needed to go.

Our first step was to build a comprehensive spatial model of the campus. We geocoded every building, mapped the complete road network, and overlaid it with class schedule data to estimate demand patterns at 15-minute intervals throughout the day. The demand profile was striking: sharp peaks at the top of each hour, near-zero demand during midday prayer times, and a persistent asymmetry between morning inbound and evening outbound flows.

Methodology: From Intuition to Optimization

Traditional bus route planning at universities tends to rely on two inputs: a campus map and the judgment of the facilities manager. Our approach replaced this with a multi-objective optimization framework that balances three competing goals: minimizing average student wait time, maximizing route coverage across campus zones, and keeping fleet utilization within the operational budget.

We formulated the problem as a variant of the Vehicle Routing Problem with Time Windows (VRPTW), adapted for the cyclical nature of academic schedules. The key innovation was incorporating real demand data rather than assuming uniform distribution. Students in the engineering and medical faculties, for example, have fundamentally different travel patterns than those in humanities, both in timing and in origin-destination pairs.

The optimization was solved using a hybrid approach combining genetic algorithms for route structure with linear programming for schedule assignment. We ran the solver against the current network as a baseline and then allowed it to propose modifications, from minor stop relocations to complete route redesigns.

Key Findings

The results were illuminating. The optimizer identified that two of the nine existing routes had over 60% overlap in coverage area, effectively duplicating service while leaving the southern campus zone underserved. Consolidating these routes and redirecting capacity freed up 4 buses during peak hours, enough to introduce a new express service connecting the main gate to the medical campus, a route that student surveys had flagged as the most painful gap in the system.

Stop placement analysis revealed that 8 of the 38 stops were located more than 200 meters from the nearest high-traffic building entrance, a distance that discourages ridership in Jeddah's climate. Relocating these stops to positions validated by pedestrian flow data could increase catchment coverage by an estimated 23%.

Perhaps the most important finding was temporal. The fixed-schedule model, where buses run the same routes all day, wastes roughly 35% of fleet capacity during off-peak periods. A demand-responsive adjustment, where two routes switch to on-demand service between 11 AM and 1 PM, could maintain service quality while allowing preventive maintenance windows that currently do not exist.

Implications for Saudi Universities

This work is not unique to KAU. Saudi Arabia's public universities are among the largest in the world by enrollment, and many operate sprawling campuses that were master-planned decades ago. As these institutions pursue sustainability targets aligned with Vision 2030, campus transportation is a natural lever. A well-optimized bus network reduces private vehicle trips, lowers emissions, improves student satisfaction, and frees up parking infrastructure for repurposing.

The methodology we developed is transferable. Any university with basic GIS data, class schedules, and a willingness to instrument its bus fleet with GPS trackers can replicate this analysis. The computational tools are open-source, and the optimization framework scales gracefully from a 10-bus system to a 100-bus fleet.

Looking Ahead

We are now in the process of validating the optimized routes through a pilot deployment planned for the next academic semester. The pilot will instrument 10 buses with passenger counters and GPS loggers to compare predicted versus actual ridership on the redesigned routes. If the results hold, KAU will have a blueprint for transit optimization that other Saudi universities can adopt, and a dataset that could support the eventual transition to electric bus fleets, a topic I plan to address in a future post.

The broader lesson is simple but often overlooked: campus infrastructure decisions that affect hundreds of thousands of daily trips deserve the same analytical rigor we apply to power grid planning or structural engineering. The data exists. The tools exist. What has been missing is the willingness to treat university transportation as an engineering problem rather than an administrative afterthought.