What is Power Factor in the real world?

[mike]

What is Power Factor in the real world?

I've been reading about devices that hook to your house power
and save you 25% on your electric bill. They're very vague,
but it looks like they add a cap across the line to correct
power factor.

I pay for WATTS delivered, so it appears that reducing V-A
doesn't save any money...until the power company starts
billing for V-A.

But that got me thinking...what is PF anyway?
It's all nice and tidy when you have a motor with a constant
load that looks inductive. You put some Capacitance
across it to shift the current back in phase with the voltage.
The place to do that is at each individual motor.

But what about the real residential world where you have
lots of switching power supplies that have current waveforms
that don't look anything like sine waves. And they all have
current peaks at about the same time.

I did some experiments.
I plugged a 14W muffin fan and a 13W CCFL lamp into a Valhalla
digital power analyzer.

I'm assuming volt-amps is Vrms * Irms. (I haven't done the math
to determine if this is actually correct, or whether I have to
do the integration to get the right number.) Somebody can
correct me.

The measured watts for both devices is the sum of the measured
watts for the individual devices.

The measured current for both devices is less than the sum
of the measured current for the individual devices.

If PF = watts/volt-amps,
the CCFL PF is .62
The Fan PF is .66
The PF for both is .75

From the perspective of the power company, is that a good thing?
I haven't reduced the peak current at all. I have purchased
more watts without increasing the peak current. But the phase
is still not optimal.

Looking at the current waveform, there's still almost no current
during the first part of the voltage waveform and the big transient
when the CCFL turns on. The current waveform looks more like
a square wave than a sine wave.

How bad is the residential power factor problem anyway?
Is there published data on what the current waveform looks
like at the power station where a lot of loads get summed up?

Decades from now when all our electronic devices have individual
PF correction, this will sort itself out. But, in the meantime,
is there really anything practical that can be done to correct
residential power factor?

 

[Martin Riddle]

What is Power Factor in the real world?

I've been reading about devices that hook to your house power
and save you 25% on your electric bill. They're very vague,
but it looks like they add a cap across the line to correct
power factor.

I pay for WATTS delivered, so it appears that reducing V-A
doesn't save any money...until the power company starts
billing for V-A.

But that got me thinking...what is PF anyway?
It's all nice and tidy when you have a motor with a constant
load that looks inductive. You put some Capacitance
across it to shift the current back in phase with the voltage.
The place to do that is at each individual motor.

But what about the real residential world where you have
lots of switching power supplies that have current waveforms
that don't look anything like sine waves. And they all have
current peaks at about the same time.

I did some experiments.
I plugged a 14W muffin fan and a 13W CCFL lamp into a Valhalla
digital power analyzer.

I'm assuming volt-amps is Vrms * Irms. (I haven't done the math
to determine if this is actually correct, or whether I have to
do the integration to get the right number.) Somebody can
correct me.

The measured watts for both devices is the sum of the measured
watts for the individual devices.

The measured current for both devices is less than the sum
of the measured current for the individual devices.

If PF = watts/volt-amps,
the CCFL PF is .62
The Fan PF is .66
The PF for both is .75

From the perspective of the power company, is that a good thing?
I haven't reduced the peak current at all. I have purchased
more watts without increasing the peak current. But the phase
is still not optimal.

Looking at the current waveform, there's still almost no current
during the first part of the voltage waveform and the big transient
when the CCFL turns on. The current waveform looks more like
a square wave than a sine wave.

How bad is the residential power factor problem anyway?
Is there published data on what the current waveform looks
like at the power station where a lot of loads get summed up?

Decades from now when all our electronic devices have individual
PF correction, this will sort itself out. But, in the meantime,
is there really anything practical that can be done to correct
residential power factor?

Most residential home power meters will handle poor power factor loads
properly. So there is no real reason to correct it.
It becomes a problem with Industrial situations because of the high
currents that can occur, here you'll see cap banks on feeders.

