Can any power engineers solve my power flickering mystery?

[Mark R]

Just wondering.

Every now and again, the lights in my apartment flicker - sometimes, they flicker like crazy for up to half an hour at a time. It's been a lot better since switching to CFL and halogen lights - low voltage halogens are almost completely unaffected. However, standard incandescents can be really bad at times - and if you want to dim them, well, you make it twice as bad. In between times, however, the power is stable as a rock.

This is definitely a problem throughout the whole apartment block, as we've had a few discussions on our forums about it. Happens at any time, including late evening and weekends.

I lived elsewhere in the city a couple of years ago, and I happened to notice similar flickering then - but at that point I had CFLs mainly, rather than halogen - and the power wasn't as good anyway, as the voltage would sag noticeably with big loads (as it was out in the suburbs). In my new apartment, voltage doesn't budge much, even when switching on a 7 kW load (unless it's on the same circuit). [I estimate supply impedance at my service to be about 20 mOhm].

Having monitored the voltage over a few days, I think that the voltage fluctuations are within legal limits, so I'm not sure I've got a complaint. However, I'm just interested in what might be causing it. The fluctuations can be quite significant (repeated 4-5 V fluctuation in RMS voltage several times per second).

Example trace: flicker.jpg The three traces are max/mean/min cycle by cycle voltage. So you can see episodes where there is noticeable rapid fluctuation where the lines separate. This is only a mild example, but I can't find a severe example at the moment.

So, what sort of thing can cause such marked flickering? It's clearly not anything within the apartment building. Could it be something so big, that's it causes power fluctuations across the whole city? I have also recorded sudden drops in grid frequency when these fluctuations start - e.g. 0.1V flicker voltage, then suddenly 4 V flicker, simultaneous with a drop of 0.05 Hz in grid frequency. So this does lend some credibility to it being something huge.

Any ideas, among electrically oriented sleuths?

 

[dkozloski]

Have you tried talking to an EE at your power company? Around here they're only too happy to talk about stuff like this. The bean counters are clueless but somebody in dispatch might help you.

 

[Mark R]

You don't know my power company. They're a joke. I've been in dispute with them for over a year, because they can't even bill me for the correct meter - despite me writing to them with the meter serial numbers, and begging them to send an engineer round to verify what I'm telling them.

At any rate, round here, customers are supposed to contact their 'energy supplier' (this is the company that buys the power from the generators and sells it to customers), this is not the same company as the electricity 'distributor' (who delivers power from the grid operator to the customer). They'll then pass the message on. As you might imagine, I don't have much faith in this happening.

 

[oynaz]

That definitely sounds like a problem your power company should handle. If they won't, consider sending your insurance company after them. These fluctuations, though unlikely, might be the indicator of a serious problem, the kind which fries insured electronics or start fires.

 

[PowerEngineer]

Hi Mark,

Your thinking on this seems pretty sound to me.

Frequency is generally the same across the entire network, so those sudden drops in frequency would be system-wide. A dip means that the rotating masses (e.g. generators and motors) are all slowing down because the power being withdrawn from the network is greater than what the generators are putting in. So it seems likely that what you're seeing is a big load being switched in. The frequency takes a dip until the generator controls can ramp their output up to match the increased load.

Voltage is more of a local phenomenon, so the fact that you're seeing voltage flicker suggest that this load may be relatively close to you; maybe on your same feeder or fed from the same substation.

Because the voltage changes are so rapid, I'm guessing that some sort of thyristor-control drive system may be the culprit. Here's the Wikipedia link. These drives (and others like them) provide fine control for motor-driven processes by throttling the AC input power through controlled gating of the in-line silicon switching devices. By their nature, they also put large reactive loadings on the electrical system (because delayed firing to reduce power also look reactive), and these reactive loadings shift over a wide range as the power to the motors is varied during the manufacturing process (reactive load increasing as power load decreasing). Fluctuating reactive loadings will cause fluctuating voltages.

It could be something relatively small but very close to you (like the HVAC system in a big office building or a small assembly line), or something bigger in your town (like a steel rolling mill).

You're right about your "energy provider"; they have nothing to do with this. You should try to contact your distribution provider (if that's the right term), or whoever it is that takes calls for power outages. Explain to them that you have a "power quality" problem you'd like to see addressed.

