Improved Vehicle Trajectories and Powertrain Physics in Microscopic Traffic Simulation
2022-01-0502
03/29/2022
- Event
- Content
- Traffic microsimulation offers several advantages when modeling traffic patterns and traffic flow conditions and has enjoyed increased capability as computing resources continue to become cheaper and more powerful. However, the car following models that are intended to emulate the longitudinal dynamics of human-driven vehicles in microsimulation, lack realistic enough physics to enable simulation of vehicle powertrain behavior over the driving trajectories. This is a significant drawback if microsimulation results are to be used for any sort of vehicle energy efficiency analysis. The present work set out to use the Simulation of Urban MObility (SUMO) software to model the energy efficiency impacts of various connected and automated vehicle (CAV) technologies when present at varying rates in the passenger vehicle fleet. To achieve this, we proposed a novel method to automatically calibrate car following model and simulation parameters to match the aggregate driving behaviors from real-world driving data using multi-objective optimization. Improved car following model behavior in relatively uncongested conditions was also critical. The resulting SUMO trajectories now resemble real-world drive cycle statistics and are suitable to conduct powertrain simulation using tools such as NREL’s FASTSim. Findings include an ensemble of enhanced Intelligent Driver Model (eIDM) parameter sets that represent various clusters of real driver behaviors. These parameter sets are publicly available, along with the SUMO simulations run by NREL on a network in the Denver metro area.
- Citation
- Holden, J., "Improved Vehicle Trajectories and Powertrain Physics in Microscopic Traffic Simulation," SAE Technical Paper 2022-01-0502, 2022, .