Uncertainty analysis of vehicle fuel consumption in twin-roller chassis dynamometer experiments and simulation models

Chassis dynamometers are one of the main test benches applied to vehicle development and homologation, however, vehicle experiments and simulations present associated uncertainties. Thus, aiming to allow previous analyses to assist the vehicle homologation process, this research analyzes uncertainties and their impacts on fuel consumption on a vehicle-dynamometer testing simulation on Matlab/SimulinkTM, under a driving cycle, and experimentally validated. The paper analyzes the developed model inputting variations (speed profiles, tire parameters, and rolling resistance coefficient) and quantifies the fuel consumption response and its uncertain-ties, comparing them with experimental results. The results showed the experimental fuel consumption (0.9985 +/- 0.0522 (l)) compared to the ones submitted to input variations. Among these, the rolling coefficient was the most influential factor (0.9670 +/- 0.1338 (l)). The final variation analysis presented the mean value closest to the experimental (0.9951 +/- 0.0959 (l) and mean error of 0.33%). The simulation allows reproducing experiments virtually, mitigating experimental time and cost spent on real benches, and providing an analysis tool in order to help the vehicle homologation.