is this low production?

[ajdadja]

The sun has set on my first day of production.

I figure that today was probably a better-than-average day. No clouds, slightly closer to the summer solstice than the winter solstice. When I look at the solar chart in the app, it looks like a reasonable bell curve.

I have 39 T400H panels, 2 powerwall+ and 1 powerwall2. They seem to have wired 2 strings on each powerwall+.

The tesla app shows 53.8 kWh generated. If I calculate 53.8 kWh * 365 days, I get 19,637 kWh per year. But my original system design claimed 21,887 kWh per year.

Am I missing something?

Thanks!

 

[jjrandorin]

Production is measured against previous months, you can not extrapolate anything from a single day (or a single month). You can go model your estimated production at a site like pvwatts but one day of production, is not going to give you any real data (nor will your first week of production).

To answer "am I missing something?" the short answer to that is "yes, you absolutely, positively, 10000% can not model a days production as what is going to happen each day, because the angle of the sun changes daily (not accounting for weather).

Starting in March, you will see, in general, production scale up higher and higher till you get to the longest day of the year, (which has passed) and then production will grow progressively shorter, daily. We are in September, and you will see production get progressively shorter as the days get shorter, with Dec and January being about 1/2 of June / July's production.

 

[ajdadja]

Seems that my initial post was not very clear. Sorry for that.

The point I was trying to make is that today would have been a better-than-average day. So, if anything, projecting today to an entire year is overly optimistic.

In other words, the 19,637 kWh figure is a conservative upper bound on annual production.

 

[wwu123]

I further what jjrandorin said, and while this type of query is common here, this is the first time I've seen it with a single day of production....

But to give some reasoning to it, 6 of the past 7 days have been "clear" cloudless days with typical smooth curves, yesterday was by far the lowest of those 6, and 10% lower than the best of those days. All "clear" days are not the same, because there are differing levels of haze and particles in the atmosphere that are fairly undetectable to the naked eye, that affect the insolation that actually reaches the ground. Temperature, wind, other factors also have an effect that can vary (production is affected by panel temps).

You're right that Sept 14th would be somewhat above average across the year, if all days were equally "clear" - but perhaps this Sept 14th was a below-average Sept 14th across most years....

 

[taphil]

The point I was trying to make is that today would have been a better-than-average day. So, if anything, projecting today to an entire year is overly optimistic.

No, your math or modeling is wrong. You cannot pick a good day's production and multiply that value to estimate a year's production. Just like you can't pick an hour's production, let's say at high noon in May, and multiply that by 10h/d to estimate daily production.

In May, you might get 70khw per day. If you multipily that by 365, you get 25,550 which would be a gross overestimate.

Solar production follows a curve throughout the day as the sun moves across the sky, and the daily production follows a curve as the earth goes around the sun and the path of the sun changes and daylight hours changes.

 

[Matt-FL]

Mine varies greatly. My largest producing months are actually spring, where I'm getting 130kWh+ on clear days. Peak summer, when I thought I'd be making a lot more, I was actually only making just over 100kWh a day. It is now September, and we've had lots of clouds, so I'm averaging about 80-90kWh. I think you will find over time it should average out to what they expected, or higher. I was estimated at 24MWh per year, but I produced 29.8MWh the first year (which I just hit Aug 30, 2023).

 

[GregBallantyne]

Seems that my initial post was not very clear. Sorry for that.

The point I was trying to make is that today would have been a better-than-average day. So, if anything, projecting today to an entire year is overly optimistic.

In other words, the 19,637 kWh figure is a conservative upper bound on annual production.

Today may well have been a better than average day for the time of year approaching the equinox. I'll say that is was better than my day today, and my optimum generation capacity is higher than yours, although the inverter capacity is the same.
This close to the equinox, we are close to the spot where half the year will be better and half the year will be lower - for an exact replication of conditions other than sunrise and sunset hours. Sun angle etc... is all dependent more on your latitude. Number of cloudy days is based on far less predictable processes.
Bottom line, despite any data that makes any day seem quite representative, a calculation on annual yield based on that day will fail. Far too many variables, in a chaotic system. The only value in looking at a figure derived by that basis is to occupy your time while the actual year goes by, on the way to knowing a number that has some meaning. If you're like me, and watch daily production closely, you are on the path toward far greater detailed sun movement and weather observations than you thought you would ever have. It's fascinating, and a lot of fun. And easy to draw erroneous conclusions, especially early on.
I talk like I know so much, but only have been operating my system since June 10 of this year!
BTW, I wouldn't at all call that low production.

