Estimating the potential for optimized curb management to reduce delivery vehicle double parking, traffic congestion and energy consumption
Aaron Burns, Jeremy J. Michalek, Constantine Samaras
Abstract
We model an optimized curb parking reservation system and characterize conditions under which such a system can reduce delivery vehicle double parking, congestion and energy consumption. We implement an optimization model leveraging integer linear and mixed-integer linear programming parking slot assignment formulations to minimize double parking and build a queuing model to estimate lane obstruction congestion and energy effects. Using delivery data from Aspen, CO and Pittsburgh, PA, we find that, when arrival times are known at the start of the day, a single-space reservation system can eliminate 1–2 min of double parking per hour, on average, increasing 3–4 times when drivers have ±5 minutes of arrival time flexibility. Scenarios involving up to seven co-located spaces show diminishing returns but suggest reservation systems can provide the equivalent of additional parking capacity in some cases. When arrival times are uncertain and buffers are used between reservations, we find that curb reservation systems reduce congestion in high-demand, moderate uncertainty scenarios with short duration reservations but increase congestion in scenarios with high uncertainty and long duration reservations, with estimated annual congestion and emissions externality cost savings ranging from -$100,000 to +$300,000 for a single space. Our results suggest that implementation of curb reservation systems should be targeted to application environments where they can provide net benefits, and our simulations help to characterize key factors for identifying such applications.