Yet Moar Spaceplane Woes

[capi3101]

Alrighty guys...I'm up to the Auk VIII now:

And here's the craft file.

Cargo delivery system, obviously - designed to haul up one orange tank worth of fuel plus one large RCS tank worth of fuel (in the cylinder tanks). Jets on 1, Nukes on 2, Intakes on 3-5, Panels on 9. The forward RCS tank has been deliberately left empty - the plan is to use the back RCS fuel for docking purposes. After delivering her payload she's tail heavy, so the fuel in the back RCS tank gets transferred into the front RCS tank and proper balance for flight is restored when she comes into land. I've checked the fuel balancing with the tweakables in all flight modes and she should remain stable the entire way.

So on to my problem - I can get her into the air about one out of every three takeoff attempts - the other times, she suddenly pulls to one side and usually lithobrakes. When she does fly, she doesn't want to fly straight. I had originally suspected too much weight on the landing gear / wheel-barrowing going on, but as I'm seeing the behavior in flight too, that leads me to believe there's a more fundamental flaw with the design that I've repeatedly overlooked. Since this is starting to aggravate me, I figured I'd toss it over to y'all for some general assistance.

Maybe the tri-adapters?

Edited February 10, 2014 by capi3101
Thread Answered Status Change

[sdj64]

If you're worried about the tri-adapters, look at the thrust of each jet engine to see if they are equal on each side. Also, it looks like you have only those two tail fins as vertical stabilizers? Try putting control surfaces on your vertical stabilizers and also moving them as far behind the center of mass as possible.

Also I suppose Moar Struts on the wings can always help.

[ComradeGoat]

I can't help but noticing the severe lack of a rudder...

ETA: To elaborate, the two canards you've stuck vertically to try and control lateral stability are almost in line with the COL/COM, which means they have precious little leverage. You need a lateral stabiliser as far aft of the COL/COM as you can get it. Perhaps try a delta wing stuck vertically above your rearmost engine with one of the large control surfaces mounted on it, set to control yaw.

Edited February 9, 2014 by ComradeGoat

[Claw]

Capi,

Long story short, you have enough yaw authority but your wings are too "flappy" because they are overloaded and improperly strutted (not necessarily moar struts). When I flew it, one of the right wing sections broke off but it was still flyable (so plenty of authority). Eventually, all the bouncing around broke off the central TurboJet. Amazingly, it was still flying.

Anyway, here's what I did to help with wing loading/bending.

Hopefully you aren't color blind...

If you look at the red arrows, I showed where I moved the flight control surfaces to offload some of the force from the wings. Basically I moved them inboard (closer to the fuselage) in some key places.

The green arrows show a couple struts I moved around. So I moved the one between the engine nacelles so that it connects the nacelles together instead of connecting it to the wing connector. Also the oval on the wingtip is slightly misleading. You have many struts crisscrossing underneath the wings, but they are largely ineffective. If you add vertical struts in between the wing sandwich (indicated where the green oval is), it will strengthen up your wings a bit. Let me know if that part doesn't make sense. You could probably remove some of the crisscross struts, but I didn't really mess with that. (So you don't need moar struts, just play with moving them around some.)

[capi3101]

Thanks. Okay...so I need to go ahead and move the Standard Canards I'm using for vertical stabilizers / rudders further back is what y'all are telling me. I'll give that a shot when I can get a chance and see what happens. Any suggestions on where to put them? Maybe out on the rear outboard dorsal wings? Do y'all think I need moar rudders, or should two do it?

Thrust is pretty even coming out of the jets. I didn't think about having lousy yaw authority; makes sense now that y'all mention it, though.

[Claw]

Thanks. Okay...so I need to go ahead and move the Standard Canards I'm using for vertical stabilizers / rudders further back is what y'all are telling me. I'll give that a shot when I can get a chance and see what happens. Any suggestions on where to put them? Maybe out on the rear outboard dorsal wings? Do y'all think I need moar rudders, or should two do it?

Thrust is pretty even coming out of the jets. I didn't think about having lousy yaw authority; makes sense now that y'all mention it, though.

