Tesla string shenanigans or sorcery?

[OCJ]

I did a fair amount of research before getting solar this year, but not enough. I partially blame that on new house, new job, and new baby.

I have a 9.6 kW Tesla system with a Powerwall+. 24 panels x 400 W, 4 different azimuths and 2 different pitches in SoCal SDG&E territory.

Azimuth	Pitch	Panels	DC
147 (SE)	12 deg	7	2.8 kW
57 (NE)	20 deg	5	2.0 kW
327 (NW)	12 deg	6	2.4 kW
237 (SW)	20 deg	6	2.4 kW

Now when I got this initial design I ran it through PVWatts and I thought cool, the NE and SE will charge up my Powerwall in the morning, the SE will give me power into the afternoon, the NW will give me some good production into the late afternoon, and the SW panels with bring it home into On-Peak TOU especially into fall/winter. It seemed that the marginal cost of those NE and NW panels was worth the squeeze.

Some of you might already see where this is going: Tesla put those two 6-panel strings in parallel, NW and SW. Now, I didn't realize this until after it was done and I wasn't even able to talk to the team much because I was slammed with the new job. Still, no excuses, I dropped the ball.

What I was expecting was when the sun was overhead in May when I got it installed, and June, and July, the production would be close as modeled in PVWatts. As the sun gets lower in the Southern sky, I expected the NW panels to bring down my SW production, and was gearing up to call Tesla and ask if they will address it, even though the annual production will still probably be above what they told me and I don't have a leg to stand on.

BUT, what I'm finding is the SW panel production is coming down from what PV estimates, but the NW panels are coming up in production. The average of the panel production is almost the same. So is this expected? Is this sorcery?

Here are the results of August 17:

	SE	NE	NW	SW	Total
PVWatts	16.884 kWh	10.752 kWh	13.251 kWh	14.468 kWh	55.625 kWh
Actual	16.892 kWh	10.186 kWh	13.972 kWh	14.077 kWh	55.127 kWh
% Diff	0.05%	-5.26%	3.34%	-2.70%	-0.90%

At some point will the decreasing angle of the sun hitting those NW panels bring down the average production of that string? I'm morbidly curious if that will happen and am debating letting it go through the winter, especially since I'm still oversized and generating good credits.

Anybody have any thoughts, opinions, explanations? Thank you in advance and I hope this wasn't too long.

[J.P.M.]

I did a fair amount of research before getting solar this year, but not enough. I partially blame that on new house, new job, and new baby.

I have a 9.6 kW Tesla system with a Powerwall+. 24 panels x 400 W, 4 different azimuths and 2 different pitches in SoCal SDG&E territory.

Azimuth	Pitch	Panels	DC
147 (SE)	12 deg	7	2.8 kW
57 (NE)	20 deg	5	2.0 kW
327 (NW)	12 deg	6	2.4 kW
237 (SW)	20 deg	6	2.4 kW

Now when I got this initial design I ran it through PVWatts and I thought cool, the NE and SE will charge up my Powerwall in the morning, the SE will give me power into the afternoon, the NW will give me some good production into the late afternoon, and the SW panels with bring it home into On-Peak TOU especially into fall/winter. It seemed that the marginal cost of those NE and NW panels was worth the squeeze.

Some of you might already see where this is going: Tesla put those two 6-panel strings in parallel, NW and SW. Now, I didn't realize this until after it was done and I wasn't even able to talk to the team much because I was slammed with the new job. Still, no excuses, I dropped the ball.

What I was expecting was when the sun was overhead in May when I got it installed, and June, and July, the production would be close as modeled in PVWatts. As the sun gets lower in the Southern sky, I expected the NW panels to bring down my SW production, and was gearing up to call Tesla and ask if they will address it, even though the annual production will still probably be above what they told me and I don't have a leg to stand on.

BUT, what I'm finding is the SW panel production is coming down from what PV estimates, but the NW panels are coming up in production. The average of the panel production is almost the same. So is this expected? Is this sorcery?

Here are the results of August 17:

	SE	NE	NW	SW	Total
PVWatts	16.884 kWh	10.752 kWh	13.251 kWh	14.468 kWh	55.625 kWh
Actual	16.892 kWh	10.186 kWh	13.972 kWh	14.077 kWh	55.127 kWh
% Diff	0.05%	-5.26%	3.34%	-2.70%	-0.90%

At some point will the decreasing angle of the sun hitting those NW panels bring down the average production of that string? I'm morbidly curious if that will happen and am debating letting it go through the winter, especially since I'm still oversized and generating good credits.

