New Solar Install

[nwdiver]

But then if you look at this link for May 19th and click on Insolation, in between the clouds the system peaks quite a bit above the projection, ~1500W (~20%) over at times. I guess this is some combination of the tiles and/or inverter being cooler and able to output more during these periods?

It's 'cloud edge effect'; You're getting direct light from the sun AND indirect light reflected from clouds so the energy the solar panels are receiving is briefly higher than the ~1kW/m² that you would get at noon on a day with no clouds.

 

[woferry]

Nifty, never heard of that. Thanks!

 

[NuShrike]

I find the insolation data on PVOutput rather interesting, also. I have the system configured as 444x 30W panels instead of the proper 561x 23.75W, they actually didn't allow a panel spec below 50W (or non-integer values), after I emailed them they lowered the limit to 30W (and still integer-only), but that's still not correct obviously. But 444 * 30 gets 13.32kW which is basically the same. Pretty much all of the roof surfaces are 18° pitch so I entered that. Since the surfaces point 4 different directions there wasn't really anything I could put in the orientation field, so I set that to East West, and I left the azimuth at 0. Panel coefficient is blank and shading is Low with no time specified.

With this setup, on a clear day the data is pretty darn close. This link shows May 12th, if you click on Insolation below the graph (there's no direct-link to this, unfortunately) shows how close the projection is to actual. It's a bit off at the edges, probably because of the panels on the less-ideal surfaces for the current low sun position. But then if you look at this link for May 19th and click on Insolation, in between the clouds the system peaks quite a bit above the projection, ~1500W (~20%) over at times. I guess this is some combination of the tiles and/or inverter being cooler and able to output more during these periods? I haven't seen any good cases so far where things cleared-up and stayed clear, to see when it eventually comes back in-line with the insolation projection. Though looking at Insolation on the first day I flipped the switch (May 5th), it basically followed the curve from the moment it was switched-on, so I guess it's more the tiles than the inverters (as the roof would have already been in the sun and warm in this case, but the inverters had been sitting off until then, and there was no overshoot in that case).

Also took screenshots since the direct-links aren't possible, in the order I referenced them above.

View attachment 410250 View attachment 410251 View attachment 410252

You can enter some sub-arrays (up to 2) and have them point in different azimuths. Max is 3 arrays total. I have 4 arrays so it's not perfect for me, but close enough.

Cloud-edges peak production due to cooler temperature and refraction focusing.

There's a TMC team: Tesla Owners and Friends 377.818kW

It's randomly cloudy in SoCal today, so I've been bouncing off my clipping ceiling for hours due to the cloud-effect: Koko wa Greenwood 10.080kW | Live Output

 

[woferry]

Hmmm. I think PG&E screwed-up. Got my first bill since PTO, and it says absolutely nothing about net metering, and still lists my rate plan as "E1 X Residential Service". But it also shows "CLOSED" after my Service Agreement number. The 4/13-5/13 bill shows what I used up to switching-on the solar system 5/4 (noting that PTO was received a week earlier, 4/26), and no consumption since then, but no production either. Similarly the meter reading reflects my consumption through 5/4, but completely ignores the production which was enough to put the meter to a lower number than I'd started the billing cycle with by 5/13.

So maybe the "closed" message is indicating that this is my final non-NEM bill and that I'll be getting the NEM one later (I'd still really like to know which rate plan I'm actually on!), but the website gives no indication of that at all. The fact that the NEM packet PG&E said would be sent in the PTO email still hasn't arrived (1 month later) also leaves me wondering if they lost track of me or something. I guess I'll have to figure out who to call/email to see what's going on here.

Yesterday was the best day so far, 71.9kWh, BTW. Looks like I hit my first MWh of solar production yesterday, as well.

 

[miimura]

Don't worry about it. They're going to go back and recalculate from the PTO date and issue adjustments to compensate for this erroneous bill. Just wait until you see the new bill. The NEMMT billing they put Powerwall users on is nearly incomprehensible, especially in my case since I'm on a CCA and asked for the "long form" bill.

