# Harmonics in Power Systems

Fisherman199
Seeking the guidance of the Physics forums gurus, I was told this would be a good question to contribute to the community. Perhaps it will lead to an in-depth discussion of harmonics: Their creation, implications to power quality, and how to mitigate them.

I've been looking at some power quality captures I've gotten from the O&M department of my utility. We have a customer who's experiencing loss of power. I get these fairly regularly and usually it's an issue within the customer's electrical work. In some cases customers are tripping on three-phase when it's not needed or wanted, others issues have been the result of old cabling, others due to transformer insulation breakdown, and some are even caused by perturbation on the transmission system if the customer has a singularly sensitive tolerance. This particular instance is giving me some trouble mainly because they're having significant harmonic issues but I'm unfamiliar with LV filters enough so that making a recommendation is difficult. I always hesitate to place the blame on the customer's electrical work, despite that being the issue more often than not.

Does anyone have recommendations of LV harmonic filters?

Below are some details for those feeling adventurous.

Issue:

A utility customer is experiencing outages on sporadic days. The outages are complete losses of power and always occur between 11:00 and 11:15. There are not indicative brightening of lights.

Power Quality Details (55 day capture):
Current THD% (secondary of transformer): Ranges between 10% - 32%. Average of ~25% per phase
Voltage THD% (secondary of transformer): Ranges between 0.1% - 0.5%, Average of ~0.15% per phase
Harmonic component contribution (most prevalent): 3rd
Maximum return imbalance (non-faulted): ~4.2 Amps, 0.2 volts
Faulted instances (unknown cause): 2 with maximums at 700 amps at 19:32 EST and 230 amps at 12:51 EST
CBEMA impulses: 202
CBEMA impulses (outside thresholds): 15
Average Vrms (phase to phase): 282 volts rms

Details:
Service: 3 phase, 480 volts phase-phase, 277 volts phase-ground, #2/0 cable (assumed)
Transformer: 3 phase, Grounded-Wye/Grounded-Wye, core-form, non-tap changing, %z = 4.7%, x/r = 7
Location Electrical Drawings: Not Available
(can't place a percentage on it without the electrical drawings, which I don't have)
Circuit (upstream utility):
Max fault capability at service location (primary of transformer) (3 phase, bolted fault): 3200 Amps (apprx. by system computational model)
Resonant order of upstream, switched capacitor bank: ~9
Size of the capacitor bank: 900 kVAR
Conducting leading to service location (primary side of transformer): 397.5 ACSR Overhead (~4.6 miles worth, #4/0 neutral), #1000 XPLE Underground (~1045 feet worth, triplexed, conduit buried)
Station (transmission or distribution) interruptions during time-period of interest (55 days): None

EDIT: (OF NOTE): There are two other capacitor banks on this customer's circuit.

Last edited:

magoo
You really need to put together a rough one-line diagram of what the circuit looks like to get started.

Is the primary 12.5 kV or something else?
How far is the customer transformer from the substation?
Are there any interrupting devices like reclosers between the substation and the customer transformer?

When does the 900 kvar capacitor bank switch on and off? Are the other capacitor banks fixed or switched?

What is the kVA rating of the customer transformer?

Tell me more about the two faulted instances. Were they on the customer side or the primary side? Since the times reported for these two are not between 11:00 and 11:15, I'm guessing they were on the low voltage side.

"they're having significant harmonic issues"
What harmonic issues are they having? In my opinion, the THD measurements don't look that bad. What leads you to suspect that harmonics is the problem?

Fisherman199
2021 Award
What is switched, or what special thing happens at 11:00?
Is that when a train departs? or when a conveyor is loaded at the start of a shift?
Or is it on a mine site that shuts things down to blast at 11:00?

jim hardy
Staff Emeritus
I'm not sure of the economics, but the problem suggests an old fashioned motor-generator set providing 100% of the customer's load.

