Why does my SSTO spaceplane tumble at the start of rocket burn?

[ajburges]

I've been trying to make an SSTO that meets my requirements in career (Stock, no clipped intakes, pre level 3 tech tier, no practical limit on cargo bay payload weight) with a lower part count. I have since moved on a couple of iterations/concepts, but one initially promising design failure still bugs me. Anyone know what likely happened here? The plane runs fairly fine (if a little air starved) until I transition to rockets. At which point it just starts tumbling.

The only hidden parts are wing connection struts (fuel lines are underwing and drain in a manner were CoM remains in front of CoL). What you see is what you get.

Javascript is disabled. View full album

Edited January 29, 2015 by ajburges

[Janos1986]

Check alignment between center of thrust and center of mass, but remove your jet engines first, since the CoT is calculated by all your engines combined.

It seems to me that the nuclear engines need to be moved up a bit.

[KerbMav]

Rather right click the jet engines and tweak their thrust to zero - that way their mass is still accounted for when you check the alignment of COM and COT.

Did you use trimming (ALT+WASD) during ascent with the jet engines?

[LordFjord]

Some of your right side fuel tanks have disabled fuel-flow, visible in the pics where your plane points straight up. Could it be that asymmetric fuel usage causes it, paired with shifting CoM as your plane burns through the fuel?

RCS build aid helps a lot to figure out these issues.

And to solve (or at least improve) the asymmetric flameout issue, check the link in my sig. Your plane seems to have equal amount of intakes for each engine, they are just not built in the right order. The explanation why that happens is described in the mod thread.

[Sean Mirrsen]

Look at the fuel distribution. All the fuel is in the back, as well as the heavy engines, all concentrated in the same general area as the main wing.

You're being dragged off course by air resistance. While you have turbojets, their gimbaling power-steers the aircraft to keep it level. The output of the LV-Ns is too puny to do that - the most minute deviation from prograde sends you tumbling because of the canards as well as all the lightweight parts in front. You have no leverage to keep the plane flying in the atmosphere while it's tail-heavy. Add some tail stabilizers, preferably large, wide, and extending far behind the fuselage.

Basically, you're having an effect I describe as "somebody forgot that aircraft have tailplanes for a reason".

[GoSlash27]

The first thing I notice is that all of your intakes are ahead of the CoM. That can make you tumble. You should set the radial intakes as far back as possible.

Also, there's a fix for the asymmetrical flameout. Check out my tutorial here:

http://forum.kerbalspaceprogram.com/threads/102182-So-you-want-to-build-a-space-plane-%28-25-stock%29

If you can fix the asymmetrical flameout problem, it'll be much easier to control when you finally switch to rockets (not to mention the fuel savings)

Best,

-Slashy

[KerikBalm]

All that to bring up a single mk2 cargo bay's payload?

My SSTOs that launch payloads from a mk2 bay are much much smaller than that, and obey all your rules (no part clipping, tier 2 techs, etc)

[linuxgurugamer]

All that to bring up a single mk2 cargo bay's payload?

My SSTOs that launch payloads from a mk2 bay are much much smaller than that, and obey all your rules (no part clipping, tier 2 techs, etc)

So let's see one of them.

[GoSlash27]

So let's see one of them.

I have one in my tutorial above. 13 tonnes and a single turbojet.

But this isn't about how efficient the OP's spaceplane is, it's about helping to solve his pitch problem.

Best,

-Slashy

[ajburges]

Check alignment between center of thrust and center of mass, but remove your jet engines first, since the CoT is calculated by all your engines combined.

It seems to me that the nuclear engines need to be moved up a bit.

CoM and CoT are not grossly misaligned (dang orbs make it hard to be precise. Parts were placed via snap and tweaked afterwards. Only unbalanced placement of weight was radial intakes, rudder, and antenna.

Some of your right side fuel tanks have disabled fuel-flow, visible in the pics where your plane points straight up. Could it be that asymmetric fuel usage causes it, paired with shifting CoM as your plane burns through the fuel?

RCS build aid helps a lot to figure out these issues.

