Extended Modeling, Calibration and Validity Assessment of Vehicle Models in Future Automotive Systems Technology Simulator via Real-World Driving Data

Software simulation tools for vehicle fuel economy/energy efficiency can play an important role in strategic decisions about advanced powertrains. One such tool that has been developed by the National Renewable Energy Laboratory (NREL) is known as FASTSim. The philosophy of FASTSim aims to strike a difficult balance between simplifying the task of creating/editing vehicle models, fast computation time and high-fidelity simulation results. In the “baseline” version of FASTSim, which is open-source and freely available in Python or Excel, the instantaneous efficiency of an engine, motor or fuel cell is estimated via reference curves as function of power demand. The reference efficiency curve for each powertrain subsystem (e.g. for a spark-ignition engine) in baseline FASTSim has the same profile irrespective of what vehicle is being modelled, which is a compromise in accuracy in favor of ease of modeling. This paper utilizes an open-source Java implementation of FASTSim with capability for custom efficiency curves for engine and motor, along with a large dataset of real-world vehicle trips to calibrate and validate FASTSim vehicle models for three Battery Electric Vehicles (BEVs), four Plug-in Hybrid Electric Vehicles (PHEVs), one non-plug-in Hybrid Electric Vehicle (HEV) and one conventional internal combustion engine (ICE) vehicle. An ultimate goal in vehicle modeling, is for the simulation results to closely match the real-world trip data for every trip, but such a goal is difficult due to many uncertainties in real-world trips. Instead, results show that it is possible to achieve high fidelity for an aggregate of several trips, and the modeling fidelity improves with less uncertainty in trips information, such as when road slope and cabin heating/cooling loads are known.