Power Factor Correction with Software 4

[jimgineer]

All,

This is something that I am new to, but I have heard this mentioned as something that is possible to do.

It doesn't make much sense to me purely being able to do PF correction with software - I am imagining that all the software does is act in a control loop to control the amount of VARs introduced into the circuit, depending on the load.

In other words, it would simply determine what the lag of the current waveform was to the voltage waveform, and depending on that lag, vary the parallel/series combination of some capacitance to get the correct amount of VARs in the system to get the PF close to 1.

I have a couple of questions here:

1) Can anyone think of a way to impact the pf of a system without having any capacitor bank or introducing some imaginary component of impedence to the load? Or is this chasing a white rabbit?

2) What practical implementations have actually been implemented with this? For example, in a substation does a utility generally assume a typical 'range' of pf that it is feeding, and then just size a bank of capacitors appropriately, or is there actually any control logic going on there to try to get the pf as close to 1 as possible, that can vary the VARs introduced based on what loads and the pf of the load. If not, why not? I'm thinking this wouldn't be too expensive to implement, although there might be some challenges with power quality.

And, is this kind of related to switch mode power supply design? I never took any classes on those but I imagine that there are similiar design challenges feeding electonics as with large scale power systems. (Matching load impedence with source impedence I would imagine is a large consideration for SMPS?)

Thanks in advance

 

[Skogsgurra]

First, you have to understand the difference between "classical" power factor or displacement power factor and distortion power factor. In both cases, the PF is equal to real power divided by apparent power, or PF=P/S.

The former only involves sine formed voltage and current and the phase difference between them and is usually expressed as the cosine for that phase angle. PF then = cos(phi), where phi is the phase difference.

The latter may very well have voltage and current peaks coincide (i.e. being in phase in the classical sense) but still have a very poor power factor because the current is close to zero most of the time and increases steeply just before peak of the voltage and then also steeply returns to zero. The effect is that the RMS value of the current is high, but work done during the "current angle" is relatively small - so P/S is correspondingly small. A typical SMPS can have a PF as low as .5, even lower if lightly loaded.

How is software used to control PF? In the first case, it is only a question of measuring PF and switch capacitors in and out to achieve a PF close to unity.

In the second case, SW is used to measure instantaneous current and either control a switch so that current always follows voltage closely (used in many power supplies) or injects a current from a DC link to "fill out" the distorted current to get a sine formed current. The resultant current is then very close to a sine and its phase is close to that of the voltage sine.

A couple of references:

http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=1458905&isnumber=31401

(needs acess to IEEE library)

http://www.fairchildsemi.com/an/AN/AN-42047.pdf

(an application note how to use special circuitry for PFC, their artist's thinking about how a sine looks could be improved)

Most of the functions used in the references above can be implemented in software and that is what we usually understand with "Power Correction with Software".

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[jghrist]

Google static var compensator

 

[jimgineer]

I fear this will get heavy but let me trudge on:

So, power factor is an abstration.

In the case where you have a perfect sine wave voltage waveform, and a perfect sine wave current waveform with a phase shift, this is where you are saying you can use switching mechanisms to control the amount of VARs introduced to the system, trying to get your power factor close to 1.

The other case that you bring up, where we have let's say for discussion sake:

Perfect sine wave voltage waveform
Imperfect waveform for current (but no phase offset, as you describe, the peaks are matched but the shape of the current waveform is 'funky')

I am on the same page there, up to that point. I have seen waveforms much like this on an oscilliscope when driving certain loads. My current understanding of this situation is that it results from characteristics of the load (ie the load is not purely resistive so the 'response' of the system will vary based on it's impedance)

Makes sense, but that's where the grasp of power factor as a useful tool goes away. I thought the whole idea with power factor was to eliminate the phase shift between your voltage and current waveform. In the case where you couldn't describe there being a phase shift, but instead had a funky current waveform, it seems like you could effectively 'smooth' this waveform (get it closer to a sine wave) by introducing a filter to your system. Whether or not you can do this without introducing a phase shift, my filter design and theory is not solid enough to know the answer to this.

I believe that the second link that you linked to touches on this - but it is in reference to SMPS (electronics) power applications. I definitely want to learn a great deal more about these as well, but is the underlying theory and operation of what is described in that application note going to apply also for large scale (MW) kind of power levels, and at 60Hz? I think the answer is yes but I want to make sure. Additionally, it sounds like it is as simple as sizing the appropriate capacitor, and switching to get this value, which makes the current waveform match the voltage waveform (both in shape (sine wave) and in phase). I just want to make sure that conceptually I am understanding what an SMPS does correctly - and that this concept applies for ANY AC power source design... at the MW or mW levels.

Also, one question directly from the Fairchild link - given the shape of the current waveform, what processing or math would be done to come up with a power factor of 0.9?

And is this a reversible process? In other words, if someone told you that your pf was 0.9, would you know definitively what size capacitor you would need to make your current waveform a perfect sine wave with no offset?

If so, I don't really understand fully how you would differenciate between a 0.9 pf due to waveform distortion, and a 0.9 pf due to phase shift.

I am imagining a PLC or DSP chip that would actually sample the voltage and current waveform directly, so whatever software and processing you would need to do would be done there, not directly with a power factor... And that's why I'm saying that power factor is basically an abstraction.

Is this fair?

 

[sibeen]

jimgineer, no offence, but I think you need to learn a few basics.

I have included a simple document to show what hapens with the 'power' when a voltage is multiplied by a harmonic current.

