Improving SAM by adding 3 specs to module .pan files

April 23, 2019

I have just completed a kind of “calibration” or “tuning” of SAM, only for the purpose of comparing annual energy production of 3 different quality levels of panels: best, average, and worst.

In this effort, I ran into a few limitations of SAM that would be great to remove. Perhaps this can be done by getting panel manufacturers to add a few additional specs to their .pan files.

Specifically,

1. Nominal POA Reflection (IAM): this does not vary with panel model, but it should. Different qualities of panels have different quality of top glass. It can be anti-reflective or not, light-trapping or not, patterned or float, prismatic or matt finish, low-iron-oxide or not, and thicknesses of about 2.5mm to 4.0mm. These differences cause different refractive indices and solar-transmission %. Baseline commercial glass has a solar transmission ratio of 83.7%, per GreenRhinoEnergy.com, so 16.3% of the energy is filtered out before it gets to the cell. About half of this 16.3% loss is due to reflection (8.15%) and half to iron impurities within the glass (8.15%). These are not insignificant percentages to SAM. 20% of all PV-module glass is thin, cheap, float glass with high iron impurities and no A/R coating at all. While I was unable to determine the bottom-line solar-transmission % for such glass, it is almost certainly below 90% and looks as though it may be as low as 87%.

In spec sheets for solar glass from several companies, I see total solar energy transmission ranging from 90.4% to 91.7% for good glass that is neither light-trapping nor has A/R coating, with most specs spread out evenly between these two extremes. Glass manufacturers claim a 2% - 3% improvement in solar energy transmission for A/R coating. I see examples of A/R glass with specs over 94%, and GreenRhinoEnergy.com recommends 94% as the bare minimum for crystalline silicon modules.

Putting all these together, panels can vary by up to 7% in IAM, with all other factors the same: say from about 87% to about 94%, although I can’t be sure about it. The glass manufacturers know, however, and provide this spec to their customers, the solar panel manufacturers.

So if this spec were added to the panels .pan files and included in CSI’s database, wouldn’t SAM be able to calculate annual energy more precisely for each panel model, using CEC Performance with Module Database?

2. Low-light/red-spectrum performance. This is related to, but not identical with, IAM, although predominantly red-spectrum light happens when incidence angle is low (mornings and evenings). My limited research into PV glass has only turned up SunPower’s claim of significant advantages in this area, but I can’t tell whether it’s attributed to their glass or to their cells. Either way, perhaps there is some spec or curve that could also be added to .pan files to reflect this difference. It appears to be of the order of about 1%.

3. LID. SAM also seems not to differentiate among panels as to their susceptibility to LID. This is a spec panel manufacturers certainly know, and I would think could easily be added to the .pan files. It appears to have a range of about 3%.

Combining these differences in IAM, low-light, and LID, we have a possible range of 11% among panels. This is a highly significant number for SAM.

So my questions for you are these:

1. Do you think it possible for panel manufacturers to add these specs to their .pan files, if asked or compelled?
2. Could SAM be revised within a reasonable time to automatically take these 3 differences into account?
3. Do you have any suggestions about how to get the manufacturers to add these to their .pan files?

Hi Dennis,

1. SAM uses module parameter data provided by the California Energy Commission ( www.gosolarcalifornia.ca.gov/equipment/pv_modules.php ). NREL does not maintain that database, so your question would have to be directed to them. Note also that SAM does not use PVsyst .pan files, although it does come with tools for importing data from .pan files.

2. We do not have immediate plans to add IAM, spectral, or LID effects to SAM's module model. For some discussion of enhancements we are planning, please see these discussions on the SAM GitHub pages:

github.com/NREL/ssc/issues/24

github.com/NREL/ssc/issues/155

github.com/NREL/ssc/issues/205

3. Again, NREL is not involved in collecting or maintaining equipment parameter databases, so I don't have any suggestions. The PV Performance Modeling Collaborative (PVPMC) is a forum for discussion of such issues that we do participate in, you can learn about events they are holding and more at pvpmc.sandia.gov/resources-and-events/events/ .

Best regards,
Paul.

Hi Dennis,

The parameters that SAM's "CEC Module Model" uses to model reflection losses are n, K, and L for the Physical IAM model described in pvpmc.sandia.gov/modeling-steps/1-weather-design-inputs/shading-soiling-and-reflection-losses/incident-angle-reflection-losses/physical-model-of-iam/ .

For spectral mismatch, it uses the five alpha parameters from the Sandia Array Performance model described in pvpmc.sandia.gov/modeling-steps/2-dc-module-iv/effective-irradiance/spectral-mismatch-models/ .

For LID, SAM applies a DC loss factor to derate the nameplate power, and that can be provided as a single percentage of nominal DC power.

Best regards,
Paul.