Unintuitive behavior of active radiator panels?

[Snark]

This post is really more of an observation than a question-- the observation is "here's how radiators behave in a surprisingly counterintuitive way", and I guess the implied question is "what are some tips and tricks for dealing with solar heat".

Background context: I've just started gearing up for close-to-the-sun operations (I'm running New Horizons, and it has some planets orbiting toe-scorchingly close to the sun). This means I'm having to deal with extreme solar heating for the first time, and I have to say, I find myself A) disappointed with the mechanics, and wondering what tips and tricks I may be missing.

The thing I've run up against that surprised and disappointed me: The behavior of active radiator panels (the ones that auto-rotate to keep themselves edge-on to the sun).

What I expected: That they do a good job dumping heat when they're edge-on to the sun. That they would absorb heat when they're face on to the sun.

What I found: It doesn't matter how the panels appear to be facing (i.e. when they auto-rotate, it doesn't matter which way they're facing relative to the sun). Their orientation does matter, but not their actual visible orientation. Instead, what matters is their original orientation before deploying, i.e. how they're mounted on the ship. The auto-rotation doesn't do anything.

Here's a test ship that I put together, with panels deployed and retracted. Note that they're arranged so that when retracted, the long axis of the folded panels is parallel to the long axis of the ship. (Ignore the engine, it's from the Atomic Age mod, and isn't relevant to this problem-- I wasn't running it when I made these shots, and it had no residual heat. This is a stock-KSP question.)

Here's what I observe when I turn on thermal info on the debug menu, and then try orienting the ship so that its long axis is pointing straight at the sun (left image) versus when its axis is perpendicular to the sunlight (right image). Note the radiation flux pointed to by arrows; green arrows indicate "good" values (i.e. what I'd expect), the red one indicates a bad/surprising value.

When the ship is pointed directly at the sun, all four radiator panels are doing exactly what I'd expect: they have big negative radiation flux, meaning that they're dumping lots of heat (rather than absorbing it).

On the other hand, when the ship is oriented perpendicular to sunlight: two of the panels have big negative radiation flux (i.e. are dumping heat), but the other two have big positive flux (i.e. they're absorbing heat rather than radiating it).

Furthermore, one thing I observe from watching the live interaction (which doesn't come through well in these static screenshots) is that the radiation flux depends on the ship's orientation, not the panel's-- that is, if I suddenly change the ship's orientation and the panels haven't had a chance to auto-rotate yet, the flux numbers don't change much as the panels auto-rotate. The flux numbers change with the rotation of the ship.

In other words: The way the radiator panels are behaving is exactly what you'd expect to see if they were simple, fixed panels that are parallel to the long axis of the ship, arranged radially, i.e. as if they looked like this:

...and no auto-rotation makes a whit of difference. If I orient my craft towards the sun, it has a good day. If I don't, it gets barbecued.

<rant> I have to say, I'm disappointed. It would be nice if they could make the panels adjust their radiation/absorption based on the auto-rotation. But even if they can't do that because it's hard for some reason, then at least don't lie to the player by making the panels appear to rotate. Why not just have them be simple, fixed, non-rotating things that just extend/retract? Then the actual behavior would be completely consistent with what the player expects from what they see on the screen. </rant>

Anyway, ranting aside, chalk that one up to experience, I guess. So I've learned: "Ignore the auto-rotation, and always arrange all panels so that their "stowed" axis is parallel to the ship's, and keep the ship's axis parallel to sunlight."

Does anyone have any other tricks for dealing with solar heating? Or is the above moral pretty much it?

Edited October 10, 2015 by Snark

[Randazzo]

Can't see the orientation of the panel with the "bad" value.

[Snark]

Can't see the orientation of the panel with the "bad" value.

Sorry, it's cut off (I didn't want to make the screenshots bigger). It's oriented "properly" for autorotation (edge-on to the sun)-- the orientation is the same as its opposing twin, the radiator you can see pointing "down" at the bottom of the right-hand picture.

Here's some more detail so you can see its orientation. Note that it's visually edge-on to the sun, but it has a big positive rad flux (indicating that it's absorbing rather than dumping heat).

Edited October 9, 2015 by Snark

[Randazzo]

Sorry, it's cut off (I didn't want to make the screenshots bigger). It's oriented "properly" for autorotation (edge-on to the sun)-- the orientation is the same as its opposing twin, the radiator you can see pointing "down" at the bottom of the right-hand picture.

Here's some more detail so you can see its orientation. Note that it's visually edge-on to the sun, but it has a big positive rad flux (indicating that it's absorbing rather than dumping heat).

http://i.imgur.com/vT2F2k2.png

That is odd. Were all the other panels normal except that one?

I can tell you from experience that rotating edge on to the sun makes a difference, much more so as you get closer to the Sun. Anything with a high emissive constant with a large surface area facing the sun heats up tremendously fast by comparison.

