FWIW, after hanging around solar for ~ 40+ years, and having done a bit of modeling of both solar thermal and later PV systems, I've found that most PV models available today can produce reasonably good estimates of system performance given the inherent variability of the required inputs.. The biggest source of error or variability is, IMO only, user ignorance of what is that inherent variability of inputs - mostly but not entirely weather related, and the unrealistic expectations that result from a lack of understanding of those limitations.
New to Forum & Solar - I have a choice to make...
Collapse
X
-
-
Future weather isn't known, so any estimate of system performance is necessarily probabilistic. Resources are freely available to self-educate about what is an excessively rosy or sunny estimate, vs estimates that are more firmly grounded in science.
On that note, the latest release of SAM is finally including source code. Anyone who wants to look under the hood of PVWatts (which is one of the models embedded in SAM) will be able to do so.
Yes, there is lots of stuff out there but, for a newbie, it is tough to know what is actually relevant or what is just a side order of minutia. Plus, it is not uncommon to find a site that directly contradicts a previous site or simply injects more uncertainty. "West facing is better than south facing? The jury is still out!". Roughly paraphrased from a solar site and could have been in reference to some odd confluence only applicable in outer Mongolia (but they don't tell you that)
I will say that you and a few others here have been really kind and helpful in answering my (probably) stupid questions. Thanks!Comment
-
JPM, you made me realize (and to continue the analogy) that I may have painted all sales/installers with the same broad brush. As I don't know you or your company, I didn't intend to tarnish anyone's reputation blindly. I apologize and should have qualified my statement as SOME installers.
By the same token, you know the type of sales/installers I'm talking about though, right?Comment
-
Again, FWIW, all these simulations and models and ... are just best case guesses and estimates. You will find that part of the solar sales pitch is (or has been) utter snake oil. You may get close to their "estimates" for modeling and simulations vs real world conditions but they are going to likely give you the best case to make the sale. They would scare people off if they quoted the panel total watts (combined STC) @ $3.50 per watt but tell you that you will never reach that kWp (combined STC). Maybe I was a bit too naive.
As Butch said, "They can easily add 1kW to the east (would need a combiner is all), and should result in less than 0.2% clipping, though likely none depending on how much degradation of the PV modules has happened." Note the words "should", "less than", "likely", "depending" and "has happened" (future past tense?). This is much like installers painting a rosy, sunny picture but leaving many holes to disappear in.
the model is as good as the inputs to the model. It is a rather simple system to model and the primary variable is weather. For any one day you could get different weather that WILL result in different result but typically ( you do understand typical ) the system will perform as modeled. No some sales people do not set up a model, they use a rule of thumb or guess, not the same. It is easily possible to model dead on, with vary low variance.
OP has already purchased a system and not from my company, I am not trying to sell him or you anything other than give facts.
So so here is one for you, you will very very rarely ever produce at STC. bTW, we try to inform all our customers of this, and most do as well ( though not all sales people within organizations).
you have to remember that your production is based on weather, solar is a game of long haul averages not short run peaks.
since most companies give some form of production number and / or guaranty ( as limited as they are) and the sale is based on size of the array, they are unlikely to give best case numbers as you say. When I review competing quotes they are usually very low on production and oversized to make a huge sale. The ones that are accurate are usually very similar to mine ( different brands but very similar size and layout) be cause there is usually only one most productive layout.
I have seem some crazy quotes and estimates but have never seen one claiming full STC production.Last edited by ButchDeal; 09-07-2017, 08:31 AM.OutBack FP1 w/ CS6P-250P http://bit.ly/1Sg5VNHComment
-
Comment
-
To my knowledge outer Mongolia does not have TOU billing.
OutBack FP1 w/ CS6P-250P http://bit.ly/1Sg5VNHComment
-
Comment
-
Comment
-
FWIW, after hanging around solar for ~ 40+ years, and having done a bit of modeling of both solar thermal and later PV systems, I've found that most PV models available today can produce reasonably good estimates of system performance given the inherent variability of the required inputs.. The biggest source of error or variability is, IMO only, user ignorance of what is that inherent variability of inputs - mostly but not entirely weather related, and the unrealistic expectations that result from a lack of understanding of those limitations.