Anybody have any thoughts, opinions, explanations? Thank you in advance and I hope this wasn't too long.

Maybe by way of comment/explanation:

If your inputs to PVWatts are reflective of reality, the model's output(s) will, in all probability reflect what you may expect from your array(s) as an annual long term average +/- about 10% or so as the PVWatts help screens will tell you.

As you seem to have done it, a one day comparison of actual output against a model's means little. While that the 08/17 data from PVWatts seems to do a good job of matching your actual output for that day, that close match is more likely due to the string of sunny weather this time of year coupled with coincidence.

If it never rains, cloud variation is removed from a model. So, the model's output doesn't drift as much day/day as also the actual conditions making for a high(er) probability of a close(r) match between the model and actual output. For my somewhat nearby to you situation (zip 92026) except for 08/09 and 08/13, the average measured irradiance for the prior 31 days has only varied by +9.3%/-8.5% over the month. Specifically, my 08/17 output was 2.5 % above what PVWatts puked out for that day. So far for the prior 31 days my array's daily output has been 97.6% of what PVWatts modeled for the same time period with a std. dev. of 6.7 %. That's pretty stable as I'd expect from what's been a very and consistently sunny month.

1.) Is there (or was there) more room on your roof at the 147 deg. or 237 deg. azimuths ?
If there is, I'd have loaded those orientations to the max starting with the 247 deg. orientation (because under DR-SES and other SDG & E T.O.U. rate schedules, that's the best orientation of the 4 for most bill offset per STC W generated). I'd have filled that orientation first, followed by the 57 deg. azimuth orientation. I'm not sure I'd have done any northerly orientation at all as they are usually very difficult to make cost effective and so drag the cost effectiveness of the whole project down thereby reducing system cash flow and so overall cost effectiveness. That would also depend on your annual usage and how much of it you wanted to offset with PV, either in terms of bill reduction or (usage - generation).

2.) Is there any shading on any of the orientations ?

3.) On a year round basis either one of those southerly orientations would have been better for production than the other two, with the 237 azimuth favored as it will produce the most $$/kWh offset against a bill. If more PV on those roof sections was not possible then the other orientations might be used but, again, at the expense of cost6 effectiveness.

But all that's water under the bridge now.

The usual primary goal of getting residential is to generate as much electricity as needed with the idea of offsetting as much of a bill as desired by first, as much use reduction and conservation as deemed possible, and then by a mix of array size and orientation. Under most T.O.U. rate schedules, and SDG & E's in particular, northerly orientations are mostly antithetical to cost effective PV design.
Then, and even though it's probably not cost effective at this time - even with SDG & E peak summer time vs. off peak T.O.U. differential prices - maybe use some storage to time shift some of the load, but storage is usually a long way from cost effective at this time.
The more electricity that gets generated per installed STC kW by utilizing the best (that is, the most bang per installed STC kW in terms of bill offset) orientation, the more $$ saved and the more cost effective the array is.

By the way, what is your current electricity usage in kWh per year (not annual $$ billing) and what do you expect it to be in the future ?

[OCJ]

Thanks for your comments J.P.M. At the time I ordered solar we had only been living in this house a month. Since I was on SDG&E at the previous house, though, I pulled up the previous 1 year of data and made some guesstimations. The old house was 2,100 sqft, the new one was 2,800 sqft, a +33.33% difference, but the new house was also "newer" so I settled on +20%. The previous one year usage at the old house was 7,800 kWh with August being the worst at 1,259 kWh. So my guesstimate for the new home was around 9,400 kWh per year.

We also planned on an EV and a pool. Our mileage was pretty much exactly 30 miles per day on average, so around 10,950 miles/yr. I went with a 2.4 mi/kWh so that gave me an annual usage of 4,560 kWh. So now at around 13,960 kWh per year. And for the pool I believe I estimated around 1,500 kWh per year. That gives me 15,460 kWh per year expected.

Now, we have an EV. And our house turned out to be even more efficient than I thought, my annual house projection actually went down to 7,400 kWh.