 

[woferry]

Well, every time I think I've learned some pattern I only collect more data that disproves my theories. My best days were running around 71-72kWh generated (6/4 as an example at 71.15kWh on PVOutput), and suddenly two days in a row (6/6 & 6/7) were >75kWh. Those two days were significantly cooler (highs in the mid-70's instead of mid-90's), so I thought maybe I'd found the impact of cell/ambient temperature. And then the following day (6/8) the temperature returned to the mid-90's, and it was the best day yet, almost 75.5kWh. So not sure what it was, clearly not directly temperature-related, all 3 days and the good ones just before I'd say were all very clear, maybe it was more of a haze that cleared-out when the cold front came through, or I just don't know. The past few days have been even hotter, but also more cloudy so no further 'clean' data like those few days.

I did get confirmation that my roof is indeed V2, didn't have any luck finding out exactly what's different in V3 so far, though at a high-level it sounded like it's both changes to help simplify manufacturing (fewer manual steps) as well as to improve installation time. So those both sound like good things, obviously.

Was also amused to learn that within a few days of my roof being finished Google updated the street view for my street. So it already has the new roof, guess I'm lucky they didn't catch it mid-construction. I'd gotten similarly lucky when I bought the house, the satellite shot for my neighborhood was literally taken the day I had closed (I know because it was the only day that both my truck and the previous owner's PODS was in the driveway, and both were clearly visible in the Google Maps shot for a while after I'd moved in), I say it's lucky because had the picture been taken a day earlier my house would have been the red & white 'big top' on the street, as it had been tented for fumigation. Now they just need to update the overhead shot again to get rid of the old roof.

I did finally get my photo album online, which includes 3 time lapses at the very end, one of the tear-off from the back yard (I put the camera in the same spot all 3 days, though only day one was actually working on the back surface), one from the back for the first 4 days of the install (until they were almost done with the back, a.k.a. "MP1"), then I figured out a place I could put my camera to catch the front surfaces (MP2, MP3 and MP5) for the remaining 6 days of work. For consistency I went with a 720x speedup for all 3 videos. I also have like 30 minutes of drone footage flying over the roof trying to get different angles, etc., not sure how much of that is worth posting, it more shows my poor skills flying a drone (couldn't keep it going in a straight line, etc. ) It was certainly good for some stills, such as the one I've used for my avatar photo.

I also finally got confirmation from PG&E that I'm on the E-TOU-A plan now, though I haven't seen any NEM bills yet. I kinda expected the billing date would shift to line-up with the PTO date but I guess not, yesterday would be the close of the next period on my old billing schedule, so I guess in another ~2 weeks I'll get my next bill, which hopefully will start showing me what the NEM stuff is really like. Strangely my account shows a current balance of -$41, so I guess they've reversed some of what I got charged on my last non-NEM bill, but there's nothing showing the breakdown of that yet.

 

[woferry]

Holy smokes. I got my first NEM statement from PG&E, and I had absolutely no idea how confusing they could be! Even the regular bill has been bizarre, since mid-May I got charged through 5/12 as if everything was normal (PTO was on 4/26, mind you), then over the past week or two the website kept changing without a new bill, first showing PG&E's energy charges from that bill being reversed (turns out that was the $41 credit I'd mentioned above), then the SJCE (CCA) charges also got reversed, so I wound up with a $62 credit (making my May bill <$6). Last night the second regular bill posted, which basically reversed all of the previous bill's electric charges, then re-applied the charges from 4/12-4/26. It really looks like some things got double-charged in this new bill, so I'll have to review the numbers very closely.



Yesterday in the mail I got two NEM statements, for 5/12 and 6/11. These things are very poorly formatted, it's like "we used to send you nice rendered webpages, now we're going to send you the raw HTML source instead!" And I've been getting all of my PG&E contact online for years, I guess the NEM statements (14 pages each) have to be snail mailed?? Never got the "welcome kit" they promised in my PTO email either, so I guess I have a few things to call them about.

Curious, if anyone else is listening who's solar+PW provides 100% of their daily power needs (at least during the summer months), is your Gateway well-enough calibrated to actually get 0 draw from the grid? Mine averages around 20W grid consumption whenever the solar isn't producing, which the Gateway basically acknowledges (bounces between a slight negative to the grid and a bit more positive from it, averaging to ~20W), and is shown by my meter and the Rainforest Eagle that logs its data. So on my 5/12-6/11 NEM statement, it shows 6kWh consumed. I obviously would have preferred for this to actually be 0, but I guess this is as good as it gets?

