In this master thesis a development of software for simulation and optimisation of a squirrel cage induction machine is shown. The simulation itself is based on a finite element method software FEMM. An example of using created software on a machine, that has already been built, is also presented.
In its core the software accesses other, already developed software, where each contributes to the final design and experience and for this reason there is a basic description of this programs presented in chapter 2.
The core of master thesis is presented in chapter 3, where the development of this program, that consists of many modules, is described in detail.
In the initial part the development of graphical user interface (GUI) for parametrical input of machine construction data is shown (whole simulation is based on this parameters). The GUI itself is developed with the help of Microsoft Visual Basic (Express edition) and in general presents interactive connection between the user and all the modules of the software. Once the software has construction parameters in the memory, we can perform an automatic machine construction in an externally used software FEMM (runs in the background). At this point we have to be aware that the machine model is two dimensional and therefore does not comprise all the phenomena that is present in the real machine. For this reason there is a special module intended to partially include extra effects in the model. For example, winding head, short circuit ring, working winding temperature, winding fill factor, etc. are all included in the model as raised specific resistivity of the material.
Once the model is adequate, machine simulation can be performed. In FEMM it is assumed that all models are current fed, whereat in my case I want the model to be voltage fed, so special iterative method (successive approximation) is used to ensure the desired phase winding voltages. Because the model is two dimensional, winding heads leakage and short circuit ring leakage are analytically calculated and subsequently included in the model. Core losses are neglected in the simulation, they are calculated based on the FEMM postprocessor results. Data that we collect during simulation is automatically stored in a text file, for possible further analysis with the help of Scilab.
Based on the data, collected from simulation, the software can calculate and graphically show the equivalent circuit elements in function of slip and all essential characteristics of the machine.
In the last part a simple geometry optimisation of an induction machine is implemented. Optimisation is based on design of experiments. There are two methods of design of experiments, programmed into this software, namely the Taguchi method and central composite design (CCD). In combination with CCD the response surface methodology (RSM) and genetic algorithm based optimisation are also used.
Master thesis is logically concluded with the fourth chapter, where practical use of this software on already developed machine is presented.
The simulation of the existing machine is first performed in FEMM. Then comparison of results between simulation and real laboratory measurement is done and it turns out that produced software sufficiently models and analyzes presented induction machine. This finding is necessary, because there is also an optimisation done on this model and we can expect reliable result, only if the machine model is suitable.
In the first optimisation attempt, the optimisation goal is maximum efficiency at nominal output power where price of the machine and other objectives are not important. In this case it turned out, that it is possible to increase the efficiency for quite a lot (in my experiment for 2,9 %), but there is also a inevitable increase in mass of the machine and consequently in price (price is increased by 16 %). There is also a decrease in power factor.
In the second optimisation attempt also price (minimum) and power factor (maximum) were taken into consideration. In this case efficiency of the machine is not so greatly improved (only for 1,4 %), but on the other hand there is lower increase in price (12,6 %) and better power factor as in the first case.