Current: 11,960 kWh (higher efficiency and EV, 7,400 kWh house usage + 4,560 kWh EV usage)
Expected: 13,460 kWh (with pool, +1,500 kWh)

So for now I'm overproducing quite a bit, which is also why I'm looking at some electric space heating for winter for our bedroom and nursery.

As far as the roof goes, I don't think there is room for more panels SW and SE. I've linked to the diagram and pic. I think Tesla did a decent design given the limitations of my roof.



Regarding the NW and NE panels. I calculated them with a marginal cost of $2.70 per watt ($2.50 plus tax) since the SE and SW panels were going to be purchased no matter what. So the worst producing one, the NE panels, is $5,400, which is $2.70 * 5 panels * 400 watts with an annual production of 2,599 kWhs. Assuming every kWh is used, that's $1,170 per year from those panels (at $.45 per kWh). I estimate a 6-year ROI on that lowest productions set of panels.

I'm going to dig into the NW and SW data a bit more, maybe spot check a day in June and July. See if they exhibit that same variance.

I was looking forward to that SW set of panels bringing me good returns late in the evening, as you pointed out, especially going into fall/winter. I was disappointed they put it in parallel with the NW panels, but I really am curious how that NW set of panels is going to hold up towards the end of the year. I'm thinking since I'm overproducing anyway maybe see how it goes out of curiousity. Also, no shading at all. I did also get storage to give me options, even though the ROI is further out, but that's another post.

[J.P.M.]

Thanks for your comments J.P.M. At the time I ordered solar we had only been living in this house a month. Since I was on SDG&E at the previous house, though, I pulled up the previous 1 year of data and made some guesstimations. The old house was 2,100 sqft, the new one was 2,800 sqft, a +33.33% difference, but the new house was also "newer" so I settled on +20%. The previous one year usage at the old house was 7,800 kWh with August being the worst at 1,259 kWh. So my guesstimate for the new home was around 9,400 kWh per year.

We also planned on an EV and a pool. Our mileage was pretty much exactly 30 miles per day on average, so around 10,950 miles/yr. I went with a 2.4 mi/kWh so that gave me an annual usage of 4,560 kWh. So now at around 13,960 kWh per year. And for the pool I believe I estimated around 1,500 kWh per year. That gives me 15,460 kWh per year expected.

Now, we have an EV. And our house turned out to be even more efficient than I thought, my annual house projection actually went down to 7,400 kWh.

Current: 11,960 kWh (higher efficiency and EV, 7,400 kWh house usage + 4,560 kWh EV usage)
Expected: 13,460 kWh (with pool, +1,500 kWh)

So for now I'm overproducing quite a bit, which is also why I'm looking at some electric space heating for winter for our bedroom and nursery.

As far as the roof goes, I don't think there is room for more panels SW and SE. I've linked to the diagram and pic. I think Tesla did a decent design given the limitations of my roof.



Regarding the NW and NE panels. I calculated them with a marginal cost of $2.70 per watt ($2.50 plus tax) since the SE and SW panels were going to be purchased no matter what. So the worst producing one, the NE panels, is $5,400, which is $2.70 * 5 panels * 400 watts with an annual production of 2,599 kWhs. Assuming every kWh is used, that's $1,170 per year from those panels (at $.45 per kWh). I estimate a 6-year ROI on that lowest productions set of panels.

I'm going to dig into the NW and SW data a bit more, maybe spot check a day in June and July. See if they exhibit that same variance.

I was looking forward to that SW set of panels bringing me good returns late in the evening, as you pointed out, especially going into fall/winter. I was disappointed they put it in parallel with the NW panels, but I really am curious how that NW set of panels is going to hold up towards the end of the year. I'm thinking since I'm overproducing anyway maybe see how it goes out of curiousity. Also, no shading at all. I did also get storage to give me options, even though the ROI is further out, but that's another post.

Thank you for the information.
What did you pay Tesla for the array whole array and, separately, how much was the storage, both before fed. tax credit ?
At this time you've got what you've got.
Since SDG & E and their NEM tariffs and rate plans don't care which panels are where, and only about their total impact on what you draw and pay them for, I'd stop compartmentalizing which panels produce what and for what purpose at what times.
If minimizing how much you send to SDG & E and/or minimizing your bill at the gas pump are two of your priorities, your best chance to do those things now lies in use reduction and conservation efforts.

