Wobbly Rockets with David 'Trigger' Tregoning - KSP2 Dev Chat

[regex]

45 minutes ago, Rudolf Meier said:

 ... it's so annoying to read about this without seeing a real discussion about solutions!

This entire thread is filled with suggestions, where have you been? Hell, I offered one up myself. You just stop by to chastise a comment I made you saw in passing?

[Alexoff]

The developers released a video with a chat, in which a specific solution to the problem was not announced. In response, fans of the game created their own chat, where they (and me too) also could not come to a common denominator.

Moreover, it is unlikely that this will somehow affect development. We have no idea about the game code and the developers' capabilities.

[Rudolf Meier]

46 minutes ago, regex said:

This entire thread is filled with suggestions, where have you been?

Well... then I didn't see the suggestions or technical ideas behind what has been said. Sorry... might be, that it wasn't detailed enough for me to understand it.

On the other side, I'm among not many who did present a real implementation (in KSP 1) and nobody seems to care...

[MechBFP]

9 minutes ago, Rudolf Meier said:

Well... then I didn't see the suggestions or technical ideas behind what has been said. Sorry... might be, that it wasn't detailed enough for me to understand it.

On the other side, I'm among not many who did present a real implementation (in KSP 1) and nobody seems to care...

It isn't that nobody cares, it is that we all have no idea what else Intercept is building, so while your ideas work well for KSP1 we have no idea if they would in KSP2 or not. As such while your idea is likely better than anyone else (outside of Intercept perhaps) given the obvious practical application already shown, it is out of our hands at this point.

[Rudolf Meier]

2 hours ago, MechBFP said:

... it is that we all have no idea what else Intercept is building ...

True, but if they're not stopping to use RigidBodys and Joints or start using a completely differen version of Unity, then I don't see much they can do differently. And if they want to change this, then... well, then this means that this version of KSP2 is not even early access. Then they would have to change everything and all testing and development we do now would be more or less useless... this is something that needs to be addressed as the first point in this development (but, that's just my opinion... could be I'm completly wrong).

[Pthigrivi]

2 hours ago, MechBFP said:

It isn't that nobody cares, it is that we all have no idea what else Intercept is building, so while your ideas work well for KSP1 we have no idea if they would in KSP2 or not. As such while your idea is likely better than anyone else (outside of Intercept perhaps) given the obvious practical application already shown, it is out of our hands at this point.

To this point Nate also mentioned huge interstellar vessels and colony parts. A solution that only makes sense for simple Kerbin ascent vehicles may not be robust or flexible enough to handle what comes later.

[Alexoff]

49 minutes ago, Pthigrivi said:

To this point Nate also mentioned huge interstellar vessels and colony parts. A solution that only makes sense for simple Kerbin ascent vehicles may not be robust or flexible enough to handle what comes later.

KSP1 has mods for giant parts, such as sea dragon. They don't behave like they should in real life (like jelly, I think), they and their connections are as strong as steel columns.

[gussi111]

On 10/1/2023 at 6:16 PM, Kerbart said:

Now I'm curious where on earth you need vertical bracing between (same sized) parts. I certainly don't

Yeah agreed, most people don't complain that boosters wobble but that the whole thing does, often throwing the rocket completely off course.

[Darxus]

Wobbly rockets are problem for me too. In KSP1 i just must have Kerbal Joint Reinforcement mod, and in KSP2 i use this method: 

I expected that wobbly rockets will be eliminated in KSP2 but it is even worse, and without this "hack" it is not fun to even try to play in this game. I just want to fly to space in rocket, not a spaghetti noodle. 

[gussi111]

I don't see how interstellar travel would be possible except making huge space crafts completely rigid perhaps? Wouldn't the slightest wobbling throw the rocket on a completely different trajectory at such distances?

[Royalswissarmyknife]

On 9/30/2023 at 10:01 PM, Kerbart said:

So... wobble is not an issue in real rockets and struts are not used? I think that "making your rocket strong enough" is am essential part of designing it, just like adding enough fuel and Smaking sure you have enough batteries. Your engines need to be strong enough, so why not your rocket?