My solar reading on the Gateway has a similar ~20W error, when the sun is down it still thinks the solar is producing around 20W continuously, which leads to ~0.2kWh/day error in the solar production (I remove this error before I post to PVOutput, so what I put there is a bit less than what the Tesla app says). It's odd because it doesn't just seem to be an offset in the CT, when the inverters come online and go offline I'll get readings that are almost 0, in fact these are the events I look for in the PVOutput filter, all of the 10-25W readings before the first <10W reading of the day are discarded, as are all of the 10-25W readings after the last <10W reading of the day, and this has reliably worked every day since I flipped the switches. But right now, at 6am with the inverters still in Night Mode from yesterday, the Tesla app shows I've gotten 3% of my power from solar since midnight. Amazing!

I did extract a portion of my (not sped-up other than the original time lapse 1 frame/3s) installation video to post on Twitter (someone who asked about replacing a broken tile), which showed a PV Module being swapped-out in less than 3 minutes. Here's the link. In hindsight I probably should have reduced the framerate of the video to slow it down, might still do that and post a higher-res version to my album.

So I should pass the 3MWh production mark by the end of this weekend, which will be 50 days from when the switches were flipped.

 

[JohnRatsey]

Curious, if anyone else is listening who's solar+PW provides 100% of their daily power needs (at least during the summer months), is your Gateway well-enough calibrated to actually get 0 draw from the grid? Mine averages around 20W grid consumption whenever the solar isn't producing, which the Gateway basically acknowledges (bounces between a slight negative to the grid and a bit more positive from it, averaging to ~20W), and is shown by my meter and the Rainforest Eagle that logs its data.

I see a minimum of about 0.5kWh/day mains power used. Some days it is more, probably due to demand spikes more than PW + solar can accommodate.

I was hoping to see the mains power usage nearer to zero but have come to accept that it's not so easy to manage the PW perfectly. Probably the biggest problem is when it's fully charged and on standby and suddenly needs to start discharging when there's an increase in demand more than the solar can handle.

 

[miimura]

@woferry you can get digital copies of the "black and white" bills in your PG&E account, but it's pretty well hidden. Look here:
In the second heading strip on the PGE.com web site after you log in, it says "Hi <customer name>" followed by "Account" and a pull down dialog box. Choose "My Accounts and Services" after you click the pull-down. On that page, click through the links in the Service ID column in the second table on that page until you find the one that has an item "Download Detail of Bill" with a pull down box with your bills listed. For some reason I have 3 service IDs, one for gas service and two for the electrical service. This web site feature was not working last year when my bills were initially issued. I think it started working late last summer and now contains all my "NEM-Paired Storage Account" bills.

Anyway, after you verify that you can download the same bill that they send by snail mail, you can call the Solar customer service center at 1-877-743-4112 and ask them to stop the paper mailing.

 

[woferry]

Thanks, @miimura. That takes some digging to find, but the statements are there. Would be nice if they just put them on the dashboard. I have 5 service IDs shown there for my current address, 2 are stopped and 3 are active.

 

[sputnik]

But no, they ask that you never walk on it, and don't let anyone else walk on it either, they want to be present (or do the work themselves I guess) if anyone has to go on the roof.

@woferry - I'm in the process of looking over a solar roof proposal and making my decision. What they've told me (including in writing over email) is that I can walk on the roof if I choose to, as in, it does not void warranty. But they don't advise it because it can be slippery. Is this also what you heard?

I asked because as a homeowner it's kind of a pain to have to tiptoe around your own roof - if I wanted to get some debris out, or inspect the vents and exhausts or put up holiday lights.

Thank you for documenting your journey in so much detail by the way, it has been very helpful in my own process.

 

[Dan123]

@woferry - I'm in the process of looking over a solar roof proposal and making my decision. What they've told me (including in writing over email) is that I can walk on the roof if I choose to, as in, it does not void warranty. But they don't advise it because it can be slippery. Is this also what you heard?

I asked because as a homeowner it's kind of a pain to have to tiptoe around your own roof - if I wanted to get some debris out, or inspect the vents and exhausts or put up holiday lights.

Thank you for documenting your journey in so much detail by the way, it has been very helpful in my own process.

I think it depends on your panels. You can probably walk on them, but there is a risk that it will crack. Why take that risk?

 

[sputnik]

I think it depends on your panels. You can probably walk on them, but there is a risk that it will crack. Why take that risk?

So note that the tempered tiles are supposed to be so hard that they are supposed to withstand massive hailstorm. (See the videos on their website).

There is some non-tempered material used around the edges as @woferry mentioned - and that doesn't have the level of hardness that tempered glass has.

But that's the thing. The Tesla solar roof rep said that they don't expect even the weaker tiles to crack by walking, and hence it's not a warranty issue. But instead I could slip and hurt myself. I understand that risk, it always exists when walking on any roof, and I can exercise caution around it.

It's just that this is different than what @woferry narrated, so I was double-checking.

 

[woferry]

Maybe they're talking to you about v3 and they've made some changes in that regard, I don't know (seemed like Musk was implying v3 was more durable, so maybe that's it). They definitely told me to be very careful around the cut tiles (really they said they didn't want anyone on the roof period without them to at least be there or preferably do the work themselves, but they also specifically called-out avoiding putting weight on the cut tiles), and it's clear that they were avoiding them while installing the roof as well. I don't recall them ever saying anything about warranty or voiding it by getting on the roof, but they were clearly concerned about both slipping and damaging them.

Fortunately it's really not a concern for me, I don't have multiple levels of roof so it's not like I have to stand on part of it to access anything (and I can get to my new gutters with a simple A-frame), I've been on my roof less than a dozen times in the 15.5 years I've lived here, and probably all but the first time (which was more first-home-buyer curiosity) was inspecting damage to my old asphalt shingles after heavy storms (see how many more had blown-off, make sure there weren't any exposed areas), so I'm hoping those days are behind me. The chimney is the one thing that it would be more convenient to work on from the roof rather than the ground, so that's why (even though I've never used my fireplace and probably never will) I had it inspected/repaired before the roof was replaced.

 

[woferry]

Well, had a bit of excitement (not really, just unusual activity) the past week, didn't want to post about it until I knew better what was going on. Friday a week ago, I get this automated text that starts with just "Tesla: ", and says they'd scheduled a service appointment for Tuesday of this week (no reason indicated), and provided a 650- number (not the one the text came from) to call for questions or to re-schedule. Since I hadn't complained about anything I wasn't sure at first that this wasn't some sort of spam/phishing thing. After a bit of googling to try to determine if either number had fraud/robocalls associated with them I decided to give the callback number a try (the suggested time didn't work for me, anyway). The phone answered right away, and an automated voice simply said that the current wait time was over an hour and started playing music (didn't say welcome to Tesla service or anything like that). So I hung up because I didn't have an hour to spend on the phone Friday, and sent an email to my installer lead and energy advisor asking if this was legit. The advisor confirmed it was legit, though he didn't volunteer anything more about the reason for the appointment (didn't know, I guess).

So Monday I found some time where I could sit on the phone on in my office while still getting work done and called again, got the same over an hour message, but it "only" took about 30 minutes before someone connected. They rescheduled me to Friday afternoon (today), and said that they had detected a "performance issue" so that triggered the service call. I went back and took a closer look at the data I've collected and didn't see anything particularly off (I look at the app and my PVOutput data every day and extract the inverter's Production History screens every week, so I'm sure I'd have caught anything gross), but the inverter input that was affected by the "wiring issue" I referred to in post #69 was perhaps reporting some more wildly-variable output than the rest. I guess I'd never expanded upon that issue at the time, the boxes they used to bring the PV wiring from above the roof deck to inside the house are pre-wired on the outside, but were just bare leads on the inside (I assume to keep the roof hole size down). Well, when they added the MC4 connectors on one of the 6 string's the connectors were done backwards, i.e. the + wire used the MC4 connector the other 5 sets used on their - wire, and vice-versa. And the PV wiring had already been connected to the RSDs in the attic, so this one RSD was seeing around -300V, and according to its datasheet it's not OK with any negative voltage. I pointed this out to the technician at the time, and he corrected the plug genders, but he felt the RSD was probably OK, he offered to replace it if I really wanted to, but clearly didn't sound thrilled about spending more time in the attic. I figured let's just go with it, and assumed that if it was damaged it would be obvious looking at the inverter data (given that at the time we hadn't turned anything on yet, in fact I don't think the wiring from the RSDs to the inverters was finished).

Anyway, as soon as the technicians (not ones I'd dealt with before) showed up today and confirmed it was a production issue I mentioned the original mis-wiring, and they felt that was likely the issue. So they went in the attic and that particular RSD was ticking. That string is S4 from my picture on post #73, and it gets combined with S6 (post-RSD, S6 has an RSD in the garage and S4's RSD is in the attic). The technicians were a bit confused by the wiring, so they also inspected the RSDs in the garage, and noticed that S6's RSD was also ticking. So they wound up replacing both of them. The one technician was also concerned about how hot the RSDs were upon opening the service boxes they're installed in, and wound up calling first Tesla's hotline then Delta (RSD manufacturer) tech support, to see what the acceptable operating temperature really was. He said in traditional panel installs when they're used they're left unboxed mounted under the panels, so they get lots of air flow around them, but in this case they're each in their own box in a hot attic, so they're already in a hot space and might be worse-off getting next to no air flow. Well, Delta said they were aware of how they were being used with Solar Roofs, and that they're comfortable with the application (they're spec'ed to operate in ambient temperatures up to 70C [158F]), but the guy on the phone apparently said that he'd feel better if the RSDs had date codes beyond a certain number, when repeating the information to me the Tesla tech must have botched the number because the week code he mentioned (1932) would be ~2 weeks from today (we're at the end of week 29 in 2019). I didn't have the current week number handy until I got back to my computer so I didn't question it while he was around. But he had determined that none of my RSDs met the date code requirement (the two that were replaced were 1830 and 1849, so if he meant 1832 instead of 1932 one of them might have been OK, or maybe he meant to say 1923 or 1922 or something). Anyway, apparently Delta was going to send the Tesla tech 5 RSDs beyond the preferred date code, as I guess they've either expanded the temperature range even farther, or made some other change to improve high-temperature operation, perhaps without modifying the overall spec.

So I currently have 2 replacement (but apparently still not new enough, I didn't see their date codes) RSDs, and it sounds like I should expect another appointment to replace all 5 sometime in the future. The solar was out from 3-6, and the sun was far-enough down by the time the inverters were online again that it would be difficult to see any difference right now. Will have to see if over the next few days the behavior of that one inverter channel stabilizes or not. Personally I'm hoping it was just 'walking wounded' behavior from the initial reversed wiring, not that I'd object to getting newer RSDs if Delta thinks they're better.

And since I just threw the initialism around a lot above, if anyone was wondering, the "RSD" is a Rapid Shutdown Device, which is basically just a relay but it works based on a signal received by the inverter, each one is about the size of a college textbook. The point is that when the inverter stops sending the signal the RSD opens the circuit, blocking the array's voltage/current from going to the inverter. As I understand it the point is to make the PV wiring safe, presumably in case firefighters end up cutting through the conduit or the like. So they're required to be placed within a certain distance of where the PV wiring enters the house to limit the amount of wiring left at higher voltages (and able to produce lots of watts if accidentally shorted together) once everything is switched-off. One thing I learned today, I had assumed that turning off the DC disconnect switch on the inverters would cause the RSDs to shut down, but the technicians claimed that the signal makes it to the RSD regardless of the position of this switch, and that the inverters have to have their AC disconnects (i.e. breakers) thrown to stop sending the signal. This still seems a bit suspect to me, as I don't know what the DC disconnect switch in the inverter is really doing if it's not breaking the connection such that the signal would no longer make it through, I guess the inverter might be injecting the "signal" on the RSD-side of the DC disconnect switch.

 

[Gaetandp]

how are they performing when it's very warm outside. do you lose a lot of efficiency?

 

[woferry]

how are they performing when it's very warm outside. do you lose a lot of efficiency?

I'll refer back to post #86. 6/8 has still been my best production day. I haven't noticed any real change since the service appointment a week and a half ago, so either I led them in the wrong direction by mentioning the once-reverse-wired RSD (though it was clicking according to them, which isn't expected behavior), or it just didn't make a big difference. But those 3 days in early June have been the only days in the 75.x kWh range, there have been 3 days in late June that topped-out at 73.x kWh, and July 8th is the last time production exceeded 70kWh. Unfortunately I haven't succeeded in getting PVOutput to show temperature data, even though I've uploaded it along with all of my production/consumption data. I still have all of the data captured on my side, but would need to do some massaging to get them easily side-by-side. I haven't seen any direct correlation between hotter days and lower production vs. cooler days, some of the hottest days recently have not been clear days (though some have been relatively cloud-free), I suspect things may be a bit hazier now than they were back in June also (unfortunately my weather station doesn't have anything like an irradiance meter, I should look for one of those).

 

[Gaetandp]

ok thx. sorry missed that one post