The Community Aerodynamics Rebalance thread!

[Robotengineer]

This thread is intended to be a central location for the community to discuss and share/improve the physics config, with the hopes of getting a 'one config to rule them all'. I will be adding configs to this page (the OP) as configs are suggested. Provided below are the 1.0 and 1.0.2 configs.

Old values (1.0):

[FONT=monospace]dragMultiplier = 6.0
dragCubeMultiplier = 0.06
liftMultiplier = 0.038
liftDragMultiplier = 0.03
bodyLiftMultiplier = 8
[/FONT]

New (1.0.1/.2), for comparison:

[FONT=monospace]dragMultiplier = 8.0
dragCubeMultiplier = 0.1
liftMultiplier = 0.055
liftDragMultiplier = 0.025
bodyLiftMultiplier = 10.7[/FONT]

Credit goes to Norpo who provided these in this post.

Note: this thread is not for arguing whether or not 1.0 is better than 1.0.2, and vice versa. If you believe that the 1.0 or 1.0.2 configs work for you, then use them, but don't argue their accuracy/value/gameplay/etc in this thread.

[Moss]

You mention that the hope of this thread is to discuss, share and improve the physics config, but you don't want people arguing accuracy or gameplay values in this thread. What sort of input are you looking for?

[Robotengineer]

You mention that the hope of this thread is to discuss, share and improve the physics config, but you don't want people arguing accuracy or gameplay values in this thread. What sort of input are you looking for?

I don't wan this thread to turn into yet another thread about 1.0 vs 1.0.2 aero. If you wish to discuss the accuracy or gameplay value of a community created config, please do.

[BmB]

I have been fiddling with the values myself a bit after being dissatisfied with 1.0.2.

#1 change to make: DragMultiplier - this is way too high by default. They increased it by almost 50% for 1.0.2 even. A high DragMultiplier makes it impossible for Kerbals to die from falling. Since Kerbals' aerodynamics cannot be modified this should be the baseline value. I find that 4.5 is sensible value, there is a small chance for a Kerbal to bounce but overall most impacts with the ground will be certain death, without the terminal velocity being overly fast either. It is a good compromise value. As a bonus this also causes parachutes to behave much more realistically, gradually opening rather than coming to a stop. Having an opening altitude of at least 700-1000m is advisable, and the deployment feels satisfying. Downside: default opening value is always too low.

The default cubemultiplier of 0.06 seems fine, for regular parts this seems to override the dragmultiplier, causing most rockets to fly as you'd expect even with the above dramatic change.

AngularDragMultiplier should probably be lowered a bit, 2.0 seems a bit stiff. 1.6 feels more natural to how a plane should roll.

#2 change to make: LiftMultiplier - With the default value I feel like I can take off anything on a dime. Again this was upped in 1.0.2 causing the soupy feel. I find that instead a dramatic lowering to 0.023 feels reasonably good for most of the airplanes I built. I have no math to back up what this value should be for certain wings, but I also doubt the lift values were balanced that carefully to begin with. Realism is for someone else to deal with. At this level you need to get some speed before the plane will take off on its own, but not too much.

LiftDragMultiplier I set to 0.032, a little higher than the lift. I have no idea what it should be, there's probably some ratio of lift to drag that is common for wings that you could use here but I do not know it. Maybe this is too low or maybe my jet is more aerodynamic than I thought. It seems that you can get up to some ludicrous speeds with this. If in doubt, leave at default. Though the 1.0.2. value almost slows you to a stop even with fast aircraft.

BodyLiftMultiplier seems like it works best when matched with DragMultiplier, 4.5 again. It's gentler than the default but will still flip an unstable rocket even against reaction wheels all the same. Seems to have virtually no effect on airplanes no matter how high you crank it.

I also recommend setting the convection values back to 1.0 levels and using the ablator heatshield fix, so that re-entry damage is actually possible again.

Final values look like this:

dragMultiplier = 4.5
dragCubeMultiplier = 0.06
angularDragMultiplier = 1.6
liftMultiplier = 0.023
liftDragMultiplier = 0.032
bodyLiftMultiplier = 4.5

[Aanker]

For the body lift, did you try Mk 2 parts? Those are the liftiest airplane fuselages.

[klgraham1013]

I wish I knew more about this subject. Honestly, I will probably just trust in Ferram and play with nuFar. I really wanted to give Squad the benefit of the doubt, but the sudden change in 1.0.1 really raised my skepticism. I'll, of course, keep reading what the community thinks as time passes.

[BmB]

For the body lift, did you try Mk 2 parts? Those are the liftiest airplane fuselages.

Yes that is exactly what I was testing with. The body lift seems to only come into play at extreme turns. It does seem that those parts have wing lift also, which is not affected by the body lift value of course.

Also on further testing the 0.032 liftdrag value seems to be exactly right to match the speeds I was getting in 1.0 after all. Lucky guess! 0.036 can work for a bit more of a conservative choice also.

[r4pt0r]

So is the goal to choose the one config to rule them all, the convince squad to adapt that community playtested config as stock?

[regex]

Keep in mind that the more you cut drag to support spaceplanes, the higher terminal velocity for capsules will be. There's a fine balance to be had. I would suggest dragMultiplier = 7 as a good place to start compromising. 6 is good for planes, 8 is good for capsules/blunt bodies.

[BmB]

Keep in mind that the more you cut drag to support spaceplanes, the higher terminal velocity for capsules will be. There's a fine balance to be had. I would suggest dragMultiplier = 7 as a good place to start compromising. 6 is good for planes, 8 is good for capsules/blunt bodies.

The dragmultiplier does not apply to parts, the cubes override it. It only applies to certain things like chutes and kerbals. At the very least the effect on parts is so small as to be pointless to consider. You also cannot edit the kerbal aerodynamics. 8.0 drag = immortal kerbals. In that case, it is preferable to edit every other part than to edit something that cannot be edited. But you don't need to.

[regex]

All I'm saying is make sure capsules have a reasonable reentry. You can't focus entirely on planes.

[BmB]

Most of the values apply to planes. The only value that applies to capsules is the cubemodifier, and the default seem fine there. Maybe the body lift but I can't imagine a situation where that kicks in.

If you prefer the 1.0.2 capsule speed you should use the 1.0.2 cubemodifier of 0.1

For reference, this is the result of an unshielded entry starting at ~1500 m/s using my settings. Seems reasonable to me. I think if it was slowed down it would be too easy to enter without a shield

Edited May 6, 2015 by BmB

[regex]

Most of the values apply to planes. The only value that applies to capsules is the cubemodifier, and the default seem fine there. Maybe the body lift but I can't imagine a situation where that kicks in.

I don't know the ins-and-outs of how things are calculated but you can perform a lifting reentry in the new aero model, not sure if it relies on an actual lift module though, I'd have to check the capsule parts.

For reference, this is the result of an unshielded entry starting at ~1500 m/s using my settings. Seems reasonable to me. I think if it was slowed down it would be too easy to enter without a shield

http://i.cubeupload.com/oJQyJH.jpg

I'll have to test this, especially at Eve.