Investigation into new decoupler failure.

[Kosmo-not]

The purpose of this thread is to open a discussion and investigation related to the large decoupler failure problem.

I\'m sure every one of us has built a big rocket with the new parts, only to have it fall to pieces and blow up during a normal ascent.

It always seems to be the new decoupler that fails in this case.

I\'ve heard people suggest the point of failure to be when the rocket has reached maximum dynamic pressure (max q). However, the rockets that have fallen apart on me have been well past max q. I think it\'s linked to gross acceleration produced by the rocket\'s thrust. What I want to look into is what exactly causes this problem and how to design our rockets to avoid it in the future.

And now for a funny video for your enjoyment...

http://youtu.be/TERc1I6yqI0

[VincentMcConnell]

I rather thing it\'s just every part...

The radial decouplers are also so weak, that mounted parts basically fall slowly off and crash into the below stages...

[Ydoow]

I just had this issue with the regular decouplar attached to the larger Fuel Tanks.

It would accelerate too fast and the tanks would \'bend\' with the Decouplar upwards into my higher stages causing an explosion.

I solved this by adding struts between the two stages, connecting the Large Fuel Tanks to the Upper Stage.

No \'bending\' after that.

Part of the issue/\'bending\' is because of the removal of internal collisions in .16.

My fuel tanks were literally infused halfway with the upper stage before everything blew up

[bsalis]

Use struts for support. There\'s info in the How To section about this.

My experience is that there was a lot of trial and error on the learning curve in using the big parts. But now I can throw together a rocket with them and have it work without 'incident'.

[VincentMcConnell]

My experience is that there was a lot of trial and error on the learning curve in using the big parts. But now I can throw together a rocket with them and have it work without 'incident'.

It shouldn\'t be that way. We shouldn\'t end up doing all of the math to make sure the rocket has the fuel and power to get some where and then find out that the parts are poorly made and explode in random cases. This happened to me several times on a Mun Rocket during launch. The outboard boosters would jettison at the estimated time and then -- because the decouplers are weak -- crash into the TLI Stage, ruining it.

[Tim_Barrett]

Damnit, I was waiting for one of my battlecruisers to just fly out of the smoke at any moment! Great video. ;P

As for the problem, struts always seem to be the solution for me. Keep in mind that in real life, the weight isn\'t bearing on the bell of the engine at all, otherwise it would get crushed during any type of acceleration.

[maro]

First it happens to me when the ship reached max. preassure.

I dealed away with this using struts.

Second problem came up during max. acceleration whilst G going up above the green scale.

I handle that with rev down the engine to not go above 4Gs.

[segaprophet]

All stability problems I\'ve encountered so far in 0.16 have been related to thrust value vs amount of heavy/large parts still attached to the lifter.

Spreading thrust value more gradually over staging (ie multi stage SRB/LFE activation) has alleviated a lot of stability issues for me.

Also, cross bracing is your friend! :

[samstarman5]

It shouldn\'t be that way. We shouldn\'t end up doing all of the math to make sure the rocket has the fuel and power to get some where and then find out that the parts are poorly made and explode in random cases. This happened to me several times on a Mun Rocket during launch. The outboard boosters would jettison at the estimated time and then -- because the decouplers are weak -- crash into the TLI Stage, ruining it.

So you prefer your trip to the Mun just be handed to you then? Where\'s the fun in that? I knew as soon as Nova announced there would be new bigger parts that we would be starting over from the beginning learning all their tolerances and hiccups, just as we did with the original selection of parts. Part of the game is the challenge of discovering how rockets work and what you can do with them. That\'s basically the core of the whole thing, even.

If we have to add more struts, so much the better. If you lay them out right, they actually add to the look of the rocket. If you lay them out off-kilter, well that\'s okay, too, as they either still do their job or you get chaos and explosions and maybe the opportunity to get one kerbal to bail out before the command module goes boom.

