breakers overheating on single phase/ multiple inverter installation

[Jason W]

I have an installation with 13 GoodWe single phase inverters and two 400 amp services with a pad mounted 167kVa transformer. The inverters are all 11.4 k, except for one which is 7.7 k. One 400 amp panel has six 11.4 k inverters and the other has six 11.4 k and one 7.6 k.
My problem is that the breakers for each inverter are overheating and tripping. We have verified that all the wire sizing is correct, the breakers are rated for backfeeding and correctly sized, all terminals are torqued to specifications and the grounding is properly done. The inverters are never going over their rated maximum amperage on the AC output so the breaker trip is happening because of excessive heating. We have shot both panels with IR camera and the temp on the breakers goes from ambient (around 75 degrees F) to over 160 degrees in less than an hour on one panel; on the other panel it goes up but not as fast or as high and breakers trip.
When the inverters are at full power, both panels hum and vibrate as do the disconnects between the main panels and the meters. The big difference between the panel B that has the heating problem to an extreme and the panel A that heats, but much less, is the length of the AC conductor runs. The inverters that heat the most are the ones with the shortest AC runs and the ones at the top of the bus bar, in both panels.
I suspect that the problem is caused by harmonic distortion but what is casing the HD is the mystery. Maybe induction--the 5 inverters that heat the most have ac conductors in non-metallic liquid tight and run right next to each other across the mounting board to the panel. Maybe it is HD from the inverters, but GoodWe claims to have multi-inverter installations in other locations that are not having this problem.
Would love any ideas about the cause that you can think of.

 

[jaggedben]

Just to clarify, this is all 240V single phase?

 

[Jason W]

yes, 240V split phase

 

[Elect117]

What is the wire size and breaker size for each inverter?

Does the inverter provide you with output information on each or is it metered as a total? If so, are any of them outputting more than the rated current?

Have you double checked everything is torqued properly and secured? Loose connections can also cause a high amount of heat do occur.

 

[Jason W]

Yes, every lug is torqued properly. Wire is #6 AWG copper THHN for the 11.4 k inverters and the same for the 7.6K. Breakers are 60 amp for the 11.4's and 40 amp for the 7.6. We are checking amperage with clamp meters on all the inverters and they are never going over specifications--47.5 for the 11.4 k and 32 for the 7.6 k.

 

[petersonra]

I am not surprised the breakers trip when they reach 160 deg F. They are thermal magnetic breakers after all and the trip elements don't care much where the heat is coming from.

Have you been able to measure the conductor and terminal temperatures? It's possible the heat is coming in from the conductors. Were the conductors from each invertor run in separate conduits?

 

[Elect117]

Do you have a way of looking or capturing the waveform? THD can be pretty notable on a voltage and current waveform capture. It can also show up as neutral current.

 

[PWDickerson]

I had something similar happen to me on two separate occasions a few years back. One site had 5 inverters, and the other site had two inverters. When we operated just one inverter, there were no problems. When more than one inverter operated at the same time, it would make a nasty whining sound that could be heard at the inverters, at the service equipment, and even at the padmount utility transformer 100' from the building. I looked at the sine wave with an o-scope, and found that the inverters were fighting each other and creating a 5000 hz sine wave that was superimposed on the 60 hz sine wave. It was insane. The inverter manufacturer said they had seen this before, and it was caused by operating on an overly inductive grid. Their fix was to upgrade the inverters to their newest model. I think you are on the right path trying to measure THD.

 

[Jason W]

Do you have a way of looking or capturing the waveform? THD can be pretty notable on a voltage and current waveform capture. It can also show up as neutral current.

This is one of my next steps, just waiting for the meter to arrive. The AC is in separate conduit for each inverter but it is non-metallic liquid tight and they are separated by less than an inch on the run to the panel so there could be some induction going on that amplifies the HD problem. Once I get a read on the wave form then the next step was to put metal conduit in place of the NMLT.

 

[Jason W]

I am not surprised the breakers trip when they reach 160 deg F. They are thermal magnetic breakers after all and the trip elements don't care much where the heat is coming from.

Have you been able to measure the conductor and terminal temperatures? It's possible the heat is coming in from the conductors. Were the conductors from each invertor run in separate conduits?

The conductors are getting hot--hotter on the panel end than the inverter end and L1 is 10 degrees hotter than L2 with every inverter, when measured at the breaker. But the conductor temp overall is much lower than the breaker temp. The breakers is where the heat is building up the highest and fastest.

 

[Elect117]

When you have more than one source with harmonics they can add to each other at the panel. The 400A breaker / panel and transformer have a larger "VA" rating. If the system is oversized, harmonics have less of an effect. Total harmonic distortion is related to the total power.

It makes sense that you would see most of the heat at the breaker because it is barely 1.25x larger. And it trips on continuous. Meaning that the normal continuous usage heat + the harmonic heat is tripping it once the continuous heat kicks in. If the wire and breaker were oversized (not advised) then the total harmonic distortion's percentage would be lower in relation to the total power capable of dissipating the heat.

I do not believe it is induction since mutual induction is more typical to higher voltage circuits were the conductors of each circuit are run further away from each other. Or where the magnetic fields do no cancel. Like when the conductors of a specific leg are run on top of each other. I won't discounted it though. I will just say that it is unlikely.

The only other thing I could think of that would create heat in a similar way would be small arcing on the insulation but that is very unlikely since you noted it shows up in more than just one circuit and it is unlikely you knicked all of them.

 

[Jason W]

When you have more than one source with harmonics they can add to each other at the panel. The 400A breaker / panel and transformer have a larger "VA" rating. If the system is oversized, harmonics have less of an effect. Total harmonic distortion is related to the total power.

