A Time-Stepping Coupled Finite Element-State Space Modeling of Sinusoidal- and Inverter-Fed Induction Motor Drives

In this thesis, a comprehensive numerical simulation model is presented for detailed no-load performance analysis of 3-phase squirrel-cage induction motors. Motor operating conditions include sinusoidal voltage, idealized inverter voltage and actual inverter excitations, respectively. The model, under the three excitation conditions given above, incorporates the ability to account for the impact of space harmonics from the machine magnetic circuit geometry and stator winding as well as squirrel-cage layouts (topologies). The model is also able to account for the time harmonics generated from the actual power conditioner, and other time harmonics introduced by magnetic nonlinearities that are inherently present in the machine. Other than neglect of three-dimensional magnetic field end effects and the assumption of material isotropies, the model excludes all of the other conventional simplifying assumptions associated with d-q transformation-based models of induction motors. The model consists primarily of two portions: a finite element (FE) computational magnetics algorithm with energy-current perturbation subprogram, and a computational state-space time-domain algorithm, iteratively coupled to each other through the FE-computed motor winding inductance profiles. The inductance profiles are obtained from a finite element energy-current perturbation subprogram excited in a time-stepping fashion by the winding current waveforms (profiles). These current profiles are in turn obtained from the state-space algorithm. The two resulting iteratively coupled algorithms form the complete unified program, which is henceforth throughout this thesis referred to as the Coupled Finite Element-State Space (CFE-SS) algorithm. The same CFE-SS algorithm is used for both the sinusoidal and idealized inverter voltage excitations. However, for the actual inverter excitation, the statespace portion had to be reformulated to incorporate the inverter network topology, power electronic switching, and machine windings\u27 lumped parameter electrical circuits representation, into one overall global system network, by use of the modified nodal approach (MNA). Also in this thesis, attention has been focused on comparison between effects of the sinusoidal voltage, idealized inverter voltage and actual inverter excitations respectively, on the machine parameters and performance characteristics, at the same level of rms voltage. Specifically, this comparison was directed towards the effects of these excitations on motor core and ohmic losses. It was found that the actual inverter excitation case increased core losses by as much as 164.1% in this 1.2 hp case-study motor, in comparison to their value under balanced 3- phase sinusoidal excitation. Wherever necessary, the inability of the d-q model to rigorously account for space harmonics, time harmonics and the interaction between space and time harmonics has been pointed out throughout this study. The model was verified for the no-load case by comparison with readily available laboratory test results obtained at the same level of rms voltage for the casestudy 1.2 hp motor. Favourable agreement between test and simulation results was achieved