PV System Clarification

Merry Christmas

inforaj

Member
Location
Chiago
Occupation
Design Engineer
At 10 AM on a sunny day, an operating grid-connected inverter suddenly shuts down. Five minutes later, it comes back on, only to shut down again after a few seconds. It continues this cycle until 3 PM at which point it remains on for the rest of the day. The most likely cause of this problem is
1. The utility voltage has exceeded the IEEE 1547 maximum limits for this time period.
2. The utility voltage has below the IEEE 1547 maximum limits for this time period.
3. The wiring between the inverter and utility connection is undersized.
4. The wiring between the inverter and utility connection is oversized.

This was asked at the end of the year review.
 

Fred B

Senior Member
Location
Upstate, NY
Occupation
Electrician
Is this a dwelling unit related solar or a large solar farm?
3 and 4 not likely, as you indicate this is an operating existing system. 3 would have shown itself fairly quickly on first commissioning and would not be transient, or a one time. 4 not likely issue at all.
If this installation is for a dwelling unit or other occupancy related installation:
Either 1 or 2 are possible, brown out or surge events. Also intermitant disruption from POCO, wind causing tree - power line interaction, other transients affecting power quality from POCO.
Was this a one time event, or a recurring disruption? If this was a one time event that would probably eliminate more serious local electrical installation issues as being the cause. (Electrician on site would be needed to confirm any potential issue.)
If this was part of a large solar farm likely cause was faulty inverter at the sight or even issue with the monitoring equipment indicating a power disruption that didn't exist. Again, was this a one and done, or has this been repeatedly present?
Monitoring is not going to replace a hands-on electrician on site to check potential problem. If this is a recurring issue, it's time to get an electrician on site to check things out.
 

inforaj

Member
Location
Chiago
Occupation
Design Engineer
Is this a dwelling unit related solar or a large solar farm?
3 and 4 not likely, as you indicate this is an operating existing system. 3 would have shown itself fairly quickly on first commissioning and would not be transient, or a one time. 4 not likely issue at all.
If this installation is for a dwelling unit or other occupancy related installation:
Either 1 or 2 are possible, brown out or surge events. Also intermitant disruption from POCO, wind causing tree - power line interaction, other transients affecting power quality from POCO.
Was this a one time event, or a recurring disruption? If this was a one time event that would probably eliminate more serious local electrical installation issues as being the cause. (Electrician on site would be needed to confirm any potential issue.)
If this was part of a large solar farm likely cause was faulty inverter at the sight or even issue with the monitoring equipment indicating a power disruption that didn't exist. Again, was this a one and done, or has this been repeatedly present?
Monitoring is not going to replace a hands-on electrician on site to check potential problem. If this is a recurring issue, it's time to get an electrician on site to check things out.
This is a commercial solar farm.
 

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
(3) is most likely: excess voltage rise on the conductors between the utility and the inverter, which means the voltage at the inverter is only within the proper voltage window when the power output of the inverter is low enough.

Cheers, Wayne
 

steve66

Senior Member
Location
Illinois
Occupation
Engineer
I'm not sure any of the answers make sense, but I'm going with #3. But to be honest, I don't know a lot about solar.

The 10AM to 3PM time frame seems to imply the problem only happens when the sun is bright and the panels are producing a lot of power. That seems to rule out #1 and #2 because the utility voltage wouldn't necessarily correlate with the brightest part of the day. I'm assuming our solar farm is small enough that it can't affect the actual utility voltage.

I think that leaves #3 as the correct answer. When the inverters produce maximum output, their output voltage is the sum of the utility voltage and the voltage drop on the output wiring. If the wiring is too small, the voltage drop will be higher. That means the output voltage at the inverter will be too high, and the inverter will shut down because it sees the grid voltage as being too high (even though the voltage at the grid connection point may be within the allowable limits).

As soon as the inverter shuts down, the voltage drops to the grid voltage, so after 5 min, the inverter tries again. But the same thing happens. The added voltage over the wiring makes the inverter voltage too high again, and it shuts down again.

Just my guess.
 

