Power quality issue: possible causes?

[BGbreeder]

Background: for the last six months or so, we have been having "over voltage" excursions, i.e. more than 250V AC at the meter. It took me a while to realize that the system response on one of UPS units was caused by it detecting an over voltage situation and going into a protective mode to reduce the voltage. Meanwhile, Tesla support had noticed some "over voltage" codes on the Powerwalls and let me know.

We normally have line voltage in the higher end of the spectrum, something I had put down to being rural, and in a branch circuit that goes 35-50 miles or so. Before we had solar, our line voltage at the meter ran about 248V most of the time. In the past, solar might have pushed it up a volt or so, but not more. A series of things have occurred over the last year;

• A neighbor finally got PTO (after 27 months, due to their installer messing up). 12kWDC
• A second neighbor installed about 6kW DC solar.
• All of the thirty or so neighbors on the same branch medium voltage line got new transformers, almost all the same exact model. (As far as I can tell.)
• The utility meter reports highly variable power exports
  
  (4-7kW), with no change in the house load, when the voltage is up at 253V.

You can see the zigzag glitches in exports.

The utility inspector was out today on a call from me on the higher voltage, and the inspector was of the opinion that it was being caused by solar. I flipped the main breaker, and the voltage at the meter dropped from 253 to 250V. The inspector was of the opinion that perhaps we might need a larger service drop and/or a larger transformer, to lower the impedance, and planned to bring the issue up with engineering. Reducing the impedance would drop the voltage needed to get our exports back on the grid, so that made sense to me.

The follow up several hours later is that the utility had found a second customer, also having voltage issues. They took a local capacitor bank offline and seem to have resolved the issue, at least for the moment.

Has anyone seen anything like this? Does any of this make sense?

Thanks!

BG

 

[power.saver]

I had a similar problem years ago, right after getting solar in 2006. SCE didn't believe me when I first reported the problem, but after their techs came out and I showed them the data from my Fluke power line monitor they believed it. The street had a 12kV primary distribution and it was overloaded, but the immediate problem was fixed by replacing the regulator at the substation. You should ask the techs when they come out if they have checked the regulator before upgrading service. The regulator for my block was in Auto but had some faulty components, so they temporarily set it to manual and lowered the primary voltage, which held for a few months until it got replaced. The problem was completely solved when they later upgraded the primary distribution to 25kV and all new (and bigger) transformers along the block. Everything is good now, service is within 238-242v.

A faulty capacitor bank might cause this. It's main purpose it to stabilize the power factor in the grid. So disconnecting them would imply they were putting too much positive reactive power into the grid, which can raise the voltage. I'm sure this is a temporary measure (like the manual regulator adjustment) because they need to have some capacitance to prevent negative reactive power from damaging the grid.

 

[BGbreeder]

I had a similar problem years ago, right after getting solar in 2006. SCE didn't believe me when I first reported the problem, but after their techs came out and I showed them the data from my Fluke power line monitor they believed it. The street had a 12kV primary distribution and it was overloaded, but the immediate problem was fixed by replacing the regulator at the substation. You should ask the techs when they come out if they have checked the regulator before upgrading service. The regulator for my block was in Auto but had some faulty components, so they temporarily set it to manual and lowered the primary voltage, which held for a few months until it got replaced. The problem was completely solved when they later upgraded the primary distribution to 25kV and all new (and bigger) transformers along the block. Everything is good now, service is within 238-242v.

A faulty capacitor bank might cause this. It's main purpose it to stabilize the power factor in the grid. So disconnecting them would imply they were putting too much positive reactive power into the grid, which can raise the voltage. I'm sure this is a temporary measure (like the manual regulator adjustment) because they need to have some capacitance to prevent negative reactive power from damaging the grid.

That's very helpful! Any suggestions on what to look for in a power line monitor / data logger, or particular models? Since I won't be earning a living with it, I'm a little reluctant to shell out $4k+ for one. I haven't used one in decades...