The real reason to use PFC is to play nice with the power company. There
is a ton of power wasted in transmission lines because of the high peak
currents from the poor balance between real and apparent power. Its one
of the reasons why you are starting to see PFC in PC power supplies.
When you have a millions of non-PFC pc power supplies, you start to see
that it could pose a problem.

As for those power saving devices, they don’t work for the average
homeowner.

As for the CFL's, those are some wave forms! My guess is that they will
go PFC just like the pc power supply.

Cheers

 

[Tim Williams]

PF is important to the residential customer that wants more power. You
can't run 1000W of PSU on a single 15A circuit without PFC.

Tim

 

[Paul Keinanen]

If you scope the power in your house or business, the sine waves are
usually flat on top. So a lot of loads are pulling peak current, and a
lot of ohmic heating is happening in wiring everywhere.

Older non-PFC-corrected switching supplies mostly pull a lot of
current at the sinewave peak. Consider a big office full of PCs. One
configuration is to take a 3-phase Y-connected feeder and split out
thirds of the room, one zone on each phase. The resulting six current
spikes per cycle all peak at different times and return through the
shared neutral leg, not canceling like they were supposed to, making
way more RMS current than intended. Fires have started this way.

These fires have usually happened in feeder cables, in which the
neutral conductor is thinner than the phase conductor (based on the
classical assumption that most of the phase currents will cancel out
in the neutral conductor), however, this assumption is not true, if
there is a lot of 3rd harmonic on the neutral conductor due to non-PFC
power supplies.

PF correction caps probably intercept the uglies before they get all
the way back to the generator. But "modern" loads are a big problem,
which is why we have PFC rules now.

PF correction caps have traditionally used to compensate for the
inductive reactance of fluorescent light ballast chokes and electric
motors. The inductive reactance will cause a phase shift between the
sinusoid voltage and sinusoid current. The capacitive reactance of the
PF capacitor will try to pull the phase difference back to zero.

With constant inductive loads such as traditional fluorescent lamps a
fixed compensation capacitor is installed at the mains terminals, at
least in Europe. For a variable reactive device, such an electric
motor with variable load, a fixed capacitor can not generally be
used, but in large industrial sites, centralized capacitor banks are
switched in with contactors, to consume the variable amount of
(negative) reactive power and hence present a resistive load to the
utility company, to avoid the reactive bill that is metered separately
for large industrial sites.

However, I am not so sure what good a simple PF capacitor would do, if
the AC load is some uncompensated SMPS, with a high, but short current
peak _in_phase_ with the voltage.

In large 3 phase (6 pulse) rectifiers series inductors "reactors" in
series with the phase conductors to limit the peak/average ratio. An
other way to limit the peak current is with 12 and 18 pulse
rectifiers, but these require transformer windings, that produce
separate three phase constellations with for instance 30 or 45 degree
phase shift between constellations. In the 6 pulse case, one
transformer secondary would be connected to delta and the other to wye
(with voltage adjusted) and thus, the current pulses seen by the
generator would be more evenly distributed.

With distribution transformers with a larger selection of winding
options (not just delta/wye) differential residential three phase
loads could be created, with a small phase shift between the
distribution cables, thus distribute the generator current demand more
evenly.

 

[bud--]

You are correct. Utility meters accurately measure kilowatt-hours.

For industrial users, utilities often have a second meter that measures
"reactive power" - KVARh. There is a large penalty for KVAR 'use' which
'encourages' the users to correct the power factor.

Since utilities do not measure KVAR, or power factor, for residential
users, the 'devices' sold are a scam, as Tim said. They actually
slightly reduce the current which slightly reduces the I squared R
losses in the wire. This is substantially a trivial loss, and only
applies to the wire from the meter to where the 'device' is installed -
typically at the service panel - a trivial distance. As the energy star
link below points out, the 'device' really needs to adjust to the
reactive load that is present. If there is no inductive load, the
correction cap causes a higher current which also results in increased I
squared R losses.

