In today's electrical networks, power quality is important, as companies may not be able to provide adequate products when power quality decreases. Power quality is divided into several parameters, and one of them is voltage and harmonic distortion. New devices are daily installed in the electrical network, which causes harmonic distortion – they are either a harmonic distortion source or cause resonance with passive devices or compensation. Such devices are most often found in industrial consumers which have a higher-rated power and implemented reactive power compensation. Some already have an implemented active filter that compensates for harmonic distortion. In this case, only its algorithm can be updated with the newly proposed compensation function of the active filter, which determines its reference based on consumer’s contribution. The proposed compensatory function is also the main contribution to the findings of this PhD thesis.
In the first part of the PhD thesis, we focus on an overview of the theoretical background. The overview includes a general description of harmonic distortion, vector representation of three-phase quantities, sources of harmonic distortion and determination of harmonic impedance. It also describes how harmonic distortion can be reduced by different approaches. In the electrical power system, odd harmonic components are mostly present, while even harmonic components are present only during transient phenomena. Odd harmonic components are divided into three characteristic groups, which are defined by the sequence of individual harmonic components: zero, negative and positive harmonic components. They are mostly determined by fast Fourier transformation, but other approaches are used for control algorithms, such as dq-transformation, which illustrates real-time quantities in the vector space at a given frequency. This frequency is determined by a harmonic component, as the vector space needs to be synchronized to the analyzed harmonic component. The DC value of this transformation presents values of the synchronized harmonic component and is determined by low-pass filters.
Sources of harmonic distortion are also important in determining the consumer's contribution to total harmonic distortion. Harmonic distortion sources are divided into two major groups, namely commercial and industrial loads. Consumer loads mostly have lower power, and the consumer does not usually have built-in reactive power compensation. Industrial loads include larger three-phase power converters, arc furnaces and induction motors. In most cases, the consumer decides to compensate for reactive power if they have implemented devices with a higher power. In such cases, it is necessary to determine the location of the harmonic distortion source due to resonance amplification. Harmonic distortion can also be amplified by passive compensation of reactive power if it is not properly dimensioned or tuned.
To properly determine the consumer’s contribution, it is necessary to determine harmonic impedances, separately for the network and the consumer. The determination of harmonic impedances in the PhD thesis will be based on reference impedances, as the actual harmonic impedances are practically impossible to determine in a real electrical environment.
In most cases, it is necessary to reduce harmonic distortion, either because values defined by standards are exceeded or due to the provision of better-quality products. Harmonic distortion reduction can be done in several ways: by changing control algorithms of drives or passive compensators, or by active filters. Harmonic distortion is most commonly reduced using passive filters tuned to a specific harmonic component or active filters, but in the PhD thesis we will focus on active filters. Active filters can be connected to the network in parallel or in series. In the PhD thesis, we will focus on the parallel connection of active filters, since the active filter directly affects the current harmonic distortion in this connection because the reference of the active filter is based on the consumer’s current harmonic source. Nevertheless, the active filter can be connected to the mains in several different combinations with passive compensation. In the PhD thesis, we will analyze three combinations of active filter connection, namely parallel and serial connections to passive compensation and a hybrid connection, where the active filter is connected between the inductor and the capacitor.
In the main part of the PhD thesis, we focus on the new compensation function of the active filter, which is the main original contribution to science. The second main part of the PhD thesis presents the development of the compensation function of the active filter in various combinations of active and passive parts, which is the second original contribution to science. At the beginning of the main part, we provide an overview of previous implementations and methods of harmonic distortion compensation, where we found that no algorithm determines the reference of the active filter based on the consumer's contribution to total distortion.
Harmonic conditions at the consumer’s point of common coupling change with the operation of the active filter, so it is necessary to determine what harmonic distortion is like if the active filter does not work. The determination of this harmonic distortion is based on harmonic impedances, the measured values of voltage and current harmonic distortion at the point of common coupling as well as harmonic current of the active filter.
The determination of the active filter reference based on the consumer's contribution to total harmonic distortion is detailed in the following part. Determining the active filter reference is based on a mixed equivalent model, as the network part is presented as a voltage harmonic source or Thevenin's model and the consumer’s part as a current harmonic source or Norton's model. The active filter compensates only for the consumer’s emission of harmonic distortion at point of common coupling due to the consumer’s harmonic source. Consumer’s emission does not include the share of the harmonic current caused by network voltage harmonic distortion on the network side, which is named reference of the network harmonic source. The reference of the network harmonic source also shows the volume of the harmonic current that still flows at the point of common coupling. The active filter reference is the share of the harmonic current running through the network impedance, caused by the consumer’s harmonic current source. Since the reference of the active filter is in the vector space, a projection of its vector on the vector of the measured value was made, as this can directly affect the measured value of the current harmonic distortion.
The active filter reference is implemented in three different combinations of active and passive parts of the compensation or filter. When implementing the active filter reference, some changes are required in determining the parameters and calculating the active filter reference, as not all combinations have the same parameters available. The implementation of the active filter reference is also analysed based on the equivalent model, which presents how the active filter works under ideal conditions. With a parallel combination of passive and active parts of the filter, the active filter is easy to implement in an existing compensator, as it is only additionally connected to the passive compensation. In the series combination, the current of the active filter is equal to the current running through the passive compensation. Therefore, it is necessary to adjust the reference of the active filter appropriately, as the current of the passive and active parts of the compensation needs to be separated. In the hybrid combination of active and passive parts of the compensation, it is necessary to perform additional voltage measurements, as the active filter is between the inductor and the capacitor.
The last part of the PhD thesis describes the results of the analysis and simulations. The simulations were performed in the digital programming environment with the PSCAD software, the DSP development platform and a real-time simulator. In the PSCAD software environment, a test power network with added active filters was modelled. In a parallel combination of active and passive compensation, several simulations were made. The results of example 1 show that it is necessary to compensate the total harmonic current, as the harmonic current is the result of detuned passive compensation and nonlinear load. In example 2, simulations were made at different tuning values of passive compensation, since the consumer reduces the voltage harmonic distortion in the original test network.
The results with the series connection of the active part to the passive compensation are comparable with the results of parallel connection of active part to the passive compensation because it only changes the consideration of the passive part of the compensation. In this case, the impedance characteristics are similar and this is the reason why the results are comparable. The results of the hybrid connection of the active part to the passive compensation are reduced comparing to the results of the parallel combination for 4 % at 11th and 13th harmonic component, 14 % at 7th harmonic component and 30 % at 5th harmonic component. The active filter current is amplified by the passive part of the compensator and this is the reason for reduction of the required active filter current compared to the other two connections. From the comparison of the results, it can be concluded that the reference of the active part is correctly determined.
The proper operation of the active filter reference is also confirmed with the comparison of the analysis and simulation results because the results are comparable. Using the suggested active filter reference, the harmonic current of the active filter could be decreased by 5% to 40% at a separate harmonic component based on total current compensation. The active filter harmonic current could be additionally decreased applying the voltage condition, which could decrease the harmonic current by over 80%. In case of such decrease in the harmonic current, a correctly detuned passive compensator must be installed, otherwise the consumer needs to compensate total harmonic current at the point of common coupling.
|