GoldDigger

Moderator
Staff member
Location
Placerville, CA, USA
Occupation
Retired PV System Designer
Another possibility is that POCO is using a controlled tap changer at the 10am and 3pm hours. (If it is happening at repeatabe exact times every day rather than approximate times.
What happens on a cloudy day?
 

jaggedben

Senior Member
Location
Northern California
Occupation
Solar and Energy Storage Installer
If this happened once I would say it is 1 or 2. If it happens repeatedly and coincides only with the sunniest conditions and maximum inverter output, then it is probably 3, possibly in combination with some amount of 1. It is possible that other nearby solar systems are contributing to 1 at the same time your system is outputting at its max.

I would presume if you know exactly when the output stopped and started you would also have voltage data to help you analyze.
 

tallgirl

Senior Member
Location
Great White North
Occupation
Controls Systems firmware engineer
The most likely answers are 3, and then possibly 1. If you gave me the time of year and outdoor temperature, I could tell you which it likely is.

If you sized your conductors to meet the load, rather than reduce voltage rise (drop ...) the inverter terminal voltage will RISE as power output increases, and that's going to happen when the incidence angle approaches 90 degrees. Assuming you've oriented the array towards the equator, that's local solar noon. Assuming you have a flat plate collector at latitude inclination, power output increases towards the equinox and declines away from it. If the system was commissioned at either solstice the peak output won't have been seen because incidence angle was at a minimum at local solar noon.

Here's where time of year comes into play.

There are two other sources of variance - orbital and environmental, and don't y'all think I'm being weird with "orbital", I had to debug a problem in Australia because of this. The Spring equinox tends to produce more power than the Autumn one because air temperature is lower, and our dear friend -dP / dT gets involved. That is, the array will produce more power at the Spring equinox because Temperature is lower than the Autumn equinox. Whenever I monitored grid voltage, I also found that cool Spring days had the higher grid voltage. So, it could be that the higher voltage rise, combined with the higher grid voltage, is causing your IEEE 1547 trips.

That's my final answer and I'm sticking with it.
 
I would say 1 or 3, but I'm going to give 1 a little more weight than most of the others have. We have had this happen several times, when the grid impedance is fairly high due to old long 4800 volt lines with not enough tap changers.

Actually I am dealing with exactly this on a system right now. In this case we have another thing working against us and that is that the energy has to go through two utility transformers before it gets to the grid. This is because The overhead distribution line is 4800 Delta, but this house is served by an underground line probably 1,000 ft to a pad mount. So what this power company does instead of running two CN cables and using a two bushing to pad transformer, they put a transformer on the pole to change it to a 2400 volt MGN so they can only use one cable to the pad mount. So each of those transformers you're picking up several volts in addition to the already highly variable distribution line voltage which runs quite high. Unhappy face.
 

tallgirl

Senior Member
Location
Great White North
Occupation
Controls Systems firmware engineer
I would say 1 or 3, but I'm going to give 1 a little more weight than most of the others have. We have had this happen several times, when the grid impedance is fairly high due to old long 4800 volt lines with not enough tap changers.

Actually I am dealing with exactly this on a system right now. In this case we have another thing working against us and that is that the energy has to go through two utility transformers before it gets to the grid. This is because The overhead distribution line is 4800 Delta, but this house is served by an underground line probably 1,000 ft to a pad mount. So what this power company does instead of running two CN cables and using a two bushing to pad transformer, they put a transformer on the pole to change it to a 2400 volt MGN so they can only use one cable to the pad mount. So each of those transformers you're picking up several volts in addition to the already highly variable distribution line voltage which runs quite high. Unhappy face.
You really want to look at total available solar energy. This was the issue I found in Australia where they were at perihelion and irradiance was at the peak for the year. Solar will cause a positive voltage gradient and you may not be able to just weasel your way out of it with wiring and transformer changes in all situations because what makes things better for you may make things worse for the ISO and they will always win.

There are several things potentially working against you right now. One is perihelion, which is a lot like that toe nail fungus no one talks about. We're 6 weeks from perihelion which means irradiance is several percent over the average. Another is temperature, because -dP / dT is a winter problem when it comes to being over the IEEE limits. It's -8C here, right now, which is 33C below STC, which is about 16% more potential power output. Finally, there was / is the tendency of projects to have more solar than they should because it was cheap and gee, the inverters will back off as they reach their output limits.
 
You really want to look at total available solar energy. This was the issue I found in Australia where they were at perihelion and irradiance was at the peak for the year. Solar will cause a positive voltage gradient and you may not be able to just weasel your way out of it with wiring and transformer changes in all situations because what makes things better for you may make things worse for the ISO and they will always win.

