An evaluation of real-time power calculation techniques in non-sinusoidal conditions

The growing incorporation of power electronics into the electricity distribution grid presents considerable difficulties for measuring real-time power. This is due to the injection of non-sinusoidal currents by various sources such as renewables, storage systems, and electric vehicle chargers. Although the significance of precise power measurement for various purposes such as billing, fault detection, and grid support is well recognized, current techniques face challenges to accurately quantifying power other than active power in distorted conditions. Existing solutions rely on the assumption of voltage and current pure sinusoidal behavior. Others, based on spectral decomposition, produce lagged or time-sparse results. This work aims to fill the existing gap in knowledge by examining different methods deemed capable of calculating power in real-time, and confront them against the IEEE 1459 standard, which is the most recent effort to clarify reactive power offline measurement. The focus is on assessing their computational requirements and accuracy in both steady-state and transient tests. Conclusively, the Fryze's approach was found to be the most effective in producing a result that exactly matches the “non-active power” defined by the IEEE standard with maximum relative error of 0.5 %. Nevertheless, it exhibits slow response (more than one fundamental cycle) in scenarios where reactance is dominant, and it incurs the highest computational expense, requiring up to 23.5 % CPU usage, well above the 1.32 % and 2.49 % of the Hilbert Power and Instantaneous Power counterparts. This discovery identifies important traits that future methods should consider considering the increasing prevalence of renewable energy and nonlinear loads.