Page 1 of 12 12311 ... LastLast
Results 1 to 10 of 116

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

  1. #1
    Join Date
    Jun 2018
    Location
    Arcata, CA, USA
    Posts
    49

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

    Please bear with me as I'm not an electrician but a certified thermographer with more experience in simply identifying electrical anomalies that represent potential issues than actually understanding or mitigating them. I usually just point and say, “ya know, sooner or later the smoke’s gonna come out of that one. You might want to fix that while the guys are off one weekend soon. Either that or stock up on brooms so you can keep ‘em busy during your impending unscheduled shutdown.”

    I was called into a 50,000 sf metal commercial building that had a 125kW rooftop solar array installed about two years ago. Over that time a vibration with an audible hum that corresponds to how heavily the system is loaded has developed and according to building occupants continues to get worse with time. Under full sun it’s loud enough in some office spaces to drive occupants away from their desks for the afternoon.

    The installing electrician and local utility have both evaluated the system, installed power monitors, etc., and have no idea what’s causing the issue.

    The arrays feed into 11 inverters on the south side of the building. The inverters feed into a combiner box that routes 3 phase power via aluminum conductors 250 ft away to an AC Solar Disconnect (ACSD) next to the Main Service Panel (MSP) at the north side of the building. Due to the long distance and several corners the owner elected to use aluminum wire splitting the load across redundant conductors, with a pair of parallel conduits each carrying all three phases, neutral, and ground.

    It is along these conduits through the building that the vibration hum is observed.

    Array components, inverters, and combiner box panel are grounded back to the MSP via these parallel grounds. The parallel conduits, which pass through the panel boxes at both ends with metallic hardware, themselves represent yet another parallel ground path. None of the array, inverter, or combiner box components have separate grounds at their locations, only the connection back to the AC Solar Disconnect, which is then bonded to the MSP, which is in turn connected to the building frame, with a separate GEC originating just a few feet away and entering the slab, interrupted by a strap around a water pipe along its path.

    Ultrasound indicated only audible frequencies so we can remove corona or other high frequency sources of vibration.

    Using FLIR I determined that both the combiner box bus bars and main disconnect near the inverters and the knife switches in the 600A AC Solar Disconnect 250 feet away were over 72°F rise over ambient, which NEMA Thermal Severity Criteria classify as seriously deficient - and this was on a day with a thin cloud layer so we were only seeing about 75% of production capacity, or about 200 amps per phase so it’s unlikely the panel components ever see more than 50% of their rated capacity.

    Hot spots observed appeared to originate at points of contact in the switchgear and thermal gradients showed heat dissipating along the length of the conductors. I did not observe significant inductive heating where conductors exit the conduits, which is sometimes the case in certain situations.

    Significant vibrations could be tangibly felt by hand along insulated sections of conductor accessible well away from contact points within the AC Solar Main Disconnect.
    I theorize there are two possible contributors to the problem vibration and it could be either or both simple induced currents due to inductive effects and/or a ground loop owing to the multiple parallel ground paths connecting the inverter combiner box to the AC Solar Disconnect. At the terminal end it appears the conductors exit the conduit and enter the box under some lateral strain such that some but not all of them are in pretty tight but insulated contact with the edge of the conduit/panel interface. I have seen examples where conductors coming off a high current bus bar and mechanically fastened to the panel using metallic clamps or cleats can buzz like the dickens, and the problem goes away when the cleats are replaced by nonmetallic straps. Could the conduit itself be acting as the secondary in a crude linear transformer and that’s the only thing causing the vibration?

    But upon further inspection of the ground system I have trouble not believing it's at least part of the equation.

    First off, we have those multiple ground paths between far ends of the building, all running in parallel to current carrying phases and neutral. Again, with no separate ground at the point of the inverters/combiner box. All components within the system are bonded together and tied to a single ground point at the line side of the building’ MPS.
    Second, I measured what some would classify as ‘objectionable current’ * in the ground itself, totaling as much as 8amps, split unevenly between the two parallel grounds at about a 2:1 ratio. Note that these ground conductors share the exact same origination and termination lug points at each end of their run but run through separate parallel conduits along with the other parallel redundant conductors only in order to reduce the required wire size for the installation.

    (*I did enough searching for answers before posting this that I am aware “objectionable current” is poorly defined and can be a hotly debated topic but objectionable or not, I don’t run across it often and it strikes me as unusual.)

