SIMULATION STUDY OF THE ONLINE AERODYNAMIC DRAG COEFFICIENT ESTIMATION OF A HEAVY-DUTY VEHICLE

Open Access
Author:
Liu, Qifeng
Area of Honors:
Mechanical Engineering
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
  • Hosam Kadry Fathy, Thesis Supervisor
  • Stephanie Stockar, Honors Advisor
Keywords:
  • Vehicle Parameter Estimation
  • Linear Least-Squares Estimation
  • Recursive Least-Squares Estimation
  • Aerodynamic Drag Coefficient
Abstract:
This thesis investigates the estimation of heavy-duty vehicles’ aerodynamic drag coefficient onboard during driving. The motivation of this thesis is the importance of accurate aerodynamic drag coefficient estimates for the performance of the close-loop system, e.g., the adaptive controller in the context of heavy-duty vehicle platooning. A platoon is a train-like formation of a group of heavy-duty vehicles at close intervehicular distances. Potential benefits in overall fuel savings coming from platooning are the ultimate result of considerable air drag reduction, the decreasing value of aerodynamic drag coefficient CD, which can vary to a large extent based on the loading situation and the relative positions with respect to other vehicles. The thesis demonstrates an estimator that specifically identifies the aerodynamic drag coefficient of one heavy-duty vehicle. The proposed estimation algorithm builds on a longitudinal vehicle dynamics model in a highway-driving scenario in which platooning is beneficial in terms of improved fuel consumption. This model-based estimator has little prior knowledge of the aerodynamic drag coefficient and rolling resistance of the vehicle. Essentially a recursive least squares estimator, the algorithm repeatedly estimates the aerodynamic drag coefficient in order to provide the most updated vehicle dynamics parameters during on-road operation. The accuracy of the algorithm is quantified by a point cloud of 1,000 independent simulation runs involving with white and Gaussian measurement noise. The estimation quickly converges (within 50 seconds) to ±2 % of the true value of the aerodynamic drag coefficient. Simulations with different road profiles demonstrate the robustness of the estimator.