All the best,

BG

 

[BGbreeder]

Update: voltage is back up today. This feels like it is going to be an adventure...
Same glitches in the solar curve; bright sun, no clouds either day.

All the best,

BG

 

[lodar]

I’ll be keeping a close eye on this thread. My UPS regularly reports over-voltage around 5:00a in the morning.

It should be too early for solar, and I’m usually not up yet to investigate. The alert goes away after a few minutes.

 

[power.saver]

That's very helpful! Any suggestions on what to look for in a power line monitor / data logger, or particular models? Since I won't be earning a living with it, I'm a little reluctant to shell out $4k+ for one. I haven't used one in decades...

All the best,

BG

For this problem, you probably only need a voltage recorder so you can see the change and any spikes or dips. Those are less expensive than a full fledged power line monitor. Some come with a standard wall plug adapter so you don't need to tap into your service panel.

 

[BGbreeder]

I’ll be keeping a close eye on this thread. My UPS regularly reports over-voltage around 5:00a in the morning.

It should be too early for solar, and I’m usually not up yet to investigate. The alert goes away after a few minutes.

I worked at a location where the local utility used to throw a switch at six every morning, causing a 1-3 second power loss. The computers on our machines did not take the drop out at all well. I ended up putting in an enormous ferroresonant automatic voltage transformer to store enough energy for the ride through. At the time, it was much cheaper than a the equivalent line interactive UPS. As the switch over was never exactly 6:00, I always suspected it was some engineer manually flipping switches, and not quite getting the timing right.

My generally low opinion of the utility, which also supplied the local drinking water, went even lower when I discovered a mosquito larva on one of our water filters when changing the filter. A live, and twitching, mosquito larva. From a supposedly filtered and supposedly chlorinated municipal drinking water supply. Funnily enough, after I showed the engineering and maintenance crew the filter, with the still twitching mosquito larva, the team quit using tap water for their coffee...(I had previously complained about the water quality, and had gotten the brush off from the, all local, crew, but after the mosquito, they were always very helpful. Good team.)

"Do not look behind the curtain!"...

All the best,

BG

 

[darhall993]

I had a very similar experience with my small utility a couple of years ago. After 3 years of Solar+Powerwall running with no issues, I suddenly started to see overvoltage events regularly which would cause the inverter to continuously go offline and restart after 5 minutes (standard grid outage behavior). After several emails (with subsequent brush offs) describing the issue and after supplying corroborating SolarEdge and Powerwall logging the local utility finally came out and found two issues with the neutral conductor connector at the transformer. They reported that a squirrel had chewed into a section of the neutral. Utilities are certainly not accustomed to paying customers having detailed technical logs on how variable their service is.

 

[BGbreeder]

Quick update: PG&E came out and load tested the service drop, and then did a load test of the transformer and medium voltage line (by ramping up to 700A/phase, in short bursts). Both passed. He added a power analyzer to log data for a week and then it goes off to engineering, so probably no updates for at least a week.

Even exporting 2.7kW of solar creates a 2V rise in the L-L voltage, which makes me think that it is an impedance issue.

The solar is still showing the odd zig-zag glitches in the exports; there were seven today, unrelated to the local demand, on a completely cloudless day. These are microinverters, so even if one dropped off, it wouldn't look like this, and it shouldn't overshoot.

The glitch around 16:00 is my turning the solar off briefly to check voltages, and the dropout around 17:00 is the meter being pulled.

All the best,

BG

 

[BGbreeder]

Update:
PG&E still has the power logging monitor on our meter.

More odd voltage behavior;

• the house was off-grid Friday afternoon through Monday as precaution against solar storm induced voltage variations.
• With the house isolated, the voltage at the meter matches two other meters a mile and three miles away. (244V yesterday at noon)
• This morning the meter showed 243V (off-grid).
• Putting the house on grid this morning with 0.6kW load, the voltage instantly went from 243V up to 246V, as soon as the load was applied. That is contrary to everything that I understand about power systems.

Any thoughts? @wwhitney @Vines ?

