Identify surface orientation from AC power and clearsky irradiance #77
Conversation
Implements method for identifying tilt and azimuth given ac power, clearsky irradiance, and solar position.
Add entry to the list of functions for orientation inference in api.rst and fix minor issues in the docstring.
Test with missing data and with Series for temperature and wind_speed parameters.
Thinking about this some more, the capacity values are relatively sensitive to temperature model/coefficient, AOI model, and spectral model, while orientation is largely independent of those choices. Expanding the API to include all of those options could be a real mess, so perhaps a good idea to restrict the scope. |
|
I originally restricted to only the tilt and azimuth because I didn't know if the other parameters would be accurate (sounds like they might not be). There is, however, evidence in the literature that this type of approach can work for tilt and azimuth (Haghdadi N; Copper J; Bruce A; MacGill I, 2017, 'A method to estimate the location and orientation of distributed photovoltaic systems from their generation output data', Renewable Energy, vol. 108, pp. 390 - 400, http://dx.doi.org/10.1016/j.renene.2017.02.080). I would like to eventually push towards the method described in that paper, and this function seems like a good first step in that direction. |
Need this for pvlib.inverter.pvwatts()
Keeps the names consistent within the system module.
| Series | ||
| Difference between `power_ac` and the PVWatts output with the | ||
| given parameters. | ||
| """ |
There was a problem hiding this comment.
| """ | |
| Notes | |
| ------ | |
| Uses the defaults in `pvlib.irradiance.get_total_irradiance` to calculated plane-of-array | |
| irradiance, i.e., the isotropic model for sky diffuse irradiance, and the Perez irradiance | |
| transposition model. | |
| """ |
There was a problem hiding this comment.
@cwhanse I don't think this code is actually using Perez. Is that something we should update? Seems like a shame to deviate from PVWatts in a way that will presumably make the fits worse.
cc @kperrynrel since she's getting bad fits :)
There was a problem hiding this comment.
", and the Perez irradiance transposition model." That text is in error and should be deleted.
There was a problem hiding this comment.
Alternatively, should we switch the code to use Perez? I think the extra inputs could be calculated automatically. It would be more aligned with PVWatts itself, plus hopefully return better results.
There was a problem hiding this comment.
I don't object, but on this second look we should also clarify what is meant here by 'PVWatts'. It's not referring to the PVWatts application (modelchain), but rather to the PVWatts DC and AC power functions. To that point, this pvanalytics function uses the Sandia model to get cell temperature.
If we decide we want to match PVWatts application, maybe we should use the pvlib.modelchain.ModelChain.with_pvwatts.
There was a problem hiding this comment.
Fair point. Note that ModelChain.with_pvwatts currently uses the Sandia temperature model too: https://github.com/pvlib/pvlib-python/blob/master/pvlib/modelchain.py#L49-L59
I'll open an issue.
There was a problem hiding this comment.
@kanderso-nrel let me know if you need help with these edits. I put in a PR for the system documentation but I'll hold off until we resolve the issues with the functions themselves.
Co-authored-by: Cliff Hansen <cwhanse@sandia.gov>
Description
Implements method for identifying tilt and azimuth given AC power, clear sky irradiance, and solar position.
Checklist
docs/api.rst