Date of Award

6-2018

Degree Name

Doctor of Philosophy

Department

Electrical and Computer Engineering

First Advisor

Dr. Pablo Gomez

Second Advisor

Dr. Johnson A. Asumadu

Third Advisor

Dr. Richard T. Meyer

Abstract

AC rotating machines are exposed to fast front pulses when fed from inverters. This type of excitation produces dielectric stress in the machine’s insulation system that can result in premature deterioration or even failure.

In this dissertation, a non-uniform transmission line model to predict the transient voltage distribution in a machine coil under steep-fronted surge conditions is described and implemented. This model is a function not only of frequency but also of space to represent the variation of electrical parameters in the overhang and slot regions of the coil. These parameters are computed using the finite element method. In order to increase the computational efficiency of the model, three strategies are proposed to represent the rest of the winding after the first coil as a function of frequency: Kron reduction, equivalent  circuit, and chain matrix. In addition, the amplifying effect of the feeder cable in the transient response of the machine winding is considered including such cable model in the transient simulations by means of a distributed parameter representation in the frequency domain.

Using the machine winding model, an optimization method to identify improvements in the insulation design of machine winding coils fed by fast front excitation is proposed. Different optimization algorithms are applied to identify and evaluate geometrical and electrical modifications to the winding design aimed at minimizing the dielectric stress. Test cases considering different geometrical configurations show that the optimized designs considerably reduce the dielectric stress in the machine coils.

Furthermore, this dissertation proposes a method to reduce the effect of the feeder cable in an inverter-cable-motor setup by using an optimized RLC passive filter. The proposed method consists of finding the optimal tuning of the filter to mitigate the transient overvoltages and dielectric stress in the machine winding. Multi-objective optimization algorithms are used to find the optimal parameters of the filter that minimize the objective functions (rise time, setting time, peak time and overshoot) in the output voltage. The optimal parameter tuning of the proposed RLC filter circuit is compared with a conventional RLC filter when using original and optimized machine geometry, demonstrating a substantial reduction of transient overvoltages while avoiding overdamping and/or distortion of the output voltage.

Finally, the machine winding model is verified experimentally by measurements on a line-end stator coil. The effect of adding the feeder cable in the machine coil and the inclusion of RLC filters are also considered for a complete experimental validation of the modeling approach.

Access Setting

Dissertation-Campus Only

Restricted to Campus until

12-2018

Share

COinS