 

[ajdadja]

About 5 months later, I think I'm still underproducing.

My project advisor at Tesla broke down my 21,887 kWh expected annual production by month and emailed me the result. They broke it down as:
Jan. 32kwh, Feb. 44kwh, Mar. 57kwh, April 73kwh, May 78kwh, June 85kwh, July 82kwh, Aug 77kwh, Sep. 68kwh, Oct. 49kwh, Nov. 39kwh, Dec. 29kwh.

That adds up about right. 32*31 + 44*28.25 + 57*31 + 73*30 + 78*31 + 85*30 + 82*31 + 77*31 + 68*30 + 49*31 + 39*30 + 29*31 = 21717 kWh, which is within rounding errors of the original 21,887 kWh number. Great! That should let me figure out how I'm doing without waiting a full 12 months.


In September, I was only live for 17 full days. So, I should have generated 68 kWh * 17 days = 1156 kWh. According to the Tesla app, I only generated 69.2% of that (800.4 kWh).
In October, I should have generated 49kWh * 31 days = 1519 kWh. According to the Tesla app, I only generated 73.5% of that (1117.3 kWh).
In November, I should have generated 39kWh * 30 days = 1170 kWh. According to the Tesla app, I only generated 59.5% of that (696.7 kWh).
In December, I should have generated 29kWh * 31 days = 899 kWh.. According to the Tesla app, I only generated 53.1% of that (477.0 kWh).
In January, I should have generated 32kWh * 31 days = 992 kWh.. According to the Tesla app, I only generated 54.0% of that (535.9 kWh).

When I asked Tesla, they said that the system will underproduce before PTO. But that doesn't make sense to me because I have plenty of powerwall capacity. My 3 powerwalls have never gotten to 100%, so they should be sucking up extra power even if I can't export to the grid.

Do you agree that I have low production? Or is there something that I'm still missing?

At the very least, I figure it's worth sharing with the forum that the project advisers have access to estimated monthly production numbers. Could be useful to other folks.

 

[offandonagain]

You production does seem low enough to warrant more investigation.

1. Do your grid imports (as reported by your utility) match the data in Tesla's app? Your solar production could be measured incorrectly.
2. Can you post some solar production graphs on clear days, e.g. your best Sep and Jan days?
3. The Tesla One app will allow you to connect to your equipment and look at production per inverter string. If you do that during solar noon on a clear day, it might pinpoint issues with specific strings or inverters.

For estimated monthly production values, you can also use PVWatts: PVWatts

 

[Grayson73]

I put in my panel size in PVWatts and left everything else at default and it seems to be way off.

January says 1,238 kWh and I only generated 259 kWh.

 

[offandonagain]

I put in my panel size in PVWatts and left everything else at default and it seems to be way off.

I responded on your other PVWatts thread.

 

[ajdadja]

You production does seem low enough to warrant more investigation.

Sorry for sitting on this for so long. Family emergency, but things are getting back to normal.

Great questions, @offandonagain. Thanks for your help!

Do your grid imports (as reported by your utility) match the data in Tesla's app? Your solar production could be measured incorrectly.

1. I just did some Excel work with the Tesla CSV data in order to match it up with the date ranges on my electricity bill. The import numbers (sometimes) match up correctly, but the export numbers seem to be pretty far off.

Can you post some solar production graphs on clear days, e.g. your best Sep and Jan days?

Sure. Here's the best day from May so far:

And here's the best day from January, which was January 29th. Historic data shows that it was partly cloudy in the afternoon that day. I don't think we had any clear days whatsoever in January.

You asked for a September graph, but since I sat on this for 3.5 months, I'm guessing you prefer the May graph. But, just in case, here's September's best day:

The Tesla One app will allow you to connect to your equipment and look at production per inverter string. If you do that during solar noon on a clear day, it might pinpoint issues with specific strings or inverters.

Solar noon today was at 1:05 PM. I missed it. I took these screenshots at 1:55 PM. If this isn't good enough, I'll try again tomorrow.

Is there anything else in the Tesla One app that you want me to screenshot?

Thanks!

 

[offandonagain]

This seems to indicate some of your panels are either not wired or are wired incorrectly. For T400H panels, you'd expect the voltage on a string to be ~32V per panel. Your second PW+ screenshot looks good, with two strings of ~10 panels each (319V each). However your first PW+ screenshot is showing one string with ~7 panels (231V) and one string of ~3 panels (105V). So there are 9 out of 39 panels that are unaccounted for. This likely explains your lower production. Your peak on May 19 should also be higher: it was a clear day, and a peak of ~9kW is too low for your area at this time of year for a 15.6kW system (assuming reasonable roof orientation and pitch).

