about plane stability

[king of nowhere]

i noticed that sometimes, when i launch a plane and all goes well, the plane will keep the orientation with just SAS set on hold attitude. it will stay perfectly still, without any input, until i touch the controls. or until it stalls for flying too high.

other times i need to touch the controls just a little bit. and then there is no way whatsoever to get the same stability. hold attitude, hold prograde, all will result in the tip of the plane going down within a few seconds. hold prograde is especially bad, because the manuevers it does to try and compensate deviations are actually causing greater deviations.

it's almost like i'm flying two different planes, one perfectly stable, one not. or perhaps in one case it is like having physical facilitation on. and in the second case flying is much more annoying, as i need to check the controls every time, and also less efficient, because i can't keep pointing the right direction and will face drag.

and i noticed this with different planes.

anyone can explain why? and if there is a way to reproduce the stability?

[Spricigo]

Oh, I'd love ta have a explanation for that too...

Some usual tricks are to use  trim, the deployment angle slider of control surfaces and briefly pressing F to disable SAS, doing small adjustments along the way. But most of it is to design the craft and pick the ascent profile to avoid those adjustments as much as you can.

[lemon cup]

Yup. If you design the craft with nearly perfect horizontal symmetry (as in it is not left or right side heavy) and if you have enough pitch authority (center of mass stays slightly forward of center of lift and you have enough control surfaces that can lift the nose with relative ease) you can get very good stability using only pitch trim.

To set pitch trim, turn SAS off and hold Alt while you slew up and down with the S and W keys. 
 

SAS is not perfect, it is a catch-all algorithm that is trying to keep a huge range of rockets and planes stable, with a lot of drawbacks. One of it’s biggest is the input delay. If SAS is holding a particularly nose heavy craft for instance, any input you make will completely dump all SAS controls the moment you hit the key and you’ll pitch down dramatically as a result, even if your only input was a roll command. That’s why it’s usually better to fly aircraft with SAS off, if possible.

[Wobbly Av8r]

I'm afraid that there is likely no silver bullet for KSP aircraft. In the spirit of conveying information on 'how to' rather than 'do this', I submit the following...

Not trying to over-simplify physics, but in the void of space, the resultant path or motion of a vehicle is a consequence of, in essence, a thrust vector and its relationship to the center of mass interacting with a single gravitational body (yes, there are localized interactions, but they are just that - localized). While in the atmosphere, however, there are numerous aerodynamic couples that complicate the forces, and thus the path of the vessel.

In steady unaccelerated flight, the forces of lift = weight and thrust = drag, but if *ANY* of those forces change, it results in a cascade of related changes that in reality even super-computers have a tough time calculating, much less your humble desk-top computer or lap-top, not to mention that the fidelity of the software cannot really approach the true dynamics of the process.

All that said, KSP attempts to retain a level of 'honesty' to their simulation and in doing so makes things rather "squirrelly", so the solution falls upon the user's ability to both make a design that is not prone to over controlling (one in which small inputs result in large movements), but also has the ability to "finely tune" the control inputs to establish and maintain the lift=weight, thrust=drag relationship. Those two concepts translate into designs that fly a specific profile that varies only minimally with the corresponding gradual change in mass due to fuel consumption.

That's the 'why', but in practice not the 'how'. The 'how' is made more possible by following certain design philosophies like not putting in an "angle of incidence" on wing/tail parts to enable a wider operating range OR building WITH various "angle's of incidence" knowing you will generally fly at one particular airspeed and will be able to set thrust and trim to maintain that condition. Also keeping the thrust vector as close in line with the center of lift so changes in thrust have minimal effect on stability is another consideration. Because take-off and landing speeds relative to the design speed are significantly different, separate lift considerations will have to be made to accommodate those flight regimes.

Anecdotally (meaning I am aware my solution is one of many) once your design has allowed for overall stability as suggested above, using the trim feature (Alt-W or S) is key to achieving the 'hands off' flight path you desire, and once established, having SAS then turned on seems to provide a solution that allows x4 physics warp to speed things along without going out of control.