At motors is a good location. Industrial users might have PF correction
caps at the service, with the amount of correction switched as the
inductive load changes.

PF correction caps aren't particularly useful. I believe for harmonic PF
distortion, low pass filters are used.

For 3 phase wiring, the 3rd (6th, 9th, ...) harmonics add on the
neutral. At a delta primary, wye secondary transformer those harmonics
cancel on the primary.

http://energystar.custhelp.com/cgi-bin/energystar.cfg/php/enduser/std_adp.php?p_faqid=4941

Nice link.

 

[Tim Williams]

PF correction caps aren't particularly useful. I believe for harmonic PF
distortion, low pass filters are used.

Indeed, if you add lots of parallel caps, you can bypass away the harmonics,
but now your 60Hz power factor is crap. So you add an inductor, which
literally resonates with the cap. So you make a 60Hz bandpass, as narrow as
you want.

Hmm...

To get 10dB attenuation at the first significant harmonic (3rd = 180Hz), you
need about Xc < 0.11 * Zsrc. I would guess the line is around 0.5 ohms
(maybe for a long run, and probably much lower most of the time). That
means Xc(180Hz) = 0.05 ohm, which is just absurd (18mF -- not uF). The
respective inductor would be quite heavy, and get quite hot, even if it
manages to have a rather high Q.

A series-parallel bandpass could get by with less, of course. Offhand,
600mH + 11.7uF series and 800uH || 9.4mF parallel would be typical,
according to some filter calculator.

Either way, active PFC is better.

Hah... revolutionary PFC idea here:
Instead of loading computers with PFC's...
Pre-PFC the line.

So you start with VAC, run it through a bigass PFC to make DC, then invert
it back to AC, square with limited harmonics. Up to ~20 harmonics let's
say. Radiation loss, harmonic heating and etc. isn't horrible when it goes
through wires and transformers (though certainly higher).

Result: distribution lines run cooler. Power company is happy. PSU diodes
and caps run cooler, saving a small amount of heat (possibly not as much as
lost in transformers though).

For 3 phase wiring, the 3rd (6th, 9th, ...) harmonics add on the neutral.
At a delta primary, wye secondary transformer those harmonics cancel on
the primary.

Wye-wye transformers usually have a tertiary winding on each leg, which is
usually connected in delta. This can be used to supply auxiliary power for
a control or whatever, but it also shorts out those harmonics. The winding
is fairly small, which they get away with because the harmonic power is
relatively small.

Tim

 

[Uwe Hercksen]

How bad is the residential power factor problem anyway?
Is there published data on what the current waveform looks
like at the power station where a lot of loads get summed up?

Hello,

as long as the residential power used thru a bridge rectifier is small
compared to the residential power used for heating with sinusodial
current it is no big problem.

Bye

 

[Uwe Hercksen]

So you start with VAC, run it through a bigass PFC to make DC, then invert
it back to AC, square with limited harmonics. Up to ~20 harmonics let's
say. Radiation loss, harmonic heating and etc. isn't horrible when it goes
through wires and transformers (though certainly higher).

Hello,

the problem of such a square wave are long transmission lines. The 20th
harmonic of 60 Hz is 1200 Hz, the resulting wavelength is about 167 km,
but the problems start with a quarter wavelength of 41 km. Therefore the
transmission lines should be not longer than only 10 or 20 km. Using
only 10 instead of 20 harmonics is not much better.

Bye

 

[Tim Williams]

the problem of such a square wave are long transmission lines. The 20th
harmonic of 60 Hz is 1200 Hz, the resulting wavelength is about 167 km,
but the problems start with a quarter wavelength of 41 km. Therefore the
transmission lines should be not longer than only 10 or 20 km. Using only
10 instead of 20 harmonics is not much better.

