search

NREL / bifacial_radiance

Toolkit for working with RADIANCE for the ray-trace modeling of Bifacial Photovoltaics
https://bifacial-radiance.readthedocs.io
BSD 3-Clause "New" or "Revised" License
90 stars 48 forks source link

Problem with scene size? Bifacial ratio is too high #80

Closed dvalencia00 closed 5 years ago

dvalencia00 commented 5 years ago

Hi, I have run several simulations including 1-axis and 5 rows. The only difference between them is the number of modules (nMods=13,26,52). The bifacial ratio for the two firsts seems ok (6-7% aprox), but the result for nMods=52 is 32.76%, which is too high! Could be related with the size of the scene? Modules are 1m width in portrait, that means trackers are 52 m long. Below you can see results from csv and all the inputs.

imagen

gcr= 30, albedo= 20, nMods= 52, nRows= 5, fullYear= True, module_width= 1, module_height= 1.98, bifi= 1, orientation= 'portrait', tilt= 0, azimut= 180, numpanels = 2, panelgap= 0.4, psx= 0.02, hub_height= 2, torque_tube= False, limit_angle = 55, angledelta = 5, backtrack = True, sensorsy = 9, modwanted= None, rowwanted= None

dvalencia00 commented 5 years ago

Ok, I have just realized that for some reason the material is not correct... It seems that the sensors are in the middle of nowhere, not linked to the proper surfaces

cdeline commented 5 years ago

Yes, it appears that your sensors are hitting the ground and sky - it should be a material name like a2.2.a0.PVmodule.6457 for both the front and backside, otherwise the rays are passing through your scene.

One interesting point - for GCR, we typically use a fractional value like 0.3. Perhaps this could be the issue?

cdeline commented 5 years ago

Albedo is also typically a fraction like 0.2

cdeline commented 5 years ago

@dvalencia00: Have you figured out more information on this issue? If you want to forward me a full copy of your script, I could try to run it and debug. Otherwise, I'm going to close this.

dvalencia00 commented 5 years ago

Althought I wrote albedo=20 and cgr=30, then I trsnform them to 0.2 and 0.3 respectively. It is extrange because I have done many more simulations just changing the number of modules, and they seems fine, even with higher number of modules. However with 52-53 modules they fail. I send you the files by email.