A bit off topic, but my condolences on choosing Tesla for your roof.

[OCJ]

Before tax credit and before sales tax, $24k for 9.6 kW and $10.5k for the Powerwall+. I checked a couple of local installers as well as https://www.californiadgstats.ca.gov/find_installer/ and decided the price was too good to pass up. Order placed 1/11/22, installed 5/11/22, PTO 6/23/22. The biggest ball drop was 6 weeks between the site visit and them telling me I needed to replace some roof tiles.

All in all the experience was fine for me, even though I hear others have major issues. I managed to generate $703 in credits through the beginning of August before getting an EV so now it's just my "hobby" to tweak whatever settings I can.

[Ampster]

Roof tiles? I have heard other stories about Tesla refusing to do roof tiles? I had a similar quote from Tesla in June 2021 but I knew I wanted to upgrade my service panel and was concerned that would delay the install. Interestingly their proposal included some panels facing ENE My proposal was with no Powerwall and they had recently changed their policy and I also thought my install would get delayed in favor of more profitable ones with Powerwalls. I did a self install with the help of some professionals and the cost, including the service panel upgrade was about the same as their proposal.

[oregon_phil]

Thanks for your comments J.P.M. At the time I ordered solar we had only been living in this house a month. Since I was on SDG&E at the previous house, though, I pulled up the previous 1 year of data and made some guesstimations. The old house was 2,100 sqft, the new one was 2,800 sqft, a +33.33% difference, but the new house was also "newer" so I settled on +20%. The previous one year usage at the old house was 7,800 kWh with August being the worst at 1,259 kWh. So my guesstimate for the new home was around 9,400 kWh per year.

We also planned on an EV and a pool. Our mileage was pretty much exactly 30 miles per day on average, so around 10,950 miles/yr. I went with a 2.4 mi/kWh so that gave me an annual usage of 4,560 kWh. So now at around 13,960 kWh per year. And for the pool I believe I estimated around 1,500 kWh per year. That gives me 15,460 kWh per year expected.

Now, we have an EV. And our house turned out to be even more efficient than I thought, my annual house projection actually went down to 7,400 kWh.

Current: 11,960 kWh (higher efficiency and EV, 7,400 kWh house usage + 4,560 kWh EV usage)
Expected: 13,460 kWh (with pool, +1,500 kWh)

So for now I'm overproducing quite a bit, which is also why I'm looking at some electric space heating for winter for our bedroom and nursery.

As far as the roof goes, I don't think there is room for more panels SW and SE. I've linked to the diagram and pic. I think Tesla did a decent design given the limitations of my roof.



Regarding the NW and NE panels. I calculated them with a marginal cost of $2.70 per watt ($2.50 plus tax) since the SE and SW panels were going to be purchased no matter what. So the worst producing one, the NE panels, is $5,400, which is $2.70 * 5 panels * 400 watts with an annual production of 2,599 kWhs. Assuming every kWh is used, that's $1,170 per year from those panels (at $.45 per kWh). I estimate a 6-year ROI on that lowest productions set of panels.

I'm going to dig into the NW and SW data a bit more, maybe spot check a day in June and July. See if they exhibit that same variance.

I was looking forward to that SW set of panels bringing me good returns late in the evening, as you pointed out, especially going into fall/winter. I was disappointed they put it in parallel with the NW panels, but I really am curious how that NW set of panels is going to hold up towards the end of the year. I'm thinking since I'm overproducing anyway maybe see how it goes out of curiousity. Also, no shading at all. I did also get storage to give me options, even though the ROI is further out, but that's another post.

Thank you for the picture of your layout. If you don't mind a slightly different set of questions:

1) Where is the electrical parallel connection for the SW and NW panels made? In the picture SW NW Panels.jpg is the parallel connection made on the roof at point A, or at the inverter at point B? Or some other place?

2) What type of wire did they use between arrays and how is the wiring run between arrays? On the roof in conduit? Down through the roof, in the attic?

3) There is very little online documentation for the Tesla 7.6kW 4MPPT inverter. I was able to find one picture of the internal wiring. I am guessing that the DC inputs have a jumper between 1+/2+ and 1-/2- or 3+/4+ and 3-/4-. Have you ever looked inside of the inverter after it has been shutdown?

