Mysterious vibration in long conduit between solar inverters and main service panel

Status
Not open for further replies.
Redwood Infrared said:
... - it is not feasible to pull the wire and put a twist in it without replacing it or adding a splice; there's just not enough slack in the line to do that. ...
Click to expand...
There's probably enough slack for a gentle twist -- one turn per five or ten feet? -- if you neatly lay the conductors, eliminate "braids", and bundle them tightly. One or two turns per segment of conduit is probably sufficient. (if it's an electromagnetic problem at all)
 
ggunn:

I have already strongly recommended the structural support measure you suggest here and

ggunn said:
Why would a tenant's UPS have any effect on the solar installation? The solar only runs when the grid is up and the UPS only runs when the grid is down. The UPS should also have a transfer switch that disconnects it from the rest of the service (and therefore the PV) when it is running.

My guess is that the conductors are vibrating a bit, and over time the vibration has loosened the conduit attachments so that they are now rattling. Didn't you say that the conduit has fewer supports than it should and that was going to be corrected? Add the new supports and tighten the old ones and see what that gets you.
Click to expand...

But beyond that, we are detecting strong magnetic fields along the length of the conduit that we believe are anomalous and likely point to the root cause of the vibration. Increasing structural integrity might prove to be a mitigation but not resolution of a condition the building owner finds undesirable. They are a regional leader in green building development so truly understanding matters of this nature is important to them.

And as far as what the UPS has to do with it, I am told there have been cases of similar situations where it turned out that THD introduced into a building's power distribution network by a UPS was definitively identified as the root cause of undesirable vibration and noise in cable ducts. As such I have not yet concluded anything but I will absolutely be running an experiment to see if it makes any difference when we take the UPS offline while the PV array is producing and the undesirable noise condition exists. If the noise stops when the UPS goes offline, then we will have identified our root cause and can develop more meaningful mitigations from there.
 
This is exactly why I prefer microinverters installed on the roof. 500 volts dc coming off of panels to a main inverter. Which then had to have a bunch of labels on the conduit. The conduit has to be EMT or RMC. Also you could be getting vibrations from the inverters themselves.


Sent from my iPhone using Tapatalk
 
180619-2130 EDT

Redwood Infrared:

I think you go off in two many different tangents instead of focusing on the very likely causes..

You have a high sound level. That means some reasonable amount of acoustic power. Possibly much greater than 10 W, and possibly not as high as 100 W. This is a lot of random power and quite likely means it originates from a high power source. Quite possibly your sound radiator is is the conduit, but you could also have energy mechanically coupled to other radiators.

It is possible that a mechanical resonance is part of the problem.

I don't have big pipe to work with for experiments. However, I did some tests on 1/2" thinwall steel conduit. My experiments are not well controlled and values may be in substantial error, but the results have some degree of consistency.

With the conduit clamped at one end to my drill press table and the other end resting on the edge of a tool box and the accelerometer taped to the middle I measured 5.9 Hz.

Next two cantilevered tests. First, was 5 ft (60"), and result was 7.7 Hz. Second, was 90", and result was 3.3 Hz. (90/60)^2 = 2.25 . The 3.3 * 2.25 = 7.4 . Looks to vary as the square of the ratio.

When the member is initially excited to oscillate there are harmonics superimposed on the fundamental. These die out fairly quickly.

I would suggest that you see where your 4" conduit tends to vibrate. Use a span length equal to the support distance in the building.

.
 
180619-2241 EDT

In a mechanical system, if mass remains constant and the support spring gets stiffer, then the resonant frequency increases.

If two wires carry the same current, then as the wires come closer together the force between the wires increases.

In an AC circuit the force between the wires will have a constant value with a superimposed double frequency force component.

Current direction in one wire relative to the other will determine whether the force is attractive, or repulsive.

Two parallel wires close together with current flowing in the same direction produce additive magnetic fields, and thus as you move away from the wires there is a substantial magnetic field.

When the currents are in opposite directions, then the external magnetic fields tend to cancel.

You have 120 Hz sound energy. That comes from synchronized 60 Hz currents.

