Discussion on the Engineering Design Method of Harmonic Suppression

1 Introduction

With the widespread application of power electronic equipment such as high-power semiconductor power converters and frequency converters, more and more harmonic currents are injected into the power grid. Since the nonlinear working characteristics of power electronic devices determine the lag of fundamental current, and the negative impact of harmonics is becoming more and more serious, how to effectively suppress harmonics is an important part of power design.

2 Harms of Harmonics

(1) Increased reactive power consumption and copper loss

In the case of current waveform distortion, the apparent power of the power system should be:

S2=P2＋Q2＋T2(1)

Where: S is the apparent power;

P is active power;

Q is reactive power;

T is the distortion power.

Since the frequencies of harmonic voltage and current are different, their phase difference changes periodically with the frequency difference, and the sum of the accumulated power is zero, so the distorted power has the properties of reactive power.

Harmonic current will cause harmonic copper loss, harmonic stray loss and harmonic iron loss in components in the power system, such as motors. The existence of harmonic loss increases the total loss of the motor, increases the temperature rise and reduces the efficiency. The motor will absorb more reactive power, resulting in a decrease in power factor.

(2) When a voltage containing high-order harmonics is applied to both ends of a capacitor, since the impedance of the capacitor to high-order harmonics is very small, the harmonic current is applied to the fundamental wave of the capacitor, which increases the total operating current of the capacitor and increases the temperature rise. It is easy to overload and even damage the capacitor, resulting in a shortened service life. At the same time, harmonics affect the matching of capacitor parameters, which may cause high-order harmonic resonance in the power grid, exacerbating the fault.

(3) The trigger angle is offset due to control system errors caused by harmonics

The current and voltage change rates are too high, causing thyristor failure, and even causing control failure and malfunction of the converter and automatic control device, which in turn causes system failure.

(4) Continuous high harmonic content will accelerate the transformer and motor

The insulation of the machine and power cable is aged, making it easy to be broken down. In some cases, especially in transient processes, it may also cause resonant overvoltage.

(5) Harmonic voltage and harmonic current pass through the inductive coupling between lines.

When combined, considerable harmonic voltage will be induced in the communication line, thus interfering with the communication line and affecting the normal operation of the communication network.

3 Harmonic suppression measures

According to the requirements of GB/T14549-93 "Harmonics in Public Power Grids for Power Quality", the harmonic voltages and currents injected into the power grid by various nonlinear loads must be limited.

In the design of the power system, increasing the short-circuit capacity of the system, improving the power supply voltage level, increasing the pulsation number of the converter, and improving the operation mode of the system, such as keeping the three-phase load balanced as much as possible, avoiding saturation of various electromagnetic systems, staggering the system resonance point, and supplying power to the harmonic source load by a special circuit, can all reduce the harmonic components in the system. However, many of these measures will greatly increase the investment in the system and equipment, and the effects of some methods may not be ideal. Therefore, setting up an AC filter is an active measure to effectively suppress harmonics and improve the waveform. At the same time, the filter can also provide part or all of the required reactive power to the system.

Figure 1 High-order harmonic equivalent circuit

(a) Wiring system (b) Equivalent circuit

Figure 2 Single tuned filter

(a) Structure (b) y=f(δ) Characteristics

Figure 3 High pass filter

(a) Structure (b) Impedance frequency characteristics

Non-linear loads such as rectifiers and inverters can be represented as constant current sources that generate high-order harmonic currents, so Figure 1 can be used to represent the equivalent circuit of high-order harmonics.

The harmonic current IS flowing to the grid and the harmonic voltage VB of the busbar can be expressed as:

IS=InZL/(ZS＋ZL）

VB=ISZS(2)

Where: IS is the harmonic current injected into the grid;

In is the harmonic current;

VB is the harmonic voltage;

ZS is the grid impedance;

ZL is the grid load impedance.

