Angular drag in a vacuum?

[Dave Kerbin]

I'm trying to understand something about the physics simulation. There are two parts, the radial parachute and the goo container. Both have a mass of 0.15, yet in game it behaves as if the radial parachute has more mass.

First some experiments to try:

• Build a ship with a radial parachute on one side and a goo container on the other. Under thrust the ship will tip toward the side with the radial parachute (even with moderate SAS torque).
  
• Build a tower out of girders and put radial parachutes all down one side and an equal number and placement of goo containers on the other. No matter the alignment it will always tip over toward the parachutes as soon as it spawns.
  

I looked into the part files and the only difference I could find was angularDrag. The goo is 2 while the parachute is 3. This would explain the ship tipping, the parachute is creating more drag and pulling that side of the ship down. It might also explain the tower, perhaps when it spawns the small movement in which it 'settles' is enough for the imbalance in drag to tip it over.

But if you continue flying the ship out of the atmosphere it gets weird. Even without an atmosphere the ship still wants to tip toward the parachute when under thrust. Even if you first use time warp to remove all rotational velocity as soon as thrust is applied the phantom mass imbalance kicks in.

Can someone help me understand what is going on? Could the mass center of mass for the parts be different (meaning the 0.15 is farther away from the ships center on one side, moving the ships center of mass in that direction). The parachute is much closer yet it seems to be the one shifting the mass toward it.

[LethalDose]

[...]

Can someone help me understand what is going on? Could the mass center of mass for the parts be different (meaning the 0.15 is farther away from the ships center on one side, moving the ships center of mass in that direction). The parachute is much closer yet it seems to be the one shifting the mass toward it.

This would be my guess: the center of mass is in a different location. I'm not sure how CoM of a part is calculated. I think it's at each part's (0,0,0) point, and the node attachment field would essentially give the "radius" of the mass from the attachment.

It could also be that the point of attachment is somehow different between the two parts, so, even though it looks like you have them attached at the same y-level, they're not, and 'chute has a longer level arm. From what I understand, the appearance of the part isn't necessarily indicative of what the part's code says about it's dimensions (even though ideally it should be).

[SpaceSphereOfDeath]

I have a similar experience, I found that the infiglider bug actually works in a vacuum, very slowly, in a 100km orbit every 2 flaps changed the periapsis by 1m, but I made the orbit elliptical pre-hand so the floating point numbers couldn't be effecting this it had to be the flaps. I think the game, instead of saying the drag=0, it approximates the drag as =0, but it actually turns out slightly more due to errors.

[numerobis]

Angular drag in Unity is drag against angular momentum. It's not going to cause you to spin -- quite the contrary, it will dampen any tendency to spin. That said, I'm not convinced that changing that number changes anything.

I'm almost certain that lift surfaces do absolutely nothing in the vacuum. However, it's definitely the case that your orbit can change as you spin around. I've heard that be attributed to your orbit being set by the location of your cockpit. At the same time, you can change your orbit by spinning around and transferring fuel back and forth (because angular momentum isn't maintained when you transfer fuel).

[Philoman]

Like when Apollo 13 had to move things from the LEM to the CM because of the weight balance, so they would stay on the right trajectory?

[Dave Kerbin]

With more experiments it seems to be the center of mass is the issue, the difference in angular drag is likely a red herring (unless it has some effect on a part's center of mass)

The goo container's center of mass seems to be very close to where it attachs to the side of the ship. By comparison my experiments keep showing the radial parachutes center of mass to be well outside of the model - experimentally it seems like it is between 3 and 4 times it's own thickness away from the attachment point (if you floated a dot where the parachute's center of mass was a kerbal could safely stand in the space between the parachute and that dot).

[Thomas988]

The wonders of an alpha.

[numerobis]

The goo container's center of mass seems to be very close to where it attachs to the side of the ship. By comparison my experiments keep showing the radial parachutes center of mass to be well outside of the model - experimentally it seems like it is between 3 and 4 times it's own thickness away from the attachment point (if you floated a dot where the parachute's center of mass was a kerbal could safely stand in the space between the parachute and that dot).

Interesting... That kind of makes sense for establishing the center of drag when the parachute is deployed. Not so much when it's stowed.

[SRV Ron]

Parking that parachute as high as possible on the capsule will help a lot with the COM issue. (One part ship)

[KerbMav]

Interesting... That kind of makes sense for establishing the center of drag when the parachute is deployed. Not so much when it's stowed.

And can wreak havoc during launch?!

Would explain some summersaults when launching crafts with radial chutes.

[Tw1]

I think it's at each part's (0,0,0) point, and the node attachment field would essentially give the "radius" of the mass from the attachment.

It is the 0,0,0 point, of the first game object in that parts hierarchy in Unity. Or something like that; I'm not an expert that's what it seems to be.