An Automated Novel Design Technique for Optimal Harmonic Filter Design

Abstract

The paper presents a novel approach in designing liarnionic shunt filters for harmonic filtering, noise mitigation and reduction of harmonic penetration into the host electric utility grid. The technique requires actual ineasurenient of system harmonic impedance. The scheme is based on constrained selective optimization perforinance criteria, of weighted System and filter harmonic currents and the harmonic voltage, for specified continuous or discrete range of dominant liarrnonics of stipulated fi-equencies. The proposed technique cat1 replace the existing trial and error approach with a guided constrained optimization that ensure the best design offering low cost, harmonic reduction and possibly a degree of reactive power coinpensation in the system. INTRODUCTION The priinary goal of series and sliurit passive harmonic filters is to reduce harmonic injection into the host electric utility grid and/or other sensitive type loads. Secondary objectives include reactive conipensation and power factor correction. Series filters presents high impedance path to particular harmonics, hence act as a blockiiig filter. Parallel or shunt type harmonic filters are used extensively in power networks to provide a low impedance path to dominant interfering harmonics generated by the nonlinear static or dynamic load. They are classified into single tuned, double tuned, high pass damped and C-type filters [l-21. Best harmonic filters should ensure maximum absorption of interfering Irarmonics at tniriirnuni cost while ensuring adequate level of reactive power compensation. However, this goal is usually hampered with the ever varying conditions in the system due to topology changes as a result of line switchings, load operating conditions necessitating capacitor energization and deenergization. The classical design philosophy [3-51 employs correction factors to account for the detuning effect due to small frequency variations as well as (L) and (C) Component variations due to temperature and aging. Different criteria do exist for filter design including telephone interference, voltage distortion and current distortion levels or a weighted measure of allowable waveforin distortion [G-81. Usually a composite criteria based on liarnionic voltage and harmonic system currents provide the most cost effective solution. To allow for Complete Control of the filter design process, a new approach based on selective constrained optimization problem is utilized. The proposed method rely on accurate System, calculated or