11.6kw Solar Roof, but only one Inverter?

[wwhitney]

I chose a AC/DC ratio of 1.12 because that's what a possible 3.8kW plus 7.6kW remediation would give. In fact there was no clipping at a 1.12 ratio at any time during the year, according to PVWatts.

So then with no clipping on any of the 3 runs your method will give the correct answer.

But maybe I missed something [edit: likely I did and this paragraph is moot] because I'm not following closely, isn't the point to also repeat the process without the extra 3.8 kW inverter, in order to see what the energy benefit of the 3.8 kW inverter is ? [= energy cost of omitting the 3.8 kW inverter.] That would make the DC/AC ratio 1.68, and then I would expect the two methods would give you different answers. With the "sum, then clip" method being more accurate than the "clip, then sum" method.

If you do do this, it would be interesting to see how much difference in the two methods you get.

Cheers, Wayne

 

[wwhitney]

while a AC/DC ratio of 1.68 gave 15,519 kWh.

Is that an estimate from the same "sum 3 runs" method? If so, I misunderstood.

5% is a pretty small difference. Of course, if the contract says more than one inverter, there's no excuse for unilaterally changing the system on the plans and as installed.

Cheers, Wayne

 

[dareed1]

Is that an estimate from the same "sum 3 runs" method? If so, I misunderstood.

I used the sum from a clear day in August to find a nearly equivalent single orientation solar array. Then I checked annual production at 1.12 and 1.68 for this equivalent array.

 

[bswood]

If you have 3 differently oriented subarrays on a single inverter (hopefully I got that right, not following the thread super close), then modeling them as 3 separate PVWatts runs, each with a DC/AC ratio of 1.12, is going to underestimate the total production. Because the summing answer will reflect a scenario when clipping occurs whenever a subarray produces more DC power than its proportional share of the AC inverter capacity. While in reality clipping will only occur when the total DC production exceeds the AC capacity.

If PVWatts gives you hourly numbers for each of the 8760 hours of the year, then you could do 3 runs with an AC/DC ratio of 1.0, dump all the numbers, sum them in a spreadsheet, and then cap each hour's production at your AC inverter capacity. That should be close to the correct clipping.

Cheers, Wayne

Thanks Wayne and dareeed1. Curious if any of you would push them for a credit due to the discrepancy between the system delivered/installed and what is listed on the PO.

 

[wwhitney]

Thanks Wayne and dareeed1. Curious if any of you would push them for a credit due to the discrepancy between the system delivered/installed and what is listed on the PO.

Certainly if the PO has the force of a contract. In which case any changes in the equipment installed would require a revision, they can't do that unilaterally.

You can ask for them to install an additional inverter. Or you or they could accurately model the ratio of lifetime energy production, and they could prorate the entire PV portion of the contract accordingly. Although that would require separating the PV cost from the ESS cost, which would be tricky.

Cheers, Wayne

 

[jgleigh]

I think you'll have a hard time getting any traction.

Performance or Production Guarantee
Provider is not providing you with a performance or production guarantee.

Interestingly enough, my PO lists the following:
2 Battery(ies): Tesla Powerwall 2
2 Battery(ies): Tesla Powerwall+
1 Inverter(s): Tesla Powerall+ 7.6

But I clearly have two inverters since I have two Powerwall+.

 

[adspguy]

How many 7.6 inverters should Tesla have installed?
View attachment 835869

Just out of curiosity what DC wattage did they put on your SMART application?

 

[bswood]

Just out of curiosity what DC wattage did they put on your SMART application?

This is what is on the SMART application. Very low estimate for annual production IMO. Not even close to the PVWatts estimates. Seems that no loss expected from shading according to this as well - assuming blank means 0%.

 

[bswood]

Certainly if the PO has the force of a contract. In which case any changes in the equipment installed would require a revision, they can't do that unilaterally.

You can ask for them to install an additional inverter. Or you or they could accurately model the ratio of lifetime energy production, and they could prorate the entire PV portion of the contract accordingly. Although that would require separating the PV cost from the ESS cost, which would be tricky.

