Tanks pushing inward when staged

[ineon]

Can I have some help with a spacecraft please? I would like it to not explode during launch.

The Imgur album below is from me hammering the f1 key during doing the first staging. The tanks spin into the main body and things get a little explody. I don't know what the arrows all represent in the aero overlay, but I assume that the cyan arrows going inwards (from about image 3) from the staged tanks are the problems.

Obviously I have tried attaching sepratrons which are not able to overcome the inward force (they do point the right way - just triple checked). I am using the nose-cone that slants outwards hoping that it would push the tanks out. It is also nothing to do with SAS, the rocket is (surprisingly) aerodynamically stable and the SAS is turned off.

One thing that I noticed is that the one time I tried staging the other 2 tanks first, they still rolled inwards, but they did not destroy the rest of the ship and I was able to go to space that day. Although that might have been due to many other factors.

I am not using any mods that affect the physics. 

p.s. Please pay no attention to the altitude, speed and angle and what that implies about my gravity turn 

[Foxster]

I think you have a couple of things going on there:

1. Are those separatons firing at the tip of the stage being dropped? If so, then it looks like they are firing towards the central tank, which will cause that tank to overheat and explode. Try angling them so they fire so they just miss the edge of that tank. 

2. Those big fins are on the outside of the discarded tank and I think they are tending to push the tanks inward. You could try mounting those tanks much lower on the central tank, so they are attached at the top of that tank and at the bottom of the central tank. Then even if they move inwards when decoupled they should miss the bottom of the central tank. 

Edited March 5, 2016 by Foxster

[FullMetalMachinist]

I'm not sure why you're getting such big lift vectors (that's the cyan lines) driving the side boosters in, but I suspect that the slanted nose cones might be the problem. And if they're not the problem, I don't think they're helping you the way you think they are. 

I'm honestly confused where the lift is coming from to push the boosters in, but I have two ideas that might help you overcome it. 

The first is to put your radial decouplers higher up on the main stack to give an extra kick outwards on the top of the radial boosters. This takes some use of the offset tool, because (as I'm sure you know) when you place something on the radial decouplers your mouse has to be pretty much right on it. So, put the decoupler higher up on the center stack, then place the boosters. Then use the offset tool to slide the booster lower on the decoupler. 

Second idea is less optimal, so I'd only try it if the first doesn't work, because it increases your drag. What you do is take either a control surface (make sure to turn it off so it doesn't respond to any control input!) or a small wing section towards the top of the radial boosters. Place it so that the "long" axis is horizontal and tangent to the booster so that it has 0 angle of attack. Then rotate it *just* a little to have an angle that, as it's moving through the air, it will want to pull the top of the boosters away from the center stack. 

Hope this helps! 

[Snark]

I suspect it's lift coming from those extra Big-S tailfins you have on the left and right sides of each of the boosters.  Have you tried removing those?  (i.e. so that each radial booster has just a single fin on it, pointing directly radially outward from the center stack)

And if the answer is "but if I do that then it's unstable and flips," then the answer to that would be some ship redesign so that that doesn't happen, but that's another discussion.    Just as an experiment, try taking off the extra fins and (if you have stability problems) just accelerate straight up and see what happens when you stage.

Another suggestion, mount the radial boosters much lower on the central stack, as low as they will go.  When you decouple them, they'll start to slide backwards (since the center stack is still thrusting, and they're not.  If they're mounted low on the stack, then your center stack doesn't have to travel very far to slide out from between them, and you'll be safe sooner.

Here's another thought:  Mount some drogue chutes on the radially outward side of the boosters, towards the top.  Set them up so that they stage together with the decouplers.  You're only going a bit over 300 m/s, so the drogues should be safe.  They'll yank the boosters down and outwards hard.  You'll need to adjust the "full deploy" altitude upwards to its maximum of 5000m.  Your screenshots are from slightly above 5000m, but it looks like your gravity turn is a bit slow-- if you turn just a bit more on the way up, I expect your boosters will drain out before you hit 5km, and the drogues will be able to have full effect.

Edited March 5, 2016 by Snark

[Frybert]

43 minutes ago, FullMetalMachinist said:

The first is to put your radial decouplers higher up on the main stack to give an extra kick outwards on the top of the radial boosters.

I concur 100% with this.

\

19 minutes ago, Snark said:

I suspect it's lift coming from those extra Big-S tailfins you have on the left and right sides of each of the boosters.  Have you tried removing those?

Also with this. You should not be needing fins that big for stability. It may not help with this particular problem, but really, you just shouldn't need fins that big.