I'm not saying these are working properly, as that one clearly isn't, but it's not because you're being lied to.

Edit: Also, you are correct that in most cases the visual representation doesn't matter because the raycast transform is a separate invisible object parented to the rotation axis, but in the case of the stock radiators the transform is set to one of the actual panel models, so what you see is indeed what you get in their case.

Edited October 9, 2015 by Randazzo

[Snark]

That is odd. Were all the other panels normal except that one?

The two panels that were pointing directly toward and away from the sun were normal.

That panel and its opposite twin (i.e. the ones pointing "up" and "down", i.e. celestial north and south) were acting "badly" i.e. absorbing heat rather than radiating.

I can tell you from experience that rotating edge on to the sun makes a difference, much more so as you get closer to the Sun. Anything with a high emissive constant with a large surface area facing the sun heats up tremendously fast by comparison.

Agreed. Since making the above-described discovery, I've spent some time tinkering/experimenting, both at "normal" orbits (where solar heating isn't extreme) and getting up-close and personal with the Sun. This includes looking not only at these panels, but also at other objects, too, such as fuel tanks, science instruments, etc.

Leaving the auto-rotating panels aside, my general observations are that it behaves as you would expect:

1. Long skinny things (big in one dimension, small in the other two), like Jumbo-64 fuel tanks, work best (as you would expect) with their long axis pointing at the sun, so that they have minimal area exposed.

2. Flat things (big in two dimensions, small in the other), like structural panels, work best (as you would expect) when they're edge-on (in any orientation) to sunlight, and are worst when their broad face is pointing right at the sun.

3. Approximately-spherical things (all dimensions roughly the same size) don't matter much which way they point, as you would expect.

All of that makes perfect sense and is what the player would intuitively expect.

However, consider another set of examples to illustrate what I've observed. This uses a different ship that arranges the panels a bit differently. In all of these examples, the sunlight is coming from directly behind the camera, shining straight at the ship.

These are a set of views of the same ship, just oriented differently relative to the sun.

First, here it is pointing celestial north, with its panels aimed east and west. "A" is before the panels have had a chance to auto-rotate, so they're face-on to the sun. "B" is a minute later, when they've auto-rotated to be edge-on.

Now here it is rotated 90 degrees around the ship's axis. Still pointing celestial north, but the panels are now pointed "up" (away from the sun) and "down" (towards the sun):

Now here it is pitched down so that the ship is pointing directly "down" (nose of the ship aimed at the sun):

Assuming that the panels work as a player would intuitively expect, what do you suppose the behavior should be in each image? Here's what I would think:

A. The panels are face-on to the sun, so they should be absorbing heat.

B. The panels are edge-on to the sun, so they should be dumping heat.

C. The panels are edge-on to the sun, so they should be dumping heat.

D. The panels are edge-on to the sun, so they should be dumping heat.

Right?

Except that that's not what it does. The expectations for A, C, and D above are correct, and the game does what you'd expect. In A, I observe that all six panels are absorbing heat. In C and D, I observe that all six panels are dumping heat.

However, the problem is that B is behaving wrong. In B, all six panels are absorbing (not dumping) heat. In fact, case A and B behave identically in-game. I can start up case A, and watch the thermal flux as they rotate and the ship gradually transitions to B. Result? No change! The auto-rotation of the panels makes absolutely no difference to the radiation flux.

I'm not saying these are working properly, as that one clearly isn't, but it's not because you're being lied to.

Essentially, the ship is behaving as if the panels are glued in the same position you see in example A-- regardless of how the ship is facing, regardless of what the auto-rotated position of the panels is. In other words: The auto-rotation is a lie, and the game is lying to the player. You have to ignore what your eyes are telling you and think of the panels as they are in position A.

(Okay, maybe my choice of the word "lie" is a bit strong, but I'm pretty darn frustrated after investing a lot of engineering effort and flight time into a mission that doesn't work because of this problem. But the fact is, the game is putting in auto-rotation as "eye candy" that not only doesn't do anything, but is actively misleading me because it gives counterintuitive results.)

If there's some software-architectural reason why they can't make case B work as expected, then I'd much prefer that they simply turn off auto-rotation on those radiator models. If the ship always acts like A, then it should look like A. I'm as big a fan of jazzy visual f/x as the next guy, but not if it actively misleads the player.

Edited October 9, 2015 by Snark

[Specialist290]

I don't know how accurately the game is set up to model heat radiation compared to the real world, but I've noticed one problem with configuration B in the above screenshots: Your middle panel would essentially be useless, because the heat from the opposing faces of the other radiators is being dumped right back onto it, and in turn almost any heat it radiates is also being radiated back onto the fore and aft radiators.

EDIT: Nevermind, that can't be the case, because otherwise the same thing would be happening in configuration C.

Edited October 10, 2015 by Specialist290

[Randazzo]

Well, it does appear to be an issue with the large stock radiator. I didn't test the smaller stock ones.