Hope that helps... As always, YMMV. Good luck!

[Snark]

23 hours ago, king of nowhere said:

other times i need to touch the controls just a little bit. and then there is no way whatsoever to get the same stability. hold attitude, hold prograde, all will result in the tip of the plane going down within a few seconds. hold prograde is especially bad, because the manuevers it does to try and compensate deviations are actually causing greater deviations.

it's almost like i'm flying two different planes, one perfectly stable, one not. or perhaps in one case it is like having physical facilitation on. and in the second case flying is much more annoying, as i need to check the controls every time, and also less efficient, because i can't keep pointing the right direction and will face drag.

and i noticed this with different planes.

anyone can explain why? and if there is a way to reproduce the stability?

Longer explanation below, but the excutive summary is:

• CoM should be towards the front of the plane
• Pitch and yaw controls (e.g. on the aircraft tail) should be as far behind the CoM as possible
• Vertical stabilizer on the tail shouldn't stick up high above the fuselage, and should have roll authority disabled
• Show us a screenshot and we may be able to offer advice

Okay, now for the wordy part:

The common reason for an unstable plane (especially on the pitch axis) is that your center of mass is behind the center of dynamic pressure.

What I mean by that:  Imagine you have a badminton birdie.  If you throw it with the heavy weighted end in front, then it's stable and will fly through the air in that orientation just fine.  But if you try to throw it with the feathered end in front... it just aint' gonna happen.  It'll flip around; the weighted end wants to be in front.

It does that because,

• the CoM wants to be in front, and
• the draggy part wants to be in back.

So for example, if you have a plane with a big lightweight fuselage sticking way out in front, but a lot of heavy engines at the back of the plane, you end up with a CoM that's pretty far to the rear, meaning the plane kinda wants to fly backwards.

You can make up for that a little bit with control surfaces... but not only is that a bit dicey, but also the effectiveness of a control surfacve depends on its lever arm, measured to the CoM.  If the control surface is too close to the CoM, it has very little control authority on the plane.  For example, if you build a craft whose CoM is way towards the back of the plane... then that means that any pitch controls on the aircraft tail ("elevators") are pretty close to the CoM and won't actually do much for you, meaning they can't help stabilize the plane's pitch well.

Also:  If you have a vertical stabilizer on the tail to control yaw... you should make it as vertically close to the center axis of the plane as possible (i.e. not sticking way up high above the plane).  And be sure to disable roll authority on it (i.e. it should be set to address only yaw, not roll).  I could explain in detail why this is, but it would be a longer explanation and I'll save it unless you're interested. 

23 hours ago, king of nowhere said:

and i noticed this with different planes.

Yep.  It really depends a lot on how the plane is arranged-- where's the CoM, where are the wings, where are the control surfaces. 

If you could post a screenshot of a plane that's having difficulty, then we could critique it for you ("oh, see, there's your problem right there-- you need to move this thing over there," that sort of thing).  Ideally, a shot of it in the SPH with the CoM display turned on would be best.

[king of nowhere]

48 minutes ago, Snark said:

Longer explanation below, but the excutive summary is:

• CoM should be towards the front of the plane
• Pitch and yaw controls (e.g. on the aircraft tail) should be as far behind the CoM as possible
• Vertical stabilizer on the tail shouldn't stick up high above the fuselage, and should have roll authority disabled
• Show us a screenshot and we may be able to offer advice

 

yes, i know the basics. and i wouldn't say it's exactly "having difficulties". it flies fairly well, it has ssto capacity, it can go cruise around on propeller power alone and can even take off from water, all on propellers.

