Drag acting on articulated cylinders in pedalling motion : An experimental study towards data-driven modelling Département A&M (Université de Liège)

Abstract

In professional cycling, aerodynamic drag is the dominant source of power loss, even in uphill conditions. Better understanding the aerodynamic behaviour of the legs during pedalling is therefore key to improving athlete performance. This work presents an experimental study of the drag acting on two articulated cylinders replicating the pedalling motion of a cyclist's leg. The system is mounted on a collaborative robot and tested in the wind tunnel of the University of Liège. The effects of Reynolds number, reduced frequency, motion amplitude and yaw angle are investigated. The results highlight a significant unsteady behaviour, with a phase shift of the drag increasing with the reduced frequency, interpreted as a wake memory effect. Two modelling approaches are then compared: a quasi-steady model based on the Independence Principle, and a data-driven model combining an in-cycle Ridge regression with an out-of-cycle Gaussian Process regression. While the former reproduces the global drag behaviour satisfactorily, the latter struggles to generalise due to the inter-dependency of the parameters and the limited size of the dataset.