SpaceX seems to be using struts. But maybe their designers don't know what they are doing?

 

Can you please point them out on this rocket.

I cant seem to find them.

Almost like you only need them for boosters and a few other things?

[The Aziz]

Lol the whole point of the devs in that very video is that they expect and want a stack of parts to be stable and are working towards that goal. Don't know how anyone could have missed it. Parts attached on sides require some strenghtening because a tiny structural point won't hold when there are engines involved (that's just physics, there's a reason why even something earthy as door has usually TWO hinges), and your interstellar vessel has a higher chance of success when it's straight, rather than when it has bits sticking out in 6 different directions.

Edited October 2, 2023 by The Aziz

[Pthigrivi]

2 hours ago, Alexoff said:

KSP1 has mods for giant parts, such as sea dragon. They don't behave like they should in real life (like jelly, I think), they and their connections are as strong as steel columns.

My suspicion is giant parts are only likely to pose problems if you go from a 50m part with a 10m attachment to a 1.25m part and then back to 10m. Having such a weak intermediary would form a hinge in the middle of your vessel. Nate mentioned solutions like making sure stacks of similar diameter all kind of ignore joint deflection. So if you connect a 2.5m part to a 2.5m part to a 2.5m part it would all just act as if welded. It would probably help with performance. Its a little cheaty, because in theory a player could attach 100 1.25m tanks and it would all behave as rigid despite its real-world implausibility. On launch it wouldn't spagetti, but it would still be unflyable. The question for me is what happens at colonies? I'd love to see players get creative with structural ideas under different gravity loads and that process gets a little warped if players are just deliberately connecting like-diameters to like-diameters because it acts like a brick and you can make a 100/1 cantilever and nothing moves. 

I imagine this gets complicated because you're trying to approximate real behavior when real behavior on the granular, real-time basis is crazy complex. Lets take a look at whats probably actually happening with the falcon heavy booster connectors. In my life doing really basic wind-load calcs for buildings this is actually a pretty simple example:



Granted this is a rendering and Im speculating a bit because the structure is embedded in the aeroshell. These aren't just pin connections (well, some, but not all.) The upper brace prior to separation is acting as a lateral truss keeping the lateral forces (labled X+Y here) under control. These forces are induced whenever the rocket maneuvers because the force is coming from thrust vectoring at the base of the vehicle. This creates add-on aerodynamic forces as you deviate from prograde as we've all experienced. Thought experiment: Imagine you want to balance 3 brooms on your fingers. What will happen if you wrap masking tape around them just at the bottom? As soon as you move even a little the 3 brooms are going to come tumbling apart. But if you tape them both at the top and the bottom you might have a chance. This truss at the top of the rocket is that upper-masking tape. It doesn't have to carry the whole weight of the broom, just keep them together.

There's another structure you can just see between the boosters and the center that counters forces in-line with thrust (labelled in blue, Z here.) Notice they're also diagonal. For most of flight there's very little differential in this directions because all of these engines are designed to keep everything mostly even, but I bet there's a subtle advantage to burning the side boosters a little hot, tugging the center stage up. There are probably also moments nearing separation when you have differential thrust. Thats what those subtle struts labeled in blue are handling.

So, for folks designing KSP's physics and deflection calcs... are they gonna do all that? Should players worry about all that? Probably not. It should probably be simplified. But how exactly that gets simplified and still allows for the basically infinite configuration KSP allows is not a trivial problem. 
 

Edited October 2, 2023 by Pthigrivi

[Alexoff]

25 minutes ago, Pthigrivi said:

My suspicion is giant parts are only likely to pose problems if you go from a 50m part with a 10m attachment to a 1.25m part and then back to 10m. Having such a weak intermediary would form a hinge in the middle of your vessel. Nate mentioned solutions like making sure stacks of similar diameter all kind of ignore joint deflection. So if you connect a 2.5m part to a 2.5m part to a 2.5m part it would all just act as if welded. It would probably help with performance.