I just would like to understand why it has that strange behavior

 

Quote

If you could post a screenshot of a plane that's having difficulty, then we could critique it for you ("oh, see, there's your problem right there-- you need to move this thing over there," that sort of thing).  Ideally, a shot of it in the SPH with the CoM display turned on would be best.

i already asked regarding this plane for a problem on the ground (which was fixed), and the pictures are there.

i made some changes since that model; i swapped out the liquid fuel tanks with rocket fuel, and i added some extra tanks under the wings, and i added two aerospikes, because i realized going to orbit on a nerv alone was too ambitious. and since the plane was much heavier i added another couple of propellers, and a few more rtg inside the cargo bays to support it.

but it makes no difference. it always does the same, sometimes starting perfectly, but losing stability as soon as i try to manuever, to never recover it

Quote

 

 

[Snark]

So, this plane, then?
 

Spoiler

It looks to me like the CoM is probably too far to the back of the plane.  Could you post a SPH screenshot that has the CoM display turned on?  (yellow-and-black checkered sphere).  (Especially with all the changes you mention, since that could have a major effect on the plane's stability depending on what you've got where.)

I have some suggestions if my hunch is correct, but I don't want to make assumptions without actually getting an accurate picture of the CoM placement.

Also... is there any chance the plane may be bending?  You've got a long narrow plane with only 1.25m stack and a whole bunch of stacked parts, meaning a lot of part joints-- that's the sort of scenario that can easily lead to bending under stress.  If the part of the plane in front of the CoM with the front wings is bending, that could cause bad instability at speed.  Are you using anything to stiffen the plane, such as autostrutting the engine and rear wings to the cockpit?

[king of nowhere]

6 hours ago, Snark said:

So, this plane, then?
 

It looks to me like the CoM is probably too far to the back of the plane.  Could you post a SPH screenshot that has the CoM display turned on?  (yellow-and-black checkered sphere).  (Especially with all the changes you mention, since that could have a major effect on the plane's stability depending on what you've got where.)

I have some suggestions if my hunch is correct, but I don't want to make assumptions without actually getting an accurate picture of the CoM placement.

Also... is there any chance the plane may be bending?  You've got a long narrow plane with only 1.25m stack and a whole bunch of stacked parts, meaning a lot of part joints-- that's the sort of scenario that can easily lead to bending under stress.  If the part of the plane in front of the CoM with the front wings is bending, that could cause bad instability at speed.  Are you using anything to stiffen the plane, such as autostrutting the engine and rear wings to the cockpit?

here it is. yes, the CoM is a bit back, which is unavoidable when the heaviest parts are the nerv engine, which must go on the back, and the wings, which must generate lift behind the CoM and so must be put fairly backwards themselves.

still, I remark that this thing flies pretty well, as long as i never touch the control since leaving the runway.

regarding autostrutting, i did indeed autostrut everything I considered also adding manual struts, especially for the lateral tanks. they would probably be a bit more stable, and they would definitely be more stable if i attached them to the central body instead of the wings. but i like the shape now, and in all my simulations it has never been a problem

[Echo__3]

This video might interest you.

 

[Wobbly Av8r]

@king of nowhereDon't know if anything I posted was significant in your mind, but those really large "elevons"  are the kind of thing that will cause you to have a wild ride; when you state that "this thing flies pretty well, as long as i never touch the control since leaving the runway" that's where I'd *start* to make it more stable.

Keep in mind that every time you press one of the control keys (WSAD) you are giving a "full deflection" instruction to those controls - if you "tap" the control key, it may be less of an input only due to the fact that the key press duration was shorter than the time it takes the control to fully deflect. Furthering the problem is that as you get faster, the effect is increased.

A couple of solutions to try: limit the total deflection of the controls, make them smaller or place them further inboard to limit their roll authority and see if that makes it a bit more manageable.

[Edit: One additional observation, below]

Edited November 6, 2020 by Wobbly Av8r
Additional information

[FleshJeb]

The COL is definitely not telling the whole story, so you have a lot less margin than you think. CorrectCOL is an amazing mod and will account for all the drag-producing items on the craft, as well as show the stability at various angles of attack.