It makes sense that you would see most of the heat at the breaker because it is barely 1.25x larger. And it trips on continuous. Meaning that the normal continuous usage heat + the harmonic heat is tripping it once the continuous heat kicks in. If the wire and breaker were oversized (not advised) then the total harmonic distortion's percentage would be lower in relation to the total power capable of dissipating the heat.

I do not believe it is induction since mutual induction is more typical to higher voltage circuits were the conductors of each circuit are run further away from each other. Or where the magnetic fields do no cancel. Like when the conductors of a specific leg are run on top of each other. I won't discounted it though. I will just say that it is unlikely.

The only other thing I could think of that would create heat in a similar way would be small arcing on the insulation but that is very unlikely since you noted it shows up in more than just one circuit and it is unlikely you knicked all of them.

I'm curious if you think longer conductor runs could attenuate some of the HD because the panel that only has inverters with long runs--90 feet versus 20 on the other panel, does not heat nearly as much and the breakers in that panel have never gotten hot enough to trip.

 

[Elect117]

I don't believe the longer runs are attenuating the distortion, rather, the heat generated by the copper losses (I²R) are higher due to the presence of distortion.

It has to dissipate all the heat from the normal frequency and the nth order harmonics. The longer the line, the higher the impedance. The higher the losses generated from heat, the more heat to dissipate. etc.

Inductive circuits (mostly reactive, not capacitive) have a higher impedance as frequency goes up.

Z=R+jwL where as "w" (omega, being 2*pi*f) goes up, so will the impedance.

There can be other reasons why the temperature of one circuit is different than another. Like being in direct sunlight, being attached directly to a roof, running through insulation so the heat dissipation is reduced, etc.

Without being able to see the waveforms, I am just guessing at it.

 

[wwhitney]

Do you have a way of looking or capturing the waveform? THD can be pretty notable on a voltage and current waveform capture. It can also show up as neutral current.

A couple background questions:

Can the inverters change the voltage waveform when they are running, or are we just talking about distortion of the current waveform? I think of the inverters as trying to match the voltage waveform at their terminals, but I guess if they do that imperfectly, they may be producing a slightly different voltage waveform? Then given the impedance of the wiring and transformers involved, the voltage waveform measured at an intermediate point between the "grid" and the inverter terminals would be (close to?) a weighted average of the two sources?

As to THD and extra heating, for I²*R heating, all that matters is the RMS current I, for any waveform. So are you suggesting that, say, the 7.2 kW inverter when operating at capacity is putting out a current waveform whose 60 Hz component is 30A (which will transfer 7.2 kW at 240V) along with some higher frequency components (which will transfer no power), so that the RMS current is above 30A? A true-RMS meter with a sufficiently high sampling rate would then show a current above 30A, I believe.

Cheers, Wayne

 

[Elect117]

A couple background questions:

Can the inverters change the voltage waveform when they are running, or are we just talking about distortion of the current waveform? I think of the inverters as trying to match the voltage waveform at their terminals, but I guess if they do that imperfectly, they may be producing a slightly different voltage waveform? Then given the impedance of the wiring and transformers involved, the voltage waveform measured at an intermediate point between the "grid" and the inverter terminals would be (close to?) a weighted average of the two sources?

As to THD and extra heating, for I²*R heating, all that matters is the RMS current I, for any waveform. So are you suggesting that, say, the 7.2 kW inverter when operating at capacity is putting out a current waveform whose 60 Hz component is 30A (which will transfer 7.2 kW at 240V) along with some higher frequency components (which will transfer no power), so that the RMS current is above 30A? A true-RMS meter with a sufficiently high sampling rate would then show a current above 30A, I believe.

Cheers, Wayne

Inverters match the input frequency and voltage but depending on the clocking and smoothing can have distortion. It may not be a perfect sine wave but a combination of pulsed square waves smoothed in to one. I'm not making the inverters but just because it is grid tied doesn't mean the output waveform is perfectly matching the grid. It can be "close enough" to remain paralleled in. Not to mention, nothing is monitoring the distortion of the output waveform. There isn't a device checking for clean voltage or current. I am pretty sure the inverter verifies that the waveform is matching the amplitude, phase, and zero crossing (fundamental frequency).

Voltage distortion and current distortion can be observed by waveform captures.

The impedance of the wire changes with frequency of the current or voltage (reactance/capacitance).

Each component of the harmonic is present on the wire. As the frequency increases, the skin effect reduces the available area of conductance, increasing resistance. Which is heat on the wire. Your fundamental waveform might be swinging on the 60Hz frequency but the presence of nth order harmonics are layered into that fundamental waveform. The high frequency components aren't "real" similar to the way reactive power isn't "real" but we still look at apparent power.

The RMS current doesn't need to be above 30A for the power being dissipated in the wire to change. You wouldn't say the DC resistance of a wire is equal to the AC resistance at 10kHz just because the Irms is 30A.

https://inverterdrive.com/file/Understanding-effects-of-Power-Factor-Harmonics/

 

[jaggedben]

...Not to mention, nothing is monitoring the distortion of the output waveform. There isn't a device checking for clean voltage or current. I am pretty sure the inverter verifies that the waveform is matching the amplitude, phase, and zero crossing (fundamental frequency).

...

I don't think you can state this definitively without possessing proprietary knowledge of the inverter. It may depend on the manufacturer and model. There is no reason an inverter couldn't, in effect, scope it's own output as a safeguard, to one degree of precision or another.

Of course in the case at hand we have a valid hypothesis that one or more inverters aren't doing that adequately.

 

[Elect117]

Ya, it is a guess. I just don't think they do. I think they test their THD or TDD in a lab setting and allow for tolerances in parts. I don't know if sampling and applying corrections like increasing smoothing capacitance or inductance is really a normal thing.