There are several things potentially working against you right now. One is perihelion, which is a lot like that toe nail fungus no one talks about. We're 6 weeks from perihelion which means irradiance is several percent over the average. Another is temperature, because -dP / dT is a winter problem when it comes to being over the IEEE limits. It's -8C here, right now, which is 33C below STC, which is about 16% more potential power output. Finally, there was / is the tendency of projects to have more solar than they should because it was cheap and gee, the inverters will back off as they reach their output limits.
I understand what you are saying, but I don't quite buy the perihelion and cold weather arguments. Typically, we are designing systems with a DC AC ratio of about 1.2, so we do see full inverter output year round. Of course when temperatures are cool in a radiance is highest you might see longer durations of maximum inverter output, but most of my systems I see full inverter output for at least part of the day at all times of the year. Generally I would say, if you aren't reaching full inverter output at least here and there in the summer, your inverters are too big.
 

tallgirl

Senior Member
Location
Great White North
Occupation
Controls Systems firmware engineer
I understand what you are saying, but I don't quite buy the perihelion and cold weather arguments. Typically, we are designing systems with a DC AC ratio of about 1.2, so we do see full inverter output year round. Of course when temperatures are cool in a radiance is highest you might see longer durations of maximum inverter output, but most of my systems I see full inverter output for at least part of the day at all times of the year. Generally I would say, if you aren't reaching full inverter output at least here and there in the summer, your inverters are too big.

From aphelion to perihelion the difference in irradiance at the ground is 6%. That's the same as a -12C change in temperature. In the Northern Hemisphere, perihelion is the first week of January, which is often even more colder than whatever average temperature you may have used for your sizing calculation. Our air temperature difference where I live now is about 60C, which is 30% with -dP / dT. Add the 6% and you now have 36% difference from aphelion (summer) to perihelion (winter).


What that does is increase the amount of time you're running the inverters at maximum output. The ISO doesn't have to operate your connection at a voltage favorable to you, they run the grid at voltages which keep the negative voltage gradient within the window needed for them to provide power in the window residences require. You come along and feed into their distribution which either reduces negative voltage gradient, or if enough people do it, causes positive voltage gradient on the distribution line.
 

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
In other words perihelion and low temperatures contribute more to 1 than to 3.
Can you elabortate? (1) and (2) refer to the voltage at the utility connection, not the inverter, correct? [Otherwise (1) and (3) would not be independent.] So how do perihelion and low temperatures contribute to a higher voltage at the utility connection?

Cheers, Wayne
 

tallgirl

Senior Member
Location
Great White North
Occupation
Controls Systems firmware engineer
In other words perhelion and low temperatures contribute more to 1 than to 3.
Correct. If you look at available energy, and you're only seeing IEEE window problems when total available energy is high (including irradiance and temperature correction), that's a problem between you and the grid. If you're seeing problems when available energy is significantly below 100%, that's very likely a feeder problem.

Utilities can do things about the duck curve and loss of inertia. They really can't do much when someone maxes out their solar production on their house so they can store kWh for the summer when they are running their pool pump and A/C at full blast.

Most residential inverters are dumb as dirt. I've become more inclined towards banning residential solar until it can play nicer with distribution systems. I've not tracked what the IEEE has been doing in terms of shaping residential inverter output, but residential inverters should start reducing output within a few volts of the upper window. That would greatly improve grid stability, as would reducing production as the upper frequency limit is approached.
 

tallgirl

Senior Member
Location
Great White North
Occupation
Controls Systems firmware engineer
Can you elabortate? (1) and (2) refer to the voltage at the utility connection, not the inverter, correct? [Otherwise (1) and (3) would not be independent.] So how do perihelion and low temperatures contribute to a higher voltage at the utility connection?

Cheers, Wayne
Power doesn't flow from low voltage to high voltage. There's always a gradient. When you maximize power, voltage rise in the feeder increases (Ohm's Law).

If you used a superconductor as the feeder, you just move the problem. In some areas positive voltage gradient in the distribution system is already an issue, and positive voltage gradient is really, really bad news.
 

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
Power doesn't flow from low voltage to high voltage. There's always a gradient. When you maximize power, voltage rise in the feeder increases (Ohm's Law).
Of course, that's why there's voltage rise (relative to the utility) on the feeder from the utility to the PV inverter.