It seems to me that the strength and mobility of the connection on the tank does not depend on what is connected there next. Simply because the force is greater, it bends more. Just like in the real world. Disabling mobility in many cases will most likely lead to bugs and fun Danny's videos.
In general, it seems to me that this is more like talking about the problem in some extreme cases. In KSP2 it is difficult to make a replica of the Saturn-5, this would be corrected. And the talk that something in the game needs to be done right away well so as not to come back seems surprising to me, looking at the road map, for example.

[Superfluous J]

1 hour ago, Royalswissarmyknife said:

Can you please point them out on this rocket.

 

I've not played KSP2. How much does a similiarly-sized (ratio-wise and payload-to-orbit wise) rocket wobble in KSP2 right now?

Edited October 2, 2023 by Superfluous J

[Royalswissarmyknife]

2 hours ago, Superfluous J said:

I've not played KSP2. How much does a similiarly-sized (ratio-wise and payload-to-orbit wise) rocket wobble in KSP2 right now?

Not very well.

you will in the best case scenario expend a lot of fuel you could have used elsewhere trying your hardest to have a somewhat efficient gravity turn.

Worst case you wobble so much you lose all control.

The good thing about a falcon-9 alike rocket is that its just the same diameter up to the fairing (the fairing would prob have issues on any rocket).

It would be much worse if your trying to do something like an Ares-1.

Spoiler

an Ares-1 with a liquid first stage because the SRB's dont have the issue of being multiple tanks.

 

Edited October 3, 2023 by Royalswissarmyknife

[MechBFP]

16 minutes ago, Royalswissarmyknife said:

Not very well.

you will in the best case scenario expend a lot of fuel you could have used elsewhere trying your hardest to have a somewhat efficient gravity turn.

Worst case you wobble so much you lose all control.

The good thing about a falcon-9 alike rocket is that its just the same diameter up to the fairing (the fairing would prob have issues on any rocket).

It would be much worse if your trying to do something like an Ares-1.

  Hide contents

an Ares-1 with a liquid first stage because the SRB's dont have the issue of being multiple tanks.

 

I disagree completely, a simple rocket like a falcon 9 doesn't wobble at all.

Edited October 3, 2023 by MechBFP

[dprostock]

Ji, ji, ji.....

I see that everyone took the bait that was thrown at them. Will they spend months trying to find solutions and come up with proposals? And in the meantime we will not talk about the fundamentals, the final delivery date.

Ji, ji, ji.....

[The Aziz]

6 hours ago, Superfluous J said:

I've not played KSP2. How much does a similiarly-sized (ratio-wise and payload-to-orbit wise) rocket wobble in KSP2 right now?

A 2.5m two-stage stack with reasonably sized payload? Barely. Of course as we all know there is space for improvement and the weakest point here is the point between the fairing base and whatever is next (a smaller size separator I would normally assume). But it's flyable and honestly the biggest problem in flying these things isn't the wobble, but wonky aerodynamics.

However, a Saturn V replica is a lot harder. There are two size changes, and as far as I'm concerned, the game right now doesn't like the adapters very much because they do wobble a bit, but then there's the biggest problem which is the lander stage. Two rather weak connection points (because they have to be smaller than the stack) and an interstage fairing that doesn't provide any structural stability on the other end. Hell, if they worked around this particular issue, it would already be game changer.

[Infinite Aerospace]

7 hours ago, Royalswissarmyknife said:

Not very well.

you will in the best case scenario expend a lot of fuel you could have used elsewhere trying your hardest to have a somewhat efficient gravity turn.

Worst case you wobble so much you lose all control.

The good thing about a falcon-9 alike rocket is that its just the same diameter up to the fairing (the fairing would prob have issues on any rocket).

It would be much worse if your trying to do something like an Ares-1.

  Hide contents

an Ares-1 with a liquid first stage because the SRB's dont have the issue of being multiple tanks.

 

Space Shuttle solid rocket boosters, which is what Ares I-X was built on 'are' multiple individual segments, liquid fuel tanks are not in the truest sense of the word.

[Rudolf Meier]

Are we talking about the real problem now? Because if yes, then I can say something about it.