And to solve (or at least improve) the asymmetric flameout issue, check the link in my sig. Your plane seems to have equal amount of intakes for each engine, they are just not built in the right order. The explanation why that happens is described in the mod thread.

Intakes were placed as a unit 1 a time before each engines. A tweak will probably be to place the set-up such that center engine starves first. Fuel flow was verified on ground. Only asymmetric draw (WRT CoM) were the two tanks/side on the wing closest to CoM. Their CoM shift did not have CoM and CoL shift on an otherwise empty bird.

Look at the fuel distribution. All the fuel is in the back, as well as the heavy engines, all concentrated in the same general area as the main wing.

You're being dragged off course by air resistance. While you have turbojets, their gimbaling power-steers the aircraft to keep it level. The output of the LV-Ns is too puny to do that - the most minute deviation from prograde sends you tumbling because of the canards as well as all the lightweight parts in front. You have no leverage to keep the plane flying in the atmosphere while it's tail-heavy. Add some tail stabilizers, preferably large, wide, and extending far behind the fuselage.

Basically, you're having an effect I describe as "somebody forgot that aircraft have tailplanes for a reason".

I've noticed that effect before, but it seem sufficient to check CoL migration as I rotate my craft to avoid it.

The first thing I notice is that all of your intakes are ahead of the CoM. That can make you tumble. You should set the radial intakes as far back as possible.

Also, there's a fix for the asymmetrical flameout. Check out my tutorial here:

http://forum.kerbalspaceprogram.com/threads/102182-So-you-want-to-build-a-space-plane-%28-25-stock%29

If you can fix the asymmetrical flameout problem, it'll be much easier to control when you finally switch to rockets (not to mention the fuel savings)

Best,

-Slashy

It doesnt progress to full flame-out, but an engine starts flickering (mayby I just need to learn throttle control)

Am I just drastically underestimating the drag of a closed intake?

I have one in my tutorial above. 13 tonnes and a single turbojet.

But this isn't about how efficient the OP's spaceplane is, it's about helping to solve his pitch problem.

Best,

-Slashy

That seems to be a short bay not the long one. 13 Tonnes sounds close to my reference payload though (4 small RCS tanks, 4 docking ports, 2 smallest probe cores, 2 large radial RCS tanks, 4 small radial RCS tanks)

All that to bring up a single mk2 cargo bay's payload?

My SSTOs that launch payloads from a mk2 bay are much much smaller than that, and obey all your rules (no part clipping, tier 2 techs, etc)

I will confess, I probably don't follow the weight optimal design or the fuel optimal acent profile. Mayby I'm taking too much fuel? My goal with my SSTO fleet is more to for 100% recovery of low stages for satilites and mission modules (Tanker duties, large deliveries, and small craft will be saved for a Mk3 behemoth). I prefer a more agressive, quicker acent.

Acent profile:

Take-off (poor gear/tail-wheel placement means I fly-off instead on 80% of designs)

Find pitch to acend at/near terminal velocity

Acend with best timewarp possible to 16-18 Km

Set pitch to maintain 100 m/s or less climb

Correct orbital inclination before I build speed

When air gest to about .08/engine, set pitch to maintain 10-50 m/s climb (leaning towards the lower end)

When air bufer is depleated, set pitch to maintain under 10 m/s climb

When engines (or my patience) starve engage rockets

(if needed) Pitch up and get apiosis above 140 Km (aim to have it under 120Km once in orbit)

120Km is chosen because that is my station height.

[Sean Mirrsen]

I've noticed that effect before, but it seem sufficient to check CoL migration as I rotate my craft to avoid it.

Eh, the CoL has nothing to do with it. For the most part. Really, the problem is that your craft is getting unstable as it loses fuel (the CoM creeps back towards the CoL), and the control surfaces you have for pitch control lack the necessary authority to keep you facing forward.

I always add a generous helping of tail to my SSTOs, to avoid a situation where the atmosphere is too thin to keep control, but still thick enough to pull the craft off course.

You can't go wrong with more control authority in the tail section.

[GoSlash27]

It doesnt progress to full flame-out, but an engine starts flickering (mayby I just need to learn throttle control)

It would be worth your time. A few more km altitude will get you out of the region where the wonky aerodynamics affect your balance, leave you the pitch authority of your turbojets, and saves you fuel.