You are concerned about paralleling SW and NW strings. Depending on where they made the parallel connection, what wire they used between between arrays (PV or THHN) and how they ran the wire between arrays (in the attic, in conduit, type of conduit, etc, ), it might be easy to un-parallel your solar arrays. Your inverter has 4 MPPT inputs. I'm at a loss on why your installer would choose to parallel inputs other than to save money on wire, labor, maybe doesn't have room in conduit?

[J.P.M.]

Before tax credit and before sales tax, $24k for 9.6 kW and $10.5k for the Powerwall+. I checked a couple of local installers as well as https://www.californiadgstats.ca.gov/find_installer/ and decided the price was too good to pass up. Order placed 1/11/22, installed 5/11/22, PTO 6/23/22. The biggest ball drop was 6 weeks between the site visit and them telling me I needed to replace some roof tiles.

All in all the experience was fine for me, even though I hear others have major issues. I managed to generate $703 in credits through the beginning of August before getting an EV so now it's just my "hobby" to tweak whatever settings I can.

Thank you for the reply.
Yes, that is a credit against future billing, not what you'll get a check for.
Any guess as to how long it'll take to break even on the power wall using differential T.O.U. rates ?

[OCJ]

Roof tiles? I have heard other stories about Tesla refusing to do roof tiles? I had a similar quote from Tesla in June 2021 but I knew I wanted to upgrade my service panel and was concerned that would delay the install. Interestingly their proposal included some panels facing ENE My proposal was with no Powerwall and they had recently changed their policy and I also thought my install would get delayed in favor of more profitable ones with Powerwalls. I did a self install with the help of some professionals and the cost, including the service panel upgrade was about the same as their proposal.

They asked me to fix them and send proof before they would schedule the install. So in my case no, they didn't do the tiles, they asked me to fix them.

[OCJ]

Thank you for the picture of your layout. If you don't mind a slightly different set of questions:

1) Where is the electrical parallel connection for the SW and NW panels made? In the picture SW NW Panels.jpg is the parallel connection made on the roof at point A, or at the inverter at point B? Or some other place?

2) What type of wire did they use between arrays and how is the wiring run between arrays? On the roof in conduit? Down through the roof, in the attic?

3) There is very little online documentation for the Tesla 7.6kW 4MPPT inverter. I was able to find one picture of the internal wiring. I am guessing that the DC inputs have a jumper between 1+/2+ and 1-/2- or 3+/4+ and 3-/4-. Have you ever looked inside of the inverter after it has been shutdown?

You are concerned about paralleling SW and NW strings. Depending on where they made the parallel connection, what wire they used between between arrays (PV or THHN) and how they ran the wire between arrays (in the attic, in conduit, type of conduit, etc, ), it might be easy to un-parallel your solar arrays. Your inverter has 4 MPPT inputs. I'm at a loss on why your installer would choose to parallel inputs other than to save money on wire, labor, maybe doesn't have room in conduit?

Yes, that's what I'm wondering about, un-paralleling the strings. I initially was going to call them and ask them to un-parallel them, but I wanted to quantify the data first. That's when I ran across that NW string producing more than PVWatts estimate, and the SW string producing less, but the average being the same. Although as J.P.M. pointed out this could be all within the margin of error of the tools.

For 1 and 2, here are the appropriate screenshots, all run through conduit on the roof:




I set up Powerwall-Dashboard so here's the output from pyPowerwall:


As for 3, I'm wondering that myself. The other panels were grouped in 5 and 7... if they had been 6 and 6, would they have put them in parallel also?

Either they did it to save running another pair, they do it by default, there wasn't enough room, or their AI knows something we don't. I don't know.

[OCJ]

Thank you for the reply.
Yes, that is a credit against future billing, not what you'll get a check for.
Any guess as to how long it'll take to break even on the power wall using differential T.O.U. rates ?

Yes, I expect I'll use some of that credit up in Winter. I know the excess generation gets < $.04 per kWh during True-Up.

I'm still working out the ROI for the Powerwall. I know it's long, but I'm not sure how long. As I mentioned, I did get it knowing that the ROI was longer, but wanted the ability to tweak how I run the system. Also, I may be able to get all or some of the Powerwall cost offset, in that case the ROI question is moot for me.

