Stability of aircraft (CL-alpha curve?)

[paul23]

Well it's kind of awkward that as an aerospace student I just can't seem to lift an aircraft off the ground. All aircraft I tried to make turn out to be longitudinal unstable, a small change from the equilibrium means a quick looping or a full nose down stall. Ideally you wish to prevent static stability through aircraft design (so no delta wing as that's inherently unstable and thus requires a lot of fly-by-wire), and dynamic stability through means of active control.

Static stability is actually quite easy, and a simple summation of forces should be enough. (Considering only the wing as lifting body):

And (look up at wikipedia for the derivation), we wish to fullfill the following requirement:

This means that a small change in angle (alpha) means that the lifting-moment & lift forces on both the tail & wing do not cause a change in moment force.

Simple enough right? One just has to know both center of lift (tail and main wing), the center of gravity. And last but not least the CL-alpha & M-alpha curves. (And for simplicity dcl-dalpha could be considered a constant value as long as we're not near the stall, similarly for dm-dalpha). But this information isn't available in KSP.

So how do you guys design your craft? How do I prevent an aircraft from tumbling.

And even worse: why does a simple aircraft tumble so quickly? Even a naivelly build craft with a front wing & tail far behind the center of gravity should be "stable" up to the point of reaching stall in the main wing long before you start tumbling.

[cantab]

Ferram Aerospace Research. There are your curves, your stability derivatives, and even a dynamic stability simulation tool.

[capi3101]

Yeah, I have to agree with cantab on this on - if you're wanting to apply what you've learned in school to KSP, install FAR. The stock aerodynamic model, while better than it used to be, is still somewhat kludgy.

[paul23]

Well yes, but I keep reading that FAR is kind of demanding on the CPU, and I'd still like to lift my 300 object station within a (real life) hour to a neat orbit.

In stock aerodynamics how is stability achieved? Especially looking at early career opportunities. (max 30 parts, max 90 science tier...)

[Kobymaru]

Well yes, but I keep reading that FAR is kind of demanding on the CPU, and I'd still like to lift my 300 object station within a (real life) hour to a neat orbit.

FAR is not killing your 300 part stations FPS. It's the single-threaded physics and excessive ressource consumption calculation. So if that's the only thing you're worried about - you shouldn't.

In stock aerodynamics how is stability achieved? Especially looking at early career opportunities. (max 30 parts, max 90 science tier...)

Uhm... I don't know what to tell an aerospace student what he doesn't already know...

IDK, just CoL slightly behind CoM rotate craft in the SPH to assure it stays that way. Also check that the CoM doesn't change significantly due to fuel consumption.

For concrete examples, please post pictures and craft files of your craft.

[cantab]

Well yes, but I keep reading that FAR is kind of demanding on the CPU, and I'd still like to lift my 300 object station within a (real life) hour to a neat orbit.

It's really not. I expect anywhere from no change to a 10% framerate reduction from FAR. The bigger problem is KSP since 1.0 has performed horribly, it was a lot better in 0.90.

[jarmund]

Well yes, but I keep reading that FAR is kind of demanding on the CPU

My computer is somewhat potato when it comes to cpu, and i've found that FAR doesn't make much difference. When my part count becomes too high, i lag both with and without far installed. For smaller crafts, i can fly just fine with FAR.

[capi3101]

My computer has a label on its side that says "Made in Byzantium"; I began using FAR a while back and I'll tell you that I have not noticed any change in overall performance of the game.

[Hodo]

Well yes, but I keep reading that FAR is kind of demanding on the CPU, and I'd still like to lift my 300 object station within a (real life) hour to a neat orbit.

In stock aerodynamics how is stability achieved? Especially looking at early career opportunities. (max 30 parts, max 90 science tier...)

FAR is anything but demanding on the CPU. It is actually one of the least CPU intensive plugins in the game.

And if you want to apply anything you know about flight physics in this game you will WANT FAR.

[Van Disaster]

As stated, seriously use FAR. Someone in a different thread pointed out that a small wing at high AoA gives exactly the same lift and drag as twice the size wing at half the AoA ( irrespective of shape ), that will tell you a lot about stock aero.

FAR does it's calculations in a seperate thread now, it won't be that that's murdering your framerate.

[Claw]

Someone in a different thread pointed out that a small wing at high AoA gives exactly the same lift and drag as twice the size wing at half the AoA ( irrespective of shape )

That isn't exactly true.

In any case, stability in stock aero is pretty simple. Ensure the CoM is a bit behind the CoM, and you'll want enough control surfaces to provide enough authority based on how large your plane is. The stock game does not spit out stability derivatives. If you haven't found it, there are three buttons in the lower left of the editor to toggle on CoL, CoM, and CoT.

