Fuel tank rotational damage

[lajoswinkler]

In real life, rockets don't allow fast rotation around prograde vector. There were several occasions during the history of launches when a rocket was rotating too fast and the fuel tanks disintegrated.

There was this video that shows one of the early US launches where the rocket starts to spin and it doesn't take a lot before its tank section blows up. It wasn't rotating fast at all. I can't find it at the moment.

Why do they fail? Because they aren't perfectly radially symmetrical. When they rotate, the centrifugal force is trying to rip them apart. Resonant vibration building can form, either in the form of the internal structure swaying, or even fuel sloshing.

Liquid propellant fuel tanks, in general, aren't designed to withstand significant lateral forces. I think that should be implemented in KSP.

So this would mean worse conditions for launches and it would require more patience during any rotation in orbit, but not much. It would only introduce problems to begginers. It's very easy to stop rockets from rotating.

[Kyarao]

So far as I know, the physics engine handles destruction by breaking and flexing connections. Whole parts are only destroyed when they are not connected to anything. This is a coding optimization Squad seems to have done to reduce the physics load.

When your spaceship smacks into the ground, parts break apart first then (maybe as short as 1 game step later, not even a whole frame) the part itself may explode when sufficient force is applied to it.

This causes problems when parts hit water, since they break when they bounce back up, instead of breaking when they hit the water. Small difference, so it doesn't really matter.

[lajoswinkler]

In real life, fuel tanks are sensitive to lateral impacts, but implementing this would require more work and it would possibly create more lag in the game.

Instant destruction of a fuel tank after it reaches a certain speed of rotation would not be very complicated. For example, you can make a small lander or a rocket and turn it around at incredible speeds and nothing happens. That should be impossible.

[Kyarao]

if (ship.angularVelocity > tolerance)

{

foreach Component i in ship.partsList

{

if (i.getType == fuelTank && i.fuelLevel > spinningFuelLevelTolerance)

{

if (i.numConnections == 0)

{

PhysicsEngine.destroyPart (i)

}

else

{

PhysicsEngine.breakConnections (i)

}

}

}

}

Psuedocode thrown together. No optimizations. Doesn't look hard to implement at all.

I see a problem with it though. If a fuel tank is distant from the axis of rotation, it may have a small angular velocity around its own axis, but a large velocity around the ship's axis. Should this still cause it to explode?

[lajoswinkler]

There aren't many rockets that have lateral tanks, unless we include boosters. I don't remember any accident with such scenario so I'm not sure what would happen. I think the greatest stress would be at the points where the tank is attached to the central body.

Youtube has several videos showing early rocket test failures that show rockets disintegrating during rotation and even inclination from the prograde vector, if they aren't very high. They look like they were made by clumsy kids, with all their nosecone detachments and falling back to the launch pad, yet it's all because of primitive stability systems and building of resonance vibrations.

Basically, KSP is too generous when it comes to wild launches. You can make one hell of a rodeo with a rather large rocket and nothing will happen.

[Temstar]

There was this video that shows one of the early US launches where the rocket starts to spin and it doesn't take a lot before its tank section blows up. It wasn't rotating fast at all. I can't find it at the moment.

I know what video you're talking about, it's one of the Little Joe II launches that was to be a test for the Apollo launch escape system:

https://www.youtube.com/watch?v=AqeJzItldSQ

It was wired up wrong which caused the fins to go hard over and stuck in clockwise roll position.

[ChrisDaBox]

So this would mean worse conditions for launches and it would require more patience during any rotation in orbit, but not much. It would only introduce problems to begginers. It's very easy to stop rockets from rotating.

I see where you are coming from but adding this would make spaceplanes even more hard to build. They have to be able to turn easily. And also the amount of times my Space-Plane has flipped out in mid-air is a lot.

[Dunbaratu]

The problem is that to implement this in the physics engine would require modeling the fuel tank as multiple part elements meshed together (i.e. like calling, say, each 45 degrees arc around the cylinder a different panel, and then making up a tank out of 8 such parts glued together.) Doing this would then allow calculations about the stresses the connections between parts endure as the thing spins. But the reason this is a problem is that having a higher part count is the LAST thing this game needs. It's already maxing out what peoples' PC's can calculate.

In order for the physics engine to calculate the strength holding a chunk of material together it has to model the solid object as several separate elements glued together and calculate the stresses on the joins between them.

In fact, if you make a rocket body out of several separate tanks around a center, the centrifugal force already IS calculated. To test this use the 4-way splitter connector to make a rocket fuselage out of 4 clustered tanks, then DON'T strut them together, launch and hold "Q" or "E" down to spin faster and faster. It actually WILL separate the clustered tanks apart and bust the rocket that way. So the calculations you're talking about are already handled by the physics engine, BUT the only reason they don't trigger for central-axis tanks is that the entire tank is just one atomic part in the calculation which is located with center of mass ON the axis of rotation. ("Atomic" in the computer science meaning, not the nuclear physics meaning).

[lajoswinkler]

Yes, it was one of the Little Joe II launches.

I'm not saying developers should implement more parts so that we see the destruction like in the video. That would significantly increase lag. I'm just saying the tanks should explode (and vanish, just like they normally do in KSP) when angular velocity (or frequency of rotation) reaches a certain value.

You can see from the video that we're talking about significant values. The frequency is roughly 1 turn in ~2.5 s, therefore 0.4 Hz. That's a lot.

I can't remember when was the last time I had significant rotation around the prograde axis. Sometimes I get a lot of swaying, but that's it.

KSP respects lots of real world physics, so it's weird that it allows the the players to rape their vehicles by violently turning them around. There should be more patience in some segments of the whole gaming experience. (Come to think about patience, those jetpacks are too violent, but we already have a thread for that...)

[Insomn]

I think the issue becomes how much realism you put into a game before it is no longer fun to play. I mean if it were 100% realistic then we'd all be qualified to work for NASA. And maybe somebody should make an x-plane for space. But not our fun kerbal game.

[thereaverofdarkness]

I prefer having the huge difficulties in real space travel eliminated from the game. What other reason is there to play this rather than to just build real rockets?

[CosmicX1]

I like the idea of this! It would also encourage people to use the rotation stabilising reaction wheels more!

[GusTurbo]

I kind of like this idea, though it might be really frustrating. It would certainly give an incentive to design rockets that don't spin violently on launch (like some of mine have done in the past). It might be a good option to have as an additional difficulty setting -- I like the idea of multiple difficulties to make the game more or less realistic/rage-inducing.

Also, this seems like something that would be fairly easy to create as a mod. It could be called "Deadly Rotation."

[lajoswinkler]

I think the issue becomes how much realism you put into a game before it is no longer fun to play. I mean if it were 100% realistic then we'd all be qualified to work for NASA. And maybe somebody should make an x-plane for space. But not our fun kerbal game.

There is a limit, obviously, but my proposition is way below the limit. You saw what those KSP rockets can do. You can flip them around like toys. It removes the gracefulness of movement in orbit, as well as realistic (and not too heavy) difficulties at the launch. Sometimes the speed of deliberate rotation is so high that even with precision mode turned on, it's annoying.

If my rockets rotate, they make a total of three spins during the whole launch. I've only had crazy rotation during the first day of playing because I've quickly learned what SAS does.

The only people affected by this would be total begginers and people who like to spin their rockets like crazy. (coughjebcough)