When I initially got PTO, I switched from Self-Powered to Time-Based Control, which runs 100% off the Powerwall between 4pm and 9pm and sends solar to the grid during peak. I was pushing around 9 kWh peak to the grid. I was also sending around 25 kWh Off-Peak during the day, until I got an EV.

At the beginning of August, I switched to Self-Powered with 20% reserve. So during On-Peak I only send what my house doesn't use. I also started charging the EV inside my solar production, from 11am to 3pm. So for Aug 17, I generated 55.7 kWh, pulled 0.3 kWh from the grid, pushed 19.1 kWh to the grid, and charged my EV at 4.3 kW for a couple of hours to get it back up to 80%. By doing this I minimize my NBCs.

Next up I'm thinking about creating my own Time-Based Control schedule to run off battery from 4pm to midnight, and maybe switch to EV-TOU2.

I know I'm only talking about $130 in NBCs for the year, but at this point of recently having a baby, switching jobs, and getting a new house, since my other hobbies have been destroyed this is my new hobby for now.

[OCJ]

Also, those few days I actually dropped my Powerwall reserve down to 5% just to see if I could not pull anything from the grid. I'm not going to do that long term, but it is cool being "invisible" from the utility in terms usage. Also, it was a dang near perfect solar day so this isn't always reproducible and of course won't be going into Winter.

Utility view:



House view:

[oregon_phil]

Yes, that's what I'm wondering about, un-paralleling the strings. I initially was going to call them and ask them to un-parallel them, but I wanted to quantify the data first. That's when I ran across that NW string producing more than PVWatts estimate, and the SW string producing less, but the average being the same. Although as J.P.M. pointed out this could be all within the margin of error of the tools.

For 1 and 2, here are the appropriate screenshots, all run through conduit on the roof:




I set up Powerwall-Dashboard so here's the output from pyPowerwall:


As for 3, I'm wondering that myself. The other panels were grouped in 5 and 7... if they had been 6 and 6, would they have put them in parallel also?

Either they did it to save running another pair, they do it by default, there wasn't enough room, or their AI knows something we don't. I don't know.

Tesla has already proven they can run 4 qty #10 PV wires with #6 bare ground wire in a 3/4" EMT conduit (see group bundle #3).

The problem is the PV wire and the 1" conduit. I suspect there is a junction box where all the 3/4" EMT conduit comes together then transitions down into a 1" EMT conduit into your inverter. Picture please.

You currently have 1 qty #6 bare copper wire and 6 qty #10 PV wire in a 1" EMT conduit. This is perfectly fine. Fill percentage number is green. See picture 6 PV.jpg.

Since you need to add two more wires, Tesla would need to add two more #10 PV wires. 1 qty #6 bare copper wire and 8 qty #10 PV wire in a 1" EMT conduit violates fill percentage per Southwire fill percent calculator. Fill percentage number is red. See picture 8 PV.jpg

One way to fix the 1" conduit fill percentage problem is to switch from #10 PV wire to #10 THHN wire. Doing this makes the fill percentage acceptable in a 1" EMT conduit. See picture THHN.jpg

The deal is that they have to make the PV wire to THHN transition somewhere. My system has a large junction box on my roof right under my array where all PV wire is transitioned to THHN.

I have to assume there is a junction box at splice.jpg. If the installer used a junction box that is large enough where the splice is made (splice.jpg), then they can make the THHN transition there. Pulling new wire with old wire in EMT is not a good idea I am told. So in this case, I would pull out the old wire from the junction box to the inverter and pull a pull rope too. Then install new THHN #10 wire (4 qty) and one ground wire at the same time.

I'm not an installer nor an EE, so please check my work. It all depends on where the junction box is located.

[oregon_phil]

I didn't observe that the "splice" on the roof is a PV Y adapter per your schematic.

They obviously made some MC4 connections onto PV wire in the field (non factory terminations), so the question becomes how did they make the transition from wire into the end of EMT?

Maybe they put the PV Y adapter in a junction box? More questions....

[Ampster]

They asked me to fix them and send proof before they would schedule the install. So in my case no, they didn't do the tiles, they asked me to fix them.

I thought NEM 3.0 was eminent in June 2021 and I chose to self install so I could do main panel changeout, new subpanels and solar install simultaneously. It was costing me $10 per day for electricity so timing was important for a couple of reasons.