31

I've spent the past three hours committing a genocide of historical proportions against my Kerbals, and it's making me quite sad.

Valentina would really appreciate if I could get my airplane off the ground, but it keeps exploding on the runway by the exact same failure mode: The plane veers to one side or another, either I or SAS correct it, which starts a left/right wobble, which builds until one of the wings hits pavement, at which point the craft quickly brings Valentina to her untimely end.

I've covered all my bases, as far as I can tell:

  • All three wheels (one steerable in front, two fixed in back) are mounted with angle snapping, symmetrically
  • Wheels are mounted to the fuselage, rather than to the wings (which flex under load)
  • Wheel base is wide
  • Center of lift is behind center of mass, and slightly raised
  • Rear wheels are in front of tail, but not enough to make tail strikes easy
  • The two engines have been mounted every imaginable way with the same result: Over the wings, on the fuselage, on girder segments, on the wingtips, etc.
  • Wings have been flown both with and without struts
  • Wheel friction was even decreased to 0.3 for front wheel and increased to 2.0 for rear wheels, as per a tip found online

It seems if I turn off SAS, I can prevent the oscillation from starting as it's initiated by some control input, whether from me or SAS, but then the plane veers off the runway until it hits something. Intriguingly, there is one thing I've found that does prevent the wobble from starting, and that is to use a single engine mounted directly on the back of the fuselage, which prevents using a tail and is ultimately a fairly significant restriction to build around.

I've spent some time scouring the internet for ways of solving this problem, and found a number of pieces of advice on preventing wobble and improving planes in general, some of which made it into the above list, but Valentina's still dying, and I'm tired of Kerbal funerals.

  • what do you mean by "wheels are mounted with angle snapping"? do you put them vertically or are they at an angle to the ground? – Noone AtAll Aug 3 at 14:24
  • I used the angle snapping feature (to the right of the symmetry selector in the bottom left hand corner of the screen in the space plane hangar view) to mount the steerable wheel perpendicular to the ground, and the two fixed wheels symmetrically such that zoomed in, the wheels themselves are perpendicular, though the fixed wheel part itself comes with something like a 45 degree angle between the wheel and the shaft that attaches it to the craft, so the shafts naturally sit at something like a 45 degree angle out to either side so the wheels can sit perpendicular. In other words classic tripod – TheEnvironmentalist Aug 3 at 14:35
  • Have you set your control surfaces to only respond to appropriate controls? E.g. the vertical stabilizer only handles yaw? – Tashus Aug 3 at 17:03
  • Do you have a picture? Getting a plane of the round isn't that hard. Keeping it from crashing is usually the hard part. – Mast Aug 3 at 18:40
  • 8
    Standard KSP airplane-stability request: post a screenshot of your airplane in the SPH, with the center-of-mass, center-of-lift, and center-of thrust markers shown. Better still, three screenshots showing side, top, and rear views. – Mark Aug 4 at 2:10
31

I've found that all my runway wiggling planes are caused by the wheels. This happens in two instances; initial placement and plane body deformation.

Wheels need to be touching the ground exactly as they were intended to, you cannot angle them in any direction other then their intended direction when placing them. So the ones that go straight up and down can only be straight up and down, no angling left, right, backwards or forwards such as when you place the little wheel on the angled body of the cockpit part. The ones that are at an angle must be placed at that exact angle, you can't angle them to make them wider.

Plane body deformations can cause the wheels to touch the ground at a slightly different angle. Body deformation can be caused by heavy parts, many parts connected together, or forces on the plane.

From your description of your plane I would check the following:

  • Center of lift is behind center of mass, and slightly raised
    When your plane speeds up, it sounds like your lift could be pushing your nose into the runway causing deformation or an oscillation with the same effect.
  • The two engines have been mounted every imaginable way with the same result: Over the wings, on the fuselage, on girder segments, on the wingtips, etc.
    Assuming nothing is blocking the exhaust of the engines, the relation of your thrust to your center of mass could also cause the same problems as above.
  • Wheel base is wide
    If your wheels are far apart there may be deformation caused by heavy parts or many parts between them, I would try attaching struts from each wheel to a single fuselage part. Also ensure that the wheels are placed "correctly".
| improve this answer | |
  • 14
    Turns out it was wheel positioning! I had my rear wheels mounted on the tail, and it seems the taper of the tail messed up the angle of the wheels creating instability. I moved the wheels from the tail to the fuel fuselage, and now she takes off like a dream! KSP should be handing out engineering degrees :) – TheEnvironmentalist Aug 3 at 18:53
12

You seem to have done a lot of things right. I'm finding it quite surprising you didn't manage to get the plane to fly yet. I haven't encountered oscillations like that in an airplane but I have encountered them aplenty in case of lightweight probes that have too strong control authority - too good RCS or reaction wheels, so I suspect the source here is the same. Try reducing control authority (or even generally amount) of control surfaces - rudder, ailerons, the front wheel (maybe even make it a fixed one).

