PSA: Solar panels generate less power in atmospheres

[amorymeltzer]

The panels clearly lose efficiency as they heat, much more than I expected in fact.

Nice work! No more rep to give...

[Brotoro]

Solar panels SHOULD generate less energy under an atmosphere that in orbit, just because the atmosphere is going to absurd/scatter some of the light. Good job, Squad! They should also generate less when the sun is at very low elevations above the horizon...but the decrease to 6.14 seems like a bit hit there (but I haven't calculated the air mass).

- - - Updated - - -

OK...The air mass is about 5.7 times what you have straight overhead when the sun is 20 degrees above the horizon, so I guess the results seem reasonable given the loss from one airmass (the difference OP found between space and noon surface values).

[Ival70]

Interesting. I wonder then if solar panels work with indirect lighting, such as immediately after sunset/ before sunrise.

[The Yellow Dart]

That looks good bitbucket, and the variations are small enough that they could easily be attributed to temperature.

The variations in the Mun test could also be the result of a change in distance to the Sun. Since they were different times of "Mun day" we know that it moved about half way round it's orbit from morning to evening. That could be enough to change the energy flow as well.

[Volix]

Interesting. I wonder then if solar panels work with indirect lighting, such as immediately after sunset/ before sunrise.

I noticed they still worked when the sun was behind the mountain range to the west . By that point it was around 5% of the noon value though. In the morning they didn't start working until the sun was actually visible.

[Spatzimaus]

The variations in the Mun test could also be the result of a change in distance to the Sun.

Kerbin's orbit around the sun has a semimajor axis of 1.3E10m, with zero eccentricity. Mun's orbital semimajor axis around Kerbin is 1.2E7m, again with zero eccentricity. So, the closest point and furthest point are only ~0.2% different in distance, meaning a 0.4% discrepancy in insolation between opposite parts of the orbit. (This'd also change the temperature of illuminated objects by 0.1%, or about 0.3 degrees Celsius, but that's not really relevant here.) Of course, we're only talking about half that difference since it'll be nighttime for the other half of each orbit, so the distance change would explain a ~0.2% difference in solar panel power output for a body in vacuum, assuming no energy is being reflected from the Mun itself. That's not quite enough to explain the observed disparity, although it's on the same order.

(Yes, I used to teach radiative astrophysics, but this is well below that level.)

I noticed they still worked when the sun was behind the mountain range to the west

Not surprising. Finding out if there's an unobstructed line of sight between two arbitrary points is not an easy process, since you basically have to move along the line checking terrain heights at each point. Sure, most of the line will be far, far above any terrain, but those aren't the parts you worry about. The accuracy of the answer will depend on how finely detailed the terrain mesh is (since the terrain is just a series of polygons connecting a grid of discrete data points), as well as how large a step size you use on the LOS check. I'd guess they're erring on the side of caution on either end, only turning the sunlight "off" when they're SURE it's obstructed, and only turning it "on" in the morning when they're sure it's visible again. Anything more than that would probably use up too much processor time to be worth the effort, given how rarely you'll actually care about those situations. That's not to say that you'll never care; I just landed a small probe almost directly on Duna's south pole, where even minor variations in the Sun's height above the horizon will change whether or not the surrounding hills block the light. But that's a rare circumstance, so I accept the headaches it might cause.

(Yes, I used to develop simulation software involving terrain models.)

------------------

And yes, it's a lot of fun having these conversations about KSP. Where else can you learn stuff AND blow up things at the same time? Besides Detroit, I mean. I worry a bit that the new aero model will make the game a bit less accessible as a learning tool for schools, as it makes the game quite a bit less forgiving, but in general KSP is still an excellent way to learn physics.

(Yes, I used to live in Detroit.)