As hinted at, stock aero is a simplified model. Wing sweep, aspect ratios, and wing thickness are not considered in the aero model. For some reason, that makes people quite passionate.

It really depends on what you're aiming for out of the game. If you're looking for something to augment your studies, then FAR is probably a better choice. If you don't care and just want to fly around, building things in stock is quite achievable (and you can get away with some interesting designs).

So really, the choice is up to you. If you'd like help with specific stock questions, there are plenty of folks here with stock experience. If you choose FAR, there are plenty with experience there too. I will say (regardless of the areo model) it helps us if you can post pictures of your craft. That helps us give specific advice beyond "install FAR."

Cheers,

-Claw

Edited October 22, 2015 by Claw

[Van Disaster]

That isn't exactly true.

*shrug* - I was quoting, or at least paraphrasing, I'd actually quite like to know the details there. You can build utterly bizarre craft in FAR too ( IRL also, although KSP has materials aircraft manufacturers would go to war to get ). You can also just throw craft together in FAR and have them work fine, you don't *need* to use all the scary tools.

[Claw]

I'd actually quite like to know the details there.

This might be a start.

http://forum.kerbalspaceprogram.com/threads/119108-Overhauls-for-1-0?p=2182068&viewfull=1#post2182068

I am neither for nor against FAR. As with most of my advice...It just depends on what you're after, and what you prefer.

Edit: It also occurs to me thet the mix up might actually be that two small wings that have the same lift rating as one large wing produces the same lift at a given AoA/speed, regardless of shape. Perhaps that's what you read?

Cheers,

-Claw

Edited October 23, 2015 by Claw

[cantab]

To go into some detail on the different aerodynamic models:

Stock takes a part-by-part approach. For wings it treats each wing piece as making lift regardless of what's around it which means that aspect ratio doesn't matter, you can stack wings up like playing cards for more lift, and so on. For other parts it means that clipping is ignored for drag calculations, only node attachment matters, and two ships that look the same can behave very differently depending on how you put them together. Changes in aerodynamics with Mach number are largely ignored or fudged. Overall Squad's objective with stock aero seems to be to make something "fun" and playable even if it means deliberately being unrealistic.

FAR takes a whole-vessel approach. For wings it considers how each part is placed and considers the aspect ratio of the whole wing and the proximity to other wings, tailplanes, etc. For other parts FAR goes by the shape of the whole vessel, and therefore half-hiding a part in another part will affect the drag as you would expect and a ship should act how it looks. Aerodynamics change with Mach number as you'd expect. Overall Ferram's objective with FAR is realistic aerodynamics within the limits of game and computer performance.

Both share the same parts. That helps you make crazy stuff fly in either model - powerful engines can brute force through drag issues and the magic of reaction wheels can compensate for stability issues to an extent. For example I built a UFO in FAR.

For a bit of history and completeness, old FAR used a part-by-part approach for the fueselage similar to current stock but handed wings like modern FAR. And old stock just gave everything that wasn't a wing the same aerodynamic deceleration, ignoring shape, orientation, and relation to other parts completely.

[Van Disaster]

Edit: It also occurs to me thet the mix up might actually be that two small wings that have the same lift rating as one large wing produces the same lift at a given AoA/speed, regardless of shape. Perhaps that's what you read?

No, it was definitely related to AoA & drag, and it being better to use high AoA and small wings. I'm not going to elaborate or make disgruntled noises until I've read a bit more.

I tell people to use FAR when they express some interest in aeronautics, because everything will make sense to them - that's as far as my evangelizing will go. I definitely don't recommend it for everyone ( there's no point using it if all you do is launch simple rockets unless you're interested in optimising simple rockets ), but I think I probably *would* recommend it if you're into spaceplanes in anything more than a very casual manner. It's like any middling complexity simulation - the closer it is to reality the easier it is to understand, because the nearer it is to the extremely well documented world we've lived in our entire lives. Most of my FAR planes over the years have been built the same way I'd do it in stock - stick wings on a fuselage in what looks like the right place, set the controls up ( FAR wins there hands down, stock needs that control setup ) and throw it at the sky. On the other hand I'm building for a combat contest right now & looking for an edge, and I have all these analysis tools to help there too.

The only bit that might upset people is how things change when you go supersonic, and all *that* is at the gross level you generally need in KSP is just a shift in CoL.

( seriously, this flew out of the box in FAR tetrydis did an entire thread of them )

[paul23]

That isn't exactly true.

In any case, stability in stock aero is pretty simple. Ensure the CoM is a bit behind the CoM, and you'll want enough control surfaces to provide enough authority based on how large your plane is. The stock game does not spit out stability derivatives. If you haven't found it, there are three buttons in the lower left of the editor to toggle on CoL, CoM, and CoT.