    I also measured what I perceive as odd currents in the ground wire running from the MSP to the building frame at about 2 amps, then about the same 2 amps along the next ground wire a few feet away running from the building frame to where it’s interrupted by the water pipe, but it jumps to about 8 amps from the water pipe to the slab. Unfortunately I only had a Fluke T5 on hand so my CT couldn’t wrap around the pipe but I’d have to assume there’s 5 amps in that as well.

    The one other ground effect observation is that when we went on the roof (metal roof deck with a recently applied thick elastomeric treatment), every person in our party received a significant static shock upon first contact with any metal component. None of the grounding cables running from the array panels to the rooftop combiner boxes nor the combiner boxes to the inverters appeared to be carrying any measurable current.

    From that my non-electrician brain would infer that there is a significant potential difference developing at different grounded points within and on top of the building. Whether this is or is not exclusively or even partly responsible for the audible hum in the conduits remains unknown to me.

    I am about to suggest having the electrician carry out a few tests the client might carry out in order to eliminate potential sources of the problem.

    1. Going for the obvious problem first, I suggest they mitigate the heat issues identified with the FLIR survey. Lug connections had already been re-torqued before my arrival but obviously fixed nothing so I suggest removal and visual inspection with renewal of conductor endpoints with fresh application of Noalox if warranted. If upon re-inspection the combiner box main disconnect is still flagged under NEMA Thermal Severity Criteria I’ll recommend replacing that component. I'll be surprised if this eliminates the hum but it needs to be done regardless.
    2. Since we have redundant parallel ground conductors between panels, I suggest lifting one of them under controlled conditions to see if its removal has any effect. If it does not, we might need to try to figure out a way to temporarily also remove one or both of the electrical bonds represented by the conduits themselves. Logistically that will be much more challenging but I believe there must be available some sort of insulating bushing that could be installed to isolate the conduit from the panel to which it is mechanically fastened. Unfortunately this will require substantial disassembly of at least one of the panels to achieve.
    3. We might also try temporarily lifting the ground connection to the water pipe, and/or tying the MSP directly to the ground by removing its connection to the building’s steel frame.
    4. We could try installing a direct ground at the location of the inverter combiner box.


    I realize any or all of 2 through 4 might be unconventional, non-code compliant, and perhaps flat out unsafe in the long run. But for purposes of figuring this out I believe – IF it is possible to do the tests safely under tightly controlled conditions – we might be able to collect some very helpful observations.

    Or better yet, one of you geniuses who has read this far and hasn’t already decided I’m an idiot and not worth the trouble might already know exactly what’s going on here and can just tell me?

    I thank you in advance for your indulgence and any thoughtful responses this might elicit.

  2. #2
    Join Date
    Apr 2009
    Location
    Williamsburg, VA
    Posts
    5,438
    Welcome to The Forum.

    High currents develop high magnetic fields, enough to make the wires chatter in the conduits. If this noise and vibration was not present initially, I would look for loose hardware contributing to conduit movement and objectionable sound.

    Depending on what and where the vibrations are, you may be able to dampen them with rubber sheeting or wedges.
    Electricians do it until it Hertz!

  3. #3
    Join Date
    May 2011
    Posts
    4,425
    Comments, such as I can make:

    - The 8amps on the solar ground could be from capacitive leakage currents, if the DC side of the solar system(s) are grounded. That is a common feature of such systems, and 8 amps is within reason for a 125kW system.
    - The 2amps on the main premises GEC likely just represents building load having nothing to do with the solar.
    - I'm somewhat doubtful that the parallel ground paths for the solar grounding is inherently contributing to the noise problem. I would more likely suspect insufficient bonding of some metal component to contribute to the problem; that is, trying to unbond certain grounding conductors and components and not others would, to my intuition, possible make the problem worse. It might be interesting to try to unbond and unground all components for a short (minutes) test to see what happens, but it would be pointless to advise on how to do that permanently since it would be non-compliant and significantly unsafe.
    - Has the inverter manufacturer been contacted for any comment or advice?
    - Your item (4) would probably be code compliant but is unlikely to make any difference. Ground rods really don't do much at those voltages.

  4. #4
    Join Date
    Dec 2012
    Location
    Placerville, CA, USA
    Posts
    19,366
    The easiest and most likely productive option would be to pull the wires out of the conduits, twist them tightly and pull them again (hopefully you will have enough extra length to do this, otherwise replace the wire.) This will both reduce extended magnetic fields and keep the wires from vibrating against the conduit and each other.