Edited March 6, 2016 by Frybert

[Laie]

What Snark said: move the outside boosters as far down as possible. Will probably require 8 or 12 more struts, but wth. The upside is that not only will this help with staging, it will also shift drag down and weight up so you can do with fewer/no wing pieces.

If all else fails, consider Fleas in place of the parachutes as king-size sepratrons.

[FancyMouse]

Also I'm seeing you don't have enough space between stages - you should at least use TT-70 decoupler so that there's enough space.

[Snark]

36 minutes ago, FancyMouse said:

Also I'm seeing you don't have enough space between stages - you should at least use TT-70 decoupler so that there's enough space.

Depends.  If the problem is that "they're nosing inwards and there's no way to stop it and you just need to give the center stack time to get clear before they hit", then yes, that could help.

On the other hand, if it's possible to tweak the design so that they get kicked outward rather than heading in, it's not needed.  Using a Hydraulic Detachment Manifold gives a fair amount of outward "kick" that can be helpful.

[FancyMouse]

1 minute ago, Snark said:

Depends.  If the problem is that "they're nosing inwards and there's no way to stop it and you just need to give the center stack time to get clear before they hit", then yes, that could help.

On the other hand, if it's possible to tweak the design so that they get kicked outward rather than heading in, it's not needed.  Using a Hydraulic Detachment Manifold gives a fair amount of outward "kick" that can be helpful.

When a sepratron is needed, decoupler force does not make any difference (that's the reason sepratron is used at the first place, right?). And in the case of OP, it's already a biggest stock engine with a biggest stock LFO tank - 24t dry mass at least. Not what radial decoupler is meant to handle, I'm afraid.

And more importantly, more space allows some margin of error when there is some inwards torque on the discarded stages - as long as it's not too much and main stage is able to escape from the collision.

I use exclusively TT-70 when the thrown stage is >=2.5m. If I need sepratron, I don't calculate precisely where to put my sepratrons but still never had any staging problem.

[Snark]

23 minutes ago, FancyMouse said:

When a sepratron is needed, decoupler force does not make any difference (that's the reason sepratron is used at the first place, right?). And in the case of OP, it's already a biggest stock engine with a biggest stock LFO tank - 24t dry mass at least. Not what radial decoupler is meant to handle, I'm afraid.

Agreed that it's on the upper end, mass-wise.  However, as long as the decoupler is mounted as high as possible on the radial tank (and therefore far from the mass concentration that is that huge engine), it has less work to do-- it's more rotating the stage's top end outward rather than shoving the whole thing.  I've had pretty good luck with Hydraulic Detachment Manifolds on this scale.

It's probably a good idea to give it some help, yeah, but it's a nontrivial contribution.  If it's possible to get the tank to start by nosing outward rather than inward, aero forces will do the rest, and it won't need extra clearance space.

I'm not saying the TT-70 is a bad idea, that's also a useful approach.    Just that there's more than one option, here.

[Spacetraindriver]

Yeah, those tails are interfering definitely.

Those angled out cones are big design no noes.

Mammoths, and those tail fins and only TWO separations are just disastrous.

Yeah, what everyone else said...

I've done that  much worse w/ struts and survived, no I'm serious, it really restricts crazy movement.

 

[Nich]

Not 100% on KSPs aero model but the noses pointed out cause lift into the middle (faster air creates lower pressure) Points on the inside closer to the core should pull the boosters away.

Edited March 6, 2016 by Nich

[Snark]

1 minute ago, Nich said:

Not 100% on KSPs aero model but the noses pointed out cause lift into the middle (faster air creates lower pressure) Points on the inside closer to the core should pull the boosters away.

That's an interesting point, hadn't thought of that.

It's true that I use slanted nosecones all the time and always have them angled in so that their tips are close to the center... and I do seem to have pretty good "luck" with my boosters not nosing into my ship... but I always assumed that the slant was pretty much cosmetic, it didn't occur to me that it might be helping me.

Would be a quick and easy experiment for the OP to check-- try rotating the nosecones 180 degrees and see what happens.

[Nich]

did some testing and to me the most important thing was staging when going directly prograde.

 

 

[SinBad]

I build radial boosters as sub assemblies, then attach them. That way i can see the empty cot col and com and better predict how they will behave once released. If you do mods, rcs build aid is great. Instead of having to guesse how much rotation the stage will gain based on cot-com distance, it gives you a number. I also mount pairs of seperatons so their exhaust misses the main tank. if your chutes and decouplers are mounted radialy at the com (empty) they wont contribute much to any rotation you put in with seperatron thrust. 