Am I just drastically underestimating the drag of a closed intake?

Possibly. Especially with those big ram air intakes hanging out in the breeze like that.

You made mention in the OP of it being borderline "starved" but in reality you've got more intake area than you actually need for the job. 3 ram airs and 3 radials is adequate to feed a 3 engine turbojet SSTO. Variations where you substitute 2 XM-G50 radials for 1 ram air would also suffice. The important part is to get the intakes as far aft as you can.

Best,

-Slashy

[ajburges]

Made the mistake of listening to the one person who could not read and shifted the wings back to increase distance between CoL and CoM further. Result: lawn dart at sub 10 Km altitude. Leading me to believe that Sean is right an I lack pitch authority. Problem is not even re-purposing some roll surfaces to pitch solved the problem and I am already at the length limit.

I think I will drop this particular design. Intakes on the nose just seem too troublesome to deal with. I can get away with them on the wing pods, but the nose seems to never work.

I made a 4 jet engine design I'm happier with (and will be able to modify for a second bay once I upgrade my hanger for action groups). It manages just short of apoapsis placement on jet power alone. I was just hoping I could salvage this design. Given how quickly I approach 30 tonnes designing around this cargo bay, I don't think I can get away with a 2 jet design. I will likely make a short bay 2 jet SSTO.

[GoSlash27]

Made the mistake of listening to the one person who could not read and shifted the wings back to increase distance between CoL and CoM further. Result: lawn dart at sub 10 Km altitude. Leading me to believe that Sean is right an I lack pitch authority. Problem is not even re-purposing some roll surfaces to pitch solved the problem and I am already at the length limit.

I think I will drop this particular design. Intakes on the nose just seem too troublesome to deal with. I can get away with them on the wing pods, but the nose seems to never work.

I made a 4 jet engine design I'm happier with (and will be able to modify for a second bay once I upgrade my hanger for action groups). It manages just short of apoapsis placement on jet power alone. I was just hoping I could salvage this design. Given how quickly I approach 30 tonnes designing around this cargo bay, I don't think I can get away with a 2 jet design. I will likely make a short bay 2 jet SSTO.

Sounds like a winner. A couple tips to help you on your way:

*Wings aren't all created equal. Smaller panels are more efficient than large ones, and some types outperform others. http://forum.kerbalspaceprogram.com/threads/105090-KSP-90-Spaceplane-wing-comparo

*Likewise, 2 short fuselage sections are less draggy than 1 long fuselage section.

*If you're just going to a station in orbit and back, nukes aren't worth the mass and drag. Go with something light and weak, such as the 48-7S.

*Basic proportions for turbojet SSTOs: At least 1.0 total lift coefficient per tonne of aircraft. At least 1 turbojet per 15 tonnes of aircraft. At least .018 m^2 intake area per turbojet.

Good luck!

-Slashy

*edit* Of course, all of this is subject to change when KSP 1.0 comes out...

[Sean Mirrsen]

I think I will drop this particular design. Intakes on the nose just seem too troublesome to deal with. I can get away with them on the wing pods, but the nose seems to never work.

They work perfectly well on the nose. Again, you just need steering authority to keep the little bit of extra drag from intakes (at higher angles of incidence) from pulling you further off course. Don't overdo canards, and try to avoid non-moving wing surfaces on the nose if you can at all help it - those are far more likely to cause you grief than intakes that are too far forward.

[ratchet freak]

one thing to help with the tumble is to close the intakes as you shut down the jets, closed intakes have much less drag

[capi3101]

Missed the fact that the OP was just going to orbit with the design. Yeah, the nukes are overkill there. You don't necessarily want nukes unless you plan to go interplanetary with the design - I wouldn't use them for the final ascent, my own self. At best you might want a set of rocket engines that'll give you a launch TWR of 1.0 (and that's being conservatively high; I suspect by the time you need them a 0.8 TWR would do the job but I need to do more experimentation; the idea behind saying 1.0 is to have a little overlap in case you botch the jet portion of the ascent and have to have the rockets take over earlier than you'd like). A combination of 24-77 and 48-7S engines will easily do the job and save you some serious mass in the process. For your original 34 tonne design, you want somewhere between 272-340 kN rocket thrust. A pair of 48-7S engines accompanied with six 24-77s each will give you 300 kN rocket, and have a total mass of 1.28 tonnes - this compared to two nukes providing 120 kN thrust with a mass of 4.5 tonnes. Remember, after your spaceplane is done being an airplane, it's a rocket; the same considerations still apply once it's up there.