I added a second cfg file to the part directory with the sun tracking disabled so they wouldn't rotate at all and compared them to standard ones. The results were the same as Snarks.

Then I did the same with the Radiators from Heat Management. Thankfully they seem to perform as expected, though the side panels with the long edge facing the sun were radiating less than the center one with only the narrow edge facing. At no point did I use their active heat dissipation for the sake of the test.

[Snark]

I added a second cfg file to the part directory with the sun tracking disabled so they wouldn't rotate at all

Ooh, that would be perfect-- I'd love to do that myself. Could you share the config?

If I could turn the auto-rotation off so that at least the behavior would be WYSIWYG, that would remove a significant annoyance.

[edit] Ah, never mind, figured out how to do it myself with ModuleManager. First time I've authored a MM patch, never had a reason to learn before.

All I had to do was add a .cfg file of my own, which contains the following snippet (this just tinkers with the big radiators, but could do it with the smaller ones, too).

@PART[foldingRadLarge]
{
    @MODULE[ModuleDeployableRadiator]
    {
        sunTracking=false
    }
}

Edited October 10, 2015 by Snark

[Wintersdark]

Has anyone tested this with the smaller stock radiators?

[KerikBalm]

I heard an earlier report, using the smaller ones, that the panels just increased heat absorption.

They concluded that radiator panels only helpif your ship is hotter than the "ambient temperature", otherwise they warm the ship up.

My guess is that they were testing as in condition B... I'm inclined to think this is general for all the sizes

[Snark]

Has anyone tested this with the smaller stock radiators?

Yes, I've since tested with the smaller radiators. All of the auto-rotating radiators show the same behavior: Their rotation is a sham. It's just eye candy, it doesn't change their radiative behavior at all. They act as if they don't rotate at all.

I couldn't stand the visual deception (it bugs me that what I see is not what I get), so I've added a little config file to turn it off (see my earlier post above, I just added entries for the smaller radiators as well).

I've been playing KSP for a long time, but never noticed this problem before because I never needed radiators for solar heating before-- I only used them to dump heat that the ship generates, e.g. when I've got LV-Ns. And unless you're flying really close to the Sun, it doesn't really matter much-- you're not going to explode from solar heating.

It becomes an issue now because I am exploring close to the Sun, and it's a total make-or-break situation. The orientation of the ship really matters, and it's vitally important to make sure that the radiators are facing in a direction where they're dumping heat rather than absorbing it.

[Jovus]

I would be surprised if there were some technical reason that auto-rotation can't behave properly, because it seems to do a plenty fine job with solar panels. To me this sounds like a bug, and you should report it.

[Streetwind]

Yes, absolutely do bug-report it. NathanKell is working on bugfixing the thermal system right now for 1.0.5, and RoverDude (who made the radiators) is also still involved with things, so there's no better moment to bring more thermal bugs to Squad's attention!

And thank you for this whole thread and for testing this stuff, +rep to you. I'll definitely bookmark this so we can check this case when testing Near Future radiators

[Snark]

I would be surprised if there were some technical reason that auto-rotation can't behave properly, because it seems to do a plenty fine job with solar panels. To me this sounds like a bug, and you should report it.

Thanks, that's a great idea! Here ya go:

http://bugs.kerbalspaceprogram.com/issues/5540

[Edit] Hmm. I created the issue and naively gave it a priority of "High". Having done so, and looked at other bugs people have logged, it occurs to me that perhaps it should be "Normal". However, I can't seem to find any way to edit the bug I made, now. There's an "Edit" link on the page, but if I click it, all it lets me do is change the status and add a note-- I can't actually edit the bug text, or set the priority, or do anything else to it. Is this by design? Am I missing something?

Edited October 13, 2015 by Snark

[NathanKell]

Note: We're aware of the issue and working on it.

[Snark]

Note: We're aware of the issue and working on it.

Sweet! Great to know, thanks for the quick response!

[hermes47]

[Edit] Hmm. I created the issue and naively gave it a priority of "High". Having done so, and looked at other bugs people have logged, it occurs to me that perhaps it should be "Normal". However, I can't seem to find any way to edit the bug I made, now. There's an "Edit" link on the page, but if I click it, all it lets me do is change the status and add a note-- I can't actually edit the bug text, or set the priority, or do anything else to it. Is this by design? Am I missing something?

Dropped it down to normal for you.

[Snark]

Dropped it down to normal for you.

Great, thanks.

How? (Just for future reference, in case I report any more bugs.) Does editing a bug after it's been logged require some sort of special access level that I don't have? Or is there some UI feature that I'm stupidly blind to?

(I can see why editing someone else's bugs would need some sort of admin privilege, but seems kind of odd that I wouldn't be able to edit one that I reported myself.)

[hermes47]

How?

Magical QA powers. I'm not actually sure what the general public perms are like in regards to editing bugs. For that you'd have to ask Ted probably.