Transportation Decarbonization Pathways Assessment Through Metropolitan Area Simulation
Abstract:
In this presentation, we will be discussing how agent-based transportation system simulation has been used to estimate the impact of advanced technologies and policies on mobility, energy, emissions, GHG, cost, equity… POLARIS is a high-performance agent-based modeling framework designed for simulating large-scale transportation systems. As an integrated network-demand model, all aspects of travel decisions (departure time, destination choice, planning and rescheduling as well as route choices) and travel execution (traffic flow, transit simulation, fleet movements) are modeled simultaneously to estimate the energy impact of vehicle technologies and mobility services at a range of scales (neighborhood to metropolitan region). A flexible workflow, centered around POLARIS, has been developed as part of the US DOE smart mobility Consortium. Examples of applications focusing on increased transit ridership, multi-modal travel, electrification, connectivity and automation will be discussed.
Bio:
Aymeric Rousseau is the Interim Director of the Center for Decarbonization Solutions Deployment (CDSD) and the Director of the Vehicle and Mobility Systems Department at Argonne National Laboratory. He received his engineering diploma at the Industrial System Engineering School in La Rochelle, France in 1997 and an Executive MBA at Chicago Booth in 2019. For the past 25 years, he has been evaluating the impact of advanced vehicle and transportation technologies from a mobility and energy point of view including the development of Autonomie (vehicle system simulation) and POLARIS (large-scale transportation system simulation). As part of CDSD, he coordinates stakeholder engagement, technical assistance and deployment related to decarbonization across Argonne. In addition to leading numerous projects for private companies and US Department of Energy, Aymeric Rousseau has been providing the vehicle energy consumption analysis for the US DOT NHTSA fuel economy regulations and has authored or co-authored more than 175 papers related to transportation decarbonization.
Summary:
Decarbonization of transportation
Mode choices
Simulation of Transport
System Level Energy Efficient Transport Research
Better vehicles
Smarter vehicles
Smarter roads
Smarter travelers
End-to-end System Simulation: From individual vehicles to entire metro areas
Vehicle energy consumption, performance, cost
CAV vehicles control and energy use
Vehicle trip profile generation
Focus of the talk:
Metro area transportation system
Polaris: Agent-based transportation system
Everything is an agent: drivers, cars, streetlights, roads, buses
Initiated by FHWA (highway authority), further developed by DOE
Study topics:
Multi-modal travel
Vehicles and infrastructure interactions
Connectivity and automation
Urban science
Scales:
Metro-scale: Urbansim
Long-term choices: population, home/workplace, household vehicle choice, CAV tech choice
Mid-term choice: telecommute, activity generation and planning
Within-day: activities, routing, traffic, EV charging, transit use, freight/logistics
Can understand impacts of many policies
Smart mobility
Interactions between drivers and policies
Fine-grained impacts on individual household types and communities
Tools:
Network editor (roads, intersections, traffic control)
Simulator and interactive visualizer
Web-based analyzer/postprocessor
Inputs:
Traffic analysis zones
Location/parcel data
Household surveys
Census data
Vehicle distributions
Transportation behavior surveys
Transit network: connectivity, traffic control, agencies, traffic counts, taxi data, bike/car share, charging locations, parking inventory
Data is collected by collaborating with local partners: transportation agencies, universities, etc.
Data quality is limited
Small-scale errors are not critical
Large-scale model predictions (average flows/speeds on major roads) are validated against live data
Polaris is being applied to many major cities: Chicago, Seattle, Detroit, Washington, Los Angeles, Beijing, Sydney, Austin
SmartMobility workflow
Understanding efficiency of different large-scale transportation policies
Transportation/logistics
Energy use
Traffic
Integrated network, demand and traffic assignment
Energy consumption is modeled per-vehicle, considering many different types of vehicles
Model of charging network, as well as queuing at chargers (key for transportation and impact on delivery schedules)
Impact of land use on transport and impact of travel time (bidirectional connection to UrbanSim)
Performance:
C++ implementation, integrated with efficient optimization algorithms (CPLEX, Gurobi, GLPK)
4-6hr for million agents
Optimization:
Before run:
· Design of EV Charging network
· Design of transportation network (warehouse-factory-store links)
Mid-run: ecorouting algorithms (also developed ML-based energy consumption models to support this)
Optimization of transportation policies (many POLARIS runs of alternative policies)
Use scenarios
Transit optimization in Chicago
Bus Rapid Transit
Improved transit ridership by 11%
Transit optimization in Austin
Varied charges for transportation
Partial subsidy improved transit use from 4.5%-5%
Free transit would increase to 5.6%
Impact of connectivity on delays due to traffic accidents
Connected vehicles rerouted around accidents
Can mitigate 50% of delay
Best performance achieved at 50% penetration (don’t want everyone to reroute due to every accident)
1500 detectors sufficient (Bloomington, IN)
Impact of Level 2 (driver assist) vs Level 4 automated vehicles (fully automated) on travel time
People in partially automated act the same
People in fully automated vehicles drive more, which slows all vehicles down
Self-driving vs buses
Buses are very efficient but their average utilization is low
If self-driving vehicles are very cheap, then this produces many empty miles as persona vehicles are shuttled between household members
This uses up a lot of energy
Cars used as taxis are much more efficient,
· e.g. municipal or private taxi fleet operators that are globally optimized
· Primary opportunity for self-driving cars to improve upon current system
CO2 Policies that mitigate land use impact of self-driving cars
Impact of increased travel costs on low income and exurban areas (e.g. road pricing)
Design of vehicle charging infrastructure (accounting for parking, depos, households, stores)