Cheers, Wayne

A PO that is mutually agreed upon is a contract in my world (capital equipment). In fact, they are even one of our customers. Pretty certain if we didn't deliver everything on a PO they would not pay the full amount. Perhaps it was just a fat finger mistake on their end, but it does smell of something more nefarious. It could easily be viewed that Tesla gave me a price originally based on 2 inverters, changed the design but not the price (more profit, yay!). Agree that the total annual output is not dramatically different (~5%) but as dareed1 says they don't guarantee output anyway. Really on the fence about pushing this to their resolution team and potentially arbitration. So far all my interaction has been with multiple PAs - I think I'm on number four or five now. Took 9 months from order to installation.

 

[dareed1]

If PVWatts gives you hourly numbers for each of the 8760 hours of the year, then you could do 3 runs with an AC/DC ratio of 1.0, dump all the numbers, sum them in a spreadsheet, and then cap each hour's production at your AC inverter capacity. That should be close to the correct clipping.

Wayne, here's your method. It predicts an even lower loss: For a DC/AC ratio of 1, the projected annual production is 16.089kWh. Clipping the hourly sum of the three arrays at 7.6kW give 15.707. Clipping at 11.4kW produces the same as DC/AC=1. This is about 2.4% loss.
@bswood, you can compare the projected duration of clipping. The below graph suggests perhaps 4.5 hours at this time of year, although it will depend on the real efficiencies of your panels (especially the panel temperature). Presumably the times shown by PVWatts are solar time, not EDT.
If I were in this situation, I would find the clipping to be unpleasing, but I certainly wouldn't push for any resolution. Also, I don't think this is nefarious behavior by Tesla. Note that other details in the plans weren't followed-- the location of the panels on your roof are apparently different to the original plans, probably to meet setback requirements. From my point of view, we customers are reaping the benefits of overall low cost, and implicitly agree to minor deviations from the originally stated plans--and of course the famously poor response for customer service.

 

[adspguy]

This is what is on the SMART application. Very low estimate for annual production IMO. Not even close to the PVWatts estimates. Seems that no loss expected from shading according to this as well - assuming blank means 0%.
View attachment 837034

So your AC is 11.6 which implies a 7.6 + another 4 of inverter. What they tell the power company affects what net metering plan you get. An inverter up to 10kW AC gets 1:1 but bigger inverters uses a more complicated and less lucrative formula. If they applied for SMART with 11.6 kW AC but gave you less than that you might be getting a worse deal on SMART.

My system from 7/2020 is specified 12.24kW DC 10kW AC and came with a solaredge 10kW inverter. My PO didn’t specify the AC generation value. The estimated production is listed at 11.255MWh. For 2021 I ended up with 11.73MWh and I am on track for a bit more this year. My roof has virtually no shading. In March when I get peak generation I have some clipping for an hour or so.

Here is a recent days generation. If my system was limited to 7.6kW it would have been clipping between 11:15 and 3:30.

 

[bswood]

So your AC is 11.6 which implies a 7.6 + another 4 of inverter. What they tell the power company affects what net metering plan you get. An inverter up to 10kW AC gets 1:1 but bigger inverters uses a more complicated and less lucrative formula. If they applied for SMART with 11.6 kW AC but gave you less than that you might be getting a worse deal on SMART.

My system from 7/2020 is specified 12.24kW DC 10kW AC and came with a solaredge 10kW inverter. My PO didn’t specify the AC generation value. The estimated production is listed at 11.255MWh. For 2021 I ended up with 11.73MWh and I am on track for a bit more this year. My roof has virtually no shading. In March when I get peak generation I have some clipping for an hour or so.

Here is a recent days generation. If my system was limited to 7.6kW it would have been clipping between 11:15 and 3:30.
View attachment 837545

Thanks adspguy (and Wayne and dareed1). Attached is my data from the same day. I’m just south of Boston so our location is pretty close. Haven’t hit 60kWh with this setup. Seems the optimal system here would be to add another 3.8 inverter but they have refused to do so. And given the small amount of overall gain I think I am just going to give in and move on. Thanks for all the input. Definitely helped me make some decisions here.

