Ingenuity as Infrastructure: Making Bus Systems Resilient to Heat

Reassigning buses to reduce wait times

Bus riders easily tire on hot days when they must wait for indeterminate periods atop smoldering pavements. Their experience is not only uncomfortable, but potentially dangerous: globally, an estimated half a million people each year die prematurely from extreme heat exposure.

Increasingly aware of such risks, many transit agencies now plan to plant more trees and install shade structures and benches at stops. Such laudable ideas, however, take time to implement and risk obsolescence. Suppose a station is relocated? Or trees die from neglect? What if temperatures grow so extreme that the shade no longer suffices?

My colleagues and I counter the drawbacks of such “unadaptable” hard-infrastructure in a study recently published by Sustainable and Resilient Infrastructure. We demonstrate how agencies can reallocate bus fleets so vulnerable riders wait less in heat. We use Phoenix, Arizona — the US’s fastest growing metropolitan area — as a case study.

Pinpointing high-risk riders

Our analysis is primarily informed by a model of bus demand throughout Phoenix on a typical day. This model, produced by the Maricopa Association of Governments, simulates individual trips throughout Phoenix along with demographic information for each traveler. Our group translated these trips into actual bus line usage by channeling these passengers through the stops and timetables publicly reported by Valley Metro, the main transit agency in Phoenix. This ultimately produces a snapshot of “business-as-usual” wait times and ridership to identify which lines require improvement.

In our study, we assume that wait time can be reduced in one of two ways: adding unused vehicles to popular lines or reassigning buses from those less busy. Our algorithm considers these changes for all possible lines while ensuring no route receives too many or too few buses. Notably, in this exercise, we consider the age, income and car-ownership of each trip-taker. We also use satellite imagery to measure the “greenness” of each start and end location to account for the availability of tree shading and cooling. These variables help us prioritize service to the most heat-sensitive and transit-dependent riders.

The takeaways

When we compare the heat-minimizing schedule to the “status-quo”, we find that small changes (10%) in the allocation of buses can reduce all passenger heat exposure by as much as 35%. This large return on investment stems from a select few bus lines carrying a disproportionate share of (vulnerable) passengers. Once we mitigate these high-demand routes, reallocation returns quickly diminish; each additional bus per line shrinks waits less than the one before it.* Eventually an equilibrium is reached where one line’s “gain” is not large enough to justify the other’s “pain”.

Map of Phoenix Fleet Reallocation. Blue polygons correspond to vulnerability (darker shaded areas are older and more transit-dependent). Lines correspond to the 72 local bus lines that we studied and are shaded according to the change in buses per runtime. Red routes experience a reduction in service, with a maximum reduction of 1.7 buses per runtime. Orange routes saw an increase in service with a maximum increase of 8.6 buses.

For sake of simplicity, our simulation ignores both a transit agency’s potential financial costs and rerouting’s spillover impacts on other factors that shape passenger comfort and utility. Still we hope our research helps Valley Metro and other US transit networks practically respond to heat crises — during which they’d need to reallocate lines with short notice.

More broadly, we hope to illustrate the adaptive power of operating, not just building, resiliently. That strengthening our infrastructure to withstand climate change can be accomplished using the tools and resources already at our disposal. All it takes is a little ingenuity.

*Consider a route served by only one bus that takes one hour for the driver. If a rider barely misses the bus, they’d wait an hour for the next (and only) bus to arrive. Two evenly-spaced buses servicing that route instead at half-hour intervals would halve the wait to 30 minutes. But a third bus on that line renders waits to 20 minutes — only 10 minutes less (and one-third the time) than the second.

This article discusses published work: Adaptive Transit Scheduling to Reduce Rider Vulnerability During Heatwaves, Noam Rosenthall, Mikhail Chester, Andrew Fraser, David Hondula, and David Eisenman, Sustainable and Resilient Infrastructure, 2022, doi: 10.1080/23789689.2022.2029324.