What to do with possibly-defective replacement inverter?

[chris_gulla]

Short version: a warranty replacement Sunpower inverter is providing much less power than the one it replaced, and I do not think it is because of the weather. What should I do?

Longer version: In September 2017 my Sunpower SPR 5000x died after running well for just over 9 years in a residential rooftop installation with 24 Sunpower SPR 230 panels (original install was July 2008). It was under warranty so my original installer put in a replacement after many delays (December), with zero cost to me. The replacement is a Sunpower SPR 5000m manufactured in July 2007.

A meter counting kWh sits between the inverter and the service panel, so I can easily see system production. I have taken down the readings every month since the system was new, so I have good statistics including average monthly production over several years. The monthly production since the replacement inverter was installed has been about 60% of the monthly average over the past 5 years (for example, my July 2018 production was 513 kWh and my average July production for 2013-2017 is 799 kWh. Other months are worse.). I think comparing against the 5-year average removes much of the possibility that the reduced production is due to the weather, but to be clear, the weather here in Connecticut this year has been a bit more rainy and cloudy than usual.

I asked the installer to check out the system (August) and they said everything is fine. They then asked for my production statistics and electric bills so they could investigate further, which I gave them right away. It has been over a month and they have not gotten back to me. These guys dragged their feet with the replacement and are again slow-walking me. I do not really want to deal with them but I may be running out of options.

Another complication or possibly red herring: the electric utility swapped out their meters in February. There used to be two meters, one counting outgoing power and one counting incoming. We have net metering here. Now there is just one meter but it still counts incoming and outgoing separately.

So many questions, and I do not even know if I am asking the right ones:

Should I get a 3^rd party to check out the system? Is it possible that the replacement inverter is defective and just doing a bad job? Can that be tested? Should I just buy a new inverter myself and try getting my installer to buy back the bad one? Should I let it run for another year to see if it was just a weather thing?

Any guidance would be greatly appreciated. I am happy to answer any questions you may have. Thanks!

[bcroe]

Start by checking the power output over the day, should be that bell curve. It might be cutting
off at times, or just running way low continuously. Bruce Roe

[J.P.M.]

Some things to look at/consider:

1.) Given the variability of the weather, 5 years is not a lot of 1X/month data. What is the range of the monthly production numbers by month ?

2.) Have you noticed an increase that you can't otherwise explain in your net billing of kWh (not dollar amounts, but how many kWh you have been billed for) since the inverter change out ?

3.) Have you compared your production changes with that of nearby systems using PVOutput.org ? Find systems of similar orientation near you and compare output per STC kW.

4.) How does the production as shown on the meter between your service panel and the inverter compare with the production shown on the inverter ?

Why did you wait so long ?

[chris_gulla]

bcroe Thanks for the suggestion. I am taking hourly readings today and will try and keep it up all week (I work from home). Will provide statistics later.

J.P.M. Some answers:
1. The monthly output for March through July 2018 has been less than my lowest output for the same month going back to 2009:
March 2018: 354 kWh, previous lowest 523 kWh (March 2017)
April: 406 vs 569 (2011)
May: 453 vs 657 (2011)
June: 497 vs 611 (2009)
July: 513 vs 759 (2013)

2. Yes, my net incoming power has gone up. Actually has gone from net sending power out to net bringing power in, as I am both sending less power out and bringing more power in than before. It might be interesting to see the actual differences. If you want statistics let me know.

3. I did not know about that website (thanks!) but have just registered and looked at a couple of systems near me. They do not show a noticeable decrease in production but I have only looked at a couple so far.

4. Unfortunately I did not take the readings from the old inverter so I cannot compare. I assumed the total output readings on the inverter would reflect the meter. Bad assumption?

5. I waited a few months to make sure it was not a temporary problem due to weather.

Thanks for your interest and help.

[J.P.M.]

bcroe Thanks for the suggestion. I am taking hourly readings today and will try and keep it up all week (I work from home). Will provide statistics later.

J.P.M. Some answers:
1. The monthly output for March through July 2018 has been less than my lowest output for the same month going back to 2009:
March 2018: 354 kWh, previous lowest 523 kWh (March 2017)
April: 406 vs 569 (2011)
May: 453 vs 657 (2011)
June: 497 vs 611 (2009)
July: 513 vs 759 (2013)

2. Yes, my net incoming power has gone up. Actually has gone from net sending power out to net bringing power in, as I am both sending less power out and bringing more power in than before. It might be interesting to see the actual differences. If you want statistics let me know.

