Quantitative comparison of power decompositions

This paper presents a comparison on power decompositions in a simple single phase circuit with nonsinusoidal waveforms of voltage and/or currents by giving particular emphasis to Reactive Power compensation. The experimental circuit is analysed by using exact analytical expressions for current and voltages determined via considering source impedance and nonlinearity, which is introduced due to supply side harmonics. Results demonstrate that; power decompositions proposed by Kusters and Moore, Fryze, Shepherd and Zakikhani, Sharon, and Czarnecki provide correct information regarding Power Factor improvement with passive compensation in nonsinusoidal voltage source–linear load, nonsinusoidal voltage source–nonlinear load and sinusoidal voltage source–nonlinear load cases. In these cases, Reactive component of Kusters and Moore's power decomposition can completely be compensated when Power Factor is maximum if there is no resonance or significant changes on load voltages in the case of compensation capacitance is inserted. The Reactive components of Fryze, Shepherd and Zakikhani, Sharon, Czarnecki's power decompositions attain minimum value when power factor is maximum. Furthermore, Kusters and Moore's Reactive Power could directly be related to the power of optimum compensation capacitance. On the other hand, power decompositions proposed by Budeanu, Kimbark and Depenbrock do not provide any useful information about optimum Reactive Power compensation with a basic capacitance in the cases nonsinusoidal voltage source–linear load and nonsinusoidal voltage source–nonlinear load although they can completely be compensated in these cases. An important observation is that; Distortion Powers of Budeanu and Kimbark, and Depenbrock's Residual Power have compensable parts; on the other hand, Kusters and Moore's Residual Reactive Power, Sharon's Complementary Power, Depenbrock's In Phase Power and Czarnecki's Scattered Power are almost constant when the power of compensation capacitor is varied.