SAM PV kWh values are much higher than actual values. Should I consider adding shading?

I am doing PV analysis to model expected output to compare to our data. Four months of data show that actual kWh's are consistently only about 60% SAM's expected monthly kWh. Our panels seem to be doing fine with our daily observations so I'm wondering why there is such a difference in our expected value and actual.

My thought is that SAM is not considering cloudy days? Instead it is just assuming ideal conditions every day. Is this what SAM does? I put my location, module, inverter, and system design in correctly. I have not modified the shading and snow, and losses section. Are these two necessary? If so, what would you recommend I add to these two sections for a location in Southern Wisconsin, USA?

As a sanity check, I calculated an expected value for May 2016 for one of our arrays:
Total Energy Predicted/month (kWh) = Nameplate capacity (W)/1000(W) * # of modules *31 days/month * 24 hours/day * capacity factor.

Exp:
(145/1000 [kW])*(60 [Modules])*(31 [Days/month])(24 [hours/day])(0.148 [Capacity factor given from SAM])= 957.97 kWh

SAM is showing a total energy of 1367 kWh.

Our meters show a value of 967.984 kWh. (Difference of 399.83 kWh)

Any ideas why there is such a big difference in the three numbers?

If you have any more questions for me please let me know.

Cheers,

Hello,
I would expect SAM's result to be closer to your calculated value, especially because you used SAM's capacity factor. One thing to keep in mind is that the capacity factor is an annual value, so to be more accurate, you would use your equation to calculate the total annual output in kWh and compare that value to SAM's. SAM does account for cloudy days because it uses solar irradiance data from the weather file, and that data includes the effect of cloudy days.
Would you mind [url=mailto:sam.support@nrel.gov?subject=SAM%20file%20and%20measured%20data%20for%20PV%20output%20comparison t=_self]emailng me[/url] your .sam file along with the 12 monthly kWh values you expect from your measurements so I can investigate?
By the way, you may be interested in the studies we published comparing SAM's PV model output to both measured data from installed systems and other PV models, which you can find here:  sam.nrel.gov/case-studies .
Best regards,

Paul.

Thank you for sending your files.

The weather file for your SAM simulations contains typical-year data that represents the typical solar resource over a period of many years, so I would expect there to be a difference in the monthly output calculated by the model and data that you measure for months in a specific year. For example, the TMY3 weather file you use represents the solar resource over the period between 1961 and 2005, while your measured data is for months in 2016. Also, the solar resource data is for a different location than the array, which also contributes to the difference. Ideally, to compare SAM results to measured data, you would want to use solar resource data measured at the array location and for the same time period input to SAM.

If the system is experiencing shading of the PV array, then you should account for that in SAM as well. You could either roughly estimate the shading impact by using the soiling loss input on the Losses input page, or you could enter shading data from a shading analysis tool like SunEye or SolarPathfinder, or you could use SAM's 3D Shade Calculator to calculate shade loss factors based on a 3D representation of the array and nearby shading objects.

Best regards,
Paul.