As retailers compete to offer faster delivery, companies like FedEx and UPS strive to maximize driver efficiency. With drivers making more than 100 stops per day, delivery management is key to optimizing driver performance. This is especially true in highly congested urban areas. Cities offer a number of unique challenges to logistics planners, including traffic congestion, difficulty in making left turns, and the inability to easily reverse directions. Addressing those challenges is sometimes easier said than done. When UPS began rolling out its On-Road Integrated Optimization and Navigation (ORION) software in 2012, the algorithm performed well in rural areas. However – despite calculating approximately 200,000 routing options over the course of a driver’s day – the software added many unnecessary miles and minutes to urban routes. Recent vehicle routing studies suggest that cities pose unique challenges for logistics planners, but none have identified the specific challenges. William J. Rose (Iowa State University), Diane A. Mollenkopf (University of Canterbury), Chad W. Autry (University of Tennessee), and Brent D. Williams (University of Arkansas) outline those challenges in their Decision Sciences article “Tailoring Transportation Planning Decisions to Diverse Urban Environments.” As its title suggests, the article highlights the importance of planning delivery logistics based on a city’s specific characteristics, rather than relying on a “one-size-fits-all” approach.
Rose, Mollenkopf, Autry, and Williams analyze vehicle routing decisions in eight major U.S. cities. One of those cities is Indianapolis, which the authors use as a baseline because it most closely represents the typical city modeled within vehicle routing literature. By comparing how logistics planners in other cities account for the specific challenges they encounter, the authors construct a theoretical framework to help guide managers.
The authors identify common urban-specific obstacles and the adjustments managers make when cities violate common routing assumptions. For instance, Indianapolis has high accessibility due to both its geography and infrastructure. That accessibility allows managers to alter routing boundaries as daily need dictates, since drivers can reach any part of the city with relative ease. Conversely, the Los Angeles area offers low accessibility. Though Malibu and Agoura Hills appear relatively close on a map, winding mountain roads make it impossible for a driver to efficiently cover both areas. The authors find that routers overcome accessibility issues by developing and observing consistent routing boundaries.
Another obstacle Rose and his co-authors identify is uncertainty. Indianapolis presents drivers with a relatively high degree of certainty. That is not the case on the outskirts of rapidly expanding Houston, where drivers are likely to encounter unexpected construction or new streets that do not yet appear in navigation software. The authors find that managers adapt to high uncertainty areas by assigning individual drivers to the same area consistently. This allows drivers to develop deep familiarity with a subregion and quickly choose efficient alternate routes when necessary.
Another urban environmental characteristic the authors identify is redundancy. Indianapolis exhibits high redundancy since it offers drivers multiple paths to efficiently reach a given destination. This redundancy contributes to Indianapolis routers’ flexibility in driver assignments and routing boundaries. That flexibility is not found in much of New York City, which combines low redundancy with high uncertainty and low accessibility.
Additionally, the authors examine the ways routers account for different population clustering profiles. While San Francisco is densely populated throughout, Austin has both dense core areas and less challenging rural routes. The authors show how different clustering profiles affect driver assignments, routing boundaries, route sequencing, and the likelihood that routes will finish near the distribution center.
Read the full article in Decision Sciences.
