de-tuned capacitors, PFC and harmonics filters question

[jpcqub]

Evening all,

I’ve a query regarding the use/need for de-tuned capacitors. Plant majority is VSD and there is an active harmonic filtering section to comply with G5/4. Now because we have the harmonic filter, we need de-tuned capacitors in our PFC section. I know it’s something to do with resonance but can anyone outline the finer points, as I’d like to understand why it is necessary (equations, general theory, all accepted!). On another note, our PF is in and around 0.87 which seems rather low, given the amount of VSDs in our system. I thought with so many VSDs in operation it would help to lift the PF?

Thanks in advance

 

[dpc]

Your adjustable frequency drives draw current in a non-linear fashion that in turn creates harmonics on the system voltage. The capacitive reactance goes down as the frequency goes up, so the capacitors can draw excessive current due to high frequency harmonics on the voltage input that they were not designed for. Under certain conditions, you can reach resonance conditions, creating very high voltages across the caps.

Typically, some low-pass filtering is added to the caps or the whole lot is replaced with a tuned filter.

But you probably need to do some actual harmonic measurement to determine what the situation is and if you need to do anything. You might also want to check for blown fuses on any existing cap banks - that's a good warning sign.

The 0.87 pf doesn't sound all that bad - it would be worse without the drives. If you have a lot of small horsepower motors that are lightly loaded, you can have a very low pf. But if you do have a lot of blown fuses on your caps, that could be an explanation of why the pf isn't higher.

I have worked in sawmills with power factors of 0.5.

 

[fsmyth]

I will preface this with the caveat that I have NO
experience with PFC on a large scale. Lots on little bitty.
Feel free to correct me, dpc.

Without going into excessive detail, it probably boils down
to the 60 Hz rating on the caps. Higher frequencies
cause more internal heat, and if high enough, into certain
types, dielectric breakdown. I would venture to guess that
this is the concern if you mean voltage de-rating.

Resonance is a whole 'nother subject - affected by length,
size, and placement of conductors, feed impedance, loads,
and all sorts of other variables. In this case, I suppose
the concern would be value de-rating (less capacity) to
either keep the system well below resonance, or to prevent
over-correction as the load(s) increase.


<als>

 

[Skogsgurra]

fsmyth,

It is good that you said "I will preface this with the caveat that I have NO experience with PFC on a large scale" So we should not take your statement about resonance very seriously.

The resonance in PFC capacators is a very real problem. It is mainly an affair between the feeding transformer's internal (leakage) inductance and the capacitors in the plant. A typical 1500 kVA transformer with an internal (almost purely reactive) impedance of, say, 5.6 percent has a certain inductance, which can be calculated from phase voltage, nominal load, impedance and frequency.

Given this inductance, it is easy to see that there is a very real risk of having resonance between this transformer internal inductance and the PFC capacitors. Especially if there is an aoutomatic PF controller that switches capacitors in and out of circuit. Usually, it is not a problem with fundamental (50 or 60 Hz). It is the harmonics that cause trouble. Fifth, seventh, eleventh and thirteenth harmonics are those that cause most problems.

In a resonance situation, there is series resonance. That means that voltage across the capacitors is increased and also that the corresponding current (fifth, seventh etc) is high. A risky situation that usually ends in capacitors blowing fuses or exploding until a non-resonant situation is established. Or the fire brigade arrives.

To prevent this, so-called detuned capacitors are used. They simply have an inductor in series with them so that the resulting resonant frequency always is lower than the lowest existing harmonic in the system. I.e. lower than the fifth harmonic in a normal mixed load three-phase system.

Google "detuned capacitor", "fine-tuned capacitor", "PFC", "power factor compensation", "transformer inductance" and other such combinations.



Gunnar Englund

http://www.gke.org

 

[davva]

jpcqub,

Not sure how correct this is, but I was once informed that the HF is detuned to account for the effect that ageing can have on capacitance. If designed incorrectly, over time the HF could tune itself such as to amplify unwanted harmonics. The HF filter is therefore often detuned to take account of this (I think!)

 

[fsmyth]

Boy, is my face red. The OP distinctly said "de-tuned",
but I read "de-rated". My only excuse is that, that very
thing was pretty fresh on my mind; I had just replaced
the caps to the ferro-resonant transformers in some 5kVA
UPS's using 660v caps instead of 440v in a attempt to get
more life out of 'em.

skogs, I'll have to stick by both my statements (and grant
that they don't constitute a decent response to his
question) and even have the brass to correct one of yours.
All the PFC solutions I have seen so far (and I /have/ at
one time or another googled all your suggestions, or
something very close) involve caps across the line(s).
This means that you are dealing with a parallel resonant
situation, with PFC caps being used to bring the phase
angle of the load(s) back closer to ideal, offsetting the
load inductance reflected back to the power source, which
is assumed to be neutral. In the absence of load, the
determining impedance is the supply transformer.

Adding inductance (reactors) in series with the capacitor
bank(s) IS series resonant, and will reduce the parallel
resonance. Adding inductance in series with a load, a VSD
in particular, performs a slightly different function -
it smooths the waveform edges inherent in rectified D.C.
supplies (the low-pass filtering that dpc referred to).

Give me the source and load impedances, and the frequency
of concern, and I can tell you /exactly/ how much
capacitance is required. I am still having trouble in
translating some of the terminology used in the power
industry, and readily admit to little practical experience
with anything prefaced with 'M', except ohms.