    Sent from my XT1585 using Tapatalk

  5. #5
    Join Date
    Feb 2009
    Location
    Durango, CO, 10 h 20 min without traffic from winged horses.
    Posts
    9,640
    Quote Originally Posted by GoldDigger View Post
    The easiest and most likely productive option would be to pull the wires out of the conduits, twist them tightly and pull them again (hopefully you will have enough extra length to do this, otherwise replace the wire.) This will both reduce extended magnetic fields and keep the wires from vibrating against the conduit and each other.

    Sent from my XT1585 using Tapatalk
    Number of times I've found a problem with conductors inside the conduit, 10. Number of times I've found a problem with the way things were wired, 1000.
    If you go and decide to dance with a gorilla the dance ain't over till the gorilla decides it's over.

  6. #6
    Join Date
    Dec 2012
    Location
    Placerville, CA, USA
    Posts
    19,366
    Quote Originally Posted by ActionDave View Post
    Number of times I've found a problem with conductors inside the conduit, 10. Number of times I've found a problem with the way things were wired, 1000.
    A valid point. Assuming, for the moment, that there is some kind of problem with the wiring, I would probably put a clamp-on ammeter (amp-clamp) around the three phase conductors from each conduit to make sure that the net current is near zero. A relatively small extra termination resistance or other problem on just one wire could lead to very different currents between the two elements of what should be a parallel path. That imbalance could then cause non-zero net current in both conduits. Just one example of a potential problem that would not be seen by simple continuity checks.
    My next step would be amp measurements on all of the individual conductors.

  7. #7
    Join Date
    Jul 2013
    Location
    Renton, WA
    Posts
    851
    Ultrasound indicated only audible frequencies

    So, what are those frequencies ? all 60 Hz harmonics or ? Any current waveforms - elg harmonics?

    What is the PWM operating frequency of the inverters?

    Conduit span between clamps?

    For instance, if the objectionable noise is at say 180 or 540 Hz, one could infer in phse harmonics on the core wires leading to the magnetic forces already mentioned. A 540 Hz whine sure could drive folks out of an office.

  8. #8
    Join Date
    Aug 2008
    Location
    San Francisco
    Posts
    715

    Net Currents

    Based on 2 jobs we have done several years ago, My gut feeling on this is that the root cause of the vibration and humming noise is due to net currents (as others have mentioned); currents that return to the source thru system grounding paths (conduits, building steel, metal enclosures, metal raceways, metal roofs, etc) instead of current carrying conductors. In both cases the symptoms were the same: screen jitter on the old CRT monitors of all the work stations. One we found using a Gauss Meter to measure the unusually high EMI in the affected areas. Also very high ground currents measured on the ground cables terminated on the ground bus of the main power panel. This one turned out to be loose bolt connections on the neutral bus (4000A service) bars at the outside utility pad mounted transformer. The other cause turned out to be the miss-match of the connectors between work station old and new partitions. During expansion, the new type partition sections caused the 120v hot conductors to connect to the ground conductors of the adjoining partitions resulting in net currents. If these currents are high enough then it could be the source of the humming since steel is not designed to carry current.
    Just a thought....hope this helps.
    Am I in shape?? I get plenty of exercise pushing my luck!!

  9. #9
    Join Date
    Jun 2018
    Location
    Arcata, CA, USA
    Posts
    49

    Wow, thanks for all the thoughtful responses so far!

    It will take me a little time to digest all this and respond.

    For what it's worth, the utility analysis was pretty comprehensive and they provided data on current and voltage waveforms at all three of the aforementioned service panels as well as THD on both I and V. I'd have included that info in my original post but ran up against the forum's character limit as it was. In any event nobody seemed to think there was much to go on from all that, although I did notice after we'd measured the odd ground currents that they also saw about 8A on the ground but didn't flag it. They did not, as far as I can tell, try measuring net current by capturing all the phases together to see if there was anything missing that must be running outside the pipe. I think that's a very good suggestion and I'll try to get that accomplished.

    I'll see if I can roust up a Gauss meter too. It's almost too bad we don't still have the occasional CRT under a back table in some hoarder's office but I think we've past the point where any building less than 10 yrs old will not likely have ever housed a CRT to begin with.

    More later, after the wife gets some Sunday chores out of me. Thanks again!

  10. #10
    Join Date
    Jan 2016
    Location
    Earth
    Posts
    6,049
    what size conductor?
    what size conduit? what material? how as it supported?

    the fact that it proportional to load is telling

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •