Incorrectly Placed Solar Roof Tiles

[trautmane2]

A friend recently took some drone shots of my new solar roof that highlighted discrepancies between actual PV tile locations and the locations specified in my plan documents. It can be difficult to see where the PV tiles are unless you get the right camera angle and (sun) lighting. It looks like 13 out of 100 PV tiles on my south facing roof, 6 out of 42 PV tiles on my west facing garage roof, and 4 out of 28 PV tiles on my east facing garage roof are incorrectly placed.

I'm concerned that the misplaced tiles might reduce my roof's production and the obsessive/detailed side of my personality thinks this should be fixed. I'm curious what others think - is this worth worrying about? I also wondered whether any others with solar roofs have verified their PV locations.

I'm trying to decide what to do next. One idea was to ask the Tesla designer to rerun his plan with my actual layout to show me if there is any difference in expected production. I'm not sure how detailed their modeling is. In general, it seems to me that placing PV tiles higher up on the roof and further from walls or adjoining roofs would be best.

Here are photos of the roof areas with the misplaced tiles marked with red dots and their plan locations marked with white dots.

 

[wjgjr]

I think your proposed next step is reasonable, along with asking Tesla both for an explanation for the change and for updated documents. Looking at the differences, I would think that it could reduce production as I would expect higher-placed tiles to be less likely to be blocked by other things in the neighborhood, and, depending on the angles, some of the misplaced tiles are possibly going to get more shade from your own home. Of course, the difference might be small, and there might be reasons for the change, but it seems unprofessional for them to do so without at least letting you know.

We did not attempt to verify our roof layout, though our roof really was a very simple gable roof with PV tiles in the middle and blanks around the edges. Having watched the install, I don't think it would be possible for more than a few of ours to be off in placement.

 

[aesculus]

I have not taken my drone up yet to see if I can detect passive and active tiles. During the walkthrough I asked why there was more tiles on one side of the roof than the other and they responded that it was really about how it all connected in groups etc. and that it would not exactly match the plan tile for tile.

Looking at what your changes are I doubt you will see that much of difference. In fact to correct those discrepancies I suspect you would do a lot more damage than any gain you would receive. It does look like you have the required number of active panels in an area and that it is reasonably in the right location to me.

 

[jboy210]

When they were installing ours they told us the proposed plan was just that, a plan. The conditions on-site determine the actual placement. I believe there was even something in our contract that spelled that out.

And this is just like every other remodel I have ever had done. Nothing goes exactly as planned and adjustments are made in the midst of the job due to conditions such as structural elements, previous construction, codes, etc. I have seen them get into the structure and then have to move plumbing runs from one side of the house to other, materials changed (ex. PEX versus copper), rewire circuits, etc.

 

[woferry]

Mine definitely has some placement deltas, it was obvious that their initial drawings didn't have the vents perfectly placed and the like, so in some cases they just couldn't have put the tiles where the drawings originally suggested. Fewer vertical rows should result in fewer diode blocks I'd think (more tiles in parallel, fewer in series), not sure if that ends up being net positive or not, though, the diodes still seem like a bit of black magic to me.

Asking certainly can't hurt, but if they got the right number of producing tiles on each mounting plane I suspect that's the most important part, and they may not be willing to just move tiles around since it would have to also affect the wiring underneath.

 

[trautmane2]

Thanks for the input everyone - it helps me keep things in perspective .

In fact to correct those discrepancies I suspect you would do a lot more damage than any gain you would receive.

I'm curious about this assumption. My thinking was that physical shuffling of the full size tiles would not be that difficult or potentially damaging. The piece I thought might be a problem is the (re)wiring since I'm pretty sure the number of tiles per string is important. Is that what you meant by "do a lot more damage"?

 

[jboy210]

Thanks for the input everyone - it helps me keep things in perspective .



I'm curious about this assumption. My thinking was that physical shuffling of the full size tiles would not be that difficult or potentially damaging. The piece I thought might be a problem is the (re)wiring since I'm pretty sure the number of tiles per string is important. Is that what you meant by "do a lot more damage"?

Removing a tile destroys them. They shatter.

In our install, they thought they had an issue with a connection so pulled up some tiles to check. The clips holding down the tile lock-in really tightly to ensure the tile is not lifted by hurricane-force winds required for the roof to meet its wind rating. So you really have to use a lot of force to remove a tile. Here is what the tiles they removed looked like. You do not want to go there unless you plan on redoing sections of the roof.

 

[Ampster]

I'm curious about this assumption. My thinking was that physical shuffling of the full size tiles would not be that difficult or potentially damaging.

As explained above, one needs to consider the method of attachment which is underneath the tile above it. If all the tiles above it have to be removed then it is time consuming and therefore possibly expensive.

 

[trautmane2]

Removing a tile destroys them. They shatter.

I've also seen the tempered glass get shattered on tiles, but thought it should be possible to remove them without damage. I just came across this discussion with Eric Weddle where he talks about swapping an inactive tile with a PV tile (forward to 9:22):

 

[jboy210]

I've also seen the tempered glass get shattered on tiles, but thought it should be possible to remove them without damage. I just came across this discussion with Eric Weddle where he talks about swapping an inactive tile with a PV tile (forward to 9:22):

I saw Matt's YouTube video. When they checked that row on our house they pulled up the tiles. Some broke and some did not. It might have been the technique they used, but the crew was supervised by Tesla employees. And the is definitely a likelihood of having to replace some tiles. Here is a wider shot showing the stack of the ones they broke.

 

[wjgjr]

I saw Matt's YouTube video. When they checked that row on our house they pulled up the tiles. Some broke and some did not. It might have been the technique they used, but the crew was supervised by Tesla employees. And the is definitely a likelihood of having to replace some tiles. Here is a wider shot showing the stack of the ones they broke.

View attachment 623525

We had a couple shatter during install (one non-PV was dropped in our yard and they had to essentially rake up the pieces) and also a couple I saw damaged during shipping that they set aside (seemed like a couple of corners got damaged.) So, I don't doubt that is a risk, but at the same time, I assume that removing/replacing tiles is also the solution for warranty issues, and it would be concerning for Tesla if one bad panel, or one that does somehow get damaged, causes that many others to get broken during a repair.

 

[sorka]

When they were installing ours they told us the proposed plan was just that, a plan. The conditions on-site determine the actual placement. I believe there was even something in our contract that spelled that out..

The actual placement will result in a significant reduction in production. You need to ask them why they didn't follow the plan. I don't see any thing that would have made it harder to install those tiles higher up.

 

[trautmane2]

So, I got an answer today from Tesla.

Earlier this week, I called customer care (725-867-6182) which I've learned is what I'm supposed to do now that my system is fully installed and active - no more bothering advisors, installers, or salespeople . I talked with a very helpful rep and sent her a detailed email with the pictures I included in my original post to this thread. She promised to forward my concerns along to the appropriate people and to get back to me with an answer. I received the answer via email today. It stated:

Relocation of PV tiles to different portions of the roof may cause slight changes to PV production, however, the changes in this case are negligible since the PV tiles are still facing the same pitch and azimuth.​

Our PV production estimates are just that, estimates, that are mainly based on these factors:

• Pitch and Azimuth
• Sun hours and temp coefficients for efficiency (longitude and latitude location)
• Shade

None of the above factors will be changed by the adjustment made at installation except maybe the first photo where PV tiles were located in a better sun access location at the top of the MP.​

If we were to run this back through our production estimates, we would get the same numbers though because the above factors have not changed significantly enough.

I hope the changes do truly result in negligible production differences. I'm eager to see how my production numbers look over time, which brings me to a related tidbit ...

Before bothering anyone about this, I thought I'd compare my actual production numbers to estimates to see if things were ok. I used PVWatts to determine what my current production numbers should be, calculating and then summing up estimates for my 3 different roof planes. My PVWatts estimated average daily production for December was 23.5 kWh per day. I likely made a mistake or two along the way calculating this, because my actual average daily production for the first 18 days was only 8.7 kWh.

Tesla's customer care rep explained that they use the following estimation rubric:
For winter (Nov-Jan): 1 x system size
For spring and fall (Feb-Apr, Aug-Oct): 3 x system size
For summer (May-Jul): 5 x system size

So for my 9.94 kW system, Tesla expects me to average at least 9.94 kWh per day in November, December, and January.

PVWatts seemed to estimate much less variability between December and July (23.5 to 41.4) than Tesla (9.94 to 49.7).

I'm not sure where I went wrong with PVWatts, though it seems suspicious that I was summing up 3 planes and wound up with an estimate that was almost 3 times too high.

 

[h2ofun]

So, I got an answer today from Tesla.

Earlier this week, I called customer care (725-867-6182) which I've learned is what I'm supposed to do now that my system is fully installed and active - no more bothering advisors, installers, or salespeople . I talked with a very helpful rep and sent her a detailed email with the pictures I included in my original post to this thread. She promised to forward my concerns along to the appropriate people and to get back to me with an answer. I received the answer via email today. It stated:

Relocation of PV tiles to different portions of the roof may cause slight changes to PV production, however, the changes in this case are negligible since the PV tiles are still facing the same pitch and azimuth.​

Our PV production estimates are just that, estimates, that are mainly based on these factors:

• Pitch and Azimuth
• Sun hours and temp coefficients for efficiency (longitude and latitude location)
• Shade

None of the above factors will be changed by the adjustment made at installation except maybe the first photo where PV tiles were located in a better sun access location at the top of the MP.​

If we were to run this back through our production estimates, we would get the same numbers though because the above factors have not changed significantly enough.

I hope the changes do truly result in negligible production differences. I'm eager to see how my production numbers look over time, which brings me to a related tidbit ...

Before bothering anyone about this, I thought I'd compare my actual production numbers to estimates to see if things were ok. I used PVWatts to determine what my current production numbers should be, calculating and then summing up estimates for my 3 different roof planes. My PVWatts estimated average daily production for December was 23.5 kWh per day. I likely made a mistake or two along the way calculating this, because my actual average daily production for the first 18 days was only 8.7 kWh.

Tesla's customer care rep explained that they use the following estimation rubric:
For winter (Nov-Jan): 1 x system size
For spring and fall (Feb-Apr, Aug-Oct): 3 x system size
For summer (May-Jul): 5 x system size

So for my 9.94 kW system, Tesla expects me to average at least 9.94 kWh per day in November, December, and January.

PVWatts seemed to estimate much less variability between December and July (23.5 to 41.4) than Tesla (9.94 to 49.7).

I'm not sure where I went wrong with PVWatts, though it seems suspicious that I was summing up 3 planes and wound up with an estimate that was almost 3 times too high.

For my system, with normal rain, I will be happy to get 1X system size average for the month, and most of my panels face south.. In summer yep, was getting about 5X. SO, for what I have seen, the tesla estimates seem real good.

 

[wjgjr]

Before bothering anyone about this, I thought I'd compare my actual production numbers to estimates to see if things were ok. I used PVWatts to determine what my current production numbers should be, calculating and then summing up estimates for my 3 different roof planes. My PVWatts estimated average daily production for December was 23.5 kWh per day. I likely made a mistake or two along the way calculating this, because my actual average daily production for the first 18 days was only 8.7 kWh.

Tesla's customer care rep explained that they use the following estimation rubric:
For winter (Nov-Jan): 1 x system size
For spring and fall (Feb-Apr, Aug-Oct): 3 x system size
For summer (May-Jul): 5 x system size

So for my 9.94 kW system, Tesla expects me to average at least 9.94 kWh per day in November, December, and January.

PVWatts seemed to estimate much less variability between December and July (23.5 to 41.4) than Tesla (9.94 to 49.7).

I'm not sure where I went wrong with PVWatts, though it seems suspicious that I was summing up 3 planes and wound up with an estimate that was almost 3 times too high.

I do not know the distribution of your shingles among the three planes, but it sounds like the actual production numbers you provided are lower than I might expect. As we are not that far from you, we likely had fairly similar weather, and our average production with an 8.2 kW solar roof, split 50/50 on a north and south face, was 12.0 kWh per day for December (PVWatts numbers predicted 13.9 kWh, and they have generally been a bit higher than actual - likely due to me not entering a factor correctly.) At this time of year, ~85% is produced from our south-facing array, or about 10.2 kWh.

So, all of this is to say that the 23.5 kWh number from PVWatts may be reasonable, with an actual that might be closer to 20 kWh. This does highly depend on the tile direction (I was surprised just how significant it is at this time of year,) but I would be surprised if your tiles have a worse average orientation than ours, and it is a larger system, so I would expect a higher number.

Do you have more than one inverter? Ours has only one, but others with more than one sometimes have issues with reporting. Even with one, possibly there is some sort of an issue with an individual string. One thing that you can do is check data from your utility to see if it matches what Tesla is reporting.

 

[trautmane2]

So, all of this is to say that the 23.5 kWh number from PVWatts may be reasonable, with an actual that might be closer to 20 kWh.

Thanks for the comparative data. It would be great to have some kind of daily production registry with the key attributes split out (geo location, PV size, pitch, azimuth) so that folks could easily compare production - maybe even get alerts when your production "significantly" differs from your own historical data or from others with similar setups. I wonder what Tesla has in terms of internal monitoring and whether they monitor across similar systems at all.

Do you have more than one inverter?

No - I just have one Delta M8 inverter (8kW max) - probably the same as you. It seems like the data from it is correct, but I haven't had the system long enough to be able to check it against a utility bill.

 

[wjgjr]

No - I just have one Delta M8 inverter (8kW max) - probably the same as you. It seems like the data from it is correct, but I haven't had the system long enough to be able to check it against a utility bill.

I have the 7.6 kW Delta Solivia that they were installing at the time. Does the M8 have a display so you can see the daily numbers to compare to the app? Also, for the utility, many utilities provide something like "green dot" data that has something like hourly numbers, and is usually available almost in real time. Failing that, you can just look at the meter.

I think once you get everything working and numbers properly calibrated, it will actually be pretty easy to notice when production is off (though it would be great if it was automated.) As a side-note, January has been terrible for me so far, with an average of only 5.6 kWh daily, due to all the cloudy days so far.

 

[trautmane2]

Does the M8 have a display so you can see the daily numbers to compare to the app?

No, it only has backlit indicators on the front panel. However, I can see realtime string inputs and current and prior day inverter outputs using the M Professional app. The inverter output numbers line up with the Tesla app numbers, so I think the numbers are good.

many utilities provide something like "green dot" data that has something like hourly numbers, and is usually available almost in real time

Yeah - it appears that Dominion Energy provides this kind of UI for its smart meter customers. My meter got replaced as part of the roof install process. I think my new meter is a smart meter, but I'm not sure. I haven't been able to find the UI yet. I need to follow up with Dominion to see if I should have access to realtime data and if so, where I go to view the data.

 

[trautmane2]

I just talked with Dominion Energy. The customer rep told me that because I have net metering, I only have access to monthly totals .

 

[wjgjr]

I just talked with Dominion Energy. The customer rep told me that because I have net metering, I only have access to monthly totals .

That sucks - hopefully the rep is off and something will eventually show up, because that is useful to have, both to sanity-check the numbers from Tesla and to see your usage patterns. Our data is certainly a bit weird, because we only get hourly net usage (since our meter does not record production and use separately) but it was a good way to confirm that the Tesla numbers mostly matched the utility numbers.

As far as production, if the inverter numbers match the Tesla figures, it is possible that your numbers really are that low, if the orientation and/or shading is really that bad for you this time of year. Still surprises me, though, given the production numbers on our install.