Professors J. R. Mayor and R. G.Harley with Chad Bednar, Georgia Institute of Technology
The overall goal of this research program funded through the Grainger CEME collaborative network is to create a multi-physics based design tool that integrates thermal, electromagnetic and mechanical design modules for optimal design of inverter-dedicated induction machines. Inverter-fed induction machines (IFIM) are not limited to 60Hz operation and will be designed to operate at higher excitation frequencies for enhanced torque density or other performance gains. Consequently the core losses in the back iron in the rotor will be increased, leading to increased thermo-mechanical loading of the rotor. The integrated IFIM electro-thermo-mechanical model must therefore consider the rotor heat load, due to both core loss and conduction loss (in the rotor bars), and the complex thermal transport in the air gap.
Prior work from the research team established an integrated electro-thermo-mechanical model for surface-mount permanent magnet machines by taking the simplification of negligible heat generation in the rotor and using a parametric self-segmenting finite different solution of spatio-temporal temperature field in a half-slot model. A two-phase approach to extending the technique to the solution of the complex IFIM temperature fields is being followed. In Phase 1 only rotor core losses will be considered, as would be the case in switched reluctance machines (SRMs), and a parametric mesh generation approach for rotor geometries and air-gap thermal interface will be developed and verified using 3D finite element analysis (FEA) benchmarking. The final phase will develop parametric mesh generation for typical rotor slot geometries and extend the rotor field solution to include both core losses and conduction losses in the rotor bars.
The effort to date has addressed Phase I. A parametric mesh generation approach for the rotor of an SRM has been developed. The rotor model was simplified by describing the heat transfer through the shaft with an equivalent thermal circuit, thereby reducing the total domain size of the finite difference (FD) solution. Parametric descriptions of various conventional rotor geometries have been developed considering a variety of tooth geometries and the implementation potting materials for acoustic signature reduction. An FD approach was used to solve the temperature distribution in the rotor and verified using FEA, Figure 11. The FD approach produced lower maximum temperature values and an error of ~10-20% when compared to the FEA. This error is due to inaccurate assumptions in the boundary conditions (i.e., air-gap heat transfer and the shaft thermal circuit) and will be addressed in the ongoing work. A 2D half-machine model is proposed to capture the thermal transport between the rotor and stator. The R-? section plane at the midpoint in the axial direction is used as the location to convert the 3D model into a 2D model as illustrated in Figure 12(a). Simplifying the R-? section using the adiabatic and electromagnetic line of symmetry common in most SRMs results in the half-machine model illustrated in Figure 12(b) and (c). The ongoing effort will extend the half-machine model to consider rotor slot geometries and include conduction losses in the transient temperature field solution. In addition, pathways to reduce computational complexity will be investigated, including simplifying the rotor to an equivalent disk with an effective density and effective internal heat generation based on total rotor losses.
This work is funded by the Grainger Center for Electric Machinery and Electromechanics
and the University of Illinois SURGE Fellowship.