Well, that isn't what I said at all. With two reaction wheels and the fins you have, I think you have enough yaw control. You can add more yaw authority and that might improve takeoff, but I don't think that will solve all your problems.

[Hodo]

Claw is right.

But here is a simple solution.

Add larger yaw stabilizers, also known as tail fins. The ones you currently have are to close to your center of mass which would be like putting your steering in the center of your car or riding a unicycle. So try replacing your current rudder system with a taller tail section that is further back on the craft.

[Claw]

Adding large yaw stability won't fix the wing flex, which is the real source of the instability. The wings are twisting which is causing asymmetric lift, leading to an out of control craft. You can add more rudder and it will probably allow you to control the craft, but that doesn't actually solve the problem and potentially adds more parts.

Wing flex is also costing lift and causing parts to break. Having more yaw control won't fix that.

Case in point::

It has two reaction wheels and no vertical tails and it flies just fine because I eliminated most of the wing flex. Before I put in struts, it veered off the runway or flew crooked (when I didn't crash).

[Hodo]

Adding large yaw stability won't fix the wing flex, which is the real source of the instability. The wings are twisting which is causing asymmetric lift, leading to an out of control craft. You can add more rudder and it will probably allow you to control the craft, but that doesn't actually solve the problem and potentially adds more parts.

Wing flex is also costing lift and causing parts to break. Having more yaw control won't fix that.

Case in point::

http://i.imgur.com/BSZf4JX.jpg

It has two reaction wheels and no vertical tails and it flies just fine because I eliminated most of the wing flex. Before I put in struts, it veered off the runway or flew crooked (when I didn't crash).

Trust me I know a bit about SSTOs, and I was not debating your fix, I was offering an additional fix or suggestion.

I have built all matter of SSTOs from space plane designs to lifting body designs, all worked, all had their own unique challenges.

[capi3101]

Well, that isn't what I said at all. With two reaction wheels and the fins you have, I think you have enough yaw control. You can add more yaw authority and that might improve takeoff, but I don't think that will solve all your problems.

Sorry, you ninja'd me; I was responding to ComradeGoat when I posted that. Fix the wing flex by changing the strutting outboard and moving the control surfaces to inboard. Making sure, I need to add a set of struts towards the outboard edge of the wings connecting the ends to one another. It also looks like I need to keep the dorsal and ventral wing control surfaces outboard of the nukes, but move the central surfaces to positions aft of the wing connectors.

And then possibly couple that with adjusting the stabilizer / rudder positions.

Alright y'all, I'll give it a shot and report what happens.

[capi3101]

Well, the stability problems are fixed, so thanks, y'all. Anybody care to take a guess what its next issue is?

Going to be adding moar intakes in the near future.

Actually, she ran out of gas just after 25000. Admittedly, I knew already that the design was already about one tank short of having what it needed solely from liquid fuel tanks, but I'm beginning to suspect that I may be doing the ascent wrong as well. I usually go 60 degrees to 10,000 (Auk VIII has about ten degrees AOA in the early flight), then back off to 45 degrees up to 20,000, reducing that by five degree increments to keep the ascent rate less than 100 m/s. Above 20,000, I'm usually at 5 degrees picking up speed, still keeping the ascent rate around 10 m/s. When I've got the last set of intakes open, I start throttling back once I get to around 0.07 intake air per engine; I generally avoid flame outs. Rockets get lit when I'm down to one third thrust; I keep the jets running and I still throttle back as needed to prevent flameouts. When I stop accelerating, I kill the jets, open up the throttle, and keep burning until the apoapsis is where I want it. Anything fundamentally wrong with that profile?

[Claw]

You can also try putting four engines on an action group to shut them down. Then put the last three on a different action group to shut them down. That way, when you are at the point where you would normally kill the TurboJets and just run the nukes, you can just shut down the first four engines and keep running the three (which will also be getting more intake air per engine) to get some extra thrust. It might take some practice to know when to switch. Just don't forget to shut them down (or throttle modulate some more) to prevent going out of control. Also make sure the three engines you choose to run are all in line (your two outboard and the center).

