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

[Redwood Infrared]

Is the source of the noise the conductors vibrating in the conduit or the conduit vibrating in its supports?

I would say the source of the noise is the vibrating conductors but it is amplified by the conduit vibrating in its supports. Qualitatively, you can feel the conductors at the ends of the conduit vibrating much moreso than the conduit itself. However it seems very interesting that the fields I am measuring with the Guass meter are much stronger along the conduit than along the exposed conductor at either end.

I should have mentioned by now, the electrician acknowledged late this week that he wished he'd gone with supports on every steel building frame member at 6 ft centers rather than skipping every other one and going 12 feet. I guess the code says every 10 feet so he knows this is insufficient and is probably going to add more supports but that work is not scheduled yet.

Even if that does dampen the vibration, I still don't think it will get us to the root cause. I'm still trying to answer the question, are fields of the magnitude I am measuring along the entire length of conductors normal in this type of installation?

 

[GoldDigger]

Even if that does dampen the vibration, I still don't think it will get us to the root cause. I'm still trying to answer the question, are fields of the magnitude I am measuring along the entire length of conductors normal in this type of installation?

Magnetic forces between current carrying wires, with the force proportional to the square of the current, are normal and expected. The total field at some distance from a balanced three phase bundle, on the other hand, should be quite small.
The resulting motion of the wires will be greater where there is no lateral support of the wires for long distances (i.e. long straight runs of wires that only loosely fill the conduit. Whether the magnitude of vibration that you see is unusual is a question I cannot answer.

If there is no underlying anomaly, the best way to reduce the vibration is to twist the wires, which will keep them from moving relative to each other. If the individual wires cannot vibrate there is no reason for them to cause the conduit to vibrate.

 

[gar]

180617-1415 EDT

Redwood Infrared:

You are now making a definite statement that the sound (noise) has a fundamental frequency of 120 Hz.

The force from an AC sine wave electromagnetic field working against a constant magnetic field is
force = Constant * Ipeak * sin wt

take two electromagnets and work one against the other with the same current in both and the force is (where C2 is a different constant)
force = C2 * Ipeak * Ipeak * sin wt * sin wt
or
force = C2 * Ipeak^2 * (sin wt)^2
or
force = C3 * (1 - cos 2wt)

In other words the force is unidirectional and of double frequency. For a 60 Hz current the force frequency fundamental is 120 Hz, and has an average DC component.

A single straight wire is an electromagnetic. Place two close together and with appropriate current these will force each other apart. Thus, we have from a single phase 60 Hz source a 120 Hz force forcing function.

So the questions in your case are:
What is the single phase current source?
And what is the mechanism of coupling the double frequency to the conduit?

.

 

[Ingenieur]

%THD for voltage and current at both the inverters and the ACSD is within normal range, under 5%. I'm not sure why it's so high for the MSP - The building doesn't have any big motors spinning unloaded so not sure what would cause really poor power factor. Is this unusual? Is it a function of loads or the grid? The PV is big enough they're a net producer by a wide margin under these conditions so don't understand what's going on in this chart.

what makes you think the pf is bad?

 

[GoldDigger]

what makes you think the pf is bad?

Junkhound noted " MSP currents are leading the inverter currents, and have huge THD". That does not, by itself, translate to "bad" PF, though.
Unless these inverters are units that allow the output PF to be controlled, the inverter current should be in phase with the grid voltage. If the MSP currents are capacitive while the inverter currents are (negative) resistive, that implies capactive local loads somewhere.
It is an open question whether POCO is helped or harmed by a net capacitive service connection here.

 

[Ingenieur]

Junkhound noted " MSP currents are leading the inverter currents, and have huge THD". That does not, by itself, translate to "bad" PF, though.
Unless these inverters are units that allow the output PF to be controlled, the inverter current should be in phase with the grid voltage. If the MSP currents are capacitive while the inverter currents are (negative) resistive, that implies capactive local loads somewhere.
It is an open question whether POCO is helped or harmed by a net capacitive service connection here.

just eye ballin' the first v and i plots looks close to unity

~0.0008 sec, 17 deg, 0.95

 

[Redwood Infrared]

what makes you think the pf is bad?

I didn't know whether it was when I posted that, but was wondering about it because I do know that PF will tend be very low for an unloaded motor, and I believe I've read that THD will be high in a line source with low power factor. Again, not an electrician, not an EE, trying to get my arms around this stuff, clearly out of my depth with respect to my responses to several of the questions put to me here since initiating this thread.

That said, as I continue to dig into the raw data with the Fluke software I just uncovered the attached PF log. I believe the period on 10/5/17 where it's all over the map corresponds to the inverter shutdown tests. It actually looks like total PF for all three phases ain't bad after that. Still don't really understand relationship between PF and THD.