This formula shows that when the grid impedance (ZS) is constant, the harmonic current flowing to the grid and the harmonic voltage (voltage distortion) of the busbar can be reduced by relatively reducing the system load impedance (ZL). Harmonic interference depends on the magnitude of the harmonic current or voltage distortion flowing to the grid. The purpose of harmonic suppression is to reduce the harmonic current flowing to the grid.

Therefore, the following two measures can be taken:

(1) For the power system, a harmonic low-impedance shunt circuit is set up to reduce the load impedance ZL and the harmonic current IS injected into the power grid;

(2) Provide harmonics of opposite phase to offset the harmonic current In generated by nonlinear loads, thereby achieving the purpose of eliminating harmonics.

The former is called a passive filter, which is the commonly used LC filter; the latter is called an active filter, which is an active filter.

4LC filter design

The LC filter uses the L-C resonance principle to artificially create a series resonant branch, providing a very low impedance channel for the main harmonics to be filtered out, so that they are not injected into the power grid. According to the different connection methods of the capacitor and the reactor, the main commonly used ones are single-tuned filters and high-pass filters. Their structures and impedance characteristics are shown in Figures 2 and 3.

The resonance number and quality factor of a single tuned filter are: Qn=XLn/Rfn(3)

The harmonic impedance is:

Zfn=Rfn＋j(nXL1－XC1/n)≈Rfn(1＋j2δQn)(4)

In the above two formulas: XC1 is the fundamental capacitive reactance of the capacitor bank;

XL1 is the fundamental inductive reactance of the reactor;

XLn is the inductive reactance of the reactor at the nth harmonic;

Rfn is the resistance of the filter at the nth harmonic;

δ is the relative deviation of the grid angular frequency.

Due to the fluctuation of system frequency, the deviation of relevant parameters of filter capacitors and reactors during manufacturing, the adjustment deviation of reactors, and the changes of ambient temperature and load, the actual resonant frequency of the filter may not be exactly the same as its design value, that is, it varies within a certain range of deviation from the design value. In general, the single-tuned filter has the best filtering effect when Qn=1/2δ, that is, the harmonic current injected into the power grid is the smallest.

Figure 4 Typical structure of LC filter

As shown in Figure 2(b), the filtering effect of a single tuned filter is directly related to δ and Qn. The larger the Qn, the sharper the curve, but the easier it is to detune, and the faster the filtering effect decreases; when Qn is too small, the filtering effect does not change much within a large range, but the effect is low, and the loss is also large. Therefore, the determination of Qn and δ requires comparison of multiple schemes and selection after taking into account various indicators.

For high-pass filters, since the reactor L is connected in parallel with the resistor R, there is a lower impedance frequency range. When the frequency is lower than a certain cut-off frequency f0 (fO=1/2πRC), the filter impedance increases significantly due to the increase in capacitive reactance, and low-order harmonic currents are difficult to pass through; when the frequency is higher than f0, the total impedance does not change much due to the small capacitive reactance, forming a passband.

In contrast to the single tuned filter, its quality factor is Qn=Rfn/XLn. This is because in the high-pass filter, the resistor R is connected in parallel with the reactor L. The larger the resistance, the sharper the tuning; while in the single tuned filter, the resistor R is connected in series with the reactor L. The smaller the resistance, the sharper the tuning. However, whether it is a single tuned filter or a high pass filter, the quality factor is an indicator of the sharpness of the tuning. For the high pass filter, the Qn value is generally 1 to 5. As can be seen from Figure 3(b), even near the tuning frequency, the frequency deviation has little effect.

The high-pass filter cutoff frequency should be chosen close to the main harmonics to be filtered, otherwise its loss will increase significantly.

For a certain harmonic, to achieve the same filtering effect, the use of a single tuned filter will greatly reduce the capacity, but the high-pass filter has a comprehensive filtering function, which can filter out several high-order harmonics at the same time and reduce the number of filtering circuits. Therefore, when selecting a filtering scheme, for the main harmonics, it is appropriate to use a single tuned filter; and for several higher harmonics, and the harmonic current value is not large, it is appropriate to use a group of high-pass filters. When combined with the required reactive power compensation capacity, in many cases, using several groups of single tuned filters plus a group of high-pass filters is a more economical and feasible solution.

