Overview
Cities like NYC and Seattle need to deal with significant growth of urban deliveries as a result of increasing e-commerce compounded by increased stay-at-home behavior due to COVID-19.
Even with the production of vaccines and a return to “normality” in the next year or so, it is likely that the introduction of e-commerce to so many more consumers will maintain its penetration. As a result, these cities are seeing more truck trips going to residential neighborhoods. To combat these challenges even before the emergence of COVID, city agencies have worked with major goods providers, delivery companies, and third-party logistics providers to explore alternative last-mile delivery modes like cargo bike, better curb management policies, off-hour delivery programs, neighborhood loading zones, planning designated truck routes, and evaluating truck route compliance.
With more limited data in NYC, we seek to model the freight movement in two phases.
- The first is a synthetic population portion that takes freight OD demand and converts it to multiple classes of commercial vehicles to make deliveries along routes. While there is work that generates OD matrices under a freight tour setting, they are not distribution models that use trip productions and attractions.
- Second, the truck trips are then assigned onto a time-of-day mesoscopic simulator, MATSim-NYC. This assignment relates the demand to the road network taking into account traffic dynamics.
Research Objectives
We propose to develop a citywide model of truck network flows, one that relates changes to truck
routes to changes in truck tours or to time-of-day congestion pricing policies, for example.
The research is broken down into three phases.
PHASE I – data collection and preparation
Task 1. Project Management – establish priorities, research team, etc.
Task 2. Obtain necessary truck route network datasets from NYC DOT and/or NYC Open Data needed.
PHASE II – model development
Task 3. Evaluate off-route truck route violations and identify hot spots and patterns. For this task, we will implement a Hackathon among students to help analyze this data set. In addition, we will apply advanced data analytics to create a GIS dashboard to visualize off-route truck route violations and identify hotspots and patterns.
Task 4. Create synthetic freight carrier “population”. Using transport establishment data along with the truck OD matrix provided by NYMTC, we will synthesize a sample of depots and customers with time windows. Continuous approximation models will help guide the locations of the random customer locations. K-best tour plans will be generated for this sample (by enumerating sequences of the tours and sorting out non-dominating results) and the weights of the utility functions corresponding to those tour plans will be calibrated such that the sample’s plan choice model best replicates the distribution pattern of the truck OD matrix from NYMTC. That sample is then replicated to match the total volumes. This produces the agendas for each truck, which is incorporated into the synthetic population file.
Task 5. MATSim recalibration. Update the vehicle file and road network with truck attributes. Then re-calibrate MATSim-NYC (this time at 10% sample) to create a new MATSim-NYC 2.0 model that includes truck flows. Scenarios that modify travel costs will impact both the synthetic population (by altering the distribution of plan choices for a freight company) followed by assignment onto the simulation.
Task 6. Help UW Jeff Ban’s group to create a MATSim-Seattle using the lessons learned from prior development.
PHASE III. Analysis
Task 7. Estimate Truck Route performance measures (e.g. VMT). Review truck travel patterns on local roadways and arterial streets to determine where trucks are traveling today, where they should be traveling, and any gaps or misalignments of these patterns. Use MATSim-NYC 2.0 to evaluate the truck route network considered by NYCDOT on congestion and greenhouse gas emissions, safety, and connectivity.
Task 8. Create a dashboard to visualize truck routes and evaluate system performance; develop
score/rating for truck routes based on established performance indicators: Travel Time Index,
Planning Time Index, Network Connectivity, Top Congested Freight Corridors
Related Media
Personnel
Kaan Ozbay
Director, C2SMART Professor, NYU
Kaan Ozbay is a Co-Principal Investigator on this project.