New 11kW Ground Mount in N San Diego

[thejq]

Yes, the discussion in logdin's link addresses that, but determines that sampling / averaging error wasn't contributing to that person's mismatch. In this case, if it was an averaging issue, then the error would tend to follow the slope of the curve, with more error when the generation rate was changing quickly and less when it was flat. The mismatch here is a constant 3%... no way that is just a timing thing.

The 3% is the scaling factor I put in to match pvoutput against solaredge's end of day output and inverter LCD display which I think is the most accurate.

[thejq]

https://getsatisfaction.com/pvoutput...osely-how-come


there may also be some net delay introduced in Pvoutput for m more accurate net calculations.

Yes because of this discrepancy I had the pvoutput scaled to match solaredge's total output. So now they match almost spot on at the end of the day.

[sensij]

How are you getting the data to PVOutput? if pulling directly from SolarEdge, then solarEdge only support poling every 15 minutes. This means that the data is being averaged on a 15 minute level and sent to pvoutput.org. This tends to clip off the peaks and smooth out the graph.

Yes, the discussion in logdin's link addresses that, but determines that sampling / averaging error wasn't contributing to that person's mismatch. In this case, if it was an averaging issue, then the error would tend to follow the slope of the curve, with more error when the generation rate was changing quickly and less when it was flat. The mismatch here is a constant 3%... no way that is just a timing thing.

[ButchDeal]

Glad to see the problem isn't isolated. If it was just a timing issue, I would kind of expect the sign of the error to change between when the sun is coming up and when it is going down. The nice thing is that PVOutput.org includes the ability to apply a correction factor, so taking that extra 3% out to make the data match the monitoring portal would be easy enough to do, should he want to do it.

The only meter that really matters is the poco's, so getting a good calibration against that probably affects the financial projections more than any other value, to the extent that possible error matters. Especially with a meter monitoring device like the Rainforest Eagle, but maybe if the inverter and meter are close by, a way to pin down the calibration would be to let the PV system run until its peak production during the day (when the curve is flattest), and then flip the breaker to shut it off. By watching the meter before and after the flip, it should be possible to scale the reported power to the measured power. Do it a couple of times or on a couple of days to average out any transients, and hope that the offset is proportional and constant.

How are you getting the data to PVOutput? if pulling directly from SolarEdge, then solarEdge only support poling every 15 minutes. This means that the data is being averaged on a 15 minute level and sent to pvoutput.org. This tends to clip off the peaks and smooth out the graph.

[sensij]

https://getsatisfaction.com/pvoutput...osely-how-come


there may also be some net delay introduced in Pvoutput for m more accurate net calculations.

Glad to see the problem isn't isolated. If it was just a timing issue, I would kind of expect the sign of the error to change between when the sun is coming up and when it is going down. The nice thing is that PVOutput.org includes the ability to apply a correction factor, so taking that extra 3% out to make the data match the monitoring portal would be easy enough to do, should he want to do it.

The only meter that really matters is the poco's, so getting a good calibration against that probably affects the financial projections more than any other value, to the extent that possible error matters. Especially with a meter monitoring device like the Rainforest Eagle, but maybe if the inverter and meter are close by, a way to pin down the calibration would be to let the PV system run until its peak production during the day (when the curve is flattest), and then flip the breaker to shut it off. By watching the meter before and after the flip, it should be possible to scale the reported power to the measured power. Do it a couple of times or on a couple of days to average out any transients, and hope that the offset is proportional and constant.

[logdin]

The lack of agreement between your PVOutput.org page and your SolarEdge page is interesting. PVOutput shows 4153 W at 13:00 on 3/3, while the SolarEdge public portal shows 4032 W. Any idea why that is? There is a consistent 3% disagreement across all the data that I've just checked.

https://getsatisfaction.com/pvoutput/topics/pvoutput-data-not-tracking-solaredge-data-closely-how-come

there may also be some net delay introduced in Pvoutput for m more accurate net calculations.

[sensij]

Not sure where you got the 1% from. Just for kicks, I looked at my outputs today (first sunny day of March). For my 16x300W panels (http://pvoutput.org/intraday.jsp?id=...88&dt=20150303), if I had gotten M250 micro-inverter, anything above 4KW will be clipped (in fact, since not all panels reaches 250W at the same time, the actual clipping point is slightly less, but that's beside the point). So today, I would have clipped from 11:30 to 1:30 for 0.353 KWh which's 1.2% of my total production today. According to PVWatt, during June/July/Aug the production should be 50% higher than today. For simplicity and be pessimistic, I'm going to assume the production curve is a triangle (in stead of bell curve). After a few geometric manipulations, increasing the total production by 50% would result in about 3.74% clipping in those 3 months. Since I'm expecting to produce 1000KWh / month, I'd be losing 112 KWh (or *$.49 = $55) in just those 3 months. If again pessimistically assume the rest of the year altogether, I loose 1/2 that due to clipping, then for the year I would have lost $83. Of course there're lots of hand waving here, but since I've been pretty pessimistic in my logic, that number is likely the lower bound for me. I guess I'll know more precise numbers as the summer rolls around.