[bsalis]

So you prefer your trip to the Mun just be handed to you then? Where\'s the fun in that? I knew as soon as Nova announced there would be new bigger parts that we would be starting over from the beginning learning all their tolerances and hiccups, just as we did with the original selection of parts. Part of the game is the challenge of discovering how rockets work and what you can do with them. That\'s basically the core of the whole thing, even.

It\'s an open sandbox type game like Minecraft. Different people enjoy different aspects of it. For some it\'s creating rockets, others like building things (eg. Mun bases), and others are into the Rocket Science side. So no need for hostilities by anyone. You can see this dynamic at play with the whole Mechjeb or not thing. Some see it as cheating, others a convenience because they are essentially playing different games.

Anyhow - on topic again. I too think the big decouplers are a bit weak, but don\'t mind the additional challenge of large parts in general. It makes it all the more satisfying when you get a huge craft to orbit.

Also, VincentMcConnell... if you have a plan that requires a rocket with specific delta-Vs per stage, please tell me. As you can probably tell I like building rockets. I\'d be happy to try and build rocket to a delta-V specification just for a fun and (possibly) challenge, even if you don\'t use it.

[Ziff]

I never experienced a failure at Max Q, instead I always discovered the failure in the 4-5+ G range. This is proven by the fact that many a launch with careful throttling has reached orbit only to fail when full thrust was applied during a transfer burn and G forces exceeded the 4-5 range.

Also, it is already known that the large decoupler has scaled up in size and mass but not in strength. I am sure the ongoing part revisions will take care of this oversight.

-Ziff

[rkman]

I think it\'s linked to gross acceleration produced by the rocket\'s thrust.

I think so to; as the rocket ascents it looses mass (burned fuel) and above a certain altitude depending on the design of the craft it encounters less air drag, while thrust remains the same (while 1st stage is still burning) - so acceleration and associated force increase.

What I want to look into is what exactly causes this problem and how to design our rockets to avoid it in the future.

If a part fails when acceleration force exceeds a certain value, then either the wrong part is used or the part is to weak for the job it is intended to do.

In case of the large stack decoupler i think it\'s the latter, which would make it a part balance issue, to be expected in an alpha version of the game. (as most people probably know it can be fixed by adding breakingForce and breakingTorque to the part\'s cfg.)

[rkman]

I knew as soon as Nova announced there would be new bigger parts that we would be starting over from the beginning learning all their tolerances...

It has semi-officially been acknowledged that the large stack decoupler is to weak:

'The new stack decoupler did not have it\'s stats changed from the smaller one so it\'s really weak. This is just an oversight that will be corrected in the future. Use struts to reinforce it for now.' - Capt\'n Skunky

http://kerbalspaceprogram.com/forum/index.php?topic=17168.msg251574#msg251574

Instead of adding struts we can edit the part cfg:

[.16] Fix for weak large decouplers plus reduced wobble

http://kerbalspaceprogram.com/forum/index.php?topic=17313.0

[Cykyrios]

I have edited the part.cfg too, and noticed the large stack decoupler had an ejection force of 25, while the small one has a force of 100.

IIRC, the crash tolerance was also a bit low, or the breakforce and breaktorque, which are not visible by default (btw, the pod\'s crash tolerance seems ridiculously high compared to other parts).

[sal_vager]

I have just had this failure mode affect one of my designs, I launched at full throttle and throttled back when above 100m/s to conserve fuel, then completed the remainder of the flight until meco at a reduced throttle.

It was when in orbit that I throttled up once again, and each time my craft would disintegrate, I believe it was the sheer power of three of the new large engines that were simply too much for the craft, especially with no atmosphere or gravity drag to cushion the forces.

Everything exploded, not just the decouplers, and completing the orbit at 1/3rd power avoided the problem, I will refrain from using full power except when initially launching in future.

The ship - Atacama

Edit, Had to change the ship slightly, I thought I was clever with the arrangement of the decouplers but it turns out that fitment presents some new and 'interesting' problems of it\'s own.