8 24-77s would give you 380 kN thrust - a greater than 1 TWR, and still only another 0.36 tonnes in mass.

Of course, seeing as the OP has abandoned the original design, this information might not be all that useful for this particular case. The nukes didn't have all that much to do with the original question, either...

[ajburges]

Missed the fact that the OP was just going to orbit with the design. Yeah, the nukes are overkill there. You don't necessarily want nukes unless you plan to go interplanetary with the design - I wouldn't use them for the final ascent, my own self. At best you might want a set of rocket engines that'll give you a launch TWR of 1.0 (and that's being conservatively high; I suspect by the time you need them a 0.8 TWR would do the job but I need to do more experimentation; the idea behind saying 1.0 is to have a little overlap in case you botch the jet portion of the ascent and have to have the rockets take over earlier than you'd like). A combination of 24-77 and 48-7S engines will easily do the job and save you some serious mass in the process. For your original 34 tonne design, you want somewhere between 272-340 kN rocket thrust. A pair of 48-7S engines accompanied with six 24-77s each will give you 300 kN rocket, and have a total mass of 1.28 tonnes - this compared to two nukes providing 120 kN thrust with a mass of 4.5 tonnes. Remember, after your spaceplane is done being an airplane, it's a rocket; the same considerations still apply once it's up there.

8 24-77s would give you 380 kN thrust - a greater than 1 TWR, and still only another 0.36 tonnes in mass.

Of course, seeing as the OP has abandoned the original design, this information might not be all that useful for this particular case. The nukes didn't have all that much to do with the original question, either...

Orbital ops are the primary use. My plan is to rendezvous with a dedicated space tug/command module for any beyond Kerbal operations (So I don't lug wings and jets to places they have no utility). My reason to use nukes is they were giving me better performance than the LV-909's I used to use. a longer jet burn was worth the extra 500 m/s dv in orbit

Sounds like a winner. A couple tips to help you on your way:

*Wings aren't all created equal. Smaller panels are more efficient than large ones, and some types outperform others. http://forum.kerbalspaceprogram.com/threads/105090-KSP-90-Spaceplane-wing-comparo

*Likewise, 2 short fuselage sections are less draggy than 1 long fuselage section.

*If you're just going to a station in orbit and back, nukes aren't worth the mass and drag. Go with something light and weak, such as the 48-7S.

*Basic proportions for turbojet SSTOs: At least 1.0 total lift coefficient per tonne of aircraft. At least 1 turbojet per 15 tonnes of aircraft. At least .018 m^2 intake area per turbojet.

Good luck!

-Slashy

*edit* Of course, all of this is subject to change when KSP 1.0 comes out...

I should probably reply to your wing thread, but here goes:

Why are smaller wings/parts more efficient? Drag is directly proportional to mass and drag value right? So as long as mass to lift ratio is equivalent, the efficiency is equivalent (assuming equal drag value).

I am familiar with the SSTO guidelines, though I thought the engine ratio was turbojet for roughly 10 tonnes and wing load was closer to 2:3 lift to mass (a recent learning, I aim for 1:1 still). I seem to prefer a higher takeoff TWR regardless for a faster climb to 16 Km and better throttle down/partial engine shutdown trust.