The lack of agreement between your PVOutput.org page and your SolarEdge page is interesting. PVOutput shows 4153 W at 13:00 on 3/3, while the SolarEdge public portal shows 4032 W. Any idea why that is? There is a consistent 3% disagreement across all the data that I've just checked.

Assuming the inflated values are correct, I ran a simulation in SAM and scaled the output to match the clear day peak power this time of year to what your system is actually reporting. When production is higher in the summer, most of that extra generation is in the width of the bell curve, not the height of the peak. It would not be surprising if your peak power in June is the same or slightly less than your peak power in March, around 4.4-4.5 kW based on your PVOutput data, because June tends to be warmer. By August, the peak height is even more likely to be lower.

The degree to which scaling was required suggests the 3% inflation is not real, but I am clearly not able to say for sure what is accurate. Even with the inflated output, the difference between clipped and unclipped power annually came out to 136 kWh (<1.5% of annual production), or $67 if all of that occurs during TOU peak... which is unlikely to be true for very many years as SDG&E is proposing to shift the peak times to later in the day when output would have dropped below the 4 kW clipping level.

If the SolarEdge portal values are correct, the clipped energy in this simulation drops to 79 kWh annually (<1% of annual production), or $38. There is reason to doubt that even the SolarEdge portal calibration is correct, so without a revenue grade meter attached on the inverter, it is hard to say exactly what is being produced.

Also keep in mind that the panels will degrade over time... the warranty allows 2% in the first year and 0.7% after that, so you may figure that after at most 5 years, and probably fewer, whatever clipping you would have been seeing now will be gone anyway. 5*38 = $190, which means if the M250's were more than $200 less than the SolarEdge system, either one would be worth the same (all other things being equal, which is probably not true). If the 3% higher vales are correct, the total value clipped may max out at more like $300-$400, but probably less.

None of this is intended to say that I think designing a system with clipping is a good thing. All I'm saying is that if avoiding clipping is going to be a design decision, the cost-benefit of that decision can be estimated. I believe that spending several hundred $$ to avoid clipping is rarely supported when the panels are only 1.2X the inverter capacity, as they are with 300 W on a 250 W microinverter. If there is no difference in cost, then sure, avoid clipping if you can. If you buy into the idea that operating the inverter at its rating for an hour or two a day for part of the year will shorten its life, and spending more money upfront on a larger inverter will prolong the life of the system, I won't argue with that either, but only point out that I have found no evidence that it is true.

Please do not to take this as an argument in favor of microinverters over SolarEdge. They each have their place, but in general, I would avoid microinverters for reasons that have nothing to do with clipping.

In the grand scheme of things, a few hundred dollars either way is pretty insignificant, so as long as a system is not grossly mis-sized, my interest and arguments are more weighted towards the decision making process, not so much the outcome. The outcomes are all essentially the same.

None of this is intended to suggest I can see the future. However, I continue to believe that modeling expected outcomes, informed by current data, is a reasonable way to make decisions about how I would choose to spend my money, or advise others on how to spend theirs (when asked).

[sensij]

Two quotes I received noted that the P300 was being discontinued. One of the quotes originally said P300 but was updated to P400 with no cost change due to P300 being discontinued (so I think it is probably really being discontinued)

Thanks for the information. SolarEdge collapsed what had been a P350 (for 72 cell panels) and a P500 (for 96 cell panels) into the P400 not long ago, and it sounds like they are streamlining further. If less inventories part numebrs and more standardized manufacturing brings the cost down, great. I don't think it changes my position... if the P300 is available for less money and is appropriate for the panel being installed, I would still choose it. If one fails in the future, swapping a P400 in is not a problem (the optimizers don't all have to match), and SolarEdge has already shown they may maintain backward compatibility in future upgrades, like they did when they moved from the OP series optimizers to the P series. No guarantees, of course.

I'm not sure that future proofing against installing 72 cell panels in the future is where I would choose to spend extra money; if 72 cell panels make sense in the future, they probably would have made since now in the original installation. For the most part, panel efficiency has improved over time leading to more current out of the cells, but only a very small increase in voltage. I would guess that the 48 V max of the P300 still has some room to run, but once it is no longer available, there is no longer a decision to make.

[Poway]

P300 is being discontinued (as you note the P400 provides a superset of capability)

Why spend that extra money? The P300 and P400 are rated for the exact same output power.... 15 A @ 60 V. The P400 has a wider input voltage range to accommodate the higher voltage 72 or 96 cell panels, but if the temperature corrected Voc of your panel is less than 48 V, there is nothing to be gained by going with the P400. All you are doing is forcing the optimizer to perform MPPT over a larger voltage range, which is pointless since the panel will never operate in the upper level of that range.