Figure 4 shows the typical structure of an LC filter used in a tinplate factory.

From formula (2), we can know that the filtering effect of LC filter depends on the relationship between power supply impedance and filter internal impedance. Since the filter is connected in parallel in the circuit, it is an impedance factor and is easily affected by the existing high-order harmonic distortion of the power supply. Therefore, the following factors should be fully considered in the design:

(1) Impedance conditions of the power supply. Calculate the harmonic impedance of the system based on the system wiring, transformer parameters or bus voltage and short-circuit capacity at the location where the filter is to be installed; the equivalent frequency deviation formed by factors such as the power grid frequency fluctuation range and the adjustment deviation of the filter capacitor and reactor;

(2) Within the power frequency range, the filter and the capacitor have the same function, coordinating the system's leading phase capacity, thereby effectively reducing the filter capacity and reducing the filter cost; the existing high-order harmonic voltage of the power grid may cause overload effects on the filter; the amount of high-order harmonics generated by the converter load determines the filter rating;

(3) High-order harmonic suppression index: According to the provisions of "Power Quality Public Grid Harmonics", determine the voltage distortion rate of each harmonic and the allowable value of the harmonic current injected into the grid of the corresponding voltage level.

The LC filter has a simple structure and a significant harmonic absorption effect. However, due to its structural principle, it has some insurmountable defects in application:

(1) It has a good compensation effect only on the harmonics of the natural frequency.

The compensation effect is poor when the harmonic components change;

(2) The compensation characteristics are greatly affected by the grid impedance;

(3) At a specific frequency, the impedance between the grid and the LC filter

Parallel resonance may occur, causing the harmonic current of that frequency to be amplified; or series resonance may occur, causing the harmonic voltage that may exist on the grid side to inject a larger harmonic current into the LC filter;

(4) When other harmonic sources connected to the power grid are not filtered, their harmonic currents may flow into the filter and cause overload.

The active filter can quickly track and compensate for the changing harmonics, basically overcoming the above-mentioned shortcomings of the LC filter.

5 Applications of Active Filters

With the development of power electronic devices and PWM technology, the instantaneous detection method of harmonic current based on instantaneous reactive power theory has been proposed, which has led to the rapid development of active filters.

As mentioned above, the LC filter is actually composed of filter capacitors and reactors, which presents a low impedance resonant branch to certain harmonics or certain harmonics, and filters these harmonics. The biggest difference between the active filter and the LC filter is that it is an active filter device that injects compensating harmonic current into the system to offset the harmonic current generated by the nonlinear load. It can quickly and dynamically track and compensate for the changing harmonics, and the compensation characteristics are not affected by the system impedance. Its structure is composed of a static power converter, and has the high controllability and fast response capability of semiconductor power devices.

Figure 5 Active filter working principle

The working principle of the active filter is shown in Figure 5.

The load current IL can be expanded according to the Fourier series as follows:

IL=ΣInsin(nωt＋θn)

=I1cosθ1sinωt+I1sinθ1cosωt+ΣInsin(nωt+θn)

=I1p＋I1q＋In(5)

Where: I1p is the load fundamental active current;

I1q is the load fundamental reactive current;

In is the high-order harmonic current.

Connect the filter in parallel between the harmonic source and the power supply, Is=IL+IF. Control the output current IF of the active filter to be -In, and the current on the power supply side is a sine waveform containing only the fundamental component. That is, the active filter generates a current with the same amplitude and opposite phase as the load harmonic current and injects it into the line through which the load current IL flows, offsetting the load harmonics and preventing them from flowing into the power grid. It can be seen from formula (5) that the active filter can also compensate for reactive power at the same time, that is, IF=-I1q-In, IS=-I1p, thereby improving the system power factor.

