Simulation
Planetary Gearbox for Wind Turbines
Extension of the interpolated contact shapes approach to planetary gear sets, validated on a helical wind turbine gearbox at 360 kNm carrier torque, achieving nearly two orders of magnitude speedup over CMS reference models while retaining accurate transmission error and tooth stress prediction.
Planetary Complexity
Planetary gear sets appear throughout high-torque transmission applications: wind turbine gearboxes, helicopter transmissions, automatic gearboxes, bicycle drivetrain hubs. Their compactness and torque distribution across multiple planet gears are engineering advantages. From a simulation standpoint, they are more involved than a simple external gear pair.
In a three-planet planetary set, six gear meshes are active simultaneously. Each planet meshes with both the sun and ring gear. The planets orbit around the sun gear on the carrier, whose rotation couples all kinematic constraints together. The ring-planet contacts are internal gear meshes, with different geometry and stress distributions compared to external contacts.
Method
The simulation uses the MUTANT toolbox and extends the interpolated contact shapes approach, previously validated for single parallel-axis gear pairs, to the planetary case. The reduction basis for each gear is built from classical vibration modes supplemented with precomputed contact shapes at sampled positions of the planet carrier. During the dynamic simulation, these shapes are interpolated as the carrier rotates, allowing the moving multi-mesh load pattern to be captured accurately without retaining the full finite element model at each time step.
Exploiting the periodicity of the planetary gear set reduces the number of required sampling positions to one angular pitch of the ring gear, keeping the offline preparation phase tractable.
Validation
The benchmark case is a helical planetary gear set representative of a wind turbine gearbox, loaded at 360 kNm on the planet carrier with a viscous generator torque at the sun gear. After 100 ms, a torque ripple simulates the drivetrain response during a symmetric voltage dip — a standard load case in wind turbine certification.
Results were compared against a Component Mode Synthesis reference model and a purely modal reduced model. The modal-only model underestimates dynamic transmission error because vibration modes alone cannot represent the localized tooth contact deformation. The contact-shape-enriched models correctly reproduce both transmission error and sun gear tooth stress distributions.
The CMS reference model used over 10,000 degrees of freedom; the contact-shape models used approximately 260 to 270. The speedup is close to two orders of magnitude.