Additionally if you have any off-center fuel tanks, double check the fuel is drawn from both at the same rate (same priority). Plane veering off course is not uncommon but uneven fuel draw can exacerbate it a lot.

Also hopefully you're not using any mods that use throttle (Throttle Controlled Avionics etc) as with engines as sluggish to react as jet, this is bound to cause trouble.

If that all fails, you may try building your plane lighter and with mightier (or just more) engines, so it just doesn't have the time to start misbehaving on the runway before you bring it up in the air.

one last thing to verify: look from the side, observe where your center of mass is, then how the control surfaces are located relative to it and if the way they react to your controls makes sense (just try to pull up, down, or roll, on the runway without even starting the engine or disengaging brakes and watch which way the control surfaces move). Control surfaces near center of mass sometimes get "confused" as to their role and do stupid things. You may need to change control authority to a negative value if they bend "the wrong way".

ps. Start with simplest, easiest planes you can and then increase complexity observing and mitigating problems as they arise. Starting from something with MK3 hull without experience in MK1 will be an uphill battle.

ps2. It's unlikely you're doing this but just in case: while on the ground use only yaw controls to change direction, don't touch the roll controls. Roll is the main way to change direction in flight, and best way to crash the plane on the ground.

| improve this answer | |
  • 1
    +1 for "If that all fails, you may try building your plane lighter and with mightier (or just more) engines, so it just doesn't have the time to start misbehaving on the runway before you bring it up in the air." I've always found that best way to not crash on the take off is to take off. – Shane Aug 4 at 2:38
  • There are some really useful tidbits here which deserve some credit. First, too strong control authority is not the kind of problem I'd think of, and was definitely worth looking into, and fuel tank priority is a really good point. Next, mods can and often do cause these kinds of problems. Finally, while straightforward to those with an engineering background, the idea to start with the simplest, easiest possible planes and build one's way up to a sufficiently complex craft is not obvious to everyone, and is the epitome of the engineering mindset KSP exists to instill and explore. Thanks :) – TheEnvironmentalist Aug 5 at 14:32
  • 1
    @TheEnvironmentalist There is one more method I'm sure would work in your case, although I didn't write about it because I think it's cheesy and wouldn't solve the root problem. Drive gently off the runway and use the huge grassy field to take off, without care in the world about the plane veering to a side. – SF. Aug 5 at 14:40
  • Hahaha absolutely right, but distinctly less satisfying – TheEnvironmentalist Aug 5 at 14:53
4

I haven't found a sure-fire solution to this, but I think I know what causes it. If you have more than one engine, you can get asymmetric thrust at low speeds. Thrust is determined partly by velocity. If you (or SAS) turns even just a little, the engine opposite your turn will have a higher velocity than the one towards your turn. At low speeds, this difference can be significant. This makes one engine generate more thrust than the other, exacerbating the turn. SAS sees this and turns the other way, which just causes the same problem in the other direction. Eventually you either crash, or you reach a high enough speed that SAS wiggling can't cause a significant enough difference in engine speeds.

In addition to the tips others have suggested, try locking steering on all your wheels. This should reduce ground control authority and make oscillations less powerful. They'll still happen but you'll have more time to get off the ground.

| improve this answer | |
  • While I couldn't accept this one because ultimately it was changing wheel positioning that increased stability enough to elegantly take off, I don't think it's unreasonable to imagine that the oscillations originated or were exacerbated through asymmetric thrust. Do you by chance know how much the thrust varies with these small velocity perturbations? – TheEnvironmentalist Aug 3 at 18:58
  • @TheEnvironmentalist No clue, but it doesn't need to be much because it's a feedback loop. If you have a perturbation that gives your left engine a tiny bit more thrust, you will turn slightly right. This gives your left engine more thrust again which makes you turn faster etc... I think it might be so small it starts with floating-point rounding errors; it's exponential and compounds every frame. – Ryan_L Aug 4 at 0:21
  • Interesting, though even being exponential there’s a chance it starts out so small it takes longer than a takeoff to reach noticeable levels. Even exponential growth can take a while to get going if it starts out microscopic, and I’d be fascinated to find someone who could figure out a way of measuring the growth rate – TheEnvironmentalist Aug 4 at 18:45
2

Another possibility to add to the existing answers (it's a bit hard to diagnose without seeing the plane) is that your landing gear might be insufficient for your plane's mass. Can you check what happens if you swap in larger gear?

And one other, more remote possibility to check: Where is your plane's center of drag? It may be too far forward, especially if the tail is too small relative to the plane's wings. Can you check what happens if you increase the tail size to pull the center of drag back?

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.