A traffic-flow-dependent macroscopic model of electric vehicle energy consumption

This paper shows that the assumption of a constant link speed in macroscopic energy consumption modelling of electric vehicles (EV) – common in many applications – leads to a significant underestimation of consumptions. To address this issue, a macroscopic model that accounts for the impact of traffic dynamics on EV energy consumption is developed. Model development follows an analytical investigation which demonstrates a dependency between energy consumption and acceleration variance. Such dependency is quantified by means of a large Monte Carlo simulation experiment, in which all modelling inputs including profiles of vehicle speed and road slope are generated by means of ad hoc algorithms to resemble real-world variability. A subsequent global sensitivity analysis quantifies the contribution of each model input factor – parameters and input profiles – to the accuracy of energy consumption calculations. Results confirm that the acceleration variance is by far the most influential model input, so that model accuracy is severely compromised if driving/traffic dynamics are neglected. The proposed macroscopic energy consumption model accounts for traffic dynamics by using the vehicle speed variance as explanatory variable, being it a convenient proxy of the (hardly measurable) acceleration variance. Indeed, we show that the speed variance of vehicles in a traffic stream can be expressed as a function of measurable traffic characteristics such as traffic link density and flow. Results of the model validation against laboratory and experimental data show that the proposed traffic-flow-dependent model consistently outperforms the base macroscopic model.