The basic structure of the active filter consists of harmonic current detection, control circuit, PWM inverter, DC power supply and injection transformer. According to the different energy storage elements of the inverter, the active filter can be divided into current type and voltage type. The energy storage element of the current type active filter is an inductor. Due to its large operating loss and complex charging control of the energy storage inductor, its application is limited; the energy storage element of the voltage type active filter is a capacitor, which has the advantages of low loss and easy control and is widely used. The working process of the voltage type active filter is that the capacitor constitutes an energy storage DC power supply. The inverter generates a PWM output voltage according to the detection signal, converts the DC power stored in the capacitor into a compensation current of the required frequency and waveform, and injects it into the line through the isolation transformer. The PWM inverter also has the function of providing DC power to the reactor or capacitor. This process is directly controlled by the harmonic current compensation detection and control circuit.

Active filters have the following characteristics:

(1) The device is a harmonic current source, and its access to the system

Impedance will have no effect;

(2) When the system structure changes, the device does not produce harmonics.

The risk of vibration does not affect the compensation performance;

(3) There is no overload problem.

When the compensation capacity of the device is exceeded, the filter can still play the maximum compensation role;

(4) All harmonics in the system can be effectively suppressed;

(5) One device can realize multiple harmonics and fundamental reactive power

Real-time dynamic tracking compensation of streams.

However, compared with LC filters, active filters have a relatively complex structure, large operating losses, and high equipment costs. Since active filters themselves work in a switching mode, while compensating for harmonics, they also inject new harmonics, but their switching frequency is very high (up to 3kHz or more), so the harmonic frequency is high and the amplitude is low.

Active filters can be used to suppress high-order harmonics with periodic load changes and some high-order harmonics that cannot be suppressed by LC filters. Table 1 compares the two wiring methods of active filters.

Direct access is the basic connection method between active filters and systems. At this time, the PWM inverter is equivalent to a controlled current source, which generates harmonic currents that are equal in magnitude and opposite in phase to the load harmonics, so that the current on the power supply side is compensated to be sinusoidal. In this mode, the fundamental voltage of the power supply is all added to the inverter, so the device capacity is large. The filter of this connection method has the function of continuously adjusting reactive power, and can dynamically compensate for system reactive power while compensating for harmonics.

The active filter of the injection circuit mode uses the reactor and capacitor as the inverter injection circuit, and uses the resonance characteristics of inductance and capacitance to make the active filter not bear the fundamental voltage, thereby reducing the device capacity of the inverter, reducing the volume and reducing the cost. By selecting the injection circuit constant, the device capacity of the inverter is only 1/4 to 1/5 of the direct access method, so it is suitable for large-scale filtering devices that constitute high-voltage circuits.

Active filters can also be used in combination with LC filters in parallel or series to form a hybrid structure. When used in parallel, the LC filter is used to share the compensation of harmonics of the same order, supplement the compensation effect of the active filter, and reduce the capacity of the required inverter. When used in series, the active filter is not mainly used to directly compensate for harmonics, but to suppress the parallel resonance between the LC filter and the grid impedance, the so-called harmonic amplification phenomenon, to improve the compensation effect of the LC filter. At this time, the inverter does not bear the fundamental voltage and the device capacity is small.

Figure 6 Active filter principle circuit diagram

Figure 6 is the principle circuit diagram of the active filter produced by Meidensha Co., Ltd. The device can effectively filter out 2nd to 19th harmonics, with a harmonic suppression rate of more than 85% and a dynamic response time of less than 1ms.

6 Conclusion

(1) When controlling system harmonics, the influence of various factors in the system should be fully considered, various indicators should be taken into account, and a reasonable and effective filtering solution should be selected;

(2) When using LC filters, the comprehensive filtering effect of the filter group should be taken as the principle, and the occurrence of harmonic amplification should be strictly avoided;

(3) The selection of the capacitance of the filter capacitor should not only meet the filtering requirements, but also take into account the need for reactive power compensation, and should also enable the capacitor to withstand the effects of overcurrent and overvoltage;

(4) Active filter is a new type of dynamic filter, and its harmonic suppression capability is much better than that of LC filter. With the increasing attention paid to the harmonic problem of power grid and the gradual reduction of its cost, it will have broad application prospects.