When you open the PDF yo will need to size to 75% to see the waveforms correctly.

 

[Skogsgurra]

Jim,

It may be an abstraction in your mind. And that's the main problem.

I made a very clear distinction between classical PF and distortion PF so that you should be aware of the difference and not try to understand the two using the same thinking. And then you start treating the two in a somewhat similar way. Please do not. That will not get you anywhere.

The only things they have in common are that apparent power and effective power are different if PF<1 and that it is about electricity.

You really have to sit down and do some studying of your own. We may be willing to help in the understanding, but we are not prepared to give a full course on the subject.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[opmgr1]

Jim,
It seems you are interested in pf for large systems (MW). Now what Gunnar mentioned is correct and you cannot analyse both cases in the same manner. Lets just clear the air a little. There cannot be any pf correction without VArs being added to the system. This is very important as for large systems it determines how the voltage sags on an electrical network. Now how the software plays its part is as follows:
Lets say that for loss optimization, level of sag etc it is determined that the pf should be 98%. A load profile (KW/ KVA) of the system would be determined and VArs would be added (fixed/ switched) to bring the pf to 98%. These switched VArs is what the software control to bring the system to 98% pf without any voltage violations due to overcorrection.

 

[jimgineer]

Guys,

I appreciate the help. I have done plenty of homework on this (in school and on my own time). I do not believe that anything I have written demonstrates a lack of understanding of the fundamentals.

Sibeen, I understand that P = IV. Thanks for the graph, but what have I written that would make you think that I don't understand this relationship?

I understand the power factor triangle, and its relationship to phase offset, VA, Real Power, and VARS. I have learned through this thread that this relationship apparently can only be applied for power factor in a 'classical' sense (where voltage and current wave forms are perfect sine waves).

I will ask again: how is power factor calculated and how is it useful in the case of distortion or displacement power factor being present?

 

[CJCPE]

When there is harmonic distortion, total power factor can be calculated as pf = real power / volt-amperes where volt-amperes is calculated using the total RMS value of voltage and current including the harmonic content. The phase relationship between the distorted current and voltage waveforms is of no use.

The power factor gives an indication of the amount of “extra” current that must be supplied and carried through the distribution system to the load. However, in order to mitigate the distortion and “correct” the power factor, you need to know the individual values of the distortion current and the reactive volt-amperes. That means that knowing the total power factor is not particularly useful without knowing the displacement power factor and/or the total harmonic distortion. For that reason, I prefer the term “total power factor” and believe that the term “real power factor” is misleading.

The total power factor is equal to the displacement power factor divided by the square root of (1 + total harmonic current distortion squared)

With a distorted waveform, instead of a power triangle, there is a power “three edges of a box.” Real power and reactive volt-amperes are two edges from a corner and distortion volt-amperes are the third edge out of the same corner. The total RMS KVA is a line out of the corner to the farthest corner of the box. I believe that this concept assumes that the voltage waveform is undistorted. That is not really a valid assumption, but the voltage may be relatively undistorted compared to the current distortion.

 

[jghrist]

Power factor is calculated by: real power divided by apparent power (Vrms·Irms)

It is useful as a measure of how efficiently a system provides real power, i.e. power that does work like moving things against a force or heating things up. If a system has high power factor, then less current is needed to supply the real power and there will be fewer losses.

You can use electronics and software to correct displacement power factor with static var compensators. You can use electronics and software to correct sistortion power factor with active harmonic filters.

 

[jghrist]

I think the problem that some of us have with your line of enquiry is that saying you are correcting power factor with software is a strange way of expressing the situation. It's a little like saying you are going to get to work with software because your car has a computer in it that uses software for control.

 

[slavag]

Hi.
Jghrist!!! Star.

"You can use electronics and software to correct displacement power factor with static var compensators. You can use electronics and software to correct sistortion power factor with active harmonic filters."

Say all about differnce between displacment and distortion PF.
Best Regards.
Slava.
Jim, small math formulas.
1. distortion PF calculated as PF=P/S, whre S is Urms*Irms, used also term, total apperent power
possible use also formula S=sqr (P^2+Q^2+D^2), whre D is reactive power due the harmonics.
2. For the displacment power factor or DPF, used formula
DPF=P/(sor(P^2+Q^2)

 

[jimgineer]

Beautiful.

This is exactly what I was looking for - the what do you apply and where, (and why of course) and some of the underlying math behind it.

The only follow up I have right now involves your formula for distortion power factor:

How do you actually calculate your D^2 and Q^2, let's say if you sampled your voltage and current waveforms?

I'm going to draw some of this stuff out on paper, especially CJCPEs description. Very good fellas thanks very much.

 

[sibeen]

jimgineer, if you have a distorted current waveform, then this waveform can be shown by Fourier analysis, to consist of a fundamental component as well as sine and cosine waves that are a multiple of the fundamental.

When you multiply the fundamental voltage by one of the individual harmonic currents (as per my diagram), you can see that there is no power being transferred.

Now using the relationship PF = VA/W, the W component is the fundamental voltage multiplied by the fundamental current. The VA component s the fundamental voltage multiplied by the harmonic components.

 

[CJCPE]

The total RMS current is the square root of the sum of the squares of the RMS values of the fundamental component and all of the individual harmonic components. The total RMS value of the harmonic current is the square root of the sum of the squares of the individual harmonic components. Distortion factor or the total harmonic distortion of the current (THDI) is the RMS value of the harmonic current divided by the RMS value of the fundamental current.