As hinted at, stock aero is a simplified model. Wing sweep, aspect ratios, and wing thickness are not considered in the aero model. For some reason, that makes people quite passionate.

It really depends on what you're aiming for out of the game. If you're looking for something to augment your studies, then FAR is probably a better choice. If you don't care and just want to fly around, building things in stock is quite achievable (and you can get away with some interesting designs).

So really, the choice is up to you. If you'd like help with specific stock questions, there are plenty of folks here with stock experience. If you choose FAR, there are plenty with experience there too. I will say (regardless of the areo model) it helps us if you can post pictures of your craft. That helps us give specific advice beyond "install FAR."

Cheers,

-Claw

Well I tried a simple aircraft, really simplistic:

But I don't really care for what model - as long as I can use a sheet of paper to calculate the most efficient way. (Actually changing models and figuring out the formulas, so you can't use easy real life examples is part of the fun).

[COLOR=#333333]Stock takes a part-by-part approach. For wings it treats each wing piece as making lift regardless of what's around it which means that aspect ratio doesn't matter[/COLOR]

This is utterly wrong.

The aspect ratio is in reality also completelly seperate from the location of tail wing. The aspect ratio is just a gauge for (low speed, so no shock wave drag) drag versus lift. As the drag is a function of the aspect ratio and lift:

https://upload.wikimedia.org/math/b/f/1/bf1fbff1724c4bf977e02068b183c96d.png]

So higher aspect ratio at the same lift coefficient (So same wing shape) means lower drag.

So my question is: what ARE the actual formulas for drag & lift, and how do they relate to the angle of attack?

Anyways the advice to get the center of lift behind the center of gravity is very solid. As said for real life aircraft this has to be the other way round. I now managed to create some semi-stable aircraft. Though why do the elevators point "up" when I try to pull the nose up? Pulling the nose up means I wish to generate more lift, so the elevators should point down to create the extra lift (they make the wing more curved).

Another thing: 9 out of 10 lift-offs fail due to me bumping around the runway and one of the wing tips then touching ground. Let alone landing in grass/desert. I am however creating aircraft to be able to fullfill the "take eva sample on kerbin" missions more efficiently.

Anyway my latest try (I removed the tailwing, since the center of lift is behind the center of mass anyways, maybe I should add a very small canard?):

It lifts off.... But that's about it, make angle of attack seems to be 2-3 degrees, and landing is as said impossible, since I can slow down enough in air.

EDIT: can someone tell me how what the bbcode for images is? kind of annoying I have to use the buttons which makes writing posts counter intuitive.

Edited October 23, 2015 by paul23

[Van Disaster]

EDIT: can someone tell me how what the bbcode for images is? kind of annoying I have to use the buttons which makes writing posts counter intuitive.

[img="link"][/img]

( and because I have to say something because that didn't count for char length - source code for FAR is on github )

Edit: be aware that the cockpit has a reaction wheel on by default. Wobbling off the runway sideways is usually a problem with landing gear alignment ( or an overloaded nosewheel ).

Edited October 23, 2015 by Van Disaster

[Claw]

To go into some detail on the different aerodynamic models:

Yes, I think that's a reasonable comparison. Though I would say stock aero didn't take a "deliberately unrealistic" approach. More of a, "left this piece out" approach. I can promise you that stock aero wasn't limited the way it is purely for the lulz factor.

Well I tried a simple aircraft, really simplistic:

But I don't really care for what model - as long as I can use a sheet of paper to calculate the most efficient way. (Actually changing models and figuring out the formulas, so you can't use easy real life examples is part of the fun).

Well, if you want to paper and pencil it, based off your text books...then yes, FAR would likely be a better option. Because it will give you much more fidelity and detailed information. If you want the puzzle of figuring out something new/different, I think they can probably both give you that. The other nice thing is that you don't really have to pencil and paper it if you don't want (with either model). It's sometimes easy to get hung up on things when trying to optimize too much (I'm guilty of that at times).

There are also a lot of people who stick to stock aero for a while and play/abuse that, then eventually shift over to FAR when they decide that's what they prefer. I often advice people to play with the stock game for a while, then decide which aspects of the game are important to you (though your approach to playstyle is always up to you). That usually helps people pick add-ons that suit their play style and interests (be it physics, parts, airplanes, space stations, science progression, life support, or whatever).

[COLOR=#333333]Stock takes a part-by-part approach. For wings it treats each wing piece as making lift regardless of what's around it which means that aspect ratio doesn't matter[/COLOR]

This is utterly wrong.

The aspect ratio is in reality also completelly seperate from the location of tail wing. The aspect ratio is just a gauge for (low speed, so no shock wave drag) drag versus lift. As the drag is a function of the aspect ratio and lift:

https://upload.wikimedia.org/math/b/f/1/bf1fbff1724c4bf977e02068b183c96d.png]

So higher aspect ratio at the same lift coefficient (So same wing shape) means lower drag.