I suspect that when it departs from perfectly forward flight, it's not stable at all. You can pick up the craft and rotate it a bit in all axes to see where the COL ends up.

If I were to modify that to make it fly better, I'd make the elevons flat (no AoA) to pull the COL back a little. I'd also double or triple the vertical tail, because the yaw stability is currently minimal--Especially with AoA on the wings.

[Wobbly Av8r]

Apologize, but after looking it over one more time I recognized also that (contributing to instability with any maneuvering) the amount of area you have in FRONT of the CoM is considerable - while wings and vertical fins can stabilize things, when you give it enough 'yaw', the airplane will act just like a rocket with a huge payload on top with little or no fins on the bottom (i.e. it will want to "swap ends"!) What @FleshJeb mentioned "I'd also double or triple the vertical tail, because the yaw stability is currently minimal" (more vertical fin-nage) will help solve the issue...

Edited November 6, 2020 by Wobbly Av8r

[king of nowhere]

11 hours ago, Wobbly Av8r said:

@king of nowhereDon't know if anything I posted was significant in your mind, but those really large "elevons"  are the kind of thing that will cause you to have a wild ride; when you state that "this thing flies pretty well, as long as i never touch the control since leaving the runway" that's where I'd *start* to make it more stable.

Keep in mind that every time you press one of the control keys (WSAD) you are giving a "full deflection" instruction to those controls - if you "tap" the control key, it may be less of an input only due to the fact that the key press duration was shorter than the time it takes the control to fully deflect. Furthering the problem is that as you get faster, the effect is increased.

A couple of solutions to try: limit the total deflection of the controls, make them smaller or place them further inboard to limit their roll authority and see if that makes it a bit more manageable.

[Edit: One additional observation, below]

i reduced the maximum angle to 20 degrees already, because i know the highest deflection level is very inefficient. i didn't want to do all the way down to 15, though, in case i needed some harsh manuever.

you are right that the elevons were oversized, in fact i distinctly remember swapping out those bigS for something smaller already in the past. probably i messed up with saving games and i got stuck with the old oversized elevons without realizing.

8 hours ago, FleshJeb said:

I suspect that when it departs from perfectly forward flight, it's not stable at all.

wrong. of course if i pull all up suddenly it will capsize, but moderate deviations (say, up to 10 degrees away from prograde even in the worst aerodinamic conditions) are handled just fine.

Quote

If I were to modify that to make it fly better, I'd make the elevons flat (no AoA) to pull the COL back a little. I'd also double or triple the vertical tail, because the yaw stability is currently minimal--Especially with AoA on the wings.

huh? by making the elevons flat, it pulls the CoL forward, not back

increasing the vertical tail instead worked,

8 hours ago, Wobbly Av8r said:

Apologize, but after looking it over one more time I recognized also that (contributing to instability with any maneuvering) the amount of area you have in FRONT of the CoM is considerable - while wings and vertical fins can stabilize things, when you give it enough 'yaw', the airplane will act just like a rocket with a huge payload on top with little or no fins on the bottom (i.e. it will want to "swap ends"!) What @FleshJeb mentioned "I'd also double or triple the vertical tail, because the yaw stability is currently minimal" (more vertical fin-nage) will help solve the issue...

how would I reduce the amount of area in front of the CoM? All i have there is part of the wings, and I can't exactly move the wings around without messing with the CoL...

 

anyway, i implemented the changes suggested: I swapped out the big elevons for smaller ones and i put them closer to the body, and i swapped out the single vertical fin for two of them. It is, indeed, somewat more stable. I am able to manuever around and then find again a trajectory that won't require constant correction. though that's not always easy, sometimes i point prograde and leave the controls and it keeps flying straight, and sometimes the nose starts going down and i have to manually correct until i find a new stability. So my original doubt "why, when i stop touching the controls, sometimes it keeps flying perfectly straight and sometimes it will not?" is still without answer. i guess it would depend on complex aerodinamic parameters determined by the exact trajectory and pressure

Since I was there, i also attached the lateral tanks to the main body.

