PV forecasting

[timhoffner]

Hello,
I am looking for resources for a daily PV output forecast. I have used:

http://www.intelligence.tuc.gr/renes/

and
solarserver.com (though the US option seems to be down right now)

any other free sites out there that predict output (pref in W) hourly or daily?
Thanks!
Tim

[J.P.M.]

Hello,
I am looking for resources for a daily PV output forecast. I have used:

http://www.intelligence.tuc.gr/renes/

and
solarserver.com (though the US option seems to be down right now)

any other free sites out there that predict output (pref in W) hourly or daily?
Thanks!
Tim

The site seems to be working for the U.S. at this time, but from what I've seen, I'd not put too much reliance on its predictive value beyond maybe 10-20% or so from the looks of the input required and its general treatment of weather parameters. You'll probably get as close by knowing what your system produces on a clear day this time of year, sticking your head out the window and estimating the % of sun for the day and multiplying the two numbers together. Either method will likely get as close.

The second works just fine for most of my day/day predictive needs.

Most of this stuff ain't as complicated as some want to make it.

If you need better estimates, try a site called "Solaranywhere.com". Most of it is free, to get the good stuff costs $$'s, and don't be insulted if you don't understand the required input.

Another way - run PVWatts and use the hourly output for a clear day near the date you want and multiply that by the (estimated) % sun for that day.

[sensij]

Hello,
I am looking for resources for a daily PV output forecast. I have used:

http://www.intelligence.tuc.gr/renes/

and
solarserver.com (though the US option seems to be down right now)

any other free sites out there that predict output (pref in W) hourly or daily?
Thanks!
Tim

Awesome link, thanks for sharing it. The models on which the Renes system is built appear to be well thought out and tested. The 22% ish rMAE error reported in the documentation sounds believable when diffuse fraction is determined by model, although most of the work that I've read is presented with RMSE, so it is hard to say how it compares.

For all the forum members who compare their output to PVWatts and come to erroneous conclusions about performance, the tool linked here is actually a *predictor*, and is a much better choice by which to baseline the live performance of the system. Detecting small changes in performance is likely to still be lost in the noise, I think comparing against crowd-sourced live data (PVOutput.org) is probably the best metric available when there are enough nearby systems to judge against.

[timhoffner]

Thanks J.P.M. which site seems to be working for you? Renes or solarserver? Understood about keeping it simple. um, snowing here today, so let's guess it will be a "low jiggawatt" day.
Tim

The site seems to be working for the U.S. at this time, but from what I've seen, I'd not put too much reliance on its predictive value beyond maybe 10-20% or so from the looks of the input required and its general treatment of weather parameters. You'll probably get as close by knowing what your system produces on a clear day this time of year, sticking your head out the window and estimating the % of sun for the day and multiplying the two numbers together. Either method will likely get as close.

The second works just fine for most of my day/day predictive needs.

Most of this stuff ain't as complicated as some want to make it.

If you need better estimates, try a site called "Solaranywhere.com". Most of it is free, to get the good stuff costs $$'s, and don't be insulted if you don't understand the required input.

Another way - run PVWatts and use the hourly output for a clear day near the date you want and multiply that by the (estimated) % sun for that day.

[J.P.M.]

Thanks J.P.M. which site seems to be working for you? Renes or solarserver? Understood about keeping it simple. um, snowing here today, so let's guess it will be a "low jiggawatt" day.
Tim

You're entirely welcome.

Actually, neither one. Daily forecasting is of little use to me. I've got my own stuff that estimates clear day output for any array in any orientation for any panel/inverter comb. that has published data anywhere on earth down to 1 min. time increments if I choose. The same spreadsheets will work with TMY inputs and give results pretty close to SAM's.

Climate's what you expect. Weather's what you get.

I'm skeptical of the sources you point to because of the lack of what I feel is necessary weather and atmospheric inputs, and equipment particulars.

Simply because one is not aware that some (a lot) of additional input is necessary does not make that input unimportant. Without more input, what your sources provide is, IMO, little more reliable than getting a clear day insolation estimate and guessing at a % of possible sun from a weather forecast.

Besides, estimates of 1 to several days insolation in advance are of limited use anyway. Check the weather report or local variables like humidity, wind direction and sky characteristics and take your best shot. Close enough and more fun.

As Sensij alludes however, this type of forecasting does sort of point out the futility and misapplication of using design tools such as PVWatts to predict short term system output.

[russ]

Actually, neither one.

YOu got it J.P.M. - pretty much of a crap shoot. And useless to the residential user anyway. Utilities need such tools but not the homeowner.