LG300 real world power output

[sensij]

More information about the insolation view is in this PVOutput forum thread.

A few things:

1) Despite what the help file says, temperature matters, and is included the calculation. Attempting to make a clear sky production estimate without some approximation of temperature is not going to get you very far.

2) If you look closely at that thread I linked, you can see that systems that appear to undershoot the insolation estimate in the middle of the day are probably running warming / less efficiently than the estimate expected. That can be for a few reasons... maybe panel coefficients, but also, the data source used or temperature might be misleading.

The system you are comparing against is actually managed under my PVOutput account, a very good friend of mine. The temperature at his house has been higher than what the Wunderground temp has reported, and we might try a different weather station at some point.

3) He just had his roof redone before installing solar, and put in an energy star (highly reflective) roof. It is essentially a flat roof, something less than 5 deg we think (haven't measured it yet). I guess it is possible that a flat, reflective roof could somehow be increasing the operating temp of the array (?). A picture of the installation is attached, the camera is pointed south or close to it.

20150502_163304.jpg

I agree we are splitting hairs, but I think there are still a few more to split before I would be able to agree that the LG300's are showing some kind of performance advantage. To the extent that panels may or may not be a commodity... my guess is that 10 years from now we'll have a better idea of which panels are good and which ones are not, but I have no good ideas on how to handicap that race.

Edit: There is a weather station just a couple blocks away from his house, definitely reports warmer temps than the station we had been using. I've switched the feed; we'll see if it makes a difference in the "insolation" calculation over the next couple days.

[SunEagle]

More information about the insolation view is in this PVOutput forum thread.

A few things:

1) Despite what the help file says, temperature matters, and is included the calculation. Attempting to make a clear sky production estimate without some approximation of temperature is not going to get you very far.

2) If you look closely at that thread I linked, you can see that systems that appear to undershoot the insolation estimate in the middle of the day are probably running warming / less efficiently than the estimate expected. That can be for a few reasons... maybe panel coefficients, but also, the data source used or temperature might be misleading.

The system you are comparing against is actually managed under my PVOutput account, a very good friend of mine. The temperature at his house has been higher than what the Wunderground temp has reported, and we might try a different weather station at some point.

3) He just had his roof redone before installing solar, and put in an energy star (highly reflective) roof. It is essentially a flat roof, something less than 5 deg we think (haven't measured it yet). I guess it is possible that a flat, reflective roof could somehow be increasing the operating temp of the array (?). A picture of the installation is attached, the camera is pointed south or close to it.

[ATTACH]6993[/ATTACH]

I agree we are splitting hairs, but I think there are still a few more to split before I would be able to agree that the LG300's are showing some kind of performance advantage. To the extent that panels may or may not be a commodity... my guess is that 10 years from now we'll have a better idea of which panels are good and which ones are not, but I have no good ideas on how to handicap that race.

Edit: There is a weather station just a couple blocks away from his house, definitely reports warmer temps than the station we had been using. I've switched the feed; we'll see if it makes a difference in the "insolation" calculation over the next couple days.

What about installing a "weather station" next to the pv array? It might not be as professional as the stations you are getting data from but it could add some info concerning what is happening on the roof at different times of day.

[J.P.M.]

What about installing a "weather station" next to the pv array? It might not be as professional as the stations you are getting data from but it could add some info concerning what is happening on the roof at different times of day.

Two things: I'm not going to rain on PVOutput's parade. I believe it's a very valuable source of information and a very useful tool. I reference it daily.

However, I'd cautiously comment, that due to my non participation, I'm ignorant of many of the ways that the site defines, obtains and uses data. Maybe that info is somewhere non members can't reach.

Reading the threads from PVOutput members, I get the feeling that many of them are sort of uninformed and ignorant about the solar resource, its estimation, components, how array orientation affects PV output and other things.

Point being, the way PVOutlet seems to define terms and how it uses information, in my probably ignorant opinion, may be adding to the ignorance, or at least not decreasing it by adding to the confusion and thus causing what I see as a lot of incorrect and/or faulty assumptions and conclusions about the solar resource, leading to wheel spinning and perhaps to less than optimum decisions and less than the most cost effective outcomes.

