Power factor correction experiment

[T.M.Haja Sahib]

The common factor is harmonics. The horrible waveform I posted above (post #17) is, to a fairly large extent, a result of harmonics.
Domestic residences have numerous appliances that take non-linear loads. The non-linear loads result in voltage distortion. Harmonics.

Another story which I have given here before. We installed some variable speed drives in a pumping station. One of the requirements of the project was to measure harmonic distortion before and after installation of the new drives to ensure compliance with an agreed standard.
There were eight drives, none of them very large. The two largest units were 200kW, 12-pulse units

The station had its own 11kV/400V transformer so the point of common coupling (with other consumers) was at 11kV.
When I made the before and after harmonic measurements two things surprised me.
The first was that was no measurable difference between the. It made do difference whether the drives were running in any combination including all and none.
The second thing that surprised me even more was the voltage distortion was outside the compliance requirements before any of our drives were connected.
It made me wonder about the cause of this distortion. Often, such pumping stations are in residential areas and that was the case here. No other heavy users. The conclusion was that the domestic users were the cause. How? All the non-linear loads in aggregate.

Just as an example this is the current taken at my kWh meter with a load of about 0.9kW:

Mostly lights (some CFL, some incandescent), television, computers, probably a few chargers.
Nothing out of the ordinary.
So you can see that harmonics are quite prevalent on domestic supplies.
Not just zinc mills.

A capacitor is not likely to be affected by distortion in the current wave form.It is the distortion in voltage wave form that affects it.Distorted current wave does not necessarily produce significant distortion in voltage wave.Then why have you depicted only current wave forms?

 

[Besoeker]

A capacitor is not likely to be affected by distortion in the current wave form.It is the distortion in voltage wave form that affects it.Distorted current wave does not necessarily produce significant distortion in voltage wave.Then why have you depicted only current wave forms?

Your wish is my....etc.
The circuit I was testing was the arrangement Gar mentioned right at the start - a motor in parallel with a capacitor, the pseudo power saving device.
I took voltage and current without the magic device connected:

You can see some distortion but, despite that, the motor, being an inductive load doesn't absorb much harmonic current
The current in post #17 is the current with the magic device connected.
So where do you thing that rather ugly current is going and why?

And, just for good measure, you might want to take a look at this:

Bit messy, eh?

 

[SAC]

You take specific equipment such as pump motor in a residence connect suitable size capacitor across it and conduct your study.Repeat it for other equipment if possible and report your study for economic viability of providing a new PFC capacitor.That will settle the issue.

I think you may be missing the "point" of the experiment - let me explain. If the PFC was put at the motor itself, and there was relatively significant loss in the conductors from the residential meter to the PFC'd motor (compared to the capacitor loss), then yes, such PFC correction "could" provide some savings. However, that is NOT what is being done by the devices being discussed, and is NOT what was intended to be illustrated by the experiment. Instead, these PFC devices are generally placed very close to the meter (in the service panel), and the branch circuit conductors run for some distance from there to the motor. The branch circuit is typically of much higher loss than those from the meter and the panel - I'd guess by at least a factor of 100. So the PFC is on the "wrong end" of the lossy conductors.

With Gar's experiment, I calculate that with the PFC placed at the motor itself, there would need to be around 125' of branch circuit between it and the panel to reach the break-even point (250' of #14). At that distance, the 3.4A current reduction from PFC would save about 2.1W of conductor loss - about the same as the loss from the PFC capacitors. And that is assuming that the motor is typically run un-loaded, which is unlikely (or it would be a lot smaller).

 

[T.M.Haja Sahib]

Your wish is my....etc.
The circuit I was testing was the arrangement Gar mentioned right at the start - a motor in parallel with a capacitor, the pseudo power saving device.
I took voltage and current without the magic device connected:

You can see some distortion but, despite that, the motor, being an inductive load doesn't absorb much harmonic current
The current in post #17 is the current with the magic device connected.
So where do you thing that rather ugly current is going and why?

And, just for good measure, you might want to take a look at this:

Bit messy, eh?

It appears you have misconceptions.The voltage wave is still almost sine wave,A capacitor is a passive device.It does not create any harmonics unlike an inductor,but it can amplify harmonics..So you have to look for resonanceThis is a specific case and not a general one.Do you know how to do it?If yes,please do a harmonic analysis,as you seem to be well-equipped.

 

[Besoeker]

It appears you have misconceptions.

Not very kind, but let that pass. My misconceptions, if they are such, have served me well for over four decades in the business.
Slightly more seriously, I have to commit to performance on design parameters, harmonic distortion, at the bid stage of a project. These become contractual requirements if we win the business.
Failure to meet such commitments has serious financial consequences.
To date, we have taken no such pain.
If I have misconceptions, I must have hidden them well. For a long time.....