I wasn't thinking so much for transmission as distribution. It could even
be implemented at the customer level, e.g., replace pole pigs with
converters.

It would be valuable to keep the number of transformers down, otherwise
they'll have to be overrated due to harmonic heating. They'll also
attenuate the harmonics somewhat.

Tim

 

[Grant]

In a squarewave, there are no even harmonics so neither the 10th nor
20th have any energy in the square wave case.

Most AC power is sent long distances as 3 phase power. This makes the
3rd harmonic a bit of a problem because 3*120 degrees is 360 degrees
so they all end up in phase.

Thus warming the neutral, which isn't there on delta distro, so where'd
the 3rd go?

Grant.

 

[JosephKK]

I wasn't thinking so much for transmission as distribution. It could even
be implemented at the customer level, e.g., replace pole pigs with
converters.

It would be valuable to keep the number of transformers down, otherwise
they'll have to be overrated due to harmonic heating. They'll also
attenuate the harmonics somewhat.

Tim

A rather not small part of that is that the required L and C to smooth it
to reasonable levels are about the same size as the pole pig with the
same failure rate, and the substitute (many kilohertz) electronics and
magnetics adds to the failure rate. Remember motor start and welding and
other ugly loads.

 

[JosephKK]

In a squarewave, there are no even harmonics so neither the 10th nor
20th have any energy in the square wave case.

Most AC power is sent long distances as 3 phase power. This makes the
3rd harmonic a bit of a problem because 3*120 degrees is 360 degrees
so they all end up in phase.

Naw, they use delta windings; which have the useful effect of nearly
killing the 3rd and attenuating the 5th and 7th harmonics. Turns them in
to heat in copper and some iron losses. this is why you see high K
(harmonic content) rated transformers, and the utilities are getting
tough with customers about PFC and harmonic control. Active harmonic
correctors are becoming common, especially where large variable frequency
drives (VFD) and adjustable frequency motor controls (AFMC, VFD's bigger
cousin).

 

[Raveninghorde]

What is Power Factor in the real world?

I've been reading about devices that hook to your house power
and save you 25% on your electric bill. They're very vague,
but it looks like they add a cap across the line to correct
power factor.

I pay for WATTS delivered, so it appears that reducing V-A
doesn't save any money...until the power company starts
billing for V-A.

But that got me thinking...what is PF anyway?
It's all nice and tidy when you have a motor with a constant
load that looks inductive. You put some Capacitance
across it to shift the current back in phase with the voltage.
The place to do that is at each individual motor.

But what about the real residential world where you have
lots of switching power supplies that have current waveforms
that don't look anything like sine waves. And they all have
current peaks at about the same time.

I did some experiments.
I plugged a 14W muffin fan and a 13W CCFL lamp into a Valhalla
digital power analyzer.

I'm assuming volt-amps is Vrms * Irms. (I haven't done the math
to determine if this is actually correct, or whether I have to
do the integration to get the right number.) Somebody can
correct me.

The measured watts for both devices is the sum of the measured
watts for the individual devices.

The measured current for both devices is less than the sum
of the measured current for the individual devices.

If PF = watts/volt-amps,
the CCFL PF is .62
The Fan PF is .66
The PF for both is .75

From the perspective of the power company, is that a good thing?
I haven't reduced the peak current at all. I have purchased
more watts without increasing the peak current. But the phase
is still not optimal.

Looking at the current waveform, there's still almost no current
during the first part of the voltage waveform and the big transient
when the CCFL turns on. The current waveform looks more like
a square wave than a sine wave.

How bad is the residential power factor problem anyway?
Is there published data on what the current waveform looks
like at the power station where a lot of loads get summed up?

Decades from now when all our electronic devices have individual
PF correction, this will sort itself out. But, in the meantime,
is there really anything practical that can be done to correct
residential power factor?

Useful info here:

http://www.copperinfo.co.uk/power-quality/power-quality-guide.shtml