 

EDIT: what the....?

i went back to trying some water landing, and i'm finding out it doesn't fly well anymore. I didn't touch anything since saving the plane after it did fly well... and i can't find stability anymore. and suddenly i have no roll-pitch control

it's like the game wants me to revert all the changes i made. but why was it flying correctly the first time i tried them, only to fail the second?

EDIT 2: I figured out why i needed the big elevons: it's to land on water. water landing requires careful manuevering at low speed; smaller elevons are more effective at high speed, but they're no good if i can't land. generally speaking, this plane's major goal is exploring laythe, and I can now recognize that some of the things I did that were a bit less effective for flying were meant to improve water landing survivability

EDIT 3: i am unable to land on water anymore. to check if it was the new changes, i tried with an old model, and i discovered i also can't land on water anymore. maybe i checked water landing before adding extra fuel tanks. or maybe the game is inconsistent again. anyway, while i was piloting the old plane, i found it behaved well too; i could find stable trajectories easily enough. all things considered, the only improvement is the double fin tail. anyway, i need to alter the project to land on water again. possibly without screwing up flying and going to orbit while at it

Edited November 6, 2020 by king of nowhere

[jimmymcgoochie]

Canards might help make this design more stable and improve its controllability. Put a pair of fins (they don’t have to be the specific canard parts, other winglets or fins will do as long as they have control authority) on the nose ahead of the cockpit and they will greatly improve your pitch control at the same time as adding a noticeable amount of lift at the front (if angled correctly), pushing the CoL back quite a bit and allowing the main wings to be shifted forwards which will in turn push the CoM forwards too. Canards are also very useful for water takeoffs as they pull the nose up and out of the water rather than pushing the tail down and into the water like trailing edge elevators do.

If I read this thread right, you want to use this plane as an SSTO for Laythe? Jet engines would be the best choice, a hybrid of Whiplash ramjets and a NERV would most likely be enough to get into orbit due to Laythe’s low orbital velocity and fairly thin atmosphere and would also have the big advantage of using solely liquid fuel and no oxidiser. Since that plane already has an ISRU in it (I think, looking at the screenshots) it would be easier to refuel it with just one fuel type- you could easily game the system by adding one small tank with oxidiser to run a fuel cell or with a drain valve to dump it, then run the ISRU for both LF/Ox and LF only to refuel faster.

You might want to try adding a pair of outriggers underneath the plane to make water landings easier, see the stock Osprey for one example of how to do this. They’ll add quite a bit of drag making it harder to fly into orbit, but can also hold fuel. You’ve mentioned propellers, which I’ll assume means you have Breaking Ground and which also means you could make the floats retractable for flying up to orbit and then deployable for water landings; some heavy duty hinges might be the best bet.

And back to the original point of the thread... Planes are rarely stable in flight as all the different forces acting on them change considerably depending on things like speed and altitude. You might want to consider using an autopilot mod of some kind to fly the plane for you, it makes life easier when the computer does the hard work for you.

[Snark]

13 hours ago, king of nowhere said:

here it is. yes, the CoM is a bit back

Yep, that's your problem right there.  It's really far back.  Unless you can figure out some way to move it forward, you're gonna have some problems.

You could scootch that LV-N forward a bit and clip about 2/3 of its length inside that fuel tank... that would help with the CoM, but not sure if you're okay with the aesthetics.

Probably the easiest way to move CoM forward would be to take those two "outriggers" (the ones with the aerospikes on them) and slide them way forwards, so that the aerospikes are closer to the middle of the plane rather than the rear.