All the best,

BG

 

[wwhitney]

Putting the house on grid this morning with 0.6kW load, the voltage instantly went from 243V up to 246V, as soon as the load was applied. That is contrary to everything that I understand about power systems.

When you reconnected the house with the grid, was it drawing 0.6 kW net load from the grid, or was that 0.6 kW house load plus some amount of PV export, so that it was a net export to the grid?

The reported voltage rise makes no sense to me in the context of the former case, while it would be expected in the context of the latter case.

[I'm a bit unclear on how the transition from off-grid to on-grid is done, whether from the point of view of the grid-following inverters it is seamless and they continue producing power if they were producing it while off-grid, or whether they see an excursion from their allowable voltage/frequency windows and get knocked off-line for 5 minutes.]

Cheers, Wayne

 

[Vines]

Update:
PG&E still has the power logging monitor on our meter.

More odd voltage behavior;

• the house was off-grid Friday afternoon through Monday as precaution against solar storm induced voltage variations.
• With the house isolated, the voltage at the meter matches two other meters a mile and three miles away. (244V yesterday at noon)
• This morning the meter showed 243V (off-grid).
• Putting the house on grid this morning with 0.6kW load, the voltage instantly went from 243V up to 246V, as soon as the load was applied. That is contrary to everything that I understand about power systems.

Any thoughts? @wwhitney @Vines ?

All the best,

BG

I have to start by mentioning that these types of grid issues are not my core competency, so I am guessing a lot. My first question was similar to Waynes above, and wondering if the PV inverter was actively exporting.

You mentioned that you suspected impedance to be off, so I was thinking we should measure it. If you ask our team to come check it as part of the site visit, we will bring the impedance loop tester to do so. This will at least tell us the impedance and wiring/connection quality of the loop from the transformer to the house.

Otherwise when I think about this, I have to wonder about a bad connection in the service drop, I don't have great insight, sorry.

 

[wwhitney]

You mentioned that you suspected impedance to be off, so I was thinking we should measure it. If you ask our team to come check it as part of the site visit, we will bring the impedance loop tester to do so. This will at least tell us the impedance and wiring/connection quality of the loop from the transformer to the house.

If I'm not mistaken, with a sufficiently accurate voltmeter at the meter (or close enough, e.g. the Backup Gateway), and a sufficiently accurate measurement of the PV current, it's not hard for the knowledgeable user to do an equivalent test.

I.e. pick a time when PV is exporting say 10A - 100A (depending on PV inverter size), the PWs are configured not to be exporting or charging (or disconnect them) and the house load is stable (or disconnect it). Then note the grid voltage V0 and PV current I, disconnect the PV inverter, and note the new grid voltage V1.

The effective grid resistance then is just R = (V0 - V1) / I. Knowing the length, material and size for the service drop (and other conductors up to your voltage measuring point) will let you calculate the resistance contribution from those wires. Any additional resistance is attributable to the utility transformer, the impedance of the utility's primary-side distribution network, and any bad connections or compromised conductors. [I say resistance here because the reactance of the grid connection will to first order only change the phase of the measured voltage, not its magnitude, as the PV inverter is presumably putting out current completely in phase with voltage.]

Vines, what values of loop impedance do you expect to see, and generally how much of that can be attributed to the transformer itself? And does the loop impedance tester measure both resistance and reactance?

Cheers, Wayne

 

[yblaser]

Update:
PG&E still has the power logging monitor on our meter.

More odd voltage behavior;

• the house was off-grid Friday afternoon through Monday as precaution against solar storm induced voltage variations.
• With the house isolated, the voltage at the meter matches two other meters a mile and three miles away. (244V yesterday at noon)
• This morning the meter showed 243V (off-grid).
• Putting the house on grid this morning with 0.6kW load, the voltage instantly went from 243V up to 246V, as soon as the load was applied. That is contrary to everything that I understand about power systems.

Any thoughts? @wwhitney @Vines ?