A motor, especially if overrated, would be relatively immune to damage by harmonics. The generator would provide clean power to the load. In addition, the m-g set could also be coupled to a diesel engine with a clutch to provide backup power during the outage. An induction motor can be started without synchronization. A synchronous generator with a voltage regulator could take care of variable reactive demands of the load. I'm presuming that precision frequency is not a load requirement.

Compared to schemes to filter harmonics in all conditions, plus deal with interruptions, the m-g set offers KISS simplicity. A Disadvantages would be unavoidable friction power losses in the spinning equipment, and lack of redundancy.

Fisherman199
Fisherman199
You really need to put together a rough one-line diagram of what the circuit looks like to get started.

Is the primary 12.5 kV or something else?
How far is the customer transformer from the substation?
Are there any interrupting devices like reclosers between the substation and the customer transformer?

When does the 900 kvar capacitor bank switch on and off? Are the other capacitor banks fixed or switched?

What is the kVA rating of the customer transformer?

Tell me more about the two faulted instances. Were they on the customer side or the primary side? Since the times reported for these two are not between 11:00 and 11:15, I'm guessing they were on the low voltage side.

"they're having significant harmonic issues"
What harmonic issues are they having? In my opinion, the THD measurements don't look that bad. What leads you to suspect that harmonics is the problem?

Yes, I thought of needing to add a little schematic last night. I drew one out some time ago. I'll add that below.

A current THD averaging around 25% isn't bad? That's what I meant by "harmonic issues." A poor choice of words.

• The faulted instances were on the low-voltage side of the transformer somewhere inside the facility. Their breakers didn't trip in either case.
• The capacitor bank I mentioned is a switched bank. the other two are fixed and switched and sized to 300kVAR and 1200kVAR respectively.
• There are no reclosers.
• I take the primary as 12.47kV. Either way, it's 12kV nominal.

What is switched, or what special thing happens at 11:00?
Is that when a train departs? or when a conveyor is loaded at the start of a shift?
Or is it on a mine site that shuts things down to blast at 11:00?

This is an office building for a water utility. Presumably a few server racks, desk computers, lighting, etc. This building is all electric. So heating, air conditioning, and water heating my be an issue. Again, I can't be sure. I've asked for electrical drawings and schedules. I was told those would not be provided. I have only guesses of what the load "is," how it's wired, and when it's active.

According to my contact, "nothing out-of-the ordinary happens at 11:00."

I'm not sure of the economics, but the problem suggests an old fashioned motor-generator set providing 100% of the customer's load.

A motor, especially if overrated, would be relatively immune to damage by harmonics. The generator would provide clean power to the load. In addition, the m-g set could also be coupled to a diesel engine with a clutch to provide backup power during the outage. An induction motor can be started without synchronization. A synchronous generator with a voltage regulator could take care of variable reactive demands of the load. I'm presuming that precision frequency is not a load requirement.

Compared to schemes to filter harmonics in all conditions, plus deal with interruptions, the m-g set offers KISS simplicity. A Disadvantages would be unavoidable friction power losses in the spinning equipment, and lack of redundancy.

This is a good idea. Probably not something the customer would welcome, though.

#### Attachments

• schematic for upline ciruit 2.JPG
80.9 KB · Views: 442
Last edited:
IMO - In all cases here a Power Quality study is warranted ( Disturbances, Harmonics, load chnages, voltage fluctuations - etc.) - as all the rest is just really guessing at what is going on.

Fisherman199, Klystron and anorlunda
Staff Emeritus
Does the power quality delivered to the customer (including outages) meet the minimums required by the PSC in that state? If not, then the cost of remedies falls on the utility.

If there is a dispute over who is responsible for problems, who bears the burden of proof?

