Tweet
It seems like this is sort of pulling a number out of a hat to me. PVWatts default for loss is 14%. If you click the info button or the calculator you can see what this is based on and some of the numbers do seem high. I have seen 10% and 8% used by others on this site. What do you base this on? Is it a matter of experience? Is it just common industry knowledge? I am playing around with PVWatts to try and determine the number of modules I need and it actually makes enough of a difference that I would appreciate any insight.
-
-
Rightly or wrongly, I kind of treat it as a catch all fudge factor. PVOutput.org is a good source of actual data from systems. Here is one with a couple years of history, just north of you:
That system is described as 37 deg tilt, 205 deg azimuth, 8.8 kW. If I plug that into PVWatts with standard rooftmount panels, 10% loss, the model outputs 10,459 kWh. Actual production in 2016 was 10,559 kWh. That level of agreement is, to a large extent, coincidental, as PVWatts is based on "typical" weather but any actual year may vary from that such that differences +/- 10% from year to year are not remarkable.
A finer estimate of appropriate loss can come from looking at the hourly output of PVWatts, and attempting to recreate the actual output of a clear day in the model by adjusting the model parameters. That is hazardous if differences in wind and temperature between the model and the actual are not accounted for (among other possible sources of error).
The system I linked above does not show the "live" output so the hourly matching is not possible. Here is another system, but it has some problems. First, the system says 180 deg azimuth and 23 deg tilt, but the clear day production curve looks more SE than S, so I don't think the azimuth is right. It also has some year-round morning shade, so its loss factor would be modeled as something higher than 10% to account for that. It produced about 1.05 kWh / W, and your ESE array, if shade free, should beat that.
Last edited by sensij; 10-26-2017, 03:01 AM.CS6P-260P/SE3000 - http://tiny.cc/ed5ozx -
A couple of other comments as to PVWatts and ways to use/massage it, particularly the system loss parameter:
- As the PVWatts help/info screens say, among other things, it's a model for preliminary design. Given reasonably accurate input, it'll give results of long term average annual output, say over 10 years or so, that will probably be reasonably accurate, +/- 30 % or so for any 30 day period, or +/- 10 % or so over any 365 day period.
Example: My system's running 31 day average output over 4 years compared to PVWatts 30 day running average output for the same periods (using a 6.9 % system loss parameter - see below) is 95.96 %. The max. ratio for any 31 day period is 131.1%. the min. for any 31 day period is 61.9%. Std. dev. is 9.7%. I estimate somewhere between a 3% to 5% shading loss that PVWatts does not account for.
- Because of the nature of the solar resource and a lot of other variables that will attenuate it, PVWatts does, as Sensij writes, treat a lot of those attenuating and modifying variables as a lumped fudge factor. That works fairly well given the variability of all the things going on.
- But, there are what appear to be ways to massage that system loss parameter from the PVWatts model that gives what some have found to be a better match to either actual system output data from specific systems, or groups of systems in somewhat localized areas, such as PVOutput data, or to the output from more involved models such as SAM, also from NREL (kind of like PVWatts on steroids), or TRNSYS, or others.
- When compared to those other models, some folks, including me, have found that PVWatts gives a reasonably "closer" estimate to SAM's modeled annual output when the system loss parameter is changed from 14 % to some number less than that. In my case, for example (only), if I make/force/use a system loss parameter of 6.9 %, the PVWatts annual output will match my system's current 4 year average running 365 day output, and will also SAM's annual output, both after accounting for a 4.5 % shading loss. My SAM input uses a 3 % fouling factor. That 6.9 % is probably a bit low compared to other systems I keep track of in my HOA and other sources such as PVOutput. But, the 14 % default number seems too high. I use 10% as a reasonably conservative but more realistic number when using PVWatts to model any system's possible output. Of course, if accurate data is available that shows a different number, that should always be used. If such data is available however, that may mean any modeling may be unnecessary in terms of estimating system output.
None of this is rocket science. Given the variable nature of the inputs and, just like the weather they're governed by, their unpredictability, +/- 10% or so is probably about as good as possible and where it counts, all that's necessary and reasonable to expect. Still, model modifications ought to be judiciously considered where they may be a blinding flash of the obvious.Comment
-
Thanks for your responses, they were helpful!Comment
Comment