Planes always fishtail

[DaMachinator]

15 minutes ago, swjr-swis said:

Oh my, that does sound very complicated indeed - practically defying the laws of physics to get such things to fly.

Obviously way more complicated than getting a plane to lift off without needing to pitch up. My apologies for over-simplifying. I am not worthy.

Cannot tell if sarcasm or not.

Would you mind informing me how to get a multipart wing to be properly angled up?

(Self-imposed rules regarding part clipping aside, I find putting on landing gear to my satisfaction to take way too much time. I think I have OCD or something...)

[Madscientist16180]

My 2 funds...

I personally saw both behaviors on my plane. 

I'm not knowledgeable about planes, i just throw them together using what i know about physic.

Every single time one of my plane had take off stability problems, adding more wheels (yup, 4 landing gear aligned) or using bigger ones manage to rule it out. The small landing gear does look like they can only manage very little weight before getting unstable.

Anybody else saw the same? Btw, i build taildragger whenever the design allow for it, maybe that's helping stability?

I never was able to reliably land planes prior to 1.1 release (did not try much), but now i feel confident enough to land a kerbaled craft without F5'ing first.

So, while i do see the bug (i do not know if it is one or not really), there is clearly ways to design around it.

[swjr-swis]

25 minutes ago, DaMachinator said:

Would you mind informing me how to get a multipart wing to be properly angled up?

By the root of the wing.

Or if by 'multipart' (aren't they all?) you mean several separate wing sections that do not share a root, but are placed very close together to *look* like a single wing: angling just one section by its root, can work just as well, and would even look more like real life wings.

More complicated if you really want all the separate sections at the same angle, and something that might resonate with any OCD: even with snap off, the rotate gizmo still moves discrete units. So carefully do the same number of 'ticks' to all wing section roots. To know absolutely sure, one can overlap wing sections with the offset gizmo, switch to rotate, angle until visibly aligned, switch back to offset (gizmo will keep the same part selected the whole time, no danger of losing it because of clipping) then offset the now same-angled wing section back to where you mean it to be.

If you only think you have OCD, you're not taking this seriously enough. It's practically a requirement in aerospace KSP engineering.

[DaMachinator]

17 minutes ago, swjr-swis said:

By the root of the wing.

Or if by 'multipart' (aren't they all?) you mean several separate wing sections that do not share a root, but are placed very close together to *look* like a single wing: angling just one section by its root, can work just as well, and would even look more like real life wings.

More complicated if you really want all the separate sections at the same angle, and something that might resonate with any OCD: even with snap off, the rotate gizmo still moves discrete units. So carefully do the same number of 'ticks' to all wing section roots. To know absolutely sure, one can overlap wing sections with the offset gizmo, switch to rotate, angle until visibly aligned, switch back to offset (gizmo will keep the same part selected the whole time, no danger of losing it because of clipping) then offset the now same-angled wing section back to where you mean it to be.

If you only think you have OCD, you're not taking this seriously enough. It's practically a requirement in aerospace KSP engineering.

I do mean several wing sections that do not share a root part, but are very carefully placed to look like a single wing. I typically use SHIFT+WASD for rotating parts instead of the rotate gizmo.

[Diche Bach]

On 7/28/2016 at 3:19 PM, swjr-swis said:

[OCD] practically a requirement in aerospace KSP engineering.

I've got the OCD bad . . .

Finally got a plane that I can more or less handle, but still some slight problems. Would appreciate any advice you can offer swjr (or anyone).

Aircraft Design: Help Needed

[Corona688]

On 7/15/2016 at 2:30 PM, Rocket In My Pocket said:

Having done some testing, the only way I could reproduce the "fishtailing" and "wipeouts" people are reporting is to forcibly hold the nose down as I accelerated down the runway

I've noticed this too.  I do actually have planes fishtailing which I didn't think should be fishtailing, but when I looked closely at them I noticed they all had a pile of force on the nose wheel and little to none on the tail wheels.  This is a suboptimal situation but the wheel bug amplifies it -- there's no slow loss of friction like you'd expect of a wheel mounted on a spring, as the down-force of the craft decreases due to lift, one wheel will switch from full friction to nothing because it's one floating-point digit higher off the ground than the other.  The minute friction of one landing gear wheel should not be able to throw around a 5 ton aircraft like that.

Clearly, the bug matters a lot less when the nose wheel lifts off first.  It probably snaps as well, but it's symmetrical.

I've noticed it on rover wheels as well.  Which wheels gain traction and which ones don't is grossly exaggerated.  Flat, symmetrical arrangements of springloaded wheels you'd expect to self-level just...don't.  Or act like they haven't, like half their wheels are in the air.  And when a wheel slips it causes motion which takes it further from stability, like how water balanced on oil is just waiting for a trigger to overturn itself.

So, I think it's both.  Wheel placement may not always be optimal but the wheel bug greatly exaggerates the problem.

Edited August 5, 2016 by Corona688

[Northstar1989]

The fishtailing behavior is clearly the result of known bugs.  The most important of these is probably the symmetry stiffness bug.