As for weather stations, I have about what SunEagle is writing about. It's about 3-4 ft. north of my array. It's that same as the equipment used by many Weatherunderground sites, including the one about 1 1/4 miles from me.

FWIW, for anyone serious about solar energy as a hobby - or more, for about $1K, you can get a decent weather station that borders on professional and comes with a pyranometer. Mine's been trouble free for going on 2 years now w/weather data measured/recorded at 1 mi. intervals 24/7. For serious students of solar energy, it's a real good and IMO almost essential tool.

In the bigger scheme of things, that $1K can be a worthwhile investment for those interested.

I've worked with sophisticated and expensive weather instruments, and very inexpensive ones. IMO, the quality of the data is as much a function of the user as the equipment. Weather instruments are better than in the past for the price, but like all instruments, need some care and feeding. Friendly advice to those who get one: Make sure you can get at it easily.

[gvl]

Yesterday was a noticeably warmer day and my best performing panels just touched 260W, others were in the mid-250W range. Looks like I should be expecting 250W or less when it really warms up in August-September here. I suppose as long as overall production tracks with PVWatts predictions it shouldn't matter much.

[J.P.M.]

Yesterday was a noticeably warmer day and my best performing panels just touched 260W, others were in the mid-250W range. Looks like I should be expecting 250W or less when it really warms up in August-September here. I suppose as long as overall production tracks with PVWatts predictions it shouldn't matter much.

It probably doesn't matter.

But, to clear something up: PVWatts is a tool for predicting long term average output over many years. it is not a predictor of performance. Some months/years will be more or less.

Example: May, 2015 production for my array was 813 kWh. May 2014 production was 974 kWh. PVWatts long term est. using std. (14 %) losses is 858 kWh. 2015/2014 = 813/974 = .834. I don't think my array's performance dropped that much in 1 year.

Point: incorrect assumptions (my array "missed" PVWatts "prediction" --->>> something must be amiss) can lead to incorrect conclusions and perhaps inappropriate actions. I suggest reading the PVWatts help/info screens.

Take what you want of the above. scrap the rest.

[foo1bar]

PVWatts is a tool for predicting long term average output over many years. it is not a predictor of performance.

What would you call a tool for predicting long term average output if not a predictor of performance?

I think you're wrong in saying it isn't a predictor of performance.
It *IS* a predictor of performance.
However it is using a best guess at weather for your location. (It's a guess based on historical data - but still a guess)
Therefore any particular day (or month or year) of results is not going to be a perfect match to the prediction.
(unless you are in a spot where your temperature and cloudiness for that day/month/year is very very consistent)

"Point: incorrect assumptions (my array "missed" PVWatts "prediction" --->>> something must be amiss) can lead to incorrect conclusions and perhaps inappropriate actions."
That I certainly agree with.

[sensij]

What would you call a tool for predicting long term average output if not a predictor of performance?

I would not have used the same choice of words. PVWatts is a model of PV system output. One of the inputs into the model is "weather"... when historical weather is used, the model estimates PV output for those conditions. If you want a prediction, you would need to input the predicted weather.

To the extent that past weather is predictive of future weather (averaged over long time periods), PVWatts may be considered a prediction tool. However, describing it as anything other than a model is misleading. The science involved in creating the historical weather file helps make the modeled past useful for estimating the potential production of a system in the future.

[gvl]

But, to clear something up: PVWatts is a tool for predicting long term average output over many years. it is not a predictor of performance. Some months/years will be more or less.

Understood, I meant long term.

[Silver_Is_Money]

Why not just accept that the PTC ratings are generally correct for most cases? That makes it all rather easy, no?

[sensij]

Why not just accept that the PTC ratings are generally correct for most cases? That makes it all rather easy, no?

Huh? How does the PTC rating help you size your system, without some kind of model to take that panel rating and and turn it into location and orientation specific output? The CEC calculator you've provided before is just a different front end for the PVWatts model (an older version of it), and so discussions about how to use PVWatts apply equally to it.

[J.P.M.]

What would you call a tool for predicting long term average output if not a predictor of performance?