Two quotes I received noted that the P300 was being discontinued. One of the quotes originally said P300 but was updated to P400 with no cost change due to P300 being discontinued (so I think it is probably really being discontinued)

[thejq]

Usually the clipping involved in these threads is less than 1%, and closer to 0.1%. Like every other design decision, it is a question of what pay for what you get. If minimizing losses without respect to cost was the goal, everyone would be wiring their systems with 4/0 conductors, and would be on the roof every night cleaning the panels. Those two things could easily add up to a percent or two of annual production... yet, they aren't done because it isn't worth the time or effort. Same thing with some of the clipping decisions, except people get emotional about it when they see a flat-topped output and decide they want that bell curve at any cost.

Not sure where you got the 1% from. Just for kicks, I looked at my outputs today (first sunny day of March). For my 16x300W panels (http://pvoutput.org/intraday.jsp?id=...88&dt=20150303), if I had gotten M250 micro-inverter, anything above 4KW will be clipped (in fact, since not all panels reaches 250W at the same time, the actual clipping point is slightly less, but that's beside the point). So today, I would have clipped from 11:30 to 1:30 for 0.353 KWh which's 1.2% of my total production today. According to PVWatt, during June/July/Aug the production should be 50% higher than today. For simplicity and be pessimistic, I'm going to assume the production curve is a triangle (in stead of bell curve). After a few geometric manipulations, increasing the total production by 50% would result in about 3.74% clipping in those 3 months. Since I'm expecting to produce 1000KWh / month, I'd be losing 112 KWh (or *$.49 = $55) in just those 3 months. If again pessimistically assume the rest of the year altogether, I loose 1/2 that due to clipping, then for the year I would have lost $83. Of course there're lots of hand waving here, but since I've been pretty pessimistic in my logic, that number is likely the lower bound for me. I guess I'll know more precise numbers as the summer rolls around.

[HX_Guy]

Future proofing is my thought process. It could literally costs less than $100 extra for all the optimizers but you'd have some flexibility if you ever want to switch panels to a larger version.

[sensij]

Go with the P400 optimizers even if you use 300W panels. They are only a few bucks more per unit so it won't cost much at all.

Why spend that extra money? The P300 and P400 are rated for the exact same output power.... 15 A @ 60 V. The P400 has a wider input voltage range to accommodate the higher voltage 72 or 96 cell panels, but if the temperature corrected Voc of your panel is less than 48 V, there is nothing to be gained by going with the P400. All you are doing is forcing the optimizer to perform MPPT over a larger voltage range, which is pointless since the panel will never operate in the upper level of that range.

[sensij]

Also as mentioned, I would not want any clipping at all. Even if it's only a percent or two of annual production, why have any clipping at all if it's not necessary?

Usually the clipping involved in these threads is less than 1%, and closer to 0.1%. Like every other design decision, it is a question of what pay for what you get. If minimizing losses without respect to cost was the goal, everyone would be wiring their systems with 4/0 conductors, and would be on the roof every night cleaning the panels. Those two things could easily add up to a percent or two of annual production... yet, they aren't done because it isn't worth the time or effort. Same thing with some of the clipping decisions, except people get emotional about it when they see a flat-topped output and decide they want that bell curve at any cost.

[HX_Guy]

11.95 kW DC
28x SunPower 427W Panels
Solar Edge P400 Optimizers
SE11400 inverter (is this to small for the 11.95 DC if so could drop to 26 panels 11.1KW?)

10.8 KW DC
36x LG 300W Panels
Solar Edge P300 Optimizers (or should I still use P400 optimizers?)
SE11400 inverter

10.8 KW DC
36x LG 300W Panels
Enphase M250 (will these clip ... do I need to use lower wattage panels given ideal pointing of array?)


Thanks

As others have said, I think the SunPower are overkill. They are good panels no doubt but it seems their niche is if you're short on roof space, otherwise theyll produce pretty much the same as an LG, Canadian Solar, etc panel.

Go with the P400 optimizers even if you use 300W panels. They are only a few bucks more per unit so it won't cost much at all.

Also as mentioned, I would not want any clipping at all. Even if it's only a percent or two of annual production, why have any clipping at all if it's not necessary?

[thejq]

We both may be wrong but I agree completely. SolarEdge has guidelines that allow DC panel rating to be 135% of AC output rating (they call this DC oversizing). This is probably not a problem in other areas of the country or on roofs with not the best orientation. But in my case I have unobsturcted view to the South and since I'm doing a ground mount.. can have perfect elevation/Azimuth angles.. Also the panels I'm buying are 0-+5% so I think at least in early years there would be alot of clipping so I will not do DC oversizing.

Since you have plenty of space, SW might be better for you, assuming you will be on either DR-SES (solar) or EV-TOU plan. Both pay the most in the afternoon. One reason some people might consider oversizing is to avoid having to upgrade (or derate) the switch panel, especially for bigger installations.