 

[Josh Gerritsen]

It seems like I might be the first person to offer some evidence that the whole Overdrive idea might be real!

Background: I live in Maine and I just had my PV and Powerwall system installed by a third party. I have 10.8 kW of Solar panels, One Powerwall+, and one Powerwall. They are connected to my system with a Backup Gateway 2. My Powerwall+ has a single 7.6 kW inverter.

Last Tuesday, I was looking at my Tesla app and the system was currently generating 11 kW of energy. Here were the specific conditions:

- I was charging my Model Y so the house was using 16.7 kW
- It was a perfectly clear, cold day and my yard is covered in snow. The snow must have bounced additional light into the panels to exceed their 10.8 kW rating
- I currently have permission to export OFF

You can verify most of this information from the screenshots attached below. I didn't think it was possible to exceed the inverters 7.6 kW rating, and my understanding is that the inverter and the Powerwall do not have a direct DC connection, only AC. Maybe that's incorrect, and the inverter is sending the excess DC voltage to the Powerwall to be utilized? Also, my system hasn't clipped once in my three weeks of having this system.

I'd love to hear everyone's theories as to how this is all possible. However it's occurring, I'm so glad I have the Powerwall+ system versus using a different inverter that can't go past it's inverter rating. The sun is shining again today and my system is generating 8 kW as we speak.

 

[dareed1]

I don't suppose that you will want to hear this, but this is much more likely to be incorrectly installed current transducers than actual generation. I ran PVWatts on Rockford Maine, and even with a 45 degree roof pitch with panels oriented exactly south and ground albedo at 0.87 (meaning snow covered), the highest peak power in February (the 13th in their data base) was 10.3kW. If the roof pitch is instead 30 degrees, the best peak power was 9.2kW. I didn't run different azimuths, but of course any direction other than exactly south will also lower the available solar power. All of this was calculated with a DC/AC ratio of 1.0. What is your roof pitch and azimuth?

Th graph shows a sharp increase at around 11AM. If the day were partly cloudy, this could be caused by cloud lensing, but it is unlikely to persist for an hour. Besides, other than the discontinuity, the curve looks quite smooth, which is uncharacteristic of partly cloudy days. And of course your recollection is that it was a clear day. I think is is much more plausible that the CTs are partially monitoring the house load and it changed at that time.

 

[Josh Gerritsen]

I don't suppose that you will want to hear this, but this is much more likely to be incorrectly installed current transducers than actual generation. I ran PVWatts on Rockford Maine, and even with a 45 degree roof pitch with panels oriented exactly south and ground albedo at 0.87 (meaning snow covered), the highest peak power in February (the 13th in their data base) was 10.3kW. If the roof pitch is instead 30 degrees, the best peak power was 9.2kW. I didn't run different azimuths, but of course any direction other than exactly south will also lower the available solar power. All of this was calculated with a DC/AC ratio of 1.0. What is your roof pitch and azimuth?

Th graph shows a sharp increase at around 11AM. If the day were partly cloudy, this could be caused by cloud lensing, but it is unlikely to persist for an hour. Besides, other than the discontinuity, the curve looks quite smooth, which is uncharacteristic of partly cloudy days. And of course your recollection is that it was a clear day. I think is is much more plausible that the CTs are partially monitoring the house load and it changed at that time.

That could be, but I need to share some additional information about my system. The majority of my solar panels, 20 panels totaling 8 kW, are oriented at 110 Degrees. I would estimate that roof pitch is around 45 degrees. Not ideal, but I was working with the roof I have. The remaining seven panels, totaling 2.8 kW, are oriented at 150 degrees. I estimate this roof pitch is around 30 degrees. This could explain why your numbers are not adding up to what my app reported: the sun was more directly facing the panels than your calculations accounted for. Note the system is only producing 10+ kW of energy for a short amount of time, when the sun angle was more ideal for the panel orientation.

Or I'm completely wrong and the current transducers were installed incorrectly : )

Is there anyone else out there who shows production beyond what their inverter is rated for? Anyone?