3. I did not know about that website (thanks!) but have just registered and looked at a couple of systems near me. They do not show a noticeable decrease in production but I have only looked at a couple so far.

4. Unfortunately I did not take the readings from the old inverter so I cannot compare. I assumed the total output readings on the inverter would reflect the meter. Bad assumption?

5. I waited a few months to make sure it was not a temporary problem due to weather.

Thanks for your interest and help.

Looking at #'s 1 and 2 above together, and assuming your use didn't spike after March 2018 for some reason(s), and/or has stayed close to the same level as recent years while your draw from the grid has increased, I'd conclude the power from the inverter to the grid has gone down. If that's the case (?), looks like one possibility might be that the inverter is seeing less power into it for some reason(s), one such reason being that while the array is producing at about the same conversion efficiency, the available irradiance has changed. However, that increase in what you draw from the grid sounds like it's too much to be explained by weather alone, if at all. Possible but IMO, a pretty low probability, particularly over several months. That leaves the connections to the inverter or the inverter as the source of the perceived increases in draw from the grid. If the (re)wiring to the inverter was somehow not done correctly, then all the power the array is producing may not be making it to the inverter. Or, it may be that the inverter is not functioning as it was designed.

On checing PVOutput, keep in mind to look for arrays near you that have close to the same orientation in both tilt and azimuth and concentrate on weekly or monthly totals for any comparison between systems. Also note that system shading can have effects that may not be noticeable or at least not apparent. After checking with what PVOutput and gleaning what you can surmise from that, another call to the folks who replaced the inverter may be in order. Depending on the warranty (if any) on the replacement, you may want to have the wiring to and from the inverter checked out as well as the array's output voltage and current under bright sun at the time of min. incidence angle for the day(s) you measure it.

[chris_gulla]

J.P.M. Thanks for your thoughtful analysis. Just to add a few data points:
Since you mentioned consumption and power to the grid, I can give you the differences between Jan-Jul 2017 vs 2018. For that 7-month span, in 2017 my consumption was 3,264 kWh while in 2018 it was 3,174 kWh. Consumption has been fairly consistent over the past 5-6 years. For those same months in 2018 I sent 1,720 kWh out to the grid, while in 2017 I sent out 2,889 kWh. I actually sent out less to the grid for those months in 2017 than in previous years. So yes I am definitely feeding the grid a lot less this year.

I found four similar systems on PVOutput that are fairly close to me and who had data for 2017 and 2018. The most distant one was 38 miles away. I looked for similar azimuth (I face just west of south, at around 189) with no or low shade (I have none) and any tilt. My guess as to my tilt is 20 deg, as I have a non-steep roof that I can easily walk on. Definitely under 45 deg. I looked at the change in efficiency by month. Similar system A: worst month was February 2018 with a 19% efficiency decrease; the best month was January (+28%). System B: worst month was February (-24%), best month was May (+14%). System C: worst month was February (-20%), best month was May (+18%). System D: January (-61%) and February (-52%) were pretty horrible and my guess is that it was snow/ice buildup and/or he is not reporting his azimuth/ shade accurately, their best month was July (+5%). My production stats: January 17 vs 18: -22%, Feb-Mar combined: -27%, April: -33%, May: -32%, June: -35%, July: -34%. It seems like a fairly easy conclusion that my system performed much worse year to year than three of the other similar systems and worse than the other one if you take away the ice/snow months.

[J.P.M.]

J.P.M. Thanks for your thoughtful analysis. Just to add a few data points:
Since you mentioned consumption and power to the grid, I can give you the differences between Jan-Jul 2017 vs 2018. For that 7-month span, in 2017 my consumption was 3,264 kWh while in 2018 it was 3,174 kWh. Consumption has been fairly consistent over the past 5-6 years. For those same months in 2018 I sent 1,720 kWh out to the grid, while in 2017 I sent out 2,889 kWh. I actually sent out less to the grid for those months in 2017 than in previous years. So yes I am definitely feeding the grid a lot less this year.