I would say that your initial climb of 60 degrees seems a bit steep for the surface TWR the Auk has. Maybe start out more around 45 degrees and let the speed slowly build. I recommend being around 200 m/s at 10km, but you don't have to race to get to 200m/s either. Believe it or not, the extra speed give you more lift, which leaves you more thrust to accelerate and climb faster. You can try that and see if it gives you any extra fuel. Climb profiles can sometimes be design dependent, so you can just run a few different ascents and see how it goes.

Incidentally, I tried running a test with intakes after one of our previous conversations. I tried closing unneeded intakes during the climbout and I didn't really notice any drag losses during the climb. So you might not need the extra intake complication (at least in KSP 0.23).

Edited February 10, 2014 by Claw
Clarification

[capi3101]

Claw, thanks to your most excellent advice, I was able to pilot the Auk VIII to orbit last night with most of the orange tank liquid fuel intact. 45 degrees saved fuel in the long run, and then by cutting out four turbojets and cutting in the rockets I was able to maintain full throttle the whole way up (and because I'd also added a proper number of ram intakes, I didn't have to cut in the rockets until I was up around 2100 m/s). Thank you once again. I've got a docking target up and I'll see how she handles that when I can, and then I get to see if my RCS ballast idea is going to work the way I hope it does.

Big problem now is machine lag; the 11 minute flight took about an hour real-time. But I don't know if there's much y'all can do about that...

[Claw]

Great job and congrats.

Some of the RCS ports look like they might be blocked, so be patient with docking. Good luck!

Edited February 10, 2014 by Claw

[m4rt14n]

Case in point::

http://i.imgur.com/BSZf4JX.jpg

It has two reaction wheels and no vertical tails and it flies just fine because I eliminated most of the wing flex. Before I put in struts, it veered off the runway or flew crooked (when I didn't crash).

That is one nice plane!

[m4rt14n]

Have you tried landing it from orbit into KSC? My last big plane (orbital refueler) would still not land right without either crashing or at the very least losing its engines due to tail strike. Still have one of them in orbit, cos I'm scared I'll kill the kerbal piloting it.

If you think flying these big ones are hard, wait until you have to cut its thrust and glide/land it... Nightmare.

[capi3101]

The docking and landing parts of its maiden test flight are planned for this evening. I've scrapped two cargo haulers already due to loss of control on re-entry. That was the reason why I added the RCS ballast system on the VIII; I tested the various fuel tank configurations for its entire flight and I'm optimistic about being able to land it safely. I will let y'all know how the rest of the mission goes.

[m4rt14n]

Just curious, why use two NERVAs? Are you trying to fly this all the way to Duna/Laythe? You can save a lot of fuel by switching them to RAPIERs and using them with the turbojets from 20km to 35km before rising the ap and pe.

[capi3101]

Mainly I have them for mass balance I guess. I probably could move the CoM a little further forward by swapping them out with a lighter engine. Yet were I to do so, I'd probably go with Aerospikes (lighter still, better Isp than the RAPIER in rocket mode, generates power).

This evening I'm perplexed. I've rechecked the original model in the SPH twice using the tweakables to show me where the CoM is going to be compared to the CoL for the ship in its post-delivery configuration, I've set the plane up to that configuration, everything shows like it should be good - and yet I loose control of the plane coming in around 18,000 and can't regain it. The one time I did, the plane developed this weird side-to-side oscillation. Turned SAS off and I was out of control again. Crashed the Auk VIII four times now and still haven't figured out what went wrong.

Docking was a thing of beauty.

[capi3101]

Actually, I wonder if the loss of control on landing problem stems back to poor yaw authority. I need to run to work now, but I'll post more after a bit.

[Claw]

That is one nice plane!

Thanks.

Capi,

Your yaw control might be the cause of that. I'm not at my computer now either, but look at where your rudders are in relation to the re-entry CoM. Also, if your plane is too neutrally stable (the lift marker nearly centered in the mass marker), then that will cause difficulties too.

