Harmonics

Status
Not open for further replies.
I

io748

Member
Location
Nashville, NC
I just observed something I was not expecting with my power quality monitor, hopefully someone can help me understand it. I have monitored utility power in our plant several times over the past several months and have never logged anything over 3.5% THD. I connected the monitor to a receptacle that is supplied from the plant UPS system, expecting to see something better than that. What I saw was 11% THD, and an ugly waveform. Most of the distortion was 1st order. A desktop PC is plugged into the receptacle also, via a surge suppressor.

The plant UPS system is 150 kva and is distributed throughout the plant to almost exclusively computer loads, including the server room. I know that computers are a contributor to harmonics, so maybe the fact that almost all of the load is computer load is a contributor? Still seems high though. Any ideas/comments?
 
io748 said:
I just observed something I was not expecting with my power quality monitor, hopefully someone can help me understand it. I have monitored utility power in our plant several times over the past several months and have never logged anything over 3.5% THD. I connected the monitor to a receptacle that is supplied from the plant UPS system, expecting to see something better than that. What I saw was 11% THD, and an ugly waveform. Most of the distortion was 1st order. A desktop PC is plugged into the receptacle also, via a surge suppressor.

The plant UPS system is 150 kva and is distributed throughout the plant to almost exclusively computer loads, including the server room. I know that computers are a contributor to harmonics, so maybe the fact that almost all of the load is computer load is a contributor? Still seems high though. Any ideas/comments?
Click to expand...

The UPS's of that size usually have a filtering circuit on the output that holds the voltage THD to </=5% at total nonlinear loads. Computers and servers are nonlinear loads due to their switching power supplies.
 
io748 said:
The plant UPS system is 150 kva and is distributed throughout the plant to almost exclusively computer loads, including the server room.
Click to expand...

As the other posters have pointed out, the UPS could have a problem. Some questions & Comments:

1. What vintage and brand UPS? When was the last maintenance and are there records of that?

2. Is it a single module or redundant? If more than one module, try shutting them off one at time and see what that does to your readings.

3. Do you have the same readings at the output at the UPS?

4. You mentioned the loads are almost exclusively computer loads, what are the other loads? You might try turning off each non-computer loads and note the results.

5. Is it possible that there are smaller point-of-use UPS being fed by the 150 KVA UPS? That can cause strange readings. Remember, control systems may not have a visible mini-UPS, sometimes they're built in.
 
brian john said:
11% Voltage THD, I'd have that UPS checked ASAP.

What do you measure at the UPS.
Click to expand...
I agree. I think measuring at the UPS output Voltage THD at the terminals should be the next step. The impedance of the distribution at the point where it was measured could be a contributing factor, particularly for the higher order harmonics.

If I were doing this, I'd also observe the actual waveform on an oscilloscope. A THD of 11% would make the waveform distinctly non-sinusoidal in appearance and might give some clues on where to look next.
 
John,

It is a Liebert series 600 UPS, from 1994. We have regular maintenance on it, in fact the batteries were replaced last week. It has only one module. I checked again right on the bus bars of the output and it was much better, 3.5-4%. Manual says it will be no worse than 5% for 100% non-linear loads. I am monitoring in another location right now, it is connected to the output of a small local UPS that is fed by the large UPS system. THD was 10% at that location when I left. When I pull it tomorrow I'll post a picture of the waveform.

Does any of this give any more insight into the situation, or bring up any other things I could check?
 
io748 said:
John,

Does any of this give any more insight into the situation, or bring up any other things I could check?
Click to expand...

Further to my comment in post #8, it looks like the distribution system is what's causing your higher levels of THD at the point of use.

I don't think there is a quick fix . If indeed fact a fix is needed.
The non-linear computer loads probably use switch-mode power supplies which are usually fairly tolerant of distorted waveforms, particularly given that they are the cause of the non-linearity in the first place.
 
io748 said:
John,

It is a Liebert series 600 UPS, from 1994. We have regular maintenance on it, in fact the batteries were replaced last week. It has only one module. I checked again right on the bus bars of the output and it was much better, 3.5-4%. Manual says it will be no worse than 5% for 100% non-linear loads. I am monitoring in another location right now, it is connected to the output of a small local UPS that is fed by the large UPS system. THD was 10% at that location when I left. When I pull it tomorrow I'll post a picture of the waveform.