The voltage wave is still almost sine wave,A capacitor is a passive device.It does not create any harmonics unlike an inductor,but it can amplify harmonics..So you have to look for resonanceThis is a specific case and not a general one.Do you know how to do it?If yes,please do a harmonic analysis,as you seem to be well-equipped.

Resonance actually requires two elements that can store and exchange energy. Inductance and capacitance both store energy.
0.5LI^2 = 0.5CV^2 is this exchange.
ωL=1/(ωC) gives you the conditions for resonance

When we include PFC on a VSD system we de-tune them with series reactors such that the circuit is inductive at anything below the first expected harmonic frequency. For VSDs this normally 5th, 250Hz in most of the regions I deal with. Some suppliers work it at 225 Hz. Too close for my liking. I use 177Hz. It puts the size and cost of the components up a bit but, while not being totally risk averse, taking a small hit to avoid a potentially costly seems to me to be prudent.

Back to your point:

The voltage wave is still almost sine wave

Well..........maybe not.
If you are accustomed to looking at such things you can visually see the distortion and, with a bit of experience, you can make an informed guess on the nature and extent of the distortion.
But, again, as you have requested, a harmonic analysis coming to you. No charge.
First I look one cycle out of the five previously displayed:

Notice the flattened tops? That indicates third harmonic and multiples thereof. Triple-n.

And you wanted a harmonic analysis...
No problem.

Note that every odd harmonic is included. Confirmation of just the visual observation if you are accustomed looking at such things.
And typical of distortion I have observed resulting from single phase non-linear loads.

 

[gar]

111105-1609 EDT

Besoeker:

My line voltage looks pretty much like yours.

I have photos of scope traces from about 50 years ago that visually look like good sine waves. Big change since then.

We have over 40,000 college students in the area, a lot of research, and three large hospital centers in the area. Thus, a lot of computers and other things with capacitor input filter power supplies. I suspect the flat top waveforms can be seen back at the primary lines, 345 kV.

.

 

[Besoeker]

111105-1609 EDT

Besoeker:

My line voltage looks pretty much like yours.

Probably for the same reasons.

 

[T.M.Haja Sahib]

Not very kind, but let that pass. My misconceptions, if they are such, have served me well for over four decades in the business.
Slightly more seriously, I have to commit to performance on design parameters, harmonic distortion, at the bid stage of a project. These become contractual requirements if we win the business.
Failure to meet such commitments has serious financial consequences.
To date, we have taken no such pain.
If I have misconceptions, I must have hidden them well. For a long time.....

That was just my friendly criticism because I consider you and all others in this forum my friends and obviously YOU reciprocated.

Back to your point:
well..........maybe not.
If you are accustomed to looking at such things you can visually see the distortion and, with a bit of experience, you can make an informed guess on the nature and extent of the distortion.
But, again, as you have requested, a harmonic analysis coming to you. No charge.
First I look one cycle out of the five previously displayed:

Notice the flattened tops? That indicates third harmonic and multiples thereof. Triple-n.

When would you give us the 'full picture' of current and voltage waveforms under discussion?No problem.Was the above voltage waveform before the connection of that magic device?Remember that a capacitor does not generate any harmonics but amplify those already present.

And you wanted a harmonic analysis...
No problem.

Note that every odd harmonic is included. Confirmation of just the visual observation if you are accustomed looking at such things.
And typical of distortion I have observed resulting from single phase non-linear loads.

Again,you seem to have no idea of why harmonic analysis was required,as you have not shown WHICH HARMONIC is actually causing the resonance by backing your visual data with the theoretical formula for harmonic resonance between power supply and a shunt capacitor.Why can't you do it now?
As a final point,is the current taken by the capacitor within 110%to 120% and voltage within 110%of its rating.If it is so,you need not worry at all about the way it is taking current.

 

[Besoeker]

When would you give us the 'full picture' of current and voltage waveforms under discussion?

Check post #22. The last picture shows current, voltage and power. What else would you like me to add to make it the "full picture" you have requested?

No problem.Was the above voltage waveform before the connection of that magic device?

It is just a recorded waveform of my supply voltage and unrelated to any tests.

Remember that a capacitor does not generate any harmonics but amplify those already present.

As I'm sure you must know, the impedance of a capacitor reduces with increasing frequency. Harmonic content in the supply voltage thus increases capacitor current beyond that which would flow with a purely sinusoidal voltage.

Again,you seem to have no idea of why harmonic analysis was required,

It wasn't.

as you have not shown WHICH HARMONIC is actually causing the resonance

Please don't confuse harmonics with resonance. A non-sinusoidal waveform can be expressed in terms of a fundamental and a series of harmonics.
For example, the supply current for a six pulse rectifier with level DC can be expressed thus:

I = 1.031 I [sinωt ? (sin 5ωt/5) - (sin 7ωt/) + (sin 11ωt/11) + (sin 13ωt/13) - (sin 17ωt/17) - ?]