I found four similar systems on PVOutput that are fairly close to me and who had data for 2017 and 2018. The most distant one was 38 miles away. I looked for similar azimuth (I face just west of south, at around 189) with no or low shade (I have none) and any tilt. My guess as to my tilt is 20 deg, as I have a non-steep roof that I can easily walk on. Definitely under 45 deg. I looked at the change in efficiency by month. Similar system A: worst month was February 2018 with a 19% efficiency decrease; the best month was January (+28%). System B: worst month was February (-24%), best month was May (+14%). System C: worst month was February (-20%), best month was May (+18%). System D: January (-61%) and February (-52%) were pretty horrible and my guess is that it was snow/ice buildup and/or he is not reporting his azimuth/ shade accurately, their best month was July (+5%). My production stats: January 17 vs 18: -22%, Feb-Mar combined: -27%, April: -33%, May: -32%, June: -35%, July: -34%. It seems like a fairly easy conclusion that my system performed much worse year to year than three of the other similar systems and worse than the other one if you take away the ice/snow months.

What's your zip ? Have you run PVWatts model yet ? If not, do so and compare each month with what the model produces for corresponding months. 5 years will not tell you much and PVWatts is a model, not a predictor of performance. But, if any month's output is substantially below what the model shows, - like, say 30 % less or so, and you're inputs to the model are a reasonable representation of reality (read the help/input screens a couple of times, get the array orientations as close as possible and use 10% for system losses, not the 14 % default), then there's a pretty high likelihood something is amiss.

[chris_gulla]

My ZIP is 06517. PVWatts is new to me but I plugged in my info with a 10% loss as you suggest. I guessed 20% for tilt. Their estimate was higher than my 5-yr averages for the winter months and lower for the summer months (I cannot see how they would account for snow/ice accumulation). The months from March through September are closer to my 5-yr averages than the other months. For the year the estimate was 7,296 kWh while my 5-yr average is 6,835 kWh (6% difference), so this first estimate seems pretty good. But if you compare their estimates with my 2018 production, all months are 33-56% less than the estimate. The closest was June, which is only 33% under the estimate (actual production 497 kWh vs estimate of 742 kWh).

I ran the estimate again with a 10% tilt and the estimated annual total was 6,906, much closer to my 5-yr average. Again, the closest month vs. actual 2018 was June, which was 34% under the estimate. When I use 30% for tilt (closer to what I would actually guess) the estimates get much higher and further from my actual production. I'll have to head up to the attic and measure my roof to get a more solid estimate for tilt.

Based on all this I think I am going to have to have another talk with my installer. I cannot see that going well but maybe they will be reasonable. Honestly if they just install a new, working inverter I would be pretty happy even if I had to pay a good portion of the cost. How much of that cost is reasonable is a guess but I will try and negotiate. But I guess the wiring might be screwed up too, like they didn't connect some panels. I hope the fix is that easy!

Thanks again J.P.M. and if you have any more thoughts I will be keeping tabs on this.

[J.P.M.]

My ZIP is 06517. PVWatts is new to me but I plugged in my info with a 10% loss as you suggest. I guessed 20% for tilt. Their estimate was higher than my 5-yr averages for the winter months and lower for the summer months (I cannot see how they would account for snow/ice accumulation). The months from March through September are closer to my 5-yr averages than the other months. For the year the estimate was 7,296 kWh while my 5-yr average is 6,835 kWh (6% difference), so this first estimate seems pretty good. But if you compare their estimates with my 2018 production, all months are 33-56% less than the estimate. The closest was June, which is only 33% under the estimate (actual production 497 kWh vs estimate of 742 kWh).

I ran the estimate again with a 10% tilt and the estimated annual total was 6,906, much closer to my 5-yr average. Again, the closest month vs. actual 2018 was June, which was 34% under the estimate. When I use 30% for tilt (closer to what I would actually guess) the estimates get much higher and further from my actual production. I'll have to head up to the attic and measure my roof to get a more solid estimate for tilt.

Based on all this I think I am going to have to have another talk with my installer. I cannot see that going well but maybe they will be reasonable. Honestly if they just install a new, working inverter I would be pretty happy even if I had to pay a good portion of the cost. How much of that cost is reasonable is a guess but I will try and negotiate. But I guess the wiring might be screwed up too, like they didn't connect some panels. I hope the fix is that easy!

Thanks again J.P.M. and if you have any more thoughts I will be keeping tabs on this.

Get the tilt and azimuth correct. It probably won't make a large difference in the modeled output. I'd SWAG that shading from snow may be the culprit for the old output difference vs. modeled output.

A 30 % output difference due to one bad winter is possible but the probability is very low of that happening.