[ForumHelper]

And now for a funny video for your enjoyment...

http://youtu.be/TERc1I6yqI0

I\'m watching this movie right now:)

[Markus Reese]

Like many have said, it does seem too weak, but my common strategy for wobble removal works well in this case. I would set the small structural supports around just below the coupler to give a raised surface to attach the struts to high on the next stage. This pretty much eliminates alot of wobble and the collapse.

[Ziff]

It was when in orbit that I throttled up once again, and each time my craft would disintegrate, I believe it was the sheer power of three of the new large engines that were simply too much for the craft, especially with no atmosphere or gravity drag to cushion the forces.

Not really sure how many times I have to say this, but here it is yet again. Watch your G-forces. Just because you are in orbit does not mean you are not affected by them. If you are getting this entire rocket into orbit than those 3 engines will easily put you into the 4-5 G-force range where bad things will happen.

OR, it could be the fact that your decouplers are mounted wrong. They are too high up on the engines. Basically, you somehow managed to attach the bottom node of the decoupler to the engine, and the fuel tanks are attached to the top node, so the tanks and engines are occupying the same space. This will also, most definitely, cause EVERYTHING to explode. Try just going through the stages on the launch pad and watch your command pod go flying up into the air from the explosion.

-Ziff

[sal_vager]

Thanks for that Ziff, I had actually set up the decouplers in the usual way initially, but I thought this other setup would work and look tidier, I hadn\'t gotten that setup into orbit without hitting the g-force wall before, I have changed the .craft now though.

I, like others, have gotten used to largely ignoring the g force meter as it hadn\'t had a bearing on my designs until now, the sudden explosion was decidedly unexpected but I\'ll learn from experience.

[rkman]

I have edited the part.cfg too, and noticed the large stack decoupler had an ejection force of 25, while the small one has a force of 100.

I assume '100' is a typo, ejectionForce of the small decoupler is 10.

IIRC, the crash tolerance was also a bit low, or the breakforce and breaktorque, which are not visible by default (btw, the pod\'s crash tolerance seems ridiculously high compared to other parts).

crashTolerance for both is the same: 9.

Indeed in the cfgs breakingForce and breakingTorque are not specified for either, and both have a 'Strength' of 22 (visible in the VAB), which appears to be a default value.

The large decoupler with cfg edited for breakingForce =200 and and breakingTorque =200, has a 'Strength' of 200.

(the mass of the large decoupler is 10 times that of the small one)

A little more testing reveals that 'Strength' takes the value of 'breakingForce'.

[rkman]

Another thing that contributes to problems with the large decoupler in combination with the new large engines is the fact that there is only a gimbal version of that engine.

From 015 and before we know that it is usually not wise to put only gimbal engines on the first stage (if it\'s a heavy rocket) because all that gimbaling puts very strong lateral forces on the craft, which can cause the craft to break up during ascent.

Because there is no non-gimbal version of the new large engine, the most powerful rockets that can be build in 0.16 do experience the strong lateral forces that could be avoided in 0.15 by using some non-gimbal engines instead of only gimbal engines. (though that\'s nothing that a bit of copying and cfg editing can\'t solve)

[kknight13]

in the video the rocket crashed because the decoupler was blown while the bottom stage was still firing, the top stage began to slow and the still firing detached stage hit it like a, well a rocket.

[rkman]

in the video the rocket crashed because the decoupler was blown while the bottom stage was still firing, the top stage began to slow and the still firing detached stage hit it like a, well a rocket.

Yeah the decoupler blew because it can not handle the forces that the new large parts put on it.

[sal_vager]

Out of interest, has anyone witnessed this with a craft with no decouplers?

Edited out a superfluous 'this'

[xmrsmoothx]

Out of interest, has anyone witnessed this this with a craft with no decouplers?

Sometimes, the large fuel tanks simply fail and everything goes to hell.