[GoSlash27]

It's hard for me to tell for sure, but it looks as if you've mounted your decouplers too low. This kicks the bottom of the discarded stages away from the stack, but not the top.

I'd also see about mounting those tailfins farther down. CoL in front of CoM is bad for stability.

Best,
-Slashy

[Plusck]

On 6/3/2016 at 3:06 AM, Nich said:

Not 100% on KSPs aero model but the noses pointed out cause lift into the middle (faster air creates lower pressure) Points on the inside closer to the core should pull the boosters away.

 

On 6/3/2016 at 3:09 AM, Snark said:

That's an interesting point, hadn't thought of that.

It's true that I use slanted nosecones all the time and always have them angled in so that their tips are close to the center... and I do seem to have pretty good "luck" with my boosters not nosing into my ship... but I always assumed that the slant was pretty much cosmetic, it didn't occur to me that it might be helping me.

Would be a quick and easy experiment for the OP to check-- try rotating the nosecones 180 degrees and see what happens.

 

I'm pretty sure that I read (from a developer?) that the slanted nosecones have zero aerodynamic difference compared to the ordinary ones, that they are aesthetic only. (Edit: pretty conclusively proved wrong on this point by @Rhomphaia)

The slanted adapter tanks, however, will obviously make a difference as soon as you add something above them, but I don't know the aero model well enough to say whether the bare top to the slanted tank will in itself add an offset to the drag model for the part.

Edited March 7, 2016 by Plusck
standing corrected

[Rhomphaia]

31 minutes ago, Plusck said:

 

 

I'm pretty sure that I read (from a developer?) that the slanted nosecones have zero aerodynamic difference compared to the ordinary ones, that they are aesthetic only.

That does not seem to be the case.

Two slanted nosecones, two straight.  The straight ones go straight up.

They have different Drag cubes and they also have off-center mass

 

[Snark]

1 hour ago, Rhomphaia said:

That does not seem to be the case.

Two slanted nosecones, two straight.  The straight ones go straight up.

They have different Drag cubes and they also have off-center mass

 

Nice experiment! 

Didn't know about the off-center mass.  I wonder how relevant that is, relative to the aero?  e.g. suppose you did your same experiment again, but used Kopernicus to mod away Kerbin's atmosphere and make it a vacuum planet.  (And give those lil' guys a reason to be wearing their helmets as they walk around KSC.)  Would be interesting to see how that would affect your experiment.

[ineon]

Wow everyone thanks for the feedback . I don't think I'll have a chance to try anything out until Tuesday, but I'll comment on some of what is above. 

Regarding decoupler placement, I think the decoupler is a far up as it can be. I have seen others suggest this as good advice and tried to generally follow it. But as @FancyMouse points out I've got 2 separatrons shop the decoupler force is minimal. What I didn't show in the imgur album is the one time that I got to space - when I staged the other set of engines at about 60km - the main thing that they did was spin rather than be pushed further out. 

Regarding placing them lower down and further out, I will definitely give that a go. Regarding nose cones, I changed to sloping because my intuition is that it should push them out - the nose cone pushes air into the middle and the (equal and opposite) reaction pushes the tank out. Is that not the case? And what is wrong with the intuition? I'll also try rotating them around. @rhomphaia which was did the rockets turn in you experiment - towards the direction that of the slope or away? 

Finally regarding the wings, it appears my intuition is at odds with the community here again. I would have thought that the CoM and Cod are both in the moved outwards by the tail (compared to just the tank on its own), but that the trail would move the drag more than the mass,  and this would induce a rotation outwards. What am I missing? 

Thanks once again everyone. 

[Snark]

59 minutes ago, ineon said:

Finally regarding the wings, it appears my intuition is at odds with the community here again. I would have thought that the CoM and Cod are both in the moved outwards by the tail (compared to just the tank on its own), but that the trail would move the drag more than the mass,  and this would induce a rotation outwards. What am I missing?

Aero forces are generally smooth and predictable when you have a controlled orientation and they're not totally dominating the movement of things.  But when you have an aerodynamically unstable object (as I suspect those boosters are, when you eject them), you can get some unpredictable results.  "Flutter" can happen.

Those wings might actually not be causing a problem.  We don't know, at this point.  It's not that the community is saying "those wings are your issue," it's more that "those wings MIGHT be your issue, because they're angled in a way that they have the potential to generate lift forces in the radial direction."