Both of you seem to miss the fact I am looking to lower my part count. I don't play in an optimized PC state so I see (minor) accent slowdowns at 255 parts and rendezvous gets even more interesting. I actually tried an honest air-hogger for the same payload (8 intakes/engine, 4 turbojets), but it is not comfortable to work with in rendezvous scenarios. That makes part count a bit more heavily weighted (I will still make aesthetic compromises, but minor efficiency losses are acceptable if they keep part count down)

[capi3101]

Orbital ops are the primary use. My plan is to rendezvous with a dedicated space tug/command module for any beyond Kerbal operations (So I don't lug wings and jets to places they have no utility). My reason to use nukes is they were giving me better performance than the LV-909's I used to use. a longer jet burn was worth the extra 500 m/s dv in orbit

< snip / >

Both of you seem to miss the fact I am looking to lower my part count. I don't play in an optimized PC state so I see (minor) accent slowdowns at 255 parts and rendezvous gets even more interesting. I actually tried an honest air-hogger for the same payload (8 intakes/engine, 4 turbojets), but it is not comfortable to work with in rendezvous scenarios. That makes part count a bit more heavily weighted (I will still make aesthetic compromises, but minor efficiency losses are acceptable if they keep part count down)

If rocket performance and low part count are your main concerns, may I (in lieu of the 48-7S/24-77 combo I suggested earlier) suggest you go with LV-T45s? You'll wind up with 400 kN rocket thrust, keep the part count the same as what you have, and still wind up saving about a tonne on the mass of the craft you've got. I too also play KSP on a Byzantine computer setup (2.9GHz Dual Core w/ 2 GB memory and an onboard video chip with the processing power somewhat less than that of a Dorito), so I can sympathize with wanting to keeping the part count down. Again, that doesn't address your initial issue, but it might prove useful.

Actually, I've been considering replicating your design to see if there's any other pieces of advice I can give you from trying to fly it my own self.

For the record - the recommended engine ratio for Turbojets is 15 tonnes per engine, for Basic Jets it's 10 (and to be wholly honest I've never even come close to making orbit on Basic Jets). I've heard 1.3 bandied about as an initial takeoff TWR for spaceplanes; couldn't tell you where that falls on mass-per-engine, though, as Turbojets don't hit maximum thrust until they're already 15-20k up in the atmosphere...

[xxjetterxx]

Move the COL back. Fuel has weight, and thus the COM shifts, changing the relative position of the COM to the COL. sometimes the COM will end up behind the COL because of this.

[GoSlash27]

Why are smaller wings/parts more efficient? Drag is directly proportional to mass and drag value right? So as long as mass to lift ratio is equivalent, the efficiency is equivalent (assuming equal drag value).

Aye, but the drag values are not equal. They scaled the drag coefficients to size, and they should not have. As a result, an equal mass of smaller parts exhibits much less drag. This is also why short fuselage sections exhibit less drag than larger ones.

I am familiar with the SSTO guidelines, though I thought the engine ratio was turbojet for roughly 10 tonnes and wing load was closer to 2:3 lift to mass (a recent learning, I aim for 1:1 still). I seem to prefer a higher takeoff TWR regardless for a faster climb to 16 Km and better throttle down/partial engine shutdown trust.

2:3 lift to mass is going to be problematic in the 25-32km altitude range. You can certainly overcome it given enough thrust, but you'll pay in efficiency and ultimately need to build a bigger spaceplane to achieve the same result.

Both of you seem to miss the fact I am looking to lower my part count. I don't play in an optimized PC state so I see (minor) accent slowdowns at 255 parts and rendezvous gets even more interesting. I actually tried an honest air-hogger for the same payload (8 intakes/engine, 4 turbojets), but it is not comfortable to work with in rendezvous scenarios. That makes part count a bit more heavily weighted (I will still make aesthetic compromises, but minor efficiency losses are acceptable if they keep part count down)

Certainly I'm not talking about generating part counts anywhere near a range that'll tax your system. The efficiency you gain from careful selection of parts and adherence to ideal ratios allows you to achieve the same payload capacity with a smaller spaceplane, which saves you parts in the end.

Also, I think several folks here are way overestimating the thrust required to circularize a turbojet spaceplane. Once you're in the regime where it's time to shut down the turbojets (roughly 43km altitude and over 2200 m/sec velocity) there's virtually no drag and you're so close to orbit that your inertia is throwing you upwards. Even RCS thrusters are adequate to complete the job at that point.