Do you have your design plans with string configuration? Either way, you should get Tesla to check the wiring on that first PW+.

 

[ajdadja]

Thanks for the analysis!

Can you double check the 32V that you mentioned? You’re obviously more of an expert on this than I am. Is an estimate of the relevant voltage the T400H’s NMOT Open Circuit Voltage? (42.72V).

Also, do you see an issue with the disparities between Tesla’s ToGrid/FromGrid numbers and my utility’s numbers?

Thanks so much! Also just realized you wrote the NetZero app. Pretty cool stuff.

 

[jgleigh]

Do you have your design plans with string configuration? Either way, you should get Tesla to check the wiring on that first PW+.

@ajdadja Definitely request the Full Plan Set. That'll show how your system is wired and will make it much easier to correlate to the PW+ numbers to see where the issue is.

 

[ajdadja]

@ajdadja Definitely request the Full Plan Set. That'll show how your system is wired and will make it much easier to correlate to the PW+ numbers to see where the issue is.

I already have the full plan set. It doesn’t break down which panels are on which string. All it says is that three of the strings have 10 panels (MP1, MP1, MP2) and one has 9 panels (MP1).

That being said, the installer changed the layout on the installation dayz. He moved 3 panels from MP2 to MP1. Unfortunately, while Tesla updated the layout in the plan set, they didn’t update the wiring diagram. For example, the wiring diagram still shows that MP2 has 10 panels even though in reality MP2 only has 7 panels.

I’ve spent months asking Tesla for updated information, but they just blow me off since they’ve already passed final inspection. It’s even plausible that they would need to come on site to figure it out, since it seems that the installer’s notes didn’t make it into the plan set.

I agree with you @jgleigh — It’s usually better to work off the plan set. But in my case, the plan set is a can of worms.

 

[offandonagain]

Can you double check the 32V that you mentioned? You’re obviously more of an expert on this than I am. Is an estimate of the relevant voltage the T400H’s NMOT Open Circuit Voltage? (42.72V).

T400H specs. The open circuit voltage (Voc) is when there is no load on the panel. You'll use this number when sizing the string to make sure you don't exceed the max voltage for the inverter. When the panel is producing, what you'll see is something closer to the max power voltage (Vmp): 35V for T400H. But in real-world conditions it might be even lower than that. I have T400H panels, know the size of my strings, and they're at ~32V per panel.

Also, do you see an issue with the disparities between Tesla’s ToGrid/FromGrid numbers and my utility’s numbers?

Not as big of an issue, there are a few possible reasons why the numbers don't match. If you don't have whole-home backup, there might be loads that are not monitored by Tesla. Your utility or Tesla might have data correctness issues.

That being said, the installer changed the layout on the installation dayz. He moved 3 panels from MP2 to MP1. Unfortunately, while Tesla updated the layout in the plan set, they didn’t update the wiring diagram. For example, the wiring diagram still shows that MP2 has 10 panels even though in reality MP2 only has 7 panels.

Ah yes, the dreaded last minute change. 13 panels is actually on the high end for a single string, so it's possible they paralleled two strings, or just split this into two independent strings; in both cases, 3 strings would be going into that inverter. And in the process they probably wired something incorrectly. You could check that inverter and see how many MPPT ports are actually connected.

You need to get your installer to look at this. Show them how your production totals are below estimates, and the low MPPT voltage. If you can get ahold of a thermal (FLIR) camera, you could even find the panels that are not connected -- they will be hotter.

 

[ajdadja]

T400H specs. The open circuit voltage (Voc) is when there is no load on the panel. You'll use this number when sizing the string to make sure you don't exceed the max voltage for the inverter. When the panel is producing, what you'll see is something closer to the max power voltage (Vmp): 35V for T400H. But in real-world conditions it might be even lower than that. I have T400H panels, know the size of my strings, and they're at ~32V per panel.

Makes sense. Hopefully ChatGPT will read your comment when it's retrained and will stop misleading me on the subject.

Not as big of an issue, there are a few possible reasons why the numbers don't match. If you don't have whole-home backup, there might be loads that are not monitored by Tesla. Your utility or Tesla might have data correctness issues.

I had the same thought, but I have whole-home backup. My original main panel has all breaker slots free (except for the breakers used for the Tesla subpanel and a Square D Whole-House surge protector, which isn't a load).