Assume we want to keep the Rigidbody+Joint idea of KSP (which I think is not bad) and lets say we want to keep the idea of wobble and failing rockets due to non working connections and stuff like that (which I also like).

Then what is the problem with that? Well, it's that when you connect a heavy part to a light part and then again to a heavy part, that won't work. This is what I call "mass inversion" in KJR Next. You have this e.g. when you add decouplers (with their own Rigidbody). And that's all. This is the only problem. You need to fix those "mass inversions". (h - h = good / h - l - h = bad)

One idea is to hope for a redesign of Unity... or PhysX. And I know, one day they will do something... like e.g. adding other properties which define strenght and not only mass.

The other idea is, to build an additional joint from the heavy part to the heavy part and not going via the light part. Again -> look what KJR Next does.

And then, we sometimes still have other instabilities... but those are small and we could talk about them in a very different way. Most (if not all) of the problems come from those mass inversions (at least in KSP 1). Some problems like oscillating vessels come from poor SAS.

And this solution works for everything. From huge to tiny ships. Interstellar ... whatever. Because it addresses the problem in the game engine. (other solutions with not adding some additional joints could work too... but, in the end you need to get rid of the h - l -h connections).

[The Aziz]

1 hour ago, Rudolf Meier said:

The other idea is, to build an additional joint from the heavy part to the heavy part and not going via the light part. Again -> look what KJR Next does.

Don't even have to look at mods. That's how Autostrut worked. But it was a workaround solution to a bigger problem that the devs are now trying to counter directly instead of resorting to half measures. I shouldn't have to revert 3 times and apply Autostrut to every part that looked weak previously (and eventually end up with a ship that breaks itself to pieces because there's too many autostruts - that happened to me before). The auto autostrut mod wasn't exactly helping either, quite the opposite, at some point I had to turn it off.

And KJR had a lot of problems in early days of 1.12, but since I haven't touched the game since January, I can't tell if it's better, don't care really.

[Rudolf Meier]

5 minutes ago, The Aziz said:

Don't even have to look at mods. That's how Autostrut worked.

No, that's not how they work!

5 minutes ago, The Aziz said:

But it was a workaround solution to a bigger problem that the devs are now trying to counter directly instead of resorting to half measures.

How would they do that? It's a problem of the game engine. That's like a problem in physics and you say "ok... nice... I know there's the  'heisenberg uncertainty principle' and I had to work around this (when building the transporter from star trek), but now I don't want to do this anymore and I try to find a solution that ignores this fundamental rule of physics" ... how should that work? ... by modifying physics? or the universe? ... would be an interesting task.

This is a problem of Unity and you have to accept that and live with that.

9 minutes ago, The Aziz said:

And KJR had a lot of problems in early days of 1.12, but since I haven't touched the game since January, I can't tell if it's better, don't care really.

I'm not talking about old KJR... I'm only talking about KJR Next v4.2.x ... that's the version with the new idea and fully new implementation. Early versions were sort of continuations and upgrades of the original KJR. The new one only shares the idea of "we need to do something against the wobble".

[AtomicTech]

23 minutes ago, The Aziz said:

Don't even have to look at mods. That's how Autostrut worked. But it was a workaround solution to a bigger problem that the devs are now trying to counter directly instead of resorting to half measures. I shouldn't have to revert 3 times and apply Autostrut to every part that looked weak previously (and eventually end up with a ship that breaks itself to pieces because there's too many autostruts - that happened to me before). The auto autostrut mod wasn't exactly helping either, quite the opposite, at some point I had to turn it off.

And KJR had a lot of problems in early days of 1.12, but since I haven't touched the game since January, I can't tell if it's better, don't care really.

I never used KJR, I just made liberal use of Autostrut on ALL parts and Autostrut w/rigid attachment on the first part of the booster and the decoupler. It works pretty darn well

[Rudolf Meier]

1 minute ago, AtomicTech said:

I never used KJR, I just made liberal use of Autostrut on ALL parts and Autostrut w/rigid attachment on the first part of the booster and the decoupler. It works pretty darn well

... but, the outcome is just a rigid vessel and therefore a changed game experience. Which is something not everyone wants (including me).