[FEAT] Back irradiance non-uniformity model for `pvlib.bifacial.power_mismatch_deline`

[echedey-ls]

Is your feature request related to a problem? Please describe.
pvlib.bifacial.power_mismatch_deline requires RMAD input to be usable. Right now, no models in pvlib nor other similar packages provide an easy/direct way to map a system to such values.

Describe the solution you'd like
Implementation of the model described in https://doi.org/10.1016/j.solener.2026.115030 ( https://www.sciencedirect.com/science/article/pii/S0038092X2600719X ) ¹.

Describe alternatives you've considered
Add a example script that mimics the procedure to calculate non-uniformities by pvlib.bifacial.power_mismatch_deline reference http://dx.doi.org/10.1002/pip.3259 ².

Additional context
The authors open sourced an implementation at https://github.com/WildSolar/General_PV_Irradiation_Model.

Mailing list report that could benefit of this feature at https://groups.google.com/g/pvlib-python/c/5g4OKvMc0Y8.

Footnotes

1. M. Wild, S. Strebel, D. Anderegg, and J. Rohrer, “A general view-factor based irradiation distribution model for multi-row bifacial PV arrays on sloped and reflective ground,” Solar Energy, vol. 318, p. 115030, Nov. 2026, doi: 10.1016/j.solener.2026.115030. ↩

2. C. Deline, S. Ayala Pelaez, S. MacAlpine, and C. Olalla, ‘Estimating and parameterizing mismatch power loss in bifacial photovoltaic systems’, Progress in Photovoltaics: Research and Applications, vol. 28, no. 7, pp. 691-703, 2020, DOI: 10.1002/pip.3259. ↩

[kandersolar]

I'm not opposed to adding more bifacial irradiance models, but it's not clear to me that the stated motivation here is correct:

Right now, no models in pvlib nor other similar packages provide an easy/direct way to map a system to such values.

The recently added ants2d model provides irradiance for arbitrary 1-D subdivisions of the front/rear side. Doesn't that satisfy the need here?

[echedey-ls]

Oh, I was out of the loop with ants2d capabilities. Thanks for the heads up. I should have mentioned it in the mailing list thread. In any case, I think this other model, if feasible to implement with a coherent API, may help check consistency of the other models and/or benchmark the computational efficiency.