Aggregate urban truck tour synthesis from public data

Increasing complexity of urban freight policies demand agent-based simulation models that can address time-of-day dynamics. However, existing state of the art tools like SimMobility and MASS-GT require access to detailed establishment/shipper data. For agencies that lack such data, urban freight agent simulation requires a truck tour synthesis that can adequately fit to aggregate public data. We propose such a truck tour synthesis methodology that takes generated freight trips and distributes them onto a set of generated tours with an original balancing algorithm for entropy maximizing tour distribution that is scalable to citywide applications. The method is tested in a case study of New York City encompassing 47 industry groups, over 500 zones including gateways into the city, two truck classes, and a road network calibrated to road restrictions from New York City Department of Transportation and Uber Movement speed data across four different time periods of the day. A total of 470,000 tours were generated (10,000 tours per industry group) and flows distributed using the proposed algorithm. Compared to cross-borough screenlines, an average error in counts of 10.2% was achieved. The resulting synthetic truck population provides a baseline dataset for truck vehicle-miles-traveled, greenhouse gas emissions, and volumes across key corridors, that can be further disaggregated into truck type, industry served, and time of day. A counterfactual scenario examining a policy to require 20% smaller truck capacities highlights the applicability to quantify trade-offs with a 49% reduction in Equivalent Single Axel Loads while increasing emissions by 25%.