PHEV Modeling: Model Validation

To verify the accuracy of a model, the outputs predicted by the component and powertrain models need to be compared to test data. Several advanced vehicles, tested at ANL's APRF, have been validated in PSAT using the process described in the figure below.

Generic Data Quality Analysis Process

The goal of validation is to:

• Automatically realign the data when different sources are used (e.g., emission bench, dynamometer)
• Select the proper sensor when the same parameter can be measured/recorded from different sources
• Quantify the uncertainty of each sensor by comparing its values with measured or calculated parameters by using powertrain equations
• Reuse existing post-processing capabilities developed initially for simulation purposes to automatically calculate effort, flow, power, energy, and efficiencies, as well as use the analytical tools already available in PSAT
• Automate report generation so that complete analysis can be summarized in an HTML document within minutes

Several conventional vehicles (e.g., Chevy Equinox, Ford Explorer) have been validated within 1% of fuel economy. Hybrid electric vehicles (e.g., Honda Insight, Toyota Prius, Lexus RX400h) have been validated within 5% for both fuel economy and battery state-of-charge for several driving cycles.

For example, the validation results for the Hymotion Prius vehicle are shown below:

May 2008