 

[Ingenieur]

pf imo is fine >0.96 avg

dpf is displacement pf, only considers the fundamental freq
will be higher
the pf considers all freq and will typically be lower since thd is largely reactive
increases Q kvar, so therefore S kva and reduces P kw

pf = P/S
S = P + jQ
|S| = sqrt(P^2 + Q^2)

the closer pf to dpf the lower the tdh, if equal, no thd

if the difference is say 0.02 and you have P = 100 kw, you have ~ 2kva
you can calc delta Q and S and delta S/S = thd

 

[gar]

180617-1710 EDT

Redwood Infrared:

Power factor is defined as the ratio of real power to apparent power. Real power is the power that does work, what you would read on a true power meter. Apparent power is the product of the RMS voltage and the RMS current to a load. I believe this definition dates back to around 1890.

There are two bread classes of power factor, displacement, and distortion.

I believe power factor has nothing to do with your problem. Also based on the information you provided I don't believe harmonics have anything to do with your problem.

.

 

[Redwood Infrared]

180617-1710 EDT

Redwood Infrared:

Power factor is defined as the ratio of real power to apparent power. Real power is the power that does work, what you would read on a true power meter. Apparent power is the product of the RMS voltage and the RMS current to a load. I believe this definition dates back to around 1890.

There are two bread classes of power factor, displacement, and distortion.

I believe power factor has nothing to do with your problem. Also based on the information you provided I don't believe harmonics have anything to do with your problem.

.

I've got a pretty good handle on power factor basics, i.e., difference between kW and kVA. It's the distortion and relationship to THD I've been struggling with. But either way, I'm hoping and inclined to believe your correct that both PF and THD are largely irrelevant here anyway. Just putting it all out there in response to questions others have raised, crossing variables off my list as I'm able to.

 

[GoldDigger]

...
...
...
Still don't really understand relationship between PF and THD.

Displacement Power Factor corresponds to a phase shift in current relative to voltage (at the fundamental frequency, and assuming basically that the voltage waveform is a pure sine wave.) It happens as a result of linear reactive loads (motors or capacitor bank, for example) and does not involve any waveform distortion as measured by THD.
A non-linear load (whether resistive or reactive in nature) results in distortion of the current waveform (assuming again that the voltage is pure sine wave). It results in a discrepancy between the product of RMS voltage and RMS current which shows up as a reduced power factor. And that distorted waveform can be built up from a summation of the fundamental frequency and suitably phased harmonics. One measure of that array of harmonics is THD. THD does not tell you anything about relative amplitude or phase of individual harmonics. It cannot be corrected for by any combination of linear reactive loads, but can be compensated by a deliberately non-linear load that exactly cancels out the harmonic current. That can be done within the input stages of otherwise non-linear loads like the DC power supplies in an electronic ballast or switching power supply or VFD.

 

[gar]

180617-2458 EDT

Redwood Infrared:

To get an idea of instrumentation problems I ran the following experiment.

1. Made a hairpin loop of #16 insulated wire. Lashed it together at a point with fiberglass tape. Then again about 6" away.

2. Powered this from a 16 V transformer with its input from a Variac.

3. Used a quartz accelerometer in this 6" region, and scope synced to the line to monitor vibration.

4. Did not measure current, but with probably several amps I saw a relative clean double frequency sine wave waveform well above the quartz amplifier noise level. No significant harmonics. Amplitude increases with current. Thus, checks theory.

5. Moving the transducer from contact with the wire, but just about as close and no signal was seen. If the two wires were at a potential difference of 240 V would I see capacitive coupling of an unwanted 60 Hz signal? I don't know.

6. I would like to see this experiment with a 3 phase balanced load, and a significant unbalanced load.

You need to do some of your own bench experiments, and with and without conduit, and at current levels seen in your problem site.

Can you simulate the noise in a lab situation?

.

 

[gar]

180618-0925 EDT

To clarify my comment on no significant harmonics in the force (accelerometer) curve ---

1. Today and probably most places in the world there is clipping distortion at the peak of the AC supply voltage curve. This did not exist 60 years ago.

2. This clipping produces some harmonics in the voltage waveform.

3. Force between conductors is a function of the current in the conductors and the interaction.

4. In my experiment I had a transformer with series leakage inductance, and nearly a short circuit on the secondary. Thus, input voltage harmonics would be reduced in the output current by the low pass filtering action of the series LR circuit.

Theory checks out nicely.

.

 

[Redwood Infrared]

Gar,

Given the scale of our problem site - over a couple hundred amps along conductors spanning 250 feet - I have no idea how I could prepare a bench experiment that wouldn't lack verisimilitude.

180617-2458 EDT

Redwood Infrared:

To get an idea of instrumentation problems I ran the following experiment.

1. Made a hairpin loop of #16 insulated wire. Lashed it together at a point with fiberglass tape. Then again about 6" away.