.
 
180620-2425 EDT

New experiment to demonstrate electricql excitation of mechanical vibration of 10' 1/2" thinwall.

The conduit is c-clamped at one end to a drill press table. The other end freely rests on a moderately sharp toolbox edge. The accelerometer is at the 5' point. A hairpin loop of #16 (not much mass) extends thru the entire conduit length.

A 16 V pulse of energy from a 50,000 ufd capacitor thru the wire loop is used to shock excite mechanical oscillation of the conduit. For a short time 50 Hz showed up with about 1/8 this (50/8) for a longer time. Correlates with mechanical excitation in prior experiments.

Note: this is an impulse excitation of a tuned circuit. I don't have a high current source with variable frequency to continuously excite the mechanical system. I suspect your conduit system has a mechanical resonance around 120 Hz, but likely would not be the fundamental resonance.

.
 
180621-1432 EDT

Another experiment. This time with 3/4 thinwall.

Fundamental about 7.6 Hz, next major spike at 63.4 Hz, ratio 8.34 from FFT. Not an integer ratio.

When viewing the y-t trace (y axis vs time) my best count ratio between the two superimposed waveforms is 8.5 .

The 3/4 tubing is a higher resonant frequency than the 1/2 as I think I would expect.

A more controlled experiment on the 1/2 and using a cursor to measure frequency instead of eyeball the results are --- 6.6 Hz, 49.1 Hz, ratio 7.44; 6.6 Hz, 86.6 Hz, ratio 13.0 .

What happens when you go to 4"? My expectation is a higher frequency.

If I take my above diameter ratio (0.75/0.5 = 1.5), then take the sq-root of 1.5 the value is 1.225 . Multiply 6.6 by 1.225 and the result is 8.08 . This is close to 8.34, but may not be a good predictor.

4 divided by 1/2 = 8 . The sq-root of 8 is 2.82. Multiply 6.6 by 2.83 and the result is 18.7 . Then by 8 and the result is 149. High compared to 120, but my approach is just a wild estimate. Your support points are greater than 10 ft and that lowers the frequency. Need to know if your supported conduit resonates near 120 Hz.

From the previous experiments with a single turn loop it is fairly clear you can excite the tube to vibrate.

Exciting at resonance can produce a large audio signal.

.
 
Magnetic Flux in Transformer?

Magnetic Flux in Transformer?

I just stumbled on this thread and have been involved in electrical acoustic noise for 30+ years, off an on. In general, it is iron that vibrates due to magnetostriction. A step-up/down or buck/boost transformer core will "hum" at 120Hz with strong 240Hz and 480Hz tones. However, magnetistriction is due to the magnetic flux created by voltage, not the current, so it should be constant. Are there any transformers in the building that could be the source of the noise?

If the current harmonics from the inverters cause a strong change in voltage, due to a "weak" utility source, this harmonic voltage will cause every transformer, iron core ballast, and motor to be noisy. A harmonic filter can lessen the current harmonics, thus lowering the voltage harmonics, thus reducing the acoustic noise.

Also, are there any VFDs in the building driving motors? A VFD creates a messy waveform which causes the iron in a motor to vibrate noisily.
 
180623-1123 EDT

Continuing my vibration experiments.

With the conduit (basically a thinwall steel tube) I am seeing a vibration at about 8, but not exactly, times the fundamental. Most of my experiments have been with the tube supported at the ends, but only one is rigid. The fundamental vibration is with the midpoint moving up and down.

With the 10 ft thinwall (both 1/2 and 3/4) I saw a fundamental below 10 Hz, then another major spike around 8 times this, but not an integer multiple. I don't know the origin of this higher frequency component.

I have only a limited supply of different items. My next experiment was with a 1/2" diameter solid round bar stock 6 ft long of 6061 aluminum. Here I measured 9.5 Hz for fundamental, and harmonics at 19.25, 28.0, 37.6, and 49.75. These are clearly 2 nd, 3 rd, 4 th, and 5 th. Then a larger spike at 85.75 or about 9 times the fundamental. On thinwall I did not see 2 nd thru 5 th harmonics,

Where does this 8 or 9 times vibration come from?
.