It's a useful mathematical tool for calculating expected voltage distortion.
Here's an excerpt from one of my spreadsheets just showing the results:

The formula I gave above is a useful starting point and gives the amplitude of the harmonic as the reciprocal of the harmonic number. A few matters make it a bit more complex. It ignores losses, overlap angle, and that the current isn't usually level.
But none of that has to do with resonance.

by backing your visual data with the theoretical formula for harmonic resonance between power supply and a shunt capacitor.Why can't you do it now?

So now we come to resonance. If you go back to post #22 and look at the current. The positive and negative half cycles, whilst messy, are pretty much symmetrical and repeats over the five cycles shown. That's a power frequency phenomenon.
But you are right. One has to take care on potential resonance with supplies and it's something I have to consider in my day job.

When we determine that it might be difficult or not possible to comply with supply voltage distortion limits, we fit sometimes passive harmonic filters. These are designed to resonate at specific harmonic frequencies thus keeping harmonic currents local rather than polluting the supply. Shunt circuits if you like.
Again an excerpt from one of my calculation spreadsheets for such filters:

So I look for potential resonances with the supply to ensure that none will coincide with a known harmonic

If there are, it's back to the drawing board.
The particular project that I have take the data from was four 800kW inverters and a 3.3kV rather weak supply.
Not for the faint hearted.

And, just to recap a little.
I do harmonic calculations prior to award of contract to determine what kit to offer to be specification compliant.
I do harmonic measurement subsequent installation to confirm that the kit supplied is specification compliant.

 

[T.M.Haja Sahib]

Check post #22. The last picture shows current, voltage and power. What else would you like me to add to make it the "full picture" you have requested?

It changed from one post to subsequent post so that my thoughts could not be streamlined.

Please don't confuse harmonics with resonance. A non-sinusoidal waveform can be expressed in terms of a fundamental and a series of harmonics.
For example, the supply current for a six pulse rectifier with level DC can be expressed thus:

I = 1.031 I [sinωt – (sin 5ωt/5) - (sin 7ωt/) + (sin 11ωt/11) + (sin 13ωt/13) - (sin 17ωt/17) - …]

It's a useful mathematical tool for calculating expected voltage distortion.
Here's an excerpt from one of my spreadsheets just showing the results:

The formula I gave above is a useful starting point and gives the amplitude of the harmonic as the reciprocal of the harmonic number. A few matters make it a bit more complex. It ignores losses, overlap angle, and that the current isn't usually level.
But none of that has to do with resonance.


So now we come to resonance. If you go back to post #22 and look at the current. The positive and negative half cycles, whilst messy, are pretty much symmetrical and repeats over the five cycles shown. That's a power frequency phenomenon.
But you are right. One has to take care on potential resonance with supplies and it's something I have to consider in my day job.

When we determine that it might be difficult or not possible to comply with supply voltage distortion limits, we fit sometimes passive harmonic filters. These are designed to resonate at specific harmonic frequencies thus keeping harmonic currents local rather than polluting the supply. Shunt circuits if you like.
Again an excerpt from one of my calculation spreadsheets for such filters:

So I look for potential resonances with the supply to ensure that none will coincide with a known harmonic

If there are, it's back to the drawing board.
The particular project that I have take the data from was four 800kW inverters and a 3.3kV rather weak supply.
Not for the faint hearted.

Your objection to installation of a PFC capacitor in a residence is still not supported.For that purpose,you have to establish that at least in one residence,resonance has occurred.To do that here are the guidelines.
Let the size of a capacitor =X [MVAr]
and Fault level =Y [MVA]

Then resonance will occur at harmonic order n=Sq.root of Y/X
If this 'n' is present on the load side of a residence,resonance would occur.Can you do this?

 

[mivey]

Your objection to installation of a PFC capacitor in a residence is still not supported.

Can you show where there is value in a pfc device installed in a residential disconnect panel?

 

[Besoeker]

Your objection to installation of a PFC capacitor in a residence is still not supported. For that purpose,you have to establish that at least in one residence,resonance has occurred.

I don't. The simple objection is on the basis of cost. Many, maybe most residential customers are billed in kWh. Power factor is irrelevant. Fit PFC at the supply into to the residence and you might help the supply company some. It will make not a what of difference to what happens in the residence.

To do that here are the guidelines.
Let the size of a capacitor =X [MVAr]
and Fault level =Y [MVA]

At residential level??????

Then resonance will occur at harmonic order n=Sq.root of Y/X
If this 'n' is present on the load side of a residence,resonance would occur.Can you do this?

Perhaps you missed this:

So I look for potential resonances with the supply to ensure that none will coincide with a known harmonic .

