All harmonic sources are referred to as nonlinear loads because they draw non-sinusoidal currents when a sinusoidal voltage is applied. The non-sinusoidal current may be due to the inherent characteristic of the load (e.g., arc furnaces), or due to a switching circuit (e.g., a 6-pulse converter that forces conduction of currents for only certain periods).
In industrial and commercial power systems there may be many such harmonic sources distributed
throughout the system. The harmonic study requires knowledge of the harmonic currents generated by nonlinear loads.
There are three options open to the analytical engineer:
a) Measure the generated harmonics at each source,
b) Calculate the generated harmonics by a mathematical analysis where possible, such as at converters or static var compensators, and
c) Use typical values based on similar applications or published data.
In practice, all three methods are used and provide reasonable results. Since the system configuration and load continually change, the harmonics also change and it would be a formidable task to study all such conditions.
Usually, the worst operating condition is determined, and the design is based on the “worst-generated” harmonics. However, It needs to be recognized that even with the “worst generated” harmonic case, the harmonic flows within different elements of the network can be different depending upon the number of transformers or tie breakers in service.
This necessitates that for the “worst generated” case, the “worst operating cases(s)” must be analyzed.
One other difficulty in the analysis arises from the fact that when multiple harmonic sources are connected to the same bus (or different buses), the phase angles between the harmonics of the same order are usually not known.
This, generally speaking, leads to arithmetic addition of harmonic magnitudes, which may be reasonable if the harmonic sources are similar and have similar operating load points. However, this approach can lead to a more conservative filter design and distortion calculations, if the sources are different or operate at different load points.
Determination of phase angles of harmonics and vectorial addition can be quite a complex and expensive approach for general industrial application. This is often resolved by simplifying assumptions based on experience or by field measurements.
More advanced techniques are used in high-voltage dc transmission and other utility applications where accuracy is important.
Industrial harmonic studies are usually represented on a single-phase basis, i.e., based on the assumption that the system is balanced and positive sequence analysis applies. A three-phase study is warranted only if the system or the load is severely unbalanced or a four-wire system with single-phase loads exist.
In such a situation it will be very desirable to determine the harmonics generated in all three phases. If the harmonic generation is assumed to be balanced and the system is considered unbalanced, a three-phase study may not serve the full purpose.