Modelling and dynamic analysis of switched reluctance drives

Abstract

This research has been driven by the necessity for machine designers to be able to quickly estimate the performance of a Switched Reluctance Motor (SRM). Although there is a very simple mechanical structure, the modelling of these machines is complicated by large amounts of magnetic saturation within the core. This necessitates running extremely timeconsuming Finite Element Analysis, which reduces the design affordability. This work attempts to produce a rapid performance prediction tool for SRMs, with a particular focus on the modelling of the static magnetisation curves to be used in the machines’ drive and in the iron loss prediction over the whole range of operating speeds. A new analytical model for 2D static characterisation of SRMs is presented, based on a combination of 2D FEA and analytical models. The minimum amount of 2D FEA is used to complete the set of magnetisation curves for any SRM design. The characterisation has then been expanded to include 3D end-effects. This uses a reduced set of 3D FEA performed on the analysed machine, and hence a reduced computation time: this helps at the early modelling stage of an electrical machine’s design. A SRM prototype has been designed, built, and tested using an asymmetric half-bridge converter. The aforementioned static modelling technique has been validated through static measurements of flux-linkage conducted on the prototype. The prediction of iron loss in the SRM has been investigated: this is particularly important for efficiency calculations in traction drives within the automotive industry. Different sets of adjustment coefficients have been considered to capture the effect of the highly non-sinusoidal flux-density waveforms in the loss computation, mostly due to the DC-biased components of flux-density and minor hysteresis loops. The rain-flow counting method has been used to account for the minor loops present in the flux-density waveform: these greatly influence the overall hysteresis loss component in SRMs. Iron loss measurements have been conducted on toroidal samples to appreciate the influence of non-sinusoidal flux-density waveforms on simple ring geometries. From these, the use of coefficients to account for the effect of DC-biased flux-density waveforms in the iron loss of SRMs has been validated. Moreover, it has been possible to evaluate the difference between the materials’ characterisations provided by the manufacturers and those derived from iron loss measurements and measured magnetization curves, which include the effects of machining processes. Iron loss prediction in SRM models has been performed using FEA solvers and proprietary algorithms, developed to post-process the values of flux-density found in each mesh element, according to the iron loss coefficients found during the material characterisation on toroidal samples. Finally, different sets of dynamic operating conditions have been tested on the SRM prototype, to obtain the iron loss contribution. This is achieved through measurement of input electrical power and output mechanical power, along with windage and winding losses to leave the iron loss.