Another problem you have is that you have very little control authority.  With the CoM so far in the back, those ailerons on the trailing edge of the wing have very little lever arm to work with, since they're not very far behind the CoM.  So you need to figure out some way to get pitch authority from control surfaces that are not close to the CoM.  One thing you could do would be to add some canards to the front of the plane, as far forwards as you can.  AV-R8 winglets work well for this.  Set them to reduced range of movement, and I'd advise giving them pitch authority only.  Since the CoM is far to the rear of the plane, that means that canards up front will have a really big lever arm to work with and ought to help with stability quite a bit, as long as you don't let your AoA get too large.

Have you made sure that your vertical stabilizer has roll authority disabled?  (It's on by default-- you have to manually turn it off.)

 

[king of nowhere]

you know what, guys, after some more extensive testing i realize the plane is perfect as it is, for my purpose.

it is supposed to grab science from laythe in every biome, both on the ground (well, water) and flying over it. this means making a lot of distance over the sea, while still keeping enough fuel to orbit. so my choice of propellers is the correct one.

at this point, i must be able to land on water. that, i discovered, can be done even with the new version, it's just a matter of figuring out the technique. i was trying to hit the water as slowly as possible and absorb the impact with the wings, which have a decent impact tolerance; this worked with the lighter model, not with this. but then i discovered that if i enter water very carefully touching only with the wheels, those wheels will slow the plane enough to safely enter with the fuselage too, and i can do it at 70 m/s reliably. i've done it 5 times in a row without crashing, and while i'll still want to save the game before landing, it's a fairly safe manuever.

taking off from water is not strictly necessary, but it is highly desirable.  i can use my propellers to move around on the water surface and pretend i'm a boat, but i'm limited to around 70 m/s top speed. I can go above 200 m/s in the air, it's gonna make things much faster. And to take off from water, a rear CoM is desirable; it helps me tilt the plane upwards for take off. i tried to move the CoM forward by moving the big fuel tank in front of the command pod, and i discovered that I couldn't take off anymore; when i tried to tilt the tip up, the back of the plane would sink in the water, slowing me down. I also need the huge elevons, to help me get off the water.

I tried to add a third pair of wings to improve manueverability at low speed, but they didn't work as well, and they apparently created all sorts of issues. plus they pushed the center of mass way too forward.

So I'm basically going with my first model, just with 2 tail fins and the elevons closer to the center.

Your input was useful, though. it made me realize that while I could make this plane fly better, I could only do it by sacrificing something else. So there are no engineering mistakes in my design, only trade-offs.

3 hours ago, jimmymcgoochie said:

If I read this thread right, you want to use this plane as an SSTO for Laythe? Jet engines would be the best choice, a hybrid of Whiplash ramjets and a NERV would most likely be enough to get into orbit due to Laythe’s low orbital velocity and fairly thin atmosphere and would also have the big advantage of using solely liquid fuel and no oxidiser. Since that plane already has an ISRU in it (I think, looking at the screenshots) it would be easier to refuel it with just one fuel type- you could easily game the system by adding one small tank with oxidiser to run a fuel cell or with a drain valve to dump it, then run the ISRU for both LF/Ox and LF only to refuel faster.

 

Propellers are better. laythe has way too little land to make isru reliably (unless i can mine ore from water? i never tried), and anyway i realized i don't need it. i land on laythe, move around with propellers (fueled by 8 RTG), then i use the fuel only to go back in orbit. I will have a mother ship waiting me, and it will have isru capacity from the smaller moons. not that i will actually need a refueling, since for the other moons i will use a rover.

the mission actually has another part earlier, where i dip inside jool atmosphere enough to reach the low atmosphere, do science there, and start again. for this, i coupled this plane with another, bigger thing that will brake during descent (thanks to an inflatable thermal shield used as a parachute, i already tested the concept) and will then propel me in orbit again. this plane then will be the last stage, the one reaching jool orbit. there it will couple again with the main ship to get refueled and explore laythe.  my choice of engine works for jool ascent too.

the final version i picked

and the jool ascent stages. the jool ascent is a pretty poor flyer, but it does its job, and since i intend to only use it once - and it has two vectors to help stability - then it's fine

Edited November 6, 2020 by king of nowhere