Umeå University's logo

An electrical distribution network supplies power to many households at a time. Each house has various types of electrical appliances which consume different levels of power depending on its function and the load it supplies. An ideal case would be when no appliances cause power disturbances, however, in reality, many electrical appliances (e.g. refrigerators, computers, heat pumps, etc.), due to their inherent characteristics. The distribution network is now fed back with non-sinusoidal power, called harmonics, causing inefficiencies and potential failure in other appliances connected to the same network. Harmonics are the additional frequency components to the fundamental sinusoidal voltage or current (e.g. 50Hz fundamental frequency in Europe). To avoid the malfunction of the network, each electrical appliance in any household has to comply with the standard IEC 61000-3-2 (current rating $<$16A) defined by the International Electrotechnical Commission (IEC). According to these standards, current harmonics have been observed up to 40th harmonics. If the fundamental frequency is 50 Hz, then harmonic currents up to 2000 Hz have to be considered. Usually, the harmonics are quantified by the measurement when the appliance has been constructed at which point changes are costly and time-consuming. 

This thesis investigates the possibility to model these appliances at an early stage of development enabling cost-effective and quick improvements. The model is achieved by reducing a heat pump into a set of chokes, an EMI filter, a rectifier, and a load represented by a variable current source to simulate the compressor which draws the majority of power the system uses. The state-of-the-art, referred to as white box modeling, uses schematics to model these components while this thesis presents a methodology to model without schematics, referred to as black box modeling. 

For the black box model, impedance of the components is measured and equivalent circuits are modeled accordingly in Simulink to simulate the harmonics. The validity of a black box simulation model from the equivalent circuit extracted by the impedance measurements is investigated in this thesis. The black box modeling is compared with the white box modeling and measurements for the purpose of verification.  

The result is good but more work is needed. The black box model is verified using two different systems which are similarly accurate. Simulating harmonics at a higher power level yields better results. The black box model is more accurate than the white box model, mainly due to the inclusion of parasitics in the EMI filter. Possible additional harmonics generation in unmodeled components is investigated and found to be insignificant.