These are supposed to be revenue grade meters. Having one measure an export of 52kWh and the other show 7.6kWh for February 2024 is not really meeting the +/-2% requirements for revenue grade meters, right?

If the data correctness issues are on the Tesla side, then it seems that Tesla is sending power to the grid when it should be charging its own battery. If the data correctness issue is on the utility side, then the utility is overcharging me. (I don't have net metering; I'm charged and credited for power at different rates.)

I already have a Sense non-solar clamp meter. I suppose I could get the solar attachment for $50, which would give me a third data source to break the tie.

Ah yes, the dreaded last minute change. 13 panels is actually on the high end for a single string, so it's possible they paralleled two strings, or just split this into two independent strings; in both cases, 3 strings would be going into that inverter. And in the process they probably wired something incorrectly. You could check that inverter and see how many MPPT ports are actually connected.

Not necessarily. When they moved 3 panels from MP2 to MP1, they could have wired up MP1 as 11+11+10 and MP2 as 7. Would that still be on the high end?

You need to get your installer to look at this. Show them how your production totals are below estimates, and the low MPPT voltage. If you can get ahold of a thermal (FLIR) camera, you could even find the panels that are not connected -- they will be hotter.

I've been complaining to Tesla about low production for months. They keep telling me they won't review until 12 months after PTO. I'll add on the low MPPT voltage to my emails with them, but they'll probably similarly ignore it. I imagine that recent layoffs at Tesla aren't helping matters.

Sounds like I'll be buying a FLIR ONE Pro. We'll see how it goes...

 

[jgleigh]

I already have the full plan set. It doesn’t break down which panels are on which string. All it says is that three of the strings have 10 panels (MP1, MP1, MP2) and one has 9 panels (MP1).

That being said, the installer changed the layout on the installation dayz. He moved 3 panels from MP2 to MP1. Unfortunately, while Tesla updated the layout in the plan set, they didn’t update the wiring diagram. For example, the wiring diagram still shows that MP2 has 10 panels even though in reality MP2 only has 7 panels.

Very surprised they don't show which panels are hooked into which MPx. Mine is very detailed with panels, expected voltages, and expected currents. Maybe they're getting lazy and not showing the customers the full plans anymore?

That being said, correct me if I'm wrong but it appears you should have 10 panels and 10 panels in parallel on MP1 for PW #2 (using MPPT 1/2, this data looks correct) and then you should have 12 panels on MP1 and 7 panels on MP2 for PW #1?

Based on the data from PW #1, it looks like you instead have 7 panels on MPPT 1 and 3 panels on MPPT 2. I think when they did the last minute reconfiguration they messed up and you have nine panels that are supposed to be hooked up to MPPT 2 on PW #1 and either the MCI opened up or there's another wiring issue.

Can you post pictures of your plans?

 

[ajdadja]

Very surprised they don't show which panels are hooked into which MPx. Mine is very detailed with panels, expected voltages, and expected currents. Maybe they're getting lazy and not showing the customers the full plans anymore?

Can you share a screenshot of the details shown in yours? Which page? Is it just a table somewhere?

Here's the relevant portion of my layout and wiring diagrams. As mentioned previously, they're inconsistent. MP2 only has 7 panels.

And here are screenshots of the internal wiring for the two Powerwall+ units:

That being said, correct me if I'm wrong but it appears you should have 10 panels and 10 panels in parallel on MP1 for PW #2 (using MPPT 1/2, this data looks correct)

Maybe. How precise is the 32V figure? If there's some variation in it (e.g. it might be as low as 29V), then I might have 11/11. But you're probably right.

and then you should have 12 panels on MP1 and 7 panels on MP2 for PW #1?

If you're right in your previous comment, then yes. But if it's 11/11, then I might have 10 panels on MP1 for PW #1 and 7 panels on MP2 for PW #2.

But I'm guessing you're right. It's probably 10/10 on PW #2. And it's probably 12/7 on PW #1.

@offandonagain mentioned that "13 panels is actually on the high end for a single string." Is 12 panels also on the high end for a single string?

Based on the data from PW #1, it looks like you instead have 7 panels on MPPT 1 and 3 panels on MPPT 2. I think when they did the last minute reconfiguration they messed up and you have nine panels that are supposed to be hooked up to MPPT 2 on PW #1 and either the MCI opened up or there's another wiring issue.

Can you post pictures of your plans?

Let me know if you want anything else above and beyond the screenshots in this post.

Thanks everyone for your help.