Redwood Infrared it would be interesting to know what resonance points exist for your $" conduit.

.
 
gar, thanks for all your continued thoughts, experiments, and comments on my issue. My apologies for being unresponsive as I've been working away with limited internet access for the last several days. I'm going to have to read through your posts several times to get my arms around it all but will take it under advisement and incorporate where I can.

Our budget on this particular project is somewhat constrained and right now we are still waiting for the tenant with the UPS to get their protocols down so that they are 100% confident they can safely execute a shutdown test in the afternoon during high solar production conditions; they've tried at night twice and ended up having to reboot servers both times, which would be unacceptable to them during business hours (they are a medical organization with highly critical IT). We think they've got it figured out now but they still need to execute a successful dry run before we can move forward. We have our live test tentatively scheduled for next Monday afternoon. I wish we could move faster but there are a lot of players involved.

At this point since there is a significant probability the UPS is involved, we have elected to put all other elements of the investigation on hold until we have the results of that test. If it's the UPS, then we either get the tenant to install filtering devices or else replace the UPS (hopefully under warranty since it's less than 2 years old and obviously causing issues in a pre-existing PV system).

If it's not the UPS, then at that point we will resume experiments to try to identify the root cause of the fields causing the vibration. Along they way we will also be performing several minor repairs that are more likely to be symptoms rather than the root cause, and we are also recommending they reduce the EMT support span from 12 feet to 6 feet throughout its entire 250' run through the building.

I'll try to post to this thread when I can however I'll be working remotely most of next week with little or no internet so it might be July 9 before I'm able to comment much more. But trust that I will definitely not let this discussion die before the investigation is complete.

Thanks again everybody!

gar said:
180623-1123 EDT

Continuing my vibration experiments.

With the conduit (basically a thinwall steel tube) I am seeing a vibration at about 8, but not exactly, times the fundamental. Most of my experiments have been with the tube supported at the ends, but only one is rigid. The fundamental vibration is with the midpoint moving up and down.

With the 10 ft thinwall (both 1/2 and 3/4) I saw a fundamental below 10 Hz, then another major spike around 8 times this, but not an integer multiple. I don't know the origin of this higher frequency component.

I have only a limited supply of different items. My next experiment was with a 1/2" diameter solid round bar stock 6 ft long of 6061 aluminum. Here I measured 9.5 Hz for fundamental, and harmonics at 19.25, 28.0, 37.6, and 49.75. These are clearly 2 nd, 3 rd, 4 th, and 5 th. Then a larger spike at 85.75 or about 9 times the fundamental. On thinwall I did not see 2 nd thru 5 th harmonics,

Where does this 8 or 9 times vibration come from?
.

Redwood Infrared it would be interesting to know what resonance points exist for your $" conduit.

.
Click to expand...
 
Redwood Infrared said:
... we are also recommending they reduce the EMT support span from 12 feet to 6 feet throughout its entire 250' run through the building.
Click to expand...
It occurs to me that you might also suggest that they vary the spacing between supports if you think mechanical resonance could be a contributor.
 
beanland, welcome to the thread and thanks very much for your comment.

beanland, welcome to the thread and thanks very much for your comment.

While I haven't done a very sophisticated audio analysis, just trying a variety of smartphone app spectrum analyzers, we get pretty consistent results pointing to 120Hz as the fundamental audio frequency. That frequency does not appear to change; it is only the volume or amplitude that varies. It seems pretty constant during clear skies but occupants point to it being intermittently somewhat louder on windy partly cloudy days, ramping up and down in a manner that correlates pretty strongly to observable fluctuations in current. I can't say I've observed this myself, but they seem to think it will be loudest as current is ramping up but then quiet down somewhat at peak current once that current stabilizes under clear sky. To me it seems possible they erroneously perceive change in volume as peak volume but frankly it's difficult to measure and impossible to recreate unless weather conditions cooperate and so far we just haven't had a lot of luck in that regard throughout my half dozen visits to the building.

