Power Loss Estimation in Poly-V Belts for Micro and Mild Hybrid Powertrains Using the Generalized Maxwell Model

In modern automotive powertrains, the front-end accessory drive represents a crucial subsystem that guarantees the proper functioning of micro and mild hybrid configurations and auxiliary vehicle functionalities. The motor/generator (12 V or 48 V), the air conditioning compressor and other accessories rely on this subsystem. Therein, the poly-V belt is the main transmission mechanism. From an efficiency standpoint, its behavior is usually represented through slip and elastic shear phenomena. However, the viscoelastic nature of the compounds that constitute the belt layers demand a more detailed approximation of the loss mechanisms. The quantification of such losses allows evaluating the performance of the e-machine integrated in the powertrain and improving its control. This work models the belt through a lumped-parameter time-domain model, where domains are discretized into multiple elements and represented through the generalized Maxwell model. Loss contributions due to bending, stretching, compression and shear are considered in the relevant degrees of freedom and calculated by numerical integration on each belt element. The present method offers advantages in terms of scalability and incorporation with other time-domain methods where viscoelastic belt losses are neglected. Furthermore, it allows using non-periodic input excitations, which is the main limitation of frequency-domain alternatives. This tool is suitable in the design phase of belt drive systems, where a flexible and computationally affordable technique may provide insights regarding energy efficiency. Moreover, the proposed methods could be extended to systems involving belts as transmission elements, such as a continuously variable transmission (CVT). To evaluate the proposed approach, the parameters of the Maxwell model are identified for a specific belt. Then, an experimental campaign is executed on a fully electric dedicated testbed reproducing a five-pulley layout. Results demonstrate the validity of the model. Furthermore, they highlight the power loss distribution in different working points of the system.