Sometimes changing the reentry angle helps. Try starting out sort of pancaking in (30-45 AoA) to bleed speed, then flatten back out before you get to 18km. If you're already doing that, then try coming in shallow.

Also, watch what the yaw trim indicators are doing during the oscillations. With a ship that massive, all that inertia can be hard for the SAS to stop. In which case you can give it a little yaw input at the right time to keep the SAS from over correcting and fishtailing.

Edited February 11, 2014 by Claw

[capi3101]

I've revised the craft file for the Auk VIII in my Dropbox to the latest version. Controls are the same as before. For good measure, I've added the target craft as well; it's a single stage rocket with a self-cleaning booster - basically some empty fuel tanks of various types cobbled onto a custom docking pier I built to add to my Mir replica for the Constellation Program Challenge (so I could dock with something other than a Sr. docking port).

Anyways - the lift marker does come in right near the center of mass for re-entry. The end of the flight has gone something like this so far - after docking with the pier, I offload the remaining fuel in the center tank; all the jet tanks are dry, the RCS in the centerline cylinders is offloaded and I go ahead and refill the NERVA fuel tanks to full. After docking, the aft RCS tank has about 670-700 units of Monoprop still in it; this gets shifted to the forward tank and the other RCS tanks are locked down. I cut loose from the pier, do a 100 m/s burn to deorbit with the nukes. All told I have about 325 units of LiquidFuel and 397 units of Oxidizer left at that point. I shift the fuel into the central tank and lock down the NERVA tanks, lock down all the jet fuselages except for the forwardmost pair, fill them to full and set the outboard tank to transfer out (I have TAC, so I leave it like this). Reentry is generally shallow, no more than 10 degrees or so. Intakes are open for drag, I'm facing prograde, keeping it level and the solar panels come in when I hit atmo. Loss of control generally happens around 17,000 meters or so, surface velocity around 800 m/s - it starts as a pull to one side, which I attempt to correct by applying opposite rudder; this usually fails, the plane whips around, and from there it generally plummets out of the sky. PARE doesn't work; I suspect either I'm applying it wrong or that KSP's aerodynamic model neutralizes that technique.

I do think the rudders still aren't far enough back and I mentioned replacing the nukes with Aerospikes last night (since the only place I intend to go with this plane is Kerbin orbit); I'd drop 1.5 tonnes from the plane and it should throw the mass further forward unless I'm mistaken. I'd just burn more fuel getting into/out of orbit.

[Claw]

Make sure you are also looking at the CoM marker with the landing gear off the craft. They are massless in flight and hence cause the SPH to lie about the CoM location. This could be a factor if your msrkers are very close together.

The out of control you are describing sounds like a CoM issue, but I believe you when you say it's in front. In which case, you would need to move the vertical tail back more or make it bigger to offset the drag you're experiencing on reentry. (That's probably the easiest.) At 18k you might not have enoug airflow to make the rudders effective enough.

Also, as wacky as it sounds, you can fire up the TurboJets around 20k (not necessairly full throttle). The thrust vector authority may be enough to stay under control till you get down to 10-12k where the rudders are more effective.

By the way, flying wings don't spin recover very well because of the lack of rudder. Sometimes you can rock them in pitch and get the nose down to get speed. KSP areo is pretty rough here too, but then again modeling the chaos of a spin is very complex to begin with. Also, spin recovery only works if you're actually in a spin. If it just flips around and falls flat (or backward) out of the sky then it won't help much.

After it goes out of control, does the plane keep rotating? Or does it flip, then sort of hesitate for a bit. Then as the prograde marker swings around, it keeps going out of control?

I'll try flying it again when I get home tonight.

Edited February 11, 2014 by Claw

[capi3101]

Make sure you are also looking at the CoM marker with the landing gear off the craft. They are massless in flight and hence cause the SPH to lie about the CoM location. This could be a factor if your msrkers are very close together.