2. Powered this from a 16 V transformer with its input from a Variac.

3. Used a quartz accelerometer in this 6" region, and scope synced to the line to monitor vibration.

4. Did not measure current, but with probably several amps I saw a relative clean double frequency sine wave waveform well above the quartz amplifier noise level. No significant harmonics. Amplitude increases with current. Thus, checks theory.

5. Moving the transducer from contact with the wire, but just about as close and no signal was seen. If the two wires were at a potential difference of 240 V would I see capacitive coupling of an unwanted 60 Hz signal? I don't know.

6. I would like to see this experiment with a 3 phase balanced load, and a significant unbalanced load.

You need to do some of your own bench experiments, and with and without conduit, and at current levels seen in your problem site.

Can you simulate the noise in a lab situation?

.

 

[electricguy61]

Gotta throw it out there sometime, I once walked up on two electricians an and engineer trying to solve a problem on why a motor wasn't working, I asked if they checked to see if a mouse had gotten in and chewed on the windings and shorted it out, they all laughed and kept checking stuff, I asked again and they told me that if I wasn't going to be serious I could leave. I told them I was serious and they asked why? I said because I can see the mouse and the arc mark it left on the side of the motor. Sometimes the simplest answer is the one. :happyyes:

Occam's razor or a more modern expression: KISS="Keep It Simple Stupid"

 

[gar]

180618-1323 EDT

Redwood Infrared:

I see no need to work with any long length to study the vibration aspects of the cable. Nor is it probably necessary to work at more than a few volts.

Many posts back GoldDigger told you to try twisting or wrapping the wires to prevent or reduce motion.

It seems to me you need to run simple small experiments to try to simulate the problem, and then make changes to see what improvement is likely. You don't want to be doing this on 250 ft of wire.

Is the problem mechanical motion of the cable hitting the conduit, or is it magnetic induction into the conduit causing the conduit to vibrate.

I have no idea about the mechanical resonant frequency of a given length of conduit.

.

 

[gar]

190618-1429 EDT

Redwood Infrared:

I don't believe you have mentioned what kind, and size of conduit is at your site.

Get a piece like it, and put a hairpin loop of wire, like the wire being used at your site, in the conduit. This will be a single phase test.

Estimate the loop resistance. It won't be high. Estimate needed voltage at possibly 400 A, and power. Pick a little larger transformer, and remove its secondary. Make a new secondary. Power the primary with a Variac, and adjust to get enough current to make noise if it will. Assuming you can generate noise, then try to determine how to mitigate the noise.

I have no idea what will happen.

.

 

[Redwood Infrared]

While I truly appreciate all the responses, it's regrettable that this thread has grown so unwieldy. I did write a post early on characterizing the circuit with as much detail as I could, but to recap:

- the conduit is 4 inch EMT securely clamped to uni-strut which is bolted to steel roof framing on 12 foot centers
- the three phase conductors and neutral are aluminum 350 KCMil
- the grounds are XHHW-2 AL #3/0 AWG
- 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.

As far as your suggested experiment, if we're not making any headway after panel repairs are completed and any grounding issues we can figure out eliminated as potential root cause of the problem I'll try to see if I can get the electrician to work with me on that. But for the next couple days I've been directed to stand down and move on to other tasks until we get our first prescribed measures completed.

180618-1323 EDT

Redwood Infrared:

I see no need to work with any long length to study the vibration aspects of the cable. Nor is it probably necessary to work at more than a few volts.

Many posts back GoldDigger told you to try twisting or wrapping the wires to prevent or reduce motion.

It seems to me you need to run simple small experiments to try to simulate the problem, and then make changes to see what improvement is likely. You don't want to be doing this on 250 ft of wire.

Is the problem mechanical motion of the cable hitting the conduit, or is it magnetic induction into the conduit causing the conduit to vibrate.

I have no idea about the mechanical resonant frequency of a given length of conduit.

.

 

[Redwood Infrared]

Turns out there IS a UPS in the building

Turns out there IS a UPS in the building

I can't believe it but after two weeks of poking around asking every question I can think of, I just finally got hold of an electrician who hasn't worked in this building for a couple years who told me he installed a UPS for a tenant AFTER the PV system. And from what I understand the PV system was fine for at least a while before they noticed any noise.

I'm still trying to pin people down on exact dates the UPS was installed and the noise started bothering people but I have a strong suspicion we're going to see a strong correlation between these two events.

 

[ggunn]

I can't believe it but after two weeks of poking around asking every question I can think of, I just finally got hold of an electrician who hasn't worked in this building for a couple years who told me he installed a UPS for a tenant AFTER the PV system. And from what I understand the PV system was fine for at least a while before they noticed any noise.

I'm still trying to pin people down on exact dates the UPS was installed and the noise started bothering people but I have a strong suspicion we're going to see a strong correlation between these two events.

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.