A pair of 48-7S engines is perfectly fine for OMS duties on a 30 tonne spaceplane, and 400 m/sec is plenty of DV to circularize, intercept, rendezvous, and deorbit.

If you look at the twin engine example in my tutorial, it's orbiting a 14 tonne payload using a single LV-909.

Best,

-Slashy

[ajburges]

Slash do you have a reference on drag being related to part size? The only references I can find agree with the wiki which only use velocity, mass, and drag value for drag force. I guess I could also compare terminal velocities on ground for equivalent craft with Kengineer. Are the drag values incorrectly printed in game?

Edit: Tried on the ground with body segments (8 short vs 4 long) hooked up to a cockpit with computer. 9x m/s vs 12x m/s terminal velocity. What the heck Squad? Still do you have a reference on how to determine these values?

I tend to handwave my atmospheric stage because of the efficiencies of jets and the difficulty in reproducing variables. Thus I turn to the Kerbal motto "In thrust we trust." (Plus I'm not keen on a half hour accent.)

To make it clear for those who missed it: CoL is behind CoM for all stages of normal fuel drain. There is some balest in the noise that could be used for emergency's in orbit (to get to a fuel depot and restore girl and balest) and the payload could be stupidly unbalanced, but what I'm flying has fuel drain order verified and that is used to verify CoM migration. Next time I cover this will likely be done in caps since I already tired further movement of CoL only to get a lawn dart.

On rockets. Thrust is a minor concern. Most my designs use rockets mostly to combat drag, equalize, and perform orbital burns. Part count and delta-v dominate my considerations with mass being a secondary concern. My biggest requirement on thrust is reasonable maneuver burn times.

I can provide craft files on request. I didn't think anybody would actually want to fly this best.

Edited January 30, 2015 by ajburges

[capi3101]

Yeah, go ahead and provide the craft file - I did go ahead and build a replica of your craft from the series of screenies this evening and I need to verify something major about the design. Had a screenie of my own all nice and ready to go - or at least, I thought I did. Coulda swore I hit the F1 key anyways.

Anyways, I use a mod called RCS Build Aid when I build planes (someone may have mentioned it already; I'm not going back through the thread to say one way or the other). Among its many, many useful features is a marker that will show you where your craft's CoM will be once all the tanks are dry, and an "average" CoM position between dry and however you have your fuel loaded at the moment. This is accompanied by a measurement telling you just how far the CoM shifts from "wet" to "dry". Optimally you build your plane so that the CoM doesn't shift at all - and it's for this reason that I use the mod. Seriously, I'd recommend it to anybody wanting to get serious about planes; it can save you a lot of headaches.

For the replica craft - which was built using the same parts in the same* positions with the same amounts of fuel indicated in your screenshots - RCS Build Aid was showing that the CoM shifts aft during flight by close to half a meter. That's not that much mind you, but given how close you've got the CoL to the CoM in the design it is sufficient to put the CoL up front. Between that and the high drag caused by the nose intakes, that's probably what's causing your issues. Again, this is all for the replica; I might've got some things wrong in the process of construction, which would make these observations completely invalid, but does heighten the need for a second glance at the original. I should mention that I lost control of the replica at roughly the same time, around 30,000 during the process of switching over to the nukes.

[GoSlash27]

Slash do you have a reference on drag being related to part size? The only references I can find agree with the wiki which only use velocity, mass, and drag value for drag force. I guess I could also compare terminal velocities on ground for equivalent craft with Kengineer. Are the drag values incorrectly printed in game?

Edit: Tried on the ground with body segments (8 short vs 4 long) hooked up to a cockpit with computer. 9x m/s vs 12x m/s terminal velocity. What the heck Squad? Still do you have a reference on how to determine these values?

ajburges,

I pulled the values directly from the part config files and calculated IAW the chunk of code from .18.

It's not readily apparent from just looking at the numbers themselves, but my chart shows how they behave at speed.

Also, not exactly Squad's fault. Porkjet configured it that way and Squad didn't catch the error. We have told them about it, but they haven't cleaned it up yet. Hopefully they'll get it in the rebalance for 1.0

Best,

-Slashy