One thing you'll find is that although there's a lot of flexibility in KSP to build what you want, the constraints of physics mean that there's a certain amount of convergent evolution in rocket design:  successful rockets of a given size tend to gravitate (no pun intended) towards a certain common design.  Experienced KSPers are used to that.  So any time someone shows up with a design that does stuff that's not the usual "normal" way, and that person says "why is this rocket acting weird," then there's a natural knee-jerk reaction in the mind of an experienced KSPer that goes something like this:

• Here's a rocket that's built in unusual fashion that's different from what I'm used to seeing.
• The builder says he's having unusual flight problems that are different from what I'm used to seeing.
• Ergo... ?

It's not necessarily a foregone conclusion that unusual thing #1 is causing unusual thing #2, but it's a natural suspicion that leaps to mind. 

In your case, you're doing some things that are a bit unusual:

• Fins that are much bigger and aggressively aerodynamic than usual.
• Fins that are oriented in a way other than straight radially outwards, and therefore have the potential to create radial lift.
• Slanted nosecones that are rotated 180 degrees backwards from the way everyone always uses them

...If you had simply presented a picture of that rocket and asked "do you think this will fly?", I would have noticed the unusual design features, but wouldn't necessarily have had alarm bells going off.  So I probably would have said, "Sure, I guess so."  But when you then say "there are unusual flight problems" ... I have an immediate instinctive reaction about some likely candidates to check first. 

[CSE]

It may be that the Mammoths are clipping into the centre engine - KSP's clipping mechanics don't always 100% match the visuals - and pushing the back of the boosters outward. The aero forces are then generated from the booster's tank sides, and there's not much that standard sepatrons or fins can do to overpower this. (Flea sepatrons are great fun in 6.4x or RSS scales, but they do eat into your payload capactity.) For this rocket, you could try:

• Rotating the Mammoths so that the shorter dimension is facing the centre, maximizing the visual separartion.
• Or use the wider decouplers as suggested by @FancyMouse. If I use non-symmetric engines then I always do this. I've rather gone off non-symmetric engines, though - the idea that the collision geometry might not always match the visuals puts me off. I'm a bad enough rocket builder without any extra help!
• @NecroBones' SpaceY has a combined wider decoupler-sepatron.

For future rocket designs, I'd also say that 5km is a bit low for reliable separation because the atmosphere is still pretty thick. Higher-altutude separations will in general have more tolerance for misbehaving boosters. And you're just below the speed of sound, in the trans-sonic region where drag is highest and thus any aero effects will be at their most pronounced. I think most real-life applications either stage well before this speed (think RATO rather than rocket booster) or else quite a bit later where the air is very thin and the boosters' path is determined mostly by their momentum. Maybe you could carry a bit more fuel in the boosters and a bit less in the centre stack?

 

Edited March 6, 2016 by CSE
Typo.

[NecroBones]

 

Responding after summoning, but haven't read all the replies yet, so forgive me if this has already been mentioned:

 

I can't quite see where the decouplers are in the images, but based on the ejection, it looks to me like the ejection charge is happening very low on the boosters. I'm not sure how that would happen with single large fuel tanks as the attachment to the decoupler. But generally, I try to always make sure the center of mass for the empty booster is below the center of the decoupler, so that the ejection charge always gives an outward kick to the upper portion of the booster. If you attach close to the engines, and then build upward by attaching more tanks etc on those strap-on boosters, you will frequently run into this problem since the bottom of the booster will kick outward, and any aerodynamic pressure (as CSE mentioned) will push the top back into you. So it's usually better to build both up and down from the attachment point, or just downward. As CSE also mentioned, the clipping mechanics can be a little weird since the collision-meshes aren't a perfect match to the visual models. And that's more of a problem with those really narrow decouplers. In this case though, I suspect it's a combination of aerodynamic forces, and the booster's center of mass being really close to the ejection charge.

 

This is a good thing to keep in mind, even if you do use the SpaceY thruster-decouplers.

[Snark]

8 hours ago, CSE said:

SpaceY has a combined wider decoupler-sepatron.

It also has some very nifty sepratron-nosecones (in 1.25m and 2.5m, centered and slanted), which are both aerodynamic nosecones and sepratrons that thrust outward from center (and don't fry your center core with their exhaust).  I love those things.  I actually use them quite a bit more than I do the thruster-decouplers.  (And I'm a guy who practically never uses the stock sepratrons.  I just don't like 'em.  The SpaceY sepratron nosecones are just what the doctor ordered, as far as I'm concerned.)

Edited March 7, 2016 by Snark

[Nich]

I did a couple tests with the smallest decouplers and it was still smashing the center tank even with low center of mass and sepratrons.  I suspect there is some unseen clipping as mentioned before that is forcing the tanks to pop out on the bottom even though the decoupler is on top.