I think I have adequately demonstrated that, not only that can I do, but actually do.

 

[T.M.Haja Sahib]

I don't. The simple objection is on the basis of cost. Many, maybe most residential customers are billed in kWh. Power factor is irrelevant. Fit PFC at the supply into to the residence and you might help the supply company some. It will make not a what of difference to what happens in the residence.

Please refer to your post #5,wherein you recorded additional objection

At residential level??????
Yes.Do not get intimidated by MVA's.The residence is also connected to the grid.

Perhaps you missed this:


I think I have adequately demonstrated that, not only that can I do, but actually do.

No.I acknowledged your expertise and requested you to apply it in the case of a residence.

 

[T.M.Haja Sahib]

Can you show where there is value in a pfc device installed in a residential disconnect panel?

I only suggested there may be an economic value by putting a suitable size capacitor across a heavy inductive load in a residence,say a pump motor.I welcome evidence for or against this only.

 

[rattus]

I only suggested there may be an economic value by putting a suitable size capacitor across a heavy inductive load in a residence,say a pump motor.I welcome evidence for or against this only.

No need for testing the concept. Unless there is a PF penalty, the only saving would be a tiny reduction in wiring loss and then only if the cap is switched and located near the inductive load.

 

[SAC]

I only suggested there may be an economic value by putting a suitable size capacitor across a heavy inductive load in a residence,say a pump motor.I welcome evidence for or against this only.

You seem to be the only one who is talking about that case - it is not the point of the thread. See my post #23:

http://forums.mikeholt.com/showthread.php?t=140830&p=1347580#post1347580

If you want to discuss that case, why don't you show calculations for a real residential example where you think it does make sense (PFC at the motor), and then compare that to what is actually being discussed (PFC at the panel).

 

[mivey]

I only suggested there may be an economic value by putting a suitable size capacitor across a heavy inductive load in a residence,say a pump motor.I welcome evidence for or against this only.

Why don't you try it?

Take a well pump (2-3 HP?) located a couple hundred feet from the house with a conductor sized for acceptable voltage drop, put a capacitor that offsets the var at some chosen average run power factor, pick a duty cycle for the motor (10-20%?), let the cost of electricity be about 10 cents/kWh (for the savings gained by reducing heat loss), and I bet you will find it might only save you a dollar or so per year, if that. Compare that to the material & installation costs.

 

[GeorgeB]

Why don't you try it?

I _LOVE_ the way you think, mivey. I think I'll brainstorm a little ...

ASSUME
2 kW
200V
200 ft from panel to motor
20A@0.5 pf uncorrected
10A@1.0 pf perfectly corrected, perfect capacitor
#10 wire, 0.4 ohm loop
drop@10A, 4V, 2%
drop@20A, 8V, 4%
reduced voltage drop 4V, reduced current 10A

These are REDUCTIONS ...
P=I*I*R=10*10*0.4=40W
crosscheck, P=E*E/R=4*4/0.4=40W

ASSUME 2.5 hours a day (probably high for a residential water pump), 100Wh/day
at the assumed $0.10/kWh, save a penny a day, on the order of $4 a year. With a real capacitor, it'll be less. With real PFs, it'll PROBABLY be less; full load PF for a 2.2kW single phase motor I picked at random from Baldor's website is 0.87, so MUCH less. Mivey's $1.00 a year looks REALLY reasonable as a SWAG.

My electrician will charge for the capacitor and installation ... REALLY hard to get reasonable payback.

 

[rattus]

My electrician will charge for the capacitor and installation ... REALLY hard to get reasonable payback.

At this rate, the cap will likely fail before it pays for itself.

 

[jghrist]

I _LOVE_ the way you think, mivey. I think I'll brainstorm a little ...

ASSUME
2 kW
200V
200 ft from panel to motor
20A@0.5 pf uncorrected
10A@1.0 pf perfectly corrected, perfect capacitor
#10 wire, 0.4 ohm loop
drop@10A, 4V, 2%
drop@20A, 8V, 4%
reduced voltage drop 4V, reduced current 10A

These are REDUCTIONS ...
P=I*I*R=10*10*0.4=40W
crosscheck, P=E*E/R=4*4/0.4=40W

ASSUME 2.5 hours a day (probably high for a residential water pump), 100Wh/day
at the assumed $0.10/kWh, save a penny a day, on the order of $4 a year. With a real capacitor, it'll be less. With real PFs, it'll PROBABLY be less; full load PF for a 2.2kW single phase motor I picked at random from Baldor's website is 0.87, so MUCH less. Mivey's $1.00 a year looks REALLY reasonable as a SWAG.

My electrician will charge for the capacitor and installation ... REALLY hard to get reasonable payback.

Actually the reduction is P2 - P1 = I2??R - I1??R = (20? - 10?)?0.4 = 120 W

Still only $12 a year.