What we feed into it as the source of weather is usually TMY, which is notably outside the expected usage of the the TMY data. However assuming TMY for a region is remotely reasonable, the solar production models should be as well on any reasonable timed average like monthly. At the hourly level, not even close, as has been mentioned on this forum repeatedly. But for monthly or annual average production within a reasonable amount of accuracy, it is very useful. Also for clipping, inverter sizing, annual production etc, it is very effective.
For the pedantic, reasonable amount of accuracy should be taken with a measure of the value of a kWh.
Any one expecting a model of future production to be accurate to the kWh doesn't understand the effect of weather or is loopy.
Last edited by ButchDeal; 09-07-2017, 10:00 AM.OutBack FP1 w/ CS6P-250P http://bit.ly/1Sg5VNHComment
-
Since homes are usually limited to 2 or 3 roof azimuths, choosing south or more southerly array orientations will usually be more cost effective than more westerly facing.
A 270 degree azimuth will not only produce less electricity than < 270 deg. az., it will also be less cost effective than azimuthal orientation less than 270 deg. A 270 deg. array azimuth is not particularly cost effective for most applications.
Of course, the most cost effective orientation will always, and in the future under T.O.U. tariffs that will shift hour/times/rates, be a more or less changing number. That looks like a new fact of life.
However, for most applications, and most T.O.U. tariffs I've seen, or might reasonably expect to see in the near to mid term, I'd bet on a "more southerly" orientation being more cost effective than a "more westerly" orientation. Of course, each application will be at least slightly different and will need to be checked, but to knee jerk it and say, or to lead people to infer west facing is better with a T.O.U. tariff is doing a disservice.Comment
-
JPM, you made me realize (and to continue the analogy) that I may have painted all sales/installers with the same broad brush. As I don't know you or your company, I didn't intend to tarnish anyone's reputation blindly. I apologize and should have qualified my statement as SOME installers.
By the same token, you know the type of sales/installers I'm talking about though, right?
A respectful comment and a suggestion: Looks to me that you seem to have a lot more opinion about solar energy than knowledge about solar energy. Get more of the latter and you'll embarrass yourself less when expressing the former.
Take what you may want of the above. Scrap the rest.Comment
-
Butch: I'd respectfully suggest distinguishing west facing from "westerly" facing. Around most of SO. CA, the orientation that produces the greatest average offset in terms of $/kWh produced on a T.O.U. tariff is usually about and very approximately somewhere around 200 to 220 deg.
What I have the team do is fill the southern most least shaded roofs then when there is a choice between easterly and westerly facing roofs, fill the west first unless is is shadowed.
Aurora gives good information to choose and does a financial model though you have to do a design first then run the financial model after.
When you show the irradiance levels, you can see the kWh/m^2/day levels for all sections.
However, for most applications, and most T.O.U. tariffs I've seen, or might reasonably expect to see in the near to mid term, I'd bet on a "more southerly" orientation being more cost effective than a "more westerly" orientation. Of course, each application will be at least slightly different and will need to be checked, but to knee jerk it and say, or to lead people to infer west facing is better with a T.O.U. tariff is doing a disservice.
The roof is the roof, so all you can do is make the best of what is there and the rules given.
The OP has a little extra room for future expansion but hedged his current install a bit towards the west side.
The change from 7E to 11E effects his annual by 130kWh, and financial pretty much undetectable. I don't have the exact location though so if there is say a tree to the east it could effect more significantly.
I modeled it at 100 / 280 degrees and his as built of 17 modules at 280 degrees and 11 at 100 degrees comes to 9.7MWh first year
Swapping it to 17 at 100 degrees and 11 at 280 degrees changes it to 9.853MWh first year. The IRR improves ever so slightly to 10.89% for the more easterly vs 10.9% for the more westerly. so really not much difference. But then TOU could change.
