Suitability of Synthetic Driving Profiles from Traffic Micro-Simulation for Real-World Energy Analysis: Preprint

Eric Wood, Jeffrey Gonder, Evan Burton, Yunfei Hou

Research output: Contribution to conferencePaper

Abstract

A shift towards increased levels of driving automation is generally expected to result in improved safety and traffic congestion outcomes. However, little empirical data exists to estimate the impact that automated driving could have on energy consumption and greenhouse gas emissions. In the absence of empirical data on differences between drive cycles from present day vehicles (primarily operated by humans) and future vehicles (partially or fully operated by computers) one approach is to model both situations over identical traffic conditions. Such an exercise requires traffic micro-simulation to not only accurately model vehicle operation under high levels of automation, but also (and potentially more challenging) vehicle operation under present day human drivers. This work seeks to quantify the ability of a commercial traffic micro-simulation program to accurately model real-world drive cycles in vehicles operated primarily by humans in terms of driving speed, acceleration, and simulated fuel economy. Synthetic profiles from models of freeway and arterial facilities near Atlanta, Georgia, are compared to empirical data collected from real-world drivers on the same facilities. Empirical and synthetic drive cycles are then simulated in a powertrain efficiency model to enable comparison on the basis of fuel economy. Synthetic profiles from traffic micro-simulation were found to exhibit low levels of transient behavior relative to the empirical data. Even with these differences, the synthetic and empirical data in this study agree well in terms of driving speed and simulated fuel economy. The differences in transient behavior between simulated and empirical data suggest that larger stochastic contributions in traffic micro-simulation (relative to those present in the traffic micro-simulation tool used in this study) are required to fully capture the arbitrary elements of human driving. Interestingly, the lack of stochastic contributions from models of human drivers in this study did not result in a significant discrepancy between fuel economy simulations based on synthetic and empirical data; a finding with implications on the potential energy efficiency gains of automated vehicle technology.
Original languageAmerican English
Number of pages15
StatePublished - 2015
Event22nd ITS World Congress - Bordeaux, France
Duration: 5 Oct 20159 Oct 2015

Conference

Conference22nd ITS World Congress
CityBordeaux, France
Period5/10/159/10/15

NREL Publication Number

  • NREL/CP-5400-64609

Keywords

  • automated and connected vehicles
  • automotive efficiency
  • driving behavior models
  • powertrain simulations
  • real-world driving data
  • traffic micro-simulation

Fingerprint

Dive into the research topics of 'Suitability of Synthetic Driving Profiles from Traffic Micro-Simulation for Real-World Energy Analysis: Preprint'. Together they form a unique fingerprint.

Cite this