 

I can confirm that I have experienced this bug many times.  Sometimes even on a plane that worked before until I detached and re-attached the wings in 2x symmetry in the exact same spot as before in order to get a closer look at something on the fuselage...

The problem is clearly with the symmetry settings leading the wheels on one side to flex more than another, and if yoy zoom in on the wheels while driving on the runway you can often even see the bending differently...  The wings themselves also occasionally flex differently- again a known and longstanding bug...

Some designs are more vulnerable to differential wing and wheel flex than others.  Sometimes not though- there have been times where I had two identical saves of the same craft with different names (in preparation for loading a cargo onto one that I was still designing) and one would always fishtail on the runway while the other would not, despite being identical designs.  So clearly the fault somehow becomes associated with the craft file on a particular computer/  KSP install (copying the file to another computer will not always replicate it- making this an extraordinarily difficult bug to track down...)

 

And, I doubt the issue is MY design expertise.  I'm very good at plane design- I even designed a working SSTO spaceplane in Real Solar System 6.4x with FAR and RealFuels installed (I was actually the first to EVER do so without cheating- another player was the 2nd a few months later...)  Admittedly it took 6 major design iterations (and countless minor tweaks) to get to that point, and it wasn't until the 4th that I could even go orbital, and the 5th until I had some semblance of re-entry stability (revisions 4, 5, and 6 were all efforts to improve stability during re-entry.  #4 could make orbit but would spin out during re-entry and was only recoverable with a lot of save-scumming and near-perfect piloting...)  Building a RSS spaceplane is about the hardest challenge you could ever undertake (there's a reason we've never built a working true HTHL spaceplane in real life, and some question if it's even possible...  My spaceplane used slightly futuristic, next-gen fission reactors in nuclear thermal rockets and turbojets- stuff that's about 35 years out in real life, had 5 distinct engines, and even then a numerical analysis indicated I would have had to sacrifice about half the cargo to provide enough margin for pilot error to be able to make orbit in RSS full-scale...) so I don't think a lack of design skill has been behind so many of my planes spinning out or flipping on the runway- especially given how often simply re-attaching the wings could fix it...

 

My point is, give the poor kid a break.  It's not HIS fault the wheel and symmetry code is still as buggy as honey left out on an August night...

 

Regards,

Northstar

 

 

Edited August 6, 2016 by Northstar1989

[swjr-swis]

7 hours ago, Northstar1989 said:

The problem is clearly with the symmetry settings leading the wheels on one side to flex more than another, and if yoy zoom in on the wheels while driving on the runway you can often even see the bending differently...  The wings themselves also occasionally flex differently- again a known and longstanding bug...

Some designs are more vulnerable to differential wing and wheel flex than others.  Sometimes not though- there have been times where I had two identical saves of the same craft with different names (in preparation for loading a cargo onto one that I was still designing) and one would always fishtail on the runway while the other would not, despite being identical designs.  So clearly the fault somehow becomes associated with the craft file on a particular computer/  KSP install (copying the file to another computer will not always replicate it- making this an extraordinarily difficult bug to track down...)

 

And, I doubt the issue is MY design expertise.  I'm very good at plane design-

No, this is a very real thing, although my theory of the actual cause is different: I think the flex is directly caused by some parts colliding and 'pushing' others out of their alignment. Notoriously, Mammoth engines, landing gear -even retracted- and yes, Kerbals' heads/helmets (!) when in external command seats apparently have a very powerful and unfortunately very invisible collision effect on other parts and can push entire wings out of alignment.

[DaMachinator]

On 8/6/2016 at 2:41 PM, swjr-swis said:

No, this is a very real thing, although my theory of the actual cause is different: I think the flex is directly caused by some parts colliding and 'pushing' others out of their alignment. Notoriously, Mammoth engines, landing gear -even retracted- and yes, Kerbals' heads/helmets (!) when in external command seats apparently have a very powerful and unfortunately very invisible collision effect on other parts and can push entire wings out of alignment.

I've noticed parts colliding in very strange ways before, usually when coming out of timewarp or doing things that they were not meant to do. Like serving as armor panels to protect against HE anti-missile shells.

[Lego8_bit]

when i think of collisions doing somthing weird i usualy think they do this as an example 

 

[Diche Bach]

On my rig this plane has zero fish-tailing.

Seagull 104AB on KerbalX

Uses a few mods, sorry about that, but only a half dozen . . .

The main thing was: I followed the advice of the more knowledgeable forumites who responded to my "Aircraft Design: Need Help" thread (link to it in the KerbalX page).

ADDIT: if anyone requests, I'd be willing to redesign it with the modded parts left out and upload a stock version. None of them are essential to flight (just alternative engines/fuel tanks and science gizmos).

Edited August 18, 2016 by Diche Bach
more info

[Maltman]

I just went through a three hour long rage sessions over this fish tailing business. I was ready to strike my airconditioner with my bare fist... But I resisted.. barely.

 

I am so glad to see that other people have suffered from this same thing! I have been trying to build planes with the low tech landing gear. IT has been pure hell. I've tried every trick I know. I've learned a lot lurking already maybe my airconditioner will be spared next time