I guess what I'm not following is this: there's some standard (which?) that says the utility has to deliver voltage within a certain window at the service point. And then there's some standard (again which?) that says the inverter should only operate within a certain voltage window.

If the former upper limit is not less than the latter upper limit, the PV system is potentially (hah) out of luck, as if the utility is operating near/at the top of the window, there's no remaining margin for allowable voltage rise on the feeder. Otherwise, you have some allowable margin, and you size your feeder to ensure that the voltage rise at full power output of the inverter will produce a voltage rise not exceeding your allowable margin.

Issues of available DC power just speak to how often / how long the inverter will want to put out full power.

Cheers, Wayne
 

jaggedben

Senior Member
Location
Northern California
Occupation
Solar and Energy Storage Installer
...

Most residential inverters are dumb as dirt. I've become more inclined towards banning residential solar until it can play nicer with distribution systems. I've not tracked what the IEEE has been doing in terms of shaping residential inverter output, but residential inverters should start reducing output within a few volts of the upper window. That would greatly improve grid stability, as would reducing production as the upper frequency limit is approached.
I would say this is out of date and you indeed haven't been keeping track. UL1741SA is an example of expanding the standard to address these issues. California's smart inverter requirements that rolled out between 2017 and 2020 mean that currently produced inverters being sold in the US are 'smart' not 'dumb'. (Technically they are 'grid support' inverters as opposed to mere 'interactive' inverters.) I haven't nerded out on all the nitty-gritty but they are supposed to adjust output and power factor to respond to voltage and frequency conditions on the grid.

These issues are essentially a matter of programming. So while there is a large fleet of 'dumb' inverters still out there, the idea that the grid can't handle continued addition of new smart inverters doesn't hold much water in my opinion.
 

jaggedben

Senior Member
Location
Northern California
Occupation
Solar and Energy Storage Installer
Of course, that's why there's voltage rise (relative to the utility) on the feeder from the utility to the PV inverter.

I guess what I'm not following is this: there's some standard (which?) that says the utility has to deliver voltage within a certain window at the service point. And then there's some standard (again which?) that says the inverter should only operate within a certain voltage window.
Yes, utilities in general have a responsibility to keep the voltage within a window near nominal so that peoples appliances and devices function and don't fry. My understanding is that broadly speaking under IEEE standards this is within +10% and -12%. Inverters generally follow the same standards. I'm vastly oversimplifying, but that's the general idea.

If the former upper limit is not less than the latter upper limit, the PV system is potentially (hah) out of luck, as if the utility is operating near/at the top of the window, there's no remaining margin for allowable voltage rise on the feeder. Otherwise, you have some allowable margin, and you size your feeder to ensure that the voltage rise at full power output of the inverter will produce a voltage rise not exceeding your allowable margin.
In general this is correct. Note that in fact inverters have the option to limit their output to mitigate the problems being discussed here, as I alluded to in my previous post.

Issues of available DC power just speak to how often / how long the inverter will want to put out full power.
But insolation also speaks to how much interactive power is being injected into the network, and that affects the grid voltage, which is what tallgirl is getting at. Ultimately the service voltage is not totally independent of the output of your solar system and everybody else's.

Simple seasonal variation is a far larger factor than perihelion. But when perihelion coincides with summer (as is approaching in Australia but not the US), then the system operator will have even more voltage rise to contend with from interconnected solar systems.
 
A couple points:

1. Regarding #1 in the OP, I think that there is a difference between "The DG being the problem" and "The POCO having a problem". In the cases I was discussing, it is definitely the latter. Fact is voltage is all over the place even without the solar connected. I was seeing voltage go from 240 to 255 in a period of less than a minute. The line needs to be upgraded from 4.8KV and/or have more regulators. The POCO guy made some comment about the DG being the problem. No. Your voltage is all over the place and I don't see a single regulator on the line, just no one noticed it before.

2. Regarding the "residential solar should be banned" comment, that just seems like a broad over generalization. I am sure there are some lines that have a ton of DG on them and perhaps it's causing issues (probably fixable with adjusting regulator parameters, maybe @Hv&Lv can comment). Other lines probably have relatively little DG on them and a long ways to go before there are any "issues".
 
Top