Gold Member
Dearly Missed
I've never used harmonic filters so cannot give a first-hand recommendation

searching on "lv harmonic filters" turned up the usual industrial suppliers
http://www.apqpower.com/products/low-voltage/harmonics/active-harmonic-filters
https://new.abb.com/high-voltage/capacitors/lv/power-quality-filters

however, this statement
The outages are complete losses of power and always occur between 11:00 and 11:15. There are not indicative brightening of lights.
paints in my mind an event where the lights blink and the computers all crash.
It used to happen in my plant every afternoon of thunderstorm season about 2PM .
Even a barely perceptible flicker of the lights provoked an immediate and deafening uproar of secretarial voices shrieking AW, @*&^^%#! as their word processors crashed.

so I'm not at all convinced the trouble is "Harmonics"

and the descriptions "Complete losses of power" and "not indicative brightening of lights" seem pregnant with clues.
Did they mean "dimming of lights" ?
If not, what is the symptom they observed and interpreted as "complete loss of power? " ?

Given that PC's are great "missing cycle detectors" , faster than the human eye, to wit my anecdote above
i'm suspicious about switching transients especially around large capacitor banks.

We need to further characterize the nature of these 'complete power losses'. How much voltage sag, for how long...?

If it's coming from the power system a motor-generator set is a darn good answer.
We installed one on our plant simulator so it would be useable during thunderstorm season.
We powered it with a 150hp induction motor for simplicity and sized the flywheel to deliver 100kw for one second with two hz drop, about 39 inch diameter by 4 inches thick.
Since the computers used line frequency for their time of day clock they lost around thirty seconds a day due to slip.
We chose rotating machinery over a UPS so as to not have to maintain batteries and inverters.

Beware of inverters for they don't do well at providing inrush current to computers.

Which leads me back to "We need to further characterize the nature of these 'complete power losses'. How much voltage sag, for how long...?"
and "Why always late morning?"

old jim

Klystron
Gold Member
Dearly Missed
A filter can clean up a distorted line cycle but it can't replace a missing one.

2021 Award
We have a customer who's experiencing loss of power.
What is meant by, and how is a “loss of power” observed?

If the two events in the 15 minute window beginning at 11:00 are a simple coincidence then why is the time thought to be significant. Is that limited sample just adding to the confusion?

A problem cannot be solved rationally from the existing observations if the observed data is unreliable or poorly defined. Then you must ignore some of the unreliable data and guess at a cause or solution. That explains this witch hunt for mystical harmonic causes.
Third harmonic current can be explained by simple bridge rectifiers charging storage capacitors in many switching power supplies.

The only sensible alternative to guessing is to set things up to gather and log reliable data, then wait for it to happen again.
There is usually a source of historical information recorded in the logs of the UPSs in a building. Find out where all the UPSs are in the building, and which ones have been logging “brownouts” or power quality.

jim hardy and Klystron
Staff Emeritus
lv harmonic filters" turned up the usual industrial suppliers
In this morning's news, ABB sold their power sector businesses to Hitachi. But with Westinghouse gone and GE mostly gone, the number of big suppliers is dwindling.

A future crisis in supply of critical power system parts and expertise is foreseeable. Renewable advocates would like to layer 300,000 miles of HVDC in North America. But the engineering expertise in power transmission, like expertise in nuclear plant design and construction, may have dissipated by the time we need it.

jim hardy and Tom.G
Fisherman199
IMO - In all cases here a Power Quality study is warranted ( Disturbances, Harmonics, load chnages, voltage fluctuations - etc.) - as all the rest is just really guessing at what is going on.

Could you clarify what you mean by power quality study? How broad are we talking? Investigation of the whole feeder?

I've installed a power quality meter (https://powermonitors.com/product/revolution-power-quality-recorder/) and other than Current THD things were found to be within tolerances.

Gold Member
Dearly Missed
I've installed a power quality meter (https://powermonitors.com/product/revolution-power-quality-recorder/) and other than Current THD things were found to be within tolerances.

now you need to sit and wait for it to capture one of those "losses of power"

choosing trigger settings is the tricky part of using those things

Fisherman199
now you need to sit and wait for it to capture one of those "losses of power"

choosing trigger settings is the tricky part of using those things

I've done this twice thus far. I've got data spanning about 70 days. There was no blink/loss/magic within that time frame. The day after I go get the meter to look at the records they have another incident. Same time as all the others, about 11:02

Gold Member
Dearly Missed
There was no blink/loss/magic within that time frame. The day after I go get the meter to look at the records they have another incident. Same time as all the others, about 11:02

Once Again --- Murphy is an optimist !

that's how it goes in troubleshooting.

