A predictive model for reaction wheel assembly microvibration

New generations of instruments contingent on high pointing accuracy are continuously pushing the requirements for satellite stability, but the microvibration disturbances originating from high-speed rotating devices such as reaction wheel assemblies (RWAs) remain an obstacle. In order to address this issue and improve current designs, accurate mathematical models of the noise sources are necessary. This paper presents a hybrid formulation which integrates a finite element-based representation of the rolling elements within an analytical model of a typical cantilevered reaction wheel, whose shaft is supported by two ball bearings. The model is able to predict both the frequency and amplitudes of the forces and moments produced at the RWA’s interface, allowing a comprehensive assessment of the emitted noise’s effect on the host satellite platform. At the bearing level, a wide range of structural parameters, such as preload, number and diameter of balls is handled in conjunction with various localised surface imperfections. It is shown that arbitrary combinations of bearing geometry and defect distributions can be efficiently synthesised from a limited set of responses precomputed by detailed nonlinear transient simulations. Using this capability, the article identifies bearing design trends exhibiting favourable performance in terms of overall emitted microvibration. Furthermore, spatial pairings of these bearing models are introduced in the governing equations of motion of an unbalanced rotor, enabling the evaluation of system-level responses. Finally, the proposed model is validated against test measurements of a physical rotor assembly, showing good agreement between predicted and measured microvibration signatures. • A comprehensive predictive model for reaction wheel (RW) microvibration. • Semi-analytical treatment of localized surface imperfections and rotor imbalances. • Variable bearing parameters, such as stiffness, preload, number and size of balls. • Handling of arbitrary combinations of RW parameters and disturbance sources. • Validation of the proposed model against physical test data.