Degree Name

Doctor of Philosophy


Department of Electrical and Computer Engineering


This thesis is concerned with some important aspects of harmonic management in a multi-voltage level power system including transmission, sub-transmission and distribution levels. In the work presented both deterministic and statistical approaches for harmonic analysis have been used. Investigations have been made both in Time and Frequency Domains. An already available computer simulation program in Time Domain has been used to derive frequency domain models of distribution feeders with composite loads. A Frequency Domain computer program has been modified to give the statistical data of harmonic voltage distortion levels of a large power system.

The thesis examines some of the fundamental assumptions used in harmonic analysis with the aid of a Time Domain simulation program. In harmonic analysis two of the commonly made assumptions are: (i) harmonic sources are represented with ideal ac current sources having equal phase angles, (ii) non-distorting loads are represented with constant impedances.

The thesis also discusses the harmonic current distortion level of distribution feeders with different load categories (e.g., residential, commercial and industrial) based on measurements data. This data is used for modelling of harmonic sources in Frequency Domain computer program.

The major harmonic management questions addressed in this thesis are: i(i)setting of limits for harmonic voltage distortion at different voltage levels, and i(ii) the determination of the maximum permissible sizes of several new distorting loads.

Existing harmonic standards give a different limit for each voltage level. If these limits are not well co-ordinated then there will be unnecessary restrictions on harmonic injection by distorting loads. A computer-based technique which can be used to investigate this problem has been developed. The approach is statistical in nature and it accounts for variations in the Space Domain (at different busbars in the system), the Time Domain (with different capacitor connections and line outages) and combinations of both Time/Space Domain.

Variation of the statistical data with respect to the voltage level is investigated and the suitability of the 9 5 % confidence level for assessment of the harmonic voltage distortion level at each voltage level is discussed. In addition, the probability distribution as affected by the strength of the power system is investigated. Histograms and probability distribution curves are used to find the best statistics for describing the harmonic voltage distribution. The effect of distorting loads in different parts of a power system on the harmonic distortion levels in a particular subsystem is also investigated.

With existing harmonic management techniques, there is no guarantee that, after connection of a new distorting load, the harmonic voltage distortion level remains within limits at all voltage levels. O n the other hand there is no method to determine the permissible size of several new distorting loads which can be connected at different voltage levels and locations simultaneously. Two techniques which use constrained and unconstrained non-linear optimization program have been developed to overcome the above restrictions. These techniques have been applied to a typical power system to demonstrate their application. The effect of non-distorting loads and system capacitance is discussed. The effect of the harmonic voltage limit at different voltage levels on the maximum permissible harmonic current injection is also investigated.