In an attempt to save a little face, I went cruising a bit.
According to Baron Power (Square-D's docs were MIA):
"Basically the De-Tuned Filters are Tuned to a frequency lower than the predominant Harmonics. The Tuning Frequencies of 136 Hz is followed for De-Tuning Filters, where the predominant harmonics are 3rd Harmonics and Tuning Frequencies of 189 Hz is followed for De-Tuned Filters, where the predominant harmonics is 5th Harmonics."

Near as I can tell, "active filtering" refers to switching
caps and/or reactors in/out as needed for PFC/harmonic
reduction, and "de-tuning" means adjusting PFC cap banks
to avoid over-correction. No??

<als>

 

[Skogsgurra]

well...

???

Gunnar Englund

http://www.gke.org

 

[jpcqub]

There is a harmonic measurement device on the harmonic filters. Not actually sure of the readings, but they cannot go higher than the limits outlined in G5/4. We just assumed if the harmonics were kept low with the filters then the PFC caps would be ok, but now not so.

We did do calculations of the load before hand, and allowed for our VSDs to counteract it, but still not getting the 0.95 we expected. I’ll take a look at them again and I’m guessing we did the calculations at full load, which isn’t really happening at present. So small motors running light are dropping the pf more than we probably had accounted for. Can you explain that a little more for me by any chance?

 

[ScottyUK]

fsmyth,

Active filtering uses a power semiconductor bridge to draw antiphase harmonics from the supply and thus provide local cancellation of the harmonics present on the line. A number of manufacturers make these systems nowadays:

Merlin Gerin's Sinewave range is typical:

http://www.mgeups.com/products/pdt230/harmcond/

The GE document gives some good insight into how they work:

http://www.geindustrial.com/products/specs/ups/DEPDAF.pdf

----------------------------------

One day my ship will come in.
But with my luck, I'll be at the airport!

 

[fsmyth]

[According to MGE]:
' For the UPS families SG Series 400Vac CE and SitePro it is proposed in combination with the passive "5th harmonic filter" installed on the UPS side. The result of this combination passive/active filters gives the best performance in terms of harmonics compensation and power factor correction, with a negligible effect on the total UPS system efficiency.'

' Active Filter can also be used as stand alone for any other power electronic equipment, provided that the load is symmetrical and not supplied by a IT power distribution system.'

[not supplied by a IT power distribution] ????????
Isolated grounding? Gensets? UPS output? What are
they referring to?


" Basically the Active Filter is an IGBT (Insulated Gate Bipolar Transistors) power bridge, which injects harmonic currents into the electrical network with exactly the opposite phase to those that are to be filtered. The reactive power required for the harmonic compensation is supplied by the capacitors installed on the DC side of the power bridge."

[Like I said: switching caps/reactors in/out] <g>

" Power factor correction
The exceeding current the one required to compensate the harmonics pollution is used to perform a partial power factor correction, by injecting reactive currents at the fundamental frequency (50 or 60 Hz); reducing thus the reactive current flowing in the upstream electrical network, limiting therefore the total losses of the electrical system with cost saving in electrical losses and no penalty risk with Public Utilities thanks to increased power factor."

[ This caught my attention ]:
" Active Filter is three wire equipment; therefore it is able to run only with balanced loads."

<als>

 

[fsmyth]

Thanks, Scotty. MGE has a wealth of white papers.

http://www.mgeups.com/techinfo/techpap/articles/genset_en.pdf

(Page 10 gives a pictorial on the various options.)

[An eye-opener (for me!)]:
"subtransient reactance X"d of a generator is approximately 10 to 12%, compared to 3 or 4% for the short-circuit voltage of a transformer. In other words, the output impedance of a genset is 2 to 4 times higher than that of a transformer. Each harmonic order IHk, a sinusoidal current k times the fundamental frequency, is amplified compared to normal operation on utility power with the output impedance of the transformer."



[Probably the best of the bunch; addresses methodology]....

http://www.mgeups.com/techinfo/techpap/articles/0246-e.pdf

" The THDU at the rectifier input depends on the total source impedance and the SCR firing angle delay. The THDU rises rapidly to significant levels, even for low source-impedance values (see figure 6)."

(influence of partial load on THDI)
" The results presented assume a 100% load on the UPS. In fact, the THDI increases with decreasing load. This phenomenon is due to the fact that the harmonic content of the current decreases less quickly than the fundamental. As switching is faster for lower currents, the resulting waveform is squarer and the currents have a higher harmonic content."

" SUMMARY
* installation of a choke in the rectifier input to attenuate the amplitude of the harmonics (in particular high-order harmonics);
* installation of an LC passive harmonic filter in the input of the rectifier bridge, tuned to the harmonic order to be eliminated;
* use of a number of rectifiers supplied with voltages that are slightly shifted in phase to eliminate the most disturbing harmonics, through current recombination. This solution, known as double-bridge or 12-pulse filtering, is in fact a form of harmonic recombination;
* use a THM active filter, derived from active harmonic conditioners such as SineWave, that compensates for harmonics in real time to eliminate distortion."


[ Leibert/Emerson likes to do it with magnetics]:

http://www.liebert.com/support/whitepapers/documents/sl_24220.asp

<als>