Interview the people for clues - do lights blink? do PC's crash ? Any UPS's nearby ? Any big computers in the building?(old ones of 80's vintage , before the days of PFC circuitry, had first half cycles inrush in the hundreds of amps enough to seriously clip the peaks off incoming sinewave if not on a real stout source)

First half cycle inrush to a transformer load is real sneaky, and you as a power guy know that.
It depends so strongly on where in the line cycle you close the power switch.
If you close real near the voltage zero crossing the core will saturate just before the end of first half cycle and current goes sky high
If you're unlucky that'll trip the overcurrent protection of an inverter type UPS so equipped, resulting in an outage of probably a cycle or two while it waits for the fault to clear

that it's within minutes of 11AM suggests to me it's the result of some manual action prescribed by procedure
i'd want to know what controls those big capacitor banks

and what machinery gets turned on in the water plant about that time of day ?.

A humble Sola* transformer, which somebody might have added to protect a computer someplace in that treatment plant,
when energized at the zero crossing will pull 20X nameplate amps when it saturates.
But it's only about 1 in 10 energizations that happen that close to zero crossing.
So it can be a real bear to figure out.
Luck is a factor in capturing the transient. I did it in the old days before digital storage with .a Polaroid 'scope camera . Took a lot of tries because the first few of those traces i captured went way offscreen - they were astonishingly huge but only on ten% of the closures..

The fix for that is to change its input voltage tap so it's running at about 70%-80% rated input voltage.
You can also protect a weak source against computer inrush with a Sola - the Sola limits its output current to about 125% rated.
*( "Sola" is a trademark , their claim to fame is their ferroresonant voltage regulating transformer .
We've got sloppy and use that name to mean that generic device, sort of like calling all cola flavored soft drinks "Coke" . So i have to honor Sola's right to their name. Google Sola CVS for a description - jh)

Anyhow - put your monitor back and hopefully it won't be so long until the next transient.

Meantime query the equipment operators as to what goes on late mornings.
In our plant i found the giant coffeepot plugged into our instrument bus...

Staff Emeritus
A utility customer is experiencing outages on sporadic days. The outages are complete losses of power and always occur between 11:00 and 11:15. There are not indicative brightening of lights.

I thought at first that you meant a complete blackout. Please clarify what you mean.

Tom.G and jim hardy
magoo
I don't think the problem has been adequately described yet to suggest what the remedies ought to be.

You are asking a bunch of people who are remote from your customer's location so we're depending on you to be our eyes and ears. How big is the office building? 1 floor, 2 floors, 3 . . .

I assume the office building has been there for a number of years and this is not a fairly new installation. Is that correct?

Some harmonic sources for an office building include PCs, printers, some HVAC, some lighting, etc. Again, how big a building?

The reason I made my comment about harmonics is that the voltage distortion is minimal. So I wouldn't expect problems with extra zero crossings on the voltage or something like that. You have current distortion but that only indicates that you have some nonlinear loads. So what? The voltage is pretty stiff. That shouldn't be a problem.

Are there other circuits served from the 12.5 kV bus at the substation?

What is the basis of saying that the approximate resonant point associated with the switched capacitor bank is close to the 9th harmonic?

What type of control is used for the switched capacitor? Voltage, time clock, etc.?

CBEMA data. Of the 15 impulses outside the threshholds, how many of them are on the low voltage side and how many are on the high voltage side?

The details on the faulted instances looks like it was a rabbit we don't have to chase. However, it would be useful for you to sit down with a customer representative and see what exactly happened during these instances. Was something damaged, or was specific equipment affected, etc.

Is this "problem" unique to just one customer?