Development and validation of a dynamic energy consumption model for electric buses

Abstract

The increasing use of electric buses in the transportation sector offers a significant opportunity to reduce environmental impacts, particularly in urban areas where air pollution and carbon emissions from conventional vehicles are major concerns. However, to fully realize this potential, accurate and reliable energy consumption estimations are essential. This study aims to develop a dynamic model for electric buses and validate its energy consumption predictions by comparing them with SORT (Standardized On-Road Test) data and a dedicated road profile. For this purpose, an electric bus model has been developed specifically for the study, enabling a detailed and realistic analysis of energy consumption. Accurate modeling of an electric bus energy efficiency provides valuable insights for both manufacturers and operators. It contributes to the design of more efficient vehicles, enhances battery lifespan, and reduces operational costs. A key objective of this work is to validate the model against actual operational data, focusing on driving cycles that reflect real-world conditions. Another objective is to predict the SoC of the battery during a dedicated road profile usin couloumb counting method. Incorporating realistic driving scenarios enhances the robustness and reliability of energy consumption predictions. The dynamic bus model is developed in MATLAB, using inputs that represent energy consumption in kWh and distance traveled in kilometers. The model's output is compared to real-world data collected during SORT tests and a dedicated road profile. To accurately calculate energy consumption, the model considers parameters such as the distance traveled, average speed, and bus motion duration. These parameters are integrated with road test data to more accurately replicate real-world driving behavior. Vehicles experience several resistance forces during motion that must be overcome by the propulsion system. These include aerodynamic drag, rolling resistance, hill (gradient) resistance, and acceleration resistance. This study neglects wind effects and gradient resistance in line with SORT test conditions, which are performed on flat roads. However, the model incorporates road gradients as input parameters when the vehicle is tested on the dedicated road profile by simulating inclined driving scenarios to assess their impact on energy consumption. The modeling approach accounts for variable motor efficiency, as PMSMs used in electric buses exhibit efficiency changes with torque and speed. Motor torque was derived from resistive forces, and efficiency was calculated at each time step using torque and vehicle speed to determine cumulative energy consumption across driving cycles. Additionally, the efficiency of the DC-DC converter, which supplies power to auxiliary systems, was integrated into the model. High-voltage auxiliary components such as the HVAC system and pre-heater can significantly affect total energy consumption, particularly under extreme ambient temperature conditions. However, since these systems were inactive during road tests, their energy usage was not considered in the validation of the road tests. On the other hand, the power consumption of auxiliary components is taken into account in the model during the tests. Validation of the model against SORT duty cycles data reveals a small deviation between the modeled and real-world data. Since SORT test are conducted on flat roads, the model was further evaluated on a dedicated route that includes a variety of driving dynamics such as urban, intercity, and inclined road conditions to assess its broader applicability. For this dedicated driving cycle, the error rate in the energy consumption analysis between the model prediction and the actual measurement was calculated as 6.75%. On the other hand, SoC is estimated during the dedicated road profile. For the SoC estimation, the Mean Absolute Error (MAE) was obtained as 0.61%, while the Root Mean Square Error (RMSE) was calculated as 0.70%. Despite this minor difference both in the consumption analysis and the SoC estimation, the model demonstrates a high degree of accuracy. Such differences may arise from sensor calibration errors, varying road conditions, or real-time operational adjustments that are difficult to replicate in simulations. Nonetheless, the close agreement indicates that the model reliably estimates the energy consumption of an electric bus. In conclusion, this study presents a validated dynamic model capable of predicting the energy consumption of an electric bus with decent accuracy. The model offers valuable insights for manufacturers and fleet operators, aiding in better fleet management, cost forecasting, and supporting sustainable transportation development. Future work can enhance model accuracy by incorporating additional driving cycles and integrating auxiliary high-voltage components such as heater and HVAC. With ongoing advancements in data collection and modeling techniques, future models are expected to provide even more precise predictions, enabling further optimization of electric bus operations and reduced energy consumption.