Good luck!

 

[bendixG15]

Get out the yellow pages...

Every state has some jurisdiction that oversees these people.
Look it up in your phone book and call the state agency with your complaint...
They "should" respond.

 

[Mark R]

Thanks PowerEngineer,

I have to confess I can't think of much in the immediate locality. It's mainly apartments, and general commercial buildings. No major industry. Yes, there is a supermarket on the ground floor of my building with the accompanying HVAC and refirigeration systems - but I can't tie such big voltage fluctuations to such small equipment, given the very low supply impedance here.

I think you may well be right about steel mills - I knew there were a lot of mills around here, but hadn't considered that they might be massive power consumers. However, some scanning with google earth reveals that several of them have dedicated 275 or 400 V feeders - so, perhaps, they really are absolutely massive consumers.

I'm not that worried about the flickering, and I'm certain that it's within all legal requirements for the majority of the time (my voltage logger can also calculate 'short-term' and 'long-term flicker' so I am pretty sure), so I don't really have a complaint. It was really just curiosity.

 

[PowerEngineer]

Yes, steel mills can be very large loads -- sometimes 100's of megawatts. That's why they usually take their power at those higher voltage levels (>100kV). Doing so helps limit the impact of their real/reactive power swings on other customers.

A supermarket in your building? I'd put their refrigeration units at the top of my list of suspects.

 

[lambchops511]

bet u its ur neighbour growing marijuana turning the lights on and off

 

[BrownTown]

Meh, just put a choke on your main feeder and be done with it.

 

[imported_bearfx]

I had a similar problem in an apartment once, although my neighbors weren't having the same issue. I b***ed to the apartment management about it for a year, along with questioning why my electric bill was twice what my neighbors was. Finally, I had to have an service guy from the power company come out... Turned out two on my circuit breakers were bad. When he pulled them out, their were burn marks and melted bits all over. Once they were replaced, the problem went away.

I would start by checking the things you can check yourself, such as the circuit breakers (please shutoff the main power before trying though).

 

[MrDudeMan]

Originally posted by: BrownTown
Meh, just put a choke on your main feeder and be done with it.

How would a choke help? You use chokes to block AC current at a particular frequency and pass DC current. I don't see how that would do anything other than block his AC power.

 

[BrownTown]

Originally posted by: MrDudeMan

Originally posted by: BrownTown
Meh, just put a choke on your main feeder and be done with it.

How would a choke help? You use chokes to block AC current at a particular frequency and pass DC current. I don't see how that would do anything other than block his AC power.

A "choke" just means an inductor, that will filter out high frequency noise. They are often used in industrial applications where you are using a variable frequency drive since the pulsed DC created will have alot of high frequency noise associated with it that can damage parts. PowerEngineer speculated that such a device might be causing the voltage fluctuations, so filtering it out would relieve the problem. A properly sized inductor will be seen as only a small impedance to 60Hz power frequency, but if you have noise coming in at 10 times that frequency then the impedance will rise in proportion. Just make sure that the voltage withstand capabilities of the inductor are high enough for the application at hand since the voltage fluctuations will be seen across the inductor.

Anyways, not saying thats the problem, but if it is than thats a solution.

 

[DrPizza]

If you're certain about the meter problem, check out Bendix's suggestion above - the public utility commision or whoever they are for your state. The first thing to do though is call the electric company with the threat of doing that along with a demand that you see the last umpteen months worth of bills for the correct meter. Generally, that threat is enough to get them on the ball. If they don't, follow through with the threat.

 

[Mark R]

Originally posted by: BrownTown
A "choke" just means an inductor, that will filter out high frequency noise. They are often used in industrial applications where you are using a variable frequency drive since the pulsed DC created will have alot of high frequency noise associated with it that can damage parts. PowerEngineer speculated that such a device might be causing the voltage fluctuations, so filtering it out would relieve the problem. A properly sized inductor will be seen as only a small impedance to 60Hz power frequency, but if you have noise coming in at 10 times that frequency then the impedance will rise in proportion. Just make sure that the voltage withstand capabilities of the inductor are high enough for the application at hand since the voltage fluctuations will be seen across the inductor.

Anyways, not saying thats the problem, but if it is than thats a solution.