I think you're wrong in saying it isn't a predictor of performance.
It *IS* a predictor of performance.
However it is using a best guess at weather for your location. (It's a guess based on historical data - but still a guess)
Therefore any particular day (or month or year) of results is not going to be a perfect match to the prediction.
(unless you are in a spot where your temperature and cloudiness for that day/month/year is very very consistent)

"Point: incorrect assumptions (my array "missed" PVWatts "prediction" --->>> something must be amiss) can lead to incorrect conclusions and perhaps inappropriate actions."
That I certainly agree with.

It's not a predictor of performance any more than long term climate models based on observations are a predictor of future weather. Climate's what you expect. Weather's what you get. Especially over short periods.

I'd suggest, if you have not done so already, understand how a TMY (Typical Meteorological Year) is calculated. That's the weather file that PVWatts, and a lot of other models commonly use. A TMY for a location isn't a model as much as a concatenated string of 12 actual months with some data gaps estimated. Doing so may change your opinion about predictive properties of any PV performance model that uses a TMY.

BTW: all this says nothing about how, or even the necessity to modify the inputs relating to system parameters such as long term fouling, line losses, fouling, shading, etc. which will be different from the model assumptions, or even how those can be estimated or will change over time.

You are certainly free to think of it as a predictor. Amen. Hail the freedom. But, IMO only, to say that may lead others more ignorant about it than you down an incorrect way of thinking that will increase the chances of a less than accurate conclusion.

[J.P.M.]

Why not just accept that the PTC ratings are generally correct for most cases? That makes it all rather easy, no?

PTC rating of a panel has nothing to do with performance estimating.

[foo1bar]

It's not a predictor of performance any more than long term climate models based on observations are a predictor of future weather. Climate's what you expect. Weather's what you get. Especially over short periods.

A climate model *IS* a predictor of future weather.
It doesn't predict what the high and low will be a year from Wednesday.
BUT it does give you a prediction of what the highs and lows will be during June for next year (and the year after and the year after that.)

Doing so may change your opinion about predictive properties of any PV performance model that uses a TMY.

Don't think so. My understanding is it is basically using a "model year" that is supposed to be representative of the past years. And since it's representative of past years, it is expected to be representative for future years as well. Of course there's no way to know if a storm is going to come through on any given day/week - but probably if it's common to have 8-10 stormy days in July in that city, the TMY info will reflect that, having 8 stormy days (or 9 or 10)

You are certainly free to think of it as a predictor.

Well, AFAIK it's the best tool for making an estimate of how many kwh you're likely to get for a given size/configuration of an array. Like any prediction tool/model it is subject to the inputs you give it being different than actual reality. And of course actual weather is going to be different than any model weather - it's not prescient. BUT as YOU said it is "a tool for predicting long term average output" - And that *IS* another way of saying it's a predictor of performance. (long term average output *IS* performance) It may be inaccurate if you try to look at a given day in the model vs actual output on the same day - but if other inputs (like shading etc) are accurate, it is probably the best prediction around for what you'll get next year for output.

[J.P.M.]

BUT as YOU said it is "a tool for predicting long term average output" - And that *IS* another way of saying it's a predictor of performance.

You are correct. I did write that and it was a mistake on my part.

I meant to write: "It is a tool for estimating long term average output." I apologize for my lack of consistency, thoroughness and proof reading.

I'm not sure we have a difference of opinion about the limits or the usefulness of the PVWatts output as much as different definitions and uses of the terms "estimate" and "predict" or other terms that deal with forecasting future outcomes.

The two words are often used interchangeably in common language. I prefer the word estimate for what solar models put out because it seems to imply (to me at least) some greater sense of uncertainty in the result than the use of the word predict, which to me carries with it a bit more of a sense of some greater certainty in the outcome.

As best as I can find, it seems that even statisticians can't agree too well on the use of the terms in a statistical sense where they are supposed to have specific meanings different from one another. I don't think statistics apply here as much as probability, but that's maybe getting off on a slightly different subject.

At the end of the day, and perhaps in spite of all my tightass opinions and rants about terms, I'm of the opinion that it's more important to stress the idea that solar models that estimate/predict/forecast/(favored term here) some "future value" have limits, and to blindly hang one's hat on a number such as "some" annual output without knowing the limits on the accuracy of that output, or the consequences of seeing such a number as less variable or chaotic than it may actually be, can lead to less than good outcomes that can be avoided with a little information about those limits. That's were I believe we may be closer to agreement.