The importance of using the right detuned filter
In this article, we will explain how the installation of a capacitor bank is in itself a change in the electrical installation; a change in which a poor choice of capacitor bank could destabilise the system due to the harmonics; causing serious problems in the capacitor bank itself and in the installation, with resulting downtime and considerable financial losses.
We will try to explain the different tuning frequencies and the consequences of a poor choice in this tuning, as well as the recommendation for avoiding these possible risks.
The search for improved energy efficiency and the increases in electrical tariffs are making power factor correction with capacitor banks more and more common. However, like any electrical unit, these capacitor banks have several electrical effects on the installation where they are installed. The most important effect, as well as correcting the installation's reactive energy consumption, is the change in the behaviour with respect to the harmonics present in the electrical network. This change may have a negative impact on power factor correction in the medium-term , an electrical destabilisation in the installation, or even production downtime.
Electrical installations are becoming more and more complex, and include different inductive loads, capacitive loads and power electronics. These networks often contain significant levels of harmonic distortion, which has led the large majority of manufacturers of automatic capacitor banks to unanimously include units specifically designed for use in such networks in their catalogue.
However, where there is no such unanimity is in the choice of the tuning frequency offered as standard, both in automatic capacitor banks and in fixed compensation units fitted with detuned filters.
For the far less common case of predominant 3rd order harmonics (150 Hz in 50 Hz networks), the use of detuned filters tuned at 134 Hz is more common (overvoltage factor of p = 14%); but for the large majority of installations, where a capacitor bank has to be fitted with detuned filters that is appropriate for 5th order harmonics (250 Hz in 50 Hz networks) or higher, which are normally produced by the more usual harmonic current sources, in other words, three-phase loads fitted with a 6-pulse diode rectifier bridge in its input: speed or frequency drives, AC/DC rectifiers, induction ovens, etc., the variety of proposed tuning frequencies is significantly varied, generally moving within a range of between 170 and 215 Hz (p = 8.7% to p = 5.4%).
However, there are two tunings that stand out over the rest: those corresponding to an overvoltage factor of p = 7% (tuning frequency of 189 Hz in 50 Hz networks) and p = 5.67% (tuning frequency of 210 Hz in 50 Hz networks).
It could easily be deduced from the above that the choice of a value of p = 7% or p = 5.67% might be indifferent and that both should give the same result for the effects of their behaviour when they are connected to the electrical network, but this is not strictly true.
To follow the arguments of this last comment, we will briefly go through the operating principle of detuned filters. Observing the la impedance-frequency graph of a standard reactor-capacitor unit with p = 7% (green line in Fig. 1), we see that it offers least impedance at 189 Hz, whereas that corresponding to p = 5.67% (red line in Fig. 1) offers the least impedance at 210 Hz. In both cases, the impedance gradually increases on either side of it, with the particular feature that the impedance is capacitive at frequencies under 189 Hz, and inductive at higher frequencies. It is precisely this inductive character with harmonic frequencies of the 5th order or higher that prevents the possibility of a resonance phenomenon being produced at any of those frequencies. However, another key parameter for the correct operation of the detuned filter is the value of said impedance at the different harmonic frequencies. Therefore, at said impedance-frequency in Fig. 1 the impedance difference of each tuning can clearly be seen at a harmonic frequency of 250 Hz which, we will remember, is the predominant frequency of the voltage and/or frequency harmonics in the electrical networks. For p = 5.67%, the value of the impedance is practically half of the value for p = 7%.
What is the main consequence of this impedance difference shown by both tunings? It is easy to deduce that the absorption of harmonic currents in the mains will be higher for p = 5.67% than for p = 7%. This could be understood as beneficial for the installation if it were deduced that the 5th order harmonic current level upstream of the connection point of the capacitor bank to the mains will be lower as compared to that which there would be with a similar power capacitor bank by p = 7% type tuning; however, both experience and the reality of the nature of the majority of networks, which is far from what would be ideal network behaviour, mean that this perception is not correct on a large number of occasions.
The use of passive harmonic filters is a subject which always requires a minimal preliminary study, as their behaviour depends on the network features. Therefore, the aim to compare the use to a certain extent of a filter tuned to 210 Hz with that which one would have tuned to 225 Hz, which is the normal frequency of absorption filters for 5th order harmonic currents in 50 Hz networks, should also have said consideration, and this is rarely so. Briefly, it is more unpredictable to determine the actual harmonic current consumption that a capacitor bank can have with p = 5.67% type filters than that which an identical one would have with p = 7%, when both are installed in the same network.
There are also other points to be considered. One basic point is the fact that if, to start with, that of p = 5.67% is going to have a larger harmonic current consumption, its elements, principally the reactor and the associated capacitor, must be designed to withstand the overload to which they are to be subject on the level of intensity and temperature; and here we are faced with one of the main problems of these filters. In the particular case of the reactors, these, at an equal power of p = 7%, and, if the design criterion has been based on this value, the result is a smaller and lighter reactor, or a lower cost, and the same temptation can be applied to the capacitors, in the sense that the overvoltage to which they are subject will be 25% smaller than if p = 7%, and therefore the use of capacitors of a lower rated voltage may be justified. In short, there is a risk that the capacitor bank might have to withstand higher levels of harmonic overloading with weaker elements, which would inevitably cause faster wear than in the similar element of p = 7%.
The other essential point to be considered, which is the most important in the opinion of CIRCUTOR, is the influence of the capacitor capacity in tuning the reactor-capacitor series group according to the formula in Fig. 2.
It is easy to deduce that a decrease in the capacitor capacity will result in an increase in the unit's resonance frequency. Capacitors are elements that lose capacity with time either due to their conditions of use (voltage, temperature, connection operation rate, etc.), or due to the natural deterioration of the polypropylene of their dielectrics. A same loss of capacity in a p = 5.67% filter and in one of p = 7% , means that the first will come much closer to the 5th order frequency than the second, and the closer it comes, the greater harmonic current absorption it will present, the greater overloading it will suffer, leading to greater deterioration. In other words, the safety margin given with this loss of capacity is considerably higher in a filter with p = 7%.
The conclusion in this case is clear, and is CIRCUTOR's unequivocal recommendation of the use of filters with p = 7% instead of p = 5.67% in all installations where they have to be applied due to the level of harmonic distortion.
The purpose of this recommendation is none other than to reduce the obvious risk that a loss of capacitor capacity could cause the appearance of serious problems as a result of overcurrents in the capacitor bank much earlier, allowing a longer reaction time through the pertinent maintenance actions that are always recommendable in any unit and the application of corrective measures before the damage is definitive and, therefore, worse economic conditions.