I don't think a choke would help. All it would do is add impedance and do nothing about fluctuations in the supply voltage, which is what the problem is. The problem is low frequency, because it causes visible fluctuations - in other words, the frequency of the noise is below 100 Hz.

The high frequency harmonics from industrial drives are a problem because they cause voltage distortion. This will overheat transformers, and drastically impair motor performance reducing efficiency and torque and increasing heat and vibration. In severe cases, it may disturb electronic equipment (e.g. if the waveform is sufficiently distorted to have 4 zero crossings per cycle instead of 2). As the drive isn't terribly concerned with its own voltage waveform, by putting an inductor in series with the drive, the high frequency harmonic currents are reduced, causing less voltage distortion at the power grid, but increasing distortion at the drive. In practice simple inductors aren't always used, instead tuned LC filters are used as 'harmonic traps' with a peak in impedance at the specific frequency.

Indeed, my voltage is very clean here. Even during episodes of flickering. See voltagedistortion.png (Legend: THD: White with blue halo, other traces: odd harmonics in ROYGBIV order)

There are a couple of episodes of flickering on this recording, and there isn't that much change in the overall harmonic content - just a lot of noise - presumably a representation of the low-ish frequency voltage flicker.

---

Meter issue: I've had no response yet to a formal complaint to my supplier. So, I'll shortly be lodging a formal complaint with the local electricity supplies regulation office. I also sent them a bill for my inconvenience in not having had correspondence within the maximum time allowed by law, and for having to take leave from work for their 2 broken appointments to read the meter, and for having to do the reasearch to find out that I was offered compensation. They haven't yet paid up. [Over here, customers are entitled to automatic compensation if an issue is not resolved with 10 days, if an appointment is missed, or if the compensation is not automatically paid, and the customer has to request it]. I count a total of at least 5 violations.

 

[futuristicmonkey]

Hell of a guess here, but take a look at the peaks of your voltage waveform -- they're slightly flat-topped. As I understand it, this may be a consequence of a substantial non-linear load. Do you have any data-centres nearby? The large amount of switching power supplies associating with data centres tends to have a negative impact on power quality.

-ben

 

[Mark R]

Just an update:

Over the 2 week holiday period, my power has been pristine. All the nearby shops, and businesses have been open, so it's not them.

My guess is that it was the Christmas shutdown of all the steel mills in the nearby towns.

Meter issue: My supplier has confirmed that they have been reading the wrong meter, and it is their fault. They've told me that actually, they've not been my supplier at all, and that another company has. They've passed my details onto the 3rd party supplier, but have told me that the 3rd party 'aren't very happy about it'. Sheesh. You'd think that the electricity industry would have an established way of dealing with swapped meter details - obviously, they don't.

 

[soccerballtux]

Well they do, except why bother following them? It's not like you have a choice in power companies.

 

[canis]

Hello. Can you please explain how you obtained this graph? (equipment, time scale, etc.)

Originally posted by: Mark R
Just wondering.

Every now and again, the lights in my apartment flicker - sometimes, they flicker like crazy for up to half an hour at a time. It's been a lot better since switching to CFL and halogen lights - low voltage halogens are almost completely unaffected. However, standard incandescents can be really bad at times - and if you want to dim them, well, you make it twice as bad. In between times, however, the power is stable as a rock.

This is definitely a problem throughout the whole apartment block, as we've had a few discussions on our forums about it. Happens at any time, including late evening and weekends.

I lived elsewhere in the city a couple of years ago, and I happened to notice similar flickering then - but at that point I had CFLs mainly, rather than halogen - and the power wasn't as good anyway, as the voltage would sag noticeably with big loads (as it was out in the suburbs). In my new apartment, voltage doesn't budge much, even when switching on a 7 kW load (unless it's on the same circuit). [I estimate supply impedance at my service to be about 20 mOhm].

Having monitored the voltage over a few days, I think that the voltage fluctuations are within legal limits, so I'm not sure I've got a complaint. However, I'm just interested in what might be causing it. The fluctuations can be quite significant (repeated 4-5 V fluctuation in RMS voltage several times per second).

Example trace: flicker.jpg The three traces are max/mean/min cycle by cycle voltage. So you can see episodes where there is noticeable rapid fluctuation where the lines separate. This is only a mild example, but I can't find a severe example at the moment.

