Abstract
Magnetic gears offer several advantages over mechanical transmissions, the root of which isthe contactless transmission. However, across a broad range of research studies, their practical
performance has not matched design predictions. In addition, a very small number of magnetic
gears are employed in industrial applications. It has been reported that manufacturing error
contributes to the discrepancy between modelled and experimentally realised performance.
Efficient modelling techniques, which could be used to predict the expected performance range,
would clearly be valuable. Geometric deviations due to manufacturing error are difficult to predict
and inherently random. This thesis assesses the effect of geometric error on the performance of a
Coaxial Magnetic Gear (CMG) using a novel computationally efficient asymmetric analytical
model to conduct Monte-Carlo simulations. The analytical model is validated through the very
close agreement achieved with respect to linear FEA results. Furthermore, a hybrid stochastic
model is proposed, which can calibrate the analytical statistical data with a few non-linear FEA
instances. The scaling of the probability distributions derived using the analytical model are shown
to match the equivalent, but much more computationally onerous, non-linear FEA based solutions.
Such a statistical assessment of the effects of the modulation ring geometric deviations on the
performance of CMGs is shown to potentially be important regarding both the stall torque and the
torque ripple. Consideration of the effects on torque ripple becomes more important in applications
where a fault mitigation perspective is considered, and accurate slipping torque estimation is
essential. It is expected that as CMGs become more widely adopted, such studies will become
increasingly valuable.
| Date of Award | 25 Jan 2022 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Jason M Yon (Supervisor) & Aydin Nassehi (Supervisor) |