All the best,

BG

Far from an expert but could stray current/voltage on the neutral/ground cause the L1 to L2 voltage to rise at the meter? Just thinking that besides the addition of the load the other difference between on and off grid is that you would be utilizing the neutral if you have unbalanced 120V loads.

 

[Vines]

If I'm not mistaken, with a sufficiently accurate voltmeter at the meter (or close enough, e.g. the Backup Gateway), and a sufficiently accurate measurement of the PV current, it's not hard for the knowledgeable user to do an equivalent test.

I.e. pick a time when PV is exporting say 10A - 100A (depending on PV inverter size), the PWs are configured not to be exporting or charging (or disconnect them) and the house load is stable (or disconnect it). Then note the grid voltage V0 and PV current I, disconnect the PV inverter, and note the new grid voltage V1.

The effective grid resistance then is just R = (V0 - V1) / I. Knowing the length, material and size for the service drop (and other conductors up to your voltage measuring point) will let you calculate the resistance contribution from those wires. Any additional resistance is attributable to the utility transformer, the impedance of the utility's primary-side distribution network, and any bad connections or compromised conductors. [I say resistance here because the reactance of the grid connection will to first order only change the phase of the measured voltage, not its magnitude, as the PV inverter is presumably putting out current completely in phase with voltage.]

Vines, what values of loop impedance do you expect to see, and generally how much of that can be attributed to the transformer itself? And does the loop impedance tester measure both resistance and reactance?

Cheers, Wayne

Typical measured values of our customer's 600-200A service are between 0.08 ohms and 0.22 ohms. The larger the transformer and closer it is the lower the impedance.

I do know that the measured impedance is supposed to be a combination of resistance and reactance. Also, I think standard transformers are about 4-5% impedance and low impedance versions are about 2%. Beyond that, I do not know much about impedance, it is one of the many things in this industry I want to learn more about.

We use a Fluke 1662 to do the loop impedance test on L1 and L2 whenever we sell 5 or more Powerwall 2 to be sure the Impedance is within allowable limits. If we see high impedance that we expect is an error we sometimes recommend the customer ask PG&E to come retorque the transformer connections.

 

[BGbreeder]

When you reconnected the house with the grid, was it drawing 0.6 kW net load from the grid, or was that 0.6 kW house load plus some amount of PV export, so that it was a net export to the grid?

The reported voltage rise makes no sense to me in the context of the former case, while it would be expected in the context of the latter case.

[I'm a bit unclear on how the transition from off-grid to on-grid is done, whether from the point of view of the grid-following inverters it is seamless and they continue producing power if they were producing it while off-grid, or whether they see an excursion from their allowable voltage/frequency windows and get knocked off-line for 5 minutes.]

Cheers, Wayne

@wwhitney Thanks. To answer a few of the questions;
The house was disconnected by opening the main service panel 200A system breaker to the meter. (and it had been off for about 70 hours)
It was 6:30am, so no solar, no export, and the normal minor house loads totaled 0.6kW, double checking on the Tesla app, showed no solar, and a house load of 0.6kW being supplied by the grid.
After switching the main breaker to on, there was about a twenty second pause before the Gateway closed the internal relay with an audible clunk, and the 0.6kW load, and 246V showed on the meter.

I had previously done the applied load delta V calculation, (250V@6kW, down to 244 at 0kW, which works out to 9.9Ohms (Edit: bad math. See @wwhitney's response below) between the meter up the service drop to the transformer. But that is not a terribly accurate way to try and measure impedance, at least as I understand it.) That voltage change made sense to me, at least in direction. It makes no sense to me in magnitude. That is way, way too many ohms.

All the best,

BG

 

[wwhitney]

The house was disconnected by opening the main service panel 200A system breaker to the meter. (and it had been off for about 70 hours)
It was 6:30am, so no solar, no export, and the normal minor house loads totaled 0.6kW, double checking on the Tesla app, showed no solar, and a house load of 0.6kW being supplied by the grid.
After switching the main breaker to on, there was about a twenty second pause before the Gateway closed the internal relay with an audible clunk, and the 0.6kW load, and 246V showed on the meter.