 

[dareed1]

My above calculation assumed all of the panels are facing directly south, and that is the very best orientation to get peak power. BTW, for this time of year at 44 degree latitude, the optimal panel inclination would be around 66 degrees.

Normally, we describe north as azimuth 0 degrees, so 110 degrees is much more east than south, while 150 is still 30 degrees east of south. Using PVWatts, with your estimated roof angles and azimuths, with 5% system losses, 98% inverter efficiency, the results for the 8kW part is 6976W at a little after 10AM. The 2.8kW system reaches 2596W close to 11AM. So your peak is perhaps 9.6kW. Modeled for February 13th, with albedo = 0.87.

It's very unlikely that your reported power is accurate.

 

[Josh Gerritsen]

My above calculation assumed all of the panels are facing directly south, and that is the very best orientation to get peak power. BTW, for this time of year at 44 degree latitude, the optimal panel inclination would be around 66 degrees.

Normally, we describe north as azimuth 0 degrees, so 110 degrees is much more east than south, while 150 is still 30 degrees east of south. Using PVWatts, with your estimated roof angles and azimuths, with 5% system losses, 98% inverter efficiency, the results for the 8kW part is 6976W at a little after 10AM. The 2.8kW system reaches 2596W close to 11AM. So your peak is perhaps 9.6kW. Modeled for February 13th, with albedo = 0.87.

It's very unlikely that your reported power is accurate.

Yes, I also am describing north as azimuth 0 degrees, so we're on the same page in terms of the orientation of my panels. I know that facing directly south is the very best orientation to get peak power over an entire day. But the sun is moving throughout the day, and if the solar panels could move and track the sun, that would be more ideal than being stationary at 180 degrees.

My point being that at a given time a day depending on the time of year, being at 110 degrees could generate more electricity than 180 degrees, but only for a sliver of the day, just like what happened to me. I obsessively look at my Tesla app and watch as the energy being produced by the panels goes up and down, depending on if its clear or a cloud comes in and blocks the sun. I see it rise and fall in real time as I'm looking out the window, and production rises and falls in a logical way. When the sun sets, production drops to 0 kW. So the monitoring isn't seeing any house load anytime the sun is down, as you would expect if monitoring was set up correctly.

Respectfully, I don't think your theory has any evidence to back it up. I have my Tesla app reporting this. I also don't think its a coincidence that my production went up to almost exactly the same number as the rating of my solar panels (10.8 kW). This has happened multiple times too. And just yesterday I hit 8.5 kW. Remember, my inverter is rated at 7.6 kW. So anything beyond 7.6 kW is considered "impossible".

Considering Tesla has said to another person here that the Powerwall+ can produce more energy than the rating of the inverter, I think we all need to be open and consider that this may actually be the case.

I hope others that have seen this activity with their Powerwall+ systems chime in. I attached photos of my inverter and Backup Gateway 2 wiring configurations so you guys can see if the CTs are in the wrong spot. If they are, then I take everything I said back!

 

[taphil]

What’s your electricity meter reading during the supposed inverter overdrive period?

At peak sun, turn off the powerwalls, car charging, and everything in the house so 99% of solar goes to the grid. Does your meter read -11kWh?

No one else has mentioned overdrive, especially now that Tesla installs high ratio of solar:inverter.

 

[jgleigh]

- I was charging my Model Y so the house was using 16.7 kW

I'm assuming you're charging at 32A (7.7 kW)? Is your normal house usage at that time also 9 kW?

 

[jgleigh]

I hope others that have seen this activity with their Powerwall+ systems chime in. I attached photos of my inverter and Backup Gateway 2 wiring configurations so you guys can see if the CTs are in the wrong spot. If they are, then I take everything I said back!

You have a CT on the 50A input (which would be from the Powerwall+) but there's already a CT built into the Powerwall+ inverter. That doesn't look right. It's probably overcounting how much solar you're actually generating.

Also, is the Gateway being used as your main service panel? It looks like the bonding jumper is still installed.