The out of control you are describing sounds like a CoM issue, but I believe you when you say it's in front. In which case, you would need to move the vertical tail back more or make it bigger to offset the drag you're experiencing on reentry. (That's probably the easiest.) At 18k you might not have enoug airflow to make the rudders effective enough.

Also, as wacky as it sounds, you can fire up the TurboJets around 20k (not necessairly full throttle). The thrust vector authority may be enough to stay under control till you get down to 10-12k where the rudders are more effective.

By the way, flying wings don't spin recover very well because of the lack of rudder. Sometimes you can rock them in pitch and get the nose down to get speed. KSP areo is pretty rough here too, but then again modeling the chaos of a spin is very complex to begin with. Also, spin recovery only works if you're actually in a spin. If it just flips around and falls flat (or backward) out of the sky then it won't help much.

After it goes out of control, does the plane keep rotating? Or does it flip, then sort of hesitate for a bit. Then as the prograde marker swings around, it keeps going out of control?

I'll try flying it again when I get home tonight.

I haven't tried taking the gear off the model; I'll do that and see what it looks like. I'll also try firing up the engines at 20k though I've read elsewhere you should wait to 400 m/s before firing them up, see if that saves the current flight.

When the plane goes out of control, it flips, heads towards retrograde for a while, prograde swings around and continues going out of control. Attempting to compensate is generally ineffective. The one time I managed to briefly regain control, I was fortunate enough to be relatively level to the horizon with the nose around prograde. That's when that oscillation occured. I might've been able to land the craft had I left SAS on.

[Claw]

I haven't tried taking the gear off the model; I'll do that and see what it looks like. I'll also try firing up the engines at 20k though I've read elsewhere you should wait to 400 m/s before firing them up, see if that saves the current flight.

When the plane goes out of control, it flips, heads towards retrograde for a while, prograde swings around and continues going out of control. Attempting to compensate is generally ineffective. The one time I managed to briefly regain control, I was fortunate enough to be relatively level to the horizon with the nose around prograde. That's when that oscillation occured. I might've been able to land the craft had I left SAS on.

Okay, I had a chance to look at your plane in the VAB. Before I delve into that, I have to ask.

When you say "it flips," are you referring to it flipping over in pitch or yaw? (When I think of flip, I think pitch. When I think about it turning in yaw, I think of sliding.) I ask just so I have an understanding of what you're seeing in flight. (Yaw or pitch departures.)

That being said, I see three things:

1) The new craft file I downloaded has no vertical tail fins at all. As discussed previously, they are not strictly needed but can help.

more importantly you are being bitten by...

2) Your CoM is ending up directly above your CoL with the current weight distribution you described (forward RCS and NERVA tanks full) even with the gear removed. This will make it neutrally stable. Which means (with such a large craft), that your limited yaw authority will have a hard time compensating (which might explain the yaw oscillations).

And finally, the one that snuck up on you...

3) KSP aero has a (one of many) bug. When AoA increases, lift produced by control surfaces increases faster than lift produced by wings. (To the point that control surfaces produce maximum lift at 90 AoA.) You have a lot of canards that are way forward from your CoM. At the config described above, and given my #2 comment, your neutrally stable craft becomes unstable a soon as you put AoA on it (which is pretty much required for flight).

Go to the SPH and check this out. Turn on your CoM and CoL markers and set up your fuel config you described. Select your entire craft (shift click wherever) and rotate it with Shift+S (5 degrees, then 10 degrees). Watch the CoL bubble and see where it goes. Bam, out of control...

So, to answer my own question at the top of this post. I suspect you are going out of control in pitch and seeing yaw oscillations separately.

Lots of ways to fix this. I can ramble on some more in this post, but you can play with it and see what you prefer. If you want help with fixing it, let me know.

Good luck!

EDIT: I forgot. I can't actually see where your CoM is with the gear on because I have edited my landing gear file. I gave the landing gear zero mass because I grew tired of the SPH lying about the CoM. So I'm guessing that given the location of the gear in relation to the CoM, it looks more forward to you. Make sure you remove the gear when you play around with the CoM.

Edited February 11, 2014 by Claw