OutBack FP1 w/ CS6P-250P http://bit.ly/1Sg5VNHComment
-
Yes very much so. The solar model is highly accurate and very simple ray tracing with temperature variances etc.
What we feed into it as the source of weather is usually TMY, which is notably outside the expected usage of the the TMY data. However assuming TMY for a region is remotely reasonable, the solar production models should be as well on any reasonable timed average like monthly. At the hourly level, not even close, as has been mentioned on this forum repeatedly. But for monthly or annual average production within a reasonable amount of accuracy, it is very useful. Also for clipping, inverter sizing, annual production etc, it is very effective.
For the pedantic, reasonable amount of accuracy should be taken with a measure of the value of a kWh.
Any one expecting a model of future production to be accurate to the kWh doesn't understand the effect of weather or is loopy.
Seems to me however, all that reasonableness needs to be qualified a bit further with respect to time limits, or at least some consideration of how short a time is "reasonable". I'd agree that it's reasonable when used for long term average results, with long term being, say 10 years or so, maybe a bit less, but with a lot more variability as the time period under consideration gets shorter.
I'd refer to my recent post describing how my system's 31 day running output compares to model estimates for ~ the last 4 years seem to produce an approx. +/- 35 % variation from the PVWatts model using TMY3 data - that 35 % being a bit more than the PVWatts published number of ~ +/- 30 % or so for monthly production est.. Admittedly, one station (mine) and 4 years does not qualify as a smoking gun, but 1,380 data, each one a 31 day comparison of actual vs. modeled data using TMY3 does not seem to contradict the +/- 30 % claim that PVWatts makes for 30 day accuracy.
It looks to me that we all mostly agree that the accuracy or predictive capability and practicality of most models currently in use is inverse to the time period under consideration in some way(s), but that we may differ on when and in what ways that accuracy may be limited.Comment
-
Butch: I'd agree that the TMY data is reasonably representative of weather, including the irradiance data, even though about 90% of that irradiance data and other data is itself modeled, a factoid that most folks are unaware of (See the NREL best practices handbook, NREL/TP-5D00-63112 for confirmation on that).
Seems to me however, all that reasonableness needs to be qualified a bit further with respect to time limits, or at least some consideration of how short a time is "reasonable". I'd agree that it's reasonable when used for long term average results, with long term being, say 10 years or so, maybe a bit less, but with a lot more variability as the time period under consideration gets shorter.
I'd refer to my recent post describing how my system's 31 day running output compares to model estimates for ~ the last 4 years seem to produce an approx. +/- 35 % variation from the PVWatts model using TMY3 data - that 35 % being a bit more than the PVWatts published number of ~ +/- 30 % or so for monthly production est.. Admittedly, one station (mine) and 4 years does not qualify as a smoking gun, but 1,380 data, each one a 31 day comparison of actual vs. modeled data using TMY3 does not seem to contradict the +/- 30 % claim that PVWatts makes for 30 day accuracy.
It looks to me that we all mostly agree that the accuracy or predictive capability and practicality of most models currently in use is inverse to the time period under consideration in some way(s), but that we may differ on when and in what ways that accuracy may be limited.
I generally do the comparisons when we get complaints about site not performing as expected, and for the most part they are pretty darn close. We had one or two where a new designer of ours made a mistake and missed shadows a little farther away. A little training there.
Usually the complaints are just what you say though, or homeowner divided the annual by 12 and expected every month to be the same, we give them the full prediction and they see it increasing on different months with some explanation of temperature etc. Oh yeah or the other big complaint is that they see the bill with labels for consumption and generation which do not match the generation they get from our monitoring, why is their bill not showing all the power? ( I love that one)
Annually we are usually within 10%, though I think you have been looking at these over longer periods.
Though I will note you mentioned +/- x% have you really seen the "+" variance be that large? We rarely see the "+" side nearly as much as the "-" side. other than a little in spring and fall, there is usually not much on the "+" side.
OutBack FP1 w/ CS6P-250P http://bit.ly/1Sg5VNHComment
Comment