Does any of this give any more insight into the situation, or bring up any other things I could check?
Click to expand...

In my opinion, part of the issue is that you don't see to have any long term data on the harmonics beyond the readings you recently took. If everything seems to be working, you need to build a database to analyze.

I'd look at any data the UPS technician has noted on the maintenance report. They typically log some values (but probably not harmonics). Look for any values that have changed over a period of time, that might be useful. But as the other posters mentioned, the changes are more likely in the distribution system, downstream of the UPS,

That being said, what are the non-server loads on the UPS and have they changed over time? Or have the IT loads changed?
 
Last edited:
John,

After further review this UPS system actually does not support the servers (servers are supported by two other systems). This UPS system is distributed throughout the assembly line to the terminals that handle assembly information (what parts need installed at each station, assembly instructions, etc.) So the loads are computers, terminals, printers, bar code scanners. There is not *supposed* to be anything else on it (that is not always the case).
 
io748 said:
John,

It is a Liebert series 600 UPS, from 1994. We have regular maintenance on it, in fact the batteries were replaced last week. It has only one module. I checked again right on the bus bars of the output and it was much better, 3.5-4%. Manual says it will be no worse than 5% for 100% non-linear loads. I am monitoring in another location right now, it is connected to the output of a small local UPS that is fed by the large UPS system. THD was 10% at that location when I left. When I pull it tomorrow I'll post a picture of the waveform.

Does any of this give any more insight into the situation, or bring up any other things I could check?
Click to expand...

There's a problem right there..smaller UPS tend to have very high THD. Measure without any other UPS in line and see what you get.

Oh, and if there are several small UPS "downstream" from the larger one that can cause the THD readings you're seeing, as they can backfeed distortion into the incoming line.

Finally, ask the client if the downstream small UPS are really necessary, and if not, remove them.
 
mxslick said:
Finally, ask the client if the downstream small UPS are really necessary, and if not, remove them.
Click to expand...
In reality, downstream UPSs of the consumer and office variety will, if they are ther sole source of supply to a device, reduce the overall availability of power to that device.

Never recommended in that configuration.

For a device with dual power inlets capable of operating off either, having one fed from the "proper" UPS and the other from a "toy" UPS does seem to be a good setup as long as there is some testing and maintenance done on the little UPSs (which are often forgotten about...)
 
I monitored a UPS receptacle (from the main system, not a small local unit) and measured 11% THD. Almost exclusively 3rd harmonic. THD measured at the bus bars of the UPS unit is 3%. From the reading I have done this harmonic is generated by the power supplies of computer equipment, and can be detrimental because the 3rd harmonic currents actually mathmatically add in the neutral, and can heat up your neutrals, transformers, etc. Question is, can it cause problems with communications or data transfer? Is 11% THD, primarily 3rd harmonic, something I should be concerned about?
 
io748 said:
I monitored a UPS receptacle (from the main system, not a small local unit) and measured 11% THD. Almost exclusively 3rd harmonic. THD measured at the bus bars of the UPS unit is 3%. From the reading I have done this harmonic is generated by the power supplies of computer equipment, and can be detrimental because the 3rd harmonic currents actually mathmatically add in the neutral, and can heat up your neutrals, transformers, etc. Question is, can it cause problems with communications or data transfer? Is 11% THD, primarily 3rd harmonic, something I should be concerned about?
Click to expand...
I would have thought that 3rd harmonic was much too low a frequency to interfere with comms and I've seen no evidence that it does.
But it can be a problem because, as you say, they add arithmetically in the neutral and can (and do) cause overloading. I have come across a couple of installations where this was a major problem, not because of computers but controlled lighting. Dimmers. But it is essentially the same problem. Third and other triple n harmonics.
 
is the UPS the back-uo (offline) type or is it the online type? If the UPS is the online (line interactive) type, it will produce harmonics as it will constantly convert ac to dc to ac. especially if this does not have any filtering system.
 
Status
Not open for further replies.
Top