Other than the solar inverters and the one tenant UPS, there are no transformers nor any VFDs that I know of inside the building. There are several tenants with security concerns so I have not yet completed a survey of every subpanel/server room, etc. It is possible I might find a small buck boost or VFD in one of those suites so I've not eliminated that possibility but there was nothing highly evident in power quality monitoring at the building's main service panel pointing to anything like that. The one 10kVA UPS that we have identified remains a variable that we are actively pursuing and we hope to have its affect on the hum fully characterized by this coming Monday afternoon.

There is a utility pad transformer located just outside the building approximately 5 feet away from the building MSP closet, and that transformer does also emit a substantial 120Hz audible hum. The utility has conducted PQM at three panels within the building including the MSP and they seem confident the problem does not originate with them but I have put in a request to have a lineman open up their transformer so we can verify the integrity of the ground and neutral bonds in that unit.

I will state that my observations point to the aluminum conductors as the main mover in the vibration equation. Clearly the sound is that of the conduit but we think it's the conductor vibration being amplified by the conduit, not the conduit vibrating and causing the conductors to move. This just seems intuitively correct based on the tactile magnitude of the conductor vibration at the terminal end compared to what you can feel in the conduit at any point along its path.

beanland said:
I just stumbled on this thread and have been involved in electrical acoustic noise for 30+ years, off an on. In general, it is iron that vibrates due to magnetostriction. A step-up/down or buck/boost transformer core will "hum" at 120Hz with strong 240Hz and 480Hz tones. However, magnetistriction is due to the magnetic flux created by voltage, not the current, so it should be constant. Are there any transformers in the building that could be the source of the noise?

If the current harmonics from the inverters cause a strong change in voltage, due to a "weak" utility source, this harmonic voltage will cause every transformer, iron core ballast, and motor to be noisy. A harmonic filter can lessen the current harmonics, thus lowering the voltage harmonics, thus reducing the acoustic noise.

Also, are there any VFDs in the building driving motors? A VFD creates a messy waveform which causes the iron in a motor to vibrate noisily.
Click to expand...
 
180629-2154 EDT

Redwood Infrared:

Based on experiments I ran and theory the following points apply:

1. A conductive loop AC current will cause a mechanical vibration of the loop at twice the AC frequency.

2. Experimentally I could see 120 Hz mechanical vibration of the current loop with a 60 Hz sine wave current.

3. A 10 ft long thinwall steel conduit has a fundamental mechanical resonance in the 6 Hz range, and another around 50 Hz. I don't known the origin of the 50 Hz. Seems to be a 7 to 8 to 1 ratio of the fundamental.

4. The mechanical resonance was obtained with an accelerometer and mechanical tapping of the conduit. You can see and feel the fundamental.

I have previously suggested doing bench testing. This gets you away from the actual installation, and allows doing controlled experiments. This is simple to do and much less costly than playing around onsite.

Get 12 ft of your conduit, suspend on non-restraining supports at the ends (strings or vee like soft under supports), tap the center, and see where there are resonance points.

I am suspecting there is a resonance near 120 Hz. If this is so, then insert a loop of #12 or larger wire, and run a large 60 Hz current thru the loop from a transformer with a low output voltage. This should excite vibration at 120 Hz. Even though the mechanical resonance is not at 120 Hz you may still get substantial sound. What I saw as an about 8 times higher resonance point on the 1/2" conduit, may be at a different ratio on a larger pipe.

Rigid termination of the conduit instead of the soft termination should raise the fundamental resonant frequency. Also conduit added to each end will have an affect on resonance.

You really need to do offsite experiments to get a handle on how various parameter changes vary sound levels or resonance points.

.
 
Redwood Infrared said:
The installing electrician claims to have already been through everything and checked torque on all connections. He didn't comment however regarding tightness of structural fasteners along the conduit. They did try the old spray foam in the conduit trick but does that ever work?
Click to expand...
This is what GoldDigger was referring to. The wires themselves will vibrate inside of the conduit due to the magnetic fields surrounding them. Twisting them together (also called "triplexing") will reduce the creation of these forces and hold the conductors together.
 
Status
Not open for further replies.
Top