Evaluating Simulation Driver Model Performance Using Dynamometer Test Criteria

The influence of driver modeling and drive cycle target speed trace modification on vehicle dynamics within energy consumption simulations is studied. EPA dynamometer speed error criteria and the SAE J2951 Drive Quality Evaluation for Chassis Dynamometer Testing standard are applied to simulation outputs as proposed components of simulation validation, providing guidelines for acceptable vehicle speed outputs and allowing comparison of simulation results to reported EPA dynamometer test statistics. The combined effect of driver model tuning and drive cycle interpolation methods is investigated for the UDDS, HwFET and US06 drive cycles, with EPA-specified linearly interpolated speed trace and a PI controller driver as a baseline result. Additional benefits of driver tuning are presented including a reduction in unnecessarily-aggressive simulated accelerator and brake pedal actuation, resulting in a drop of over 70% for peak jerk, 60% for RMS jerk, 80% for ripple aggressiveness, and 20% for peak accelerator pedal actuation between analyzed cases. For the simulated vehicle, a “lazy” driver model and smoothly interpolated drive cycle better estimate a human driver response by decreasing error from EPA-reported SAE J2951 parameters while reducing harsh simulated vehicle dynamics.