So the 243V (off-grid) and 246V (on-grid) numbers are coming from the normal PG&E electrical meter on the utility side of your main breaker? That's really weird. There's no good basic electrical theory explanation for that. I can only guess that the meter isn't very accurate at low load, or a voltage regulator happened to switch during that 20 second pause, or something.

I had previously done the applied load delta V calculation, (250V@6kW, down to 244 at 0kW, which works out to 9.9Ohms between the meter up the service drop to the transformer.

I get: 6000W / 250V = 24A vs 0A; and (250V - 244V) / (24A - 0A) = 0.25 ohms source resistance.

Cheers, Wayne

 

[BGbreeder]

I have to start by mentioning that these types of grid issues are not my core competency, so I am guessing a lot. My first question was similar to Waynes above, and wondering if the PV inverter was actively exporting.

You mentioned that you suspected impedance to be off, so I was thinking we should measure it. If you ask our team to come check it as part of the site visit, we will bring the impedance loop tester to do so. This will at least tell us the impedance and wiring/connection quality of the loop from the transformer to the house.

Otherwise when I think about this, I have to wonder about a bad connection in the service drop, I don't have great insight, sorry.

It certainly could be a bad connection on the service drop. The utility has futzed with it several times; once to replace a transformer that blew (lightning, followed by a floating neutral issue a year later that killed it), and then a year later to replace some (all?) of the service drop due to insulation cracks that had lead to corrosion. And it could be a bad connection at the transformer either to the service drop, or to our second three phase generating transformer.

Typical measured values of our customer's 600-200A service are between 0.08 ohms and 0.22 ohms. The larger the transformer and closer it is the lower the impedance.

I do know that the measured impedance is supposed to be a combination of resistance and reactance. Also, I think standard transformers are about 4-5% impedance and low impedance versions are about 2%. Beyond that, I do not know much about impedance, it is one of the many things in this industry I want to learn more about.

We use a Fluke 1662 to do the loop impedance test on L1 and L2 whenever we sell 5 or more Powerwall 2 to be sure the Impedance is within allowable limits. If we see high impedance that we expect is an error we sometimes recommend the customer ask PG&E to come retorque the transformer connections.

Thanks for the suggestions and help.

I was a little surprised at the first PG&E visit when the tech wasn't surprised by a voltage of 252V, dropping to 244 without the solar backfeed. Subsequent discussions lead me to believe that impedance and backfeeds were really out of his comfort zone.

When PG&E comes to remove the monitor, I will ask about retightening the transformer lugs, and perhaps checking all of the two transformer lugs and connections. I will try to get binoculars on the transformer to see if I can see anything. The meter connections seem tight, and clean, as do the main service panel lines from the meter.

I thought that I understood impedance, but the behavior that I am seeing mystifies me, unless it is something really off, like the second split phase being supplemented by the third AC phase via the second transformer, as the impedance of the house shifts the impedance on the three phase circuit. I do not understand the behavior here.

All the best,

BG

 

[BGbreeder]

So the 243V (off-grid) and 246V (on-grid) numbers are coming from the normal PG&E electrical meter on the utility side of your main breaker? That's really weird. There's no good basic electrical theory explanation for that. I can only guess that the meter isn't very accurate at low load, or a voltage regulator happened to switch during that 20 second pause, or something.


I get: 6000W / 250V = 24A vs 0A; and (250V - 244V) / (24A - 0A) = 0.25 ohms source resistance.

Cheers, Wayne

Thanks for correcting my math!

Since the meter is "revenue grade" it is supposed to be more accurate than that, but that doesn't mean that this meter is. I can confirm that the meter voltage matches my DVMs, and matched the current of the clamp on Fluke meter when it was at 6.kW export, so I think it is ok, but there is the possibility that it is less accurate at low current. The meter and Tesla gateway were both reporting the same (0.6kW) number for imported power, so they agreed on the magnitude and direction. (Head scratching on this end...)

All the best,

BG