“…The methods used by both Beikmann et al 15 and Zhang and Zu 25 had numerical problems in computational efficiency and were not easy to model and program in a general-purpose code because the explicit characteristic equations varied with the configuration of the belt drives. 2, 26 To overcome the defects of the explicit exact expression, approximation methods such as the Galerkin method, 2, 27, 28 , the finite difference method, 8, 27, 29 the harmonic balance method, 30 the finite element method, 13, 31, 32 and the multi-body dynamics method 33, 34 have been widely used in the analysis of belt drives. These methods usually provided approximate solutions by globally or locally discretizing the continuous belt spans in spatial domains and are easy to program in a general-purpose code for arbitrary configurations of the belt drives.…”