So, what sort of thing can cause such marked flickering? It's clearly not anything within the apartment building. Could it be something so big, that's it causes power fluctuations across the whole city? I have also recorded sudden drops in grid frequency when these fluctuations start - e.g. 0.1V flicker voltage, then suddenly 4 V flicker, simultaneous with a drop of 0.05 Hz in grid frequency. So this does lend some credibility to it being something huge.

Any ideas, among electrically oriented sleuths?

 

[Mark R]

Originally posted by: canis
Hello. Can you please explain how you obtained this graph? (equipment, time scale, etc.)

The equipment is home made. For practical purposes it's a premium quality mains -> 18 V toroidal transformer, and potential divider which feeds a line-in port on my comp.

The software is just something I knocked up to perform this analysis. Some more pics: Pic1, Pic2, Pic3, Pic4

Time scale, I can't remember - but I think it's about 5 seconds per horizontal pixel, so about 1 hours for the whole graph.

 

[canis]

Do you have any real time graphs that show voltage drops severe enough to cause light flickering? If there is flickering the real time graph would show marked fluctuations.

Originally posted by: Mark R

Originally posted by: canis
Hello. Can you please explain how you obtained this graph? (equipment, time scale, etc.)

The equipment is home made. For practical purposes it's a premium quality mains -> 18 V toroidal transformer, and potential divider which feeds a line-in port on my comp.

The software is just something I knocked up to perform this analysis. Some more pics: Pic1, Pic2, Pic3, Pic4

Time scale, I can't remember - but I think it's about 5 seconds per horizontal pixel, so about 1 hours for the whole graph.

 

[Mark R]

Originally posted by: canis
Do you have any real time graphs that show voltage drops severe enough to cause light flickering? If there is flickering the real time graph would show marked fluctuations.

I don't display the realtime graphs of voltage. Mainly because, the fluctuations are short lived and difficult to see (but they are there).

I settled on measuring min/max/mean over 5 second intervals, because it shows up the fluctuations very clearly, and shows when they start and stop. It's just much easier to see the min/max lines diverging

The fluctuations are also easily seen on the the IEC short/long term flicker traces. Pic 4 is a not particularly dramatic example (that was a mild day), on more severe days, the short term flicker goes up to about 1.3 for a few minutes at a time.

 

[canis]

Originally posted by: Mark R

Originally posted by: canis
Do you have any real time graphs that show voltage drops severe enough to cause light flickering? If there is flickering the real time graph would show marked fluctuations.

I don't display the realtime graphs of voltage. Mainly because, the fluctuations are short lived and difficult to see (but they are there).

I settled on measuring min/max/mean over 5 second intervals, because it shows up the fluctuations very clearly, and shows when they start and stop. It's just much easier to see the min/max lines diverging

The fluctuations are also easily seen on the the IEC short/long term flicker traces. Pic 4 is a not particularly dramatic example (that was a mild day), on more severe days, the short term flicker goes up to about 1.3 for a few minutes at a time.

I can't prove flickering without a real time graph, but looking at the first graph the voltages don't seem to indicate an incadenscent light dimming problem assuming your voltages are rms. The largest drop in voltage I see from the mean is 4 volts. That also is momentary since the mean appears to smoothly decrease. What is the sampling rate you used for each 5 second interval? As for the mean voltages, it ranges from 246V to 240V, voltages sufficent to cause an incadescent to glow brightly without dimming. As for the flicker graph, can you please explain the vertical and horizontal axis?

 

[canis]

It is also possible your equipment is not up to snuff. The transformer is a poor choice due to the large L and significant temp dependant R. It will hide transients and cause drifting of voltage.

 

[Mark R]

Originally posted by: canis
What is the sampling rate you used for each 5 second interval? As for the mean voltages, it ranges from 246V to 240V, voltages sufficent to cause an incadescent to glow brightly without dimming. As for the flicker graph, can you please explain the vertical and horizontal axis?

Sampling rate is 96 kHz.

The flicker graph plots voltage flicker calculated according to the IEC 1000-4-15 method. Horizontal axis is 20 minutes per division. Vertical axis is irritability, where 1 is the threshold of irritability. The traces are short-term (magenta) and long-term (cream) flicker.

I accept that the transformer isn't ideal - it's not a proper potential transformer, but it's not like it's a cheap wall wart transformer.