I think what Cantab was getting at there is that the shape of the wing, or even where they are at, is irrelevant in stock aero in terms of lift/drag performance. They can be straight wing, swept forward, swept back, split so that half are out front and half are way aft. Stock aero doesn't really care about any of that, in terms of lift/drag from the wings.

So my question is: what ARE the actual formulas for drag & lift, and how do they relate to the angle of attack?

I'm not sure if there's more comprehensive info elsewhere, but here's what I have put up on the forum (linked earlier): http://forum.kerbalspaceprogram.com/threads/119108-Overhauls-for-1-0?p=2182068&viewfull=1#post2182068

Anyways the advice to get the center of lift behind the center of gravity is very solid. As said for real life aircraft this has to be the other way round. I now managed to create some semi-stable aircraft. Though why do the elevators point "up" when I try to pull the nose up? Pulling the nose up means I wish to generate more lift, so the elevators should point down to create the extra lift (they make the wing more curved).

The CoL for the wings on a real plane is in front of the CoM, but KSP shows a composite CoL. And, just like real life, if you want an inherently stable bare aircraft, the composite CoL needs to be behind the CoM (or you need some good computers to help compensate). KSP does not have "good computers" to compensate for CoL being on top of or in front of the CoM.

"elevators point 'up'" - I'd have to ask which way you mean. Typically I see this referred to as "Trailing Edge Up/Down," so I'm going to assume that's the nomenclature you are using. To pull the nose up, the elevators go trailing edge up to create LESS lift on the tail (or even a negative lift vector). You need less lift on the tail to rotate the nose of the aircraft up. Generically speaking, this causes the CoL to shift forward, which makes the plane rotate upward around the CoM. It's certainly more complicated than that, which you either already know or will be getting into soon.

Another thing: 9 out of 10 lift-offs fail due to me bumping around the runway and one of the wing tips then touching ground. Let alone landing in grass/desert. I am however creating aircraft to be able to fullfill the "take eva sample on kerbin" missions more efficiently.

Well, this could be for a variety of reasons. Sometimes it is due to slightly misaligned wheels. Sometimes they are too far aft (making it difficult for the airplane to rotate for liftoff). Sometimes it's due to having too much downforce on the nose or main gear, which causes some fuselage bending and the plane to turn.

Anyway my latest try (I removed the tailwing, since the center of lift is behind the center of mass anyways, maybe I should add a very small canard?):

So this looks okay to start with, but one thing to watch out for in KSP is there is a bug with control surfaces being attached to swept wings. If it so happens that the root of the swept wing is attached forward of the CoM, but the control surface is aft of the CoM, the flight controls will reverse. So beware of that (if you haven't run into it already). That's not an "aero" problem, it's a bug in calculating the deflection direction.

You may also wish to start out with a classic aeroplane design (with a horizontal tail or canards) and play with that for a bit to get a feel for how either aero model works. Try moving the wings forward some and putting a mirrored set of those fins on the tail too. You can also right-click and lock out specific axes for each flight control surface, which limits the amount of control coupling that can occur. (I believe FAR also includes sliders, so that you can adjust the control range of the surface.)

That's a fairly slick airplane, so yeah, I suspect it's difficult to slow down in it. You could add some more control surfaces (such as the tail I mentioned), and you can also add some "flaps." Put some control surfaces on the wings near the root and right click. There's a "deploy" option, and you can add those to the action groups for easier deployment in flight. They can take some fiddling, but can help you slow down. You can even overlap two surfaces and provide a sort of spoiler type device, which is often helpful because it doesn't much change the effective CoL vector. The other approach is to cut your engine much further out, and make sure you are on a 2 to 3 degree glidepath to help bleed off speed.

If you post more pictures, try to leave the CoM and CoL markers on. That can also help us help you better. Generally, I try to keep the center of the CoL marker near the back edge of the CoM bubble. That's about the sweet spot for control vs. stability. Though you can certainly move it around a lot from there and still have a solid design. (Again, this is stock advice.)

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

It lifts off.... But that's about it, make angle of attack seems to be 2-3 degrees, and landing is as said impossible, since I can slow down enough in air.

EDIT: can someone tell me how what the bbcode for images is? kind of annoying I have to use the buttons which makes writing posts counter intuitive.

[noparse][/noparse]

It's like any middling complexity simulation - the closer it is to reality the easier it is to understand, because the nearer it is to the extremely well documented world we've lived in our entire lives.

Yeah, I can agree with pretty much all of what you're saying. I'm starting to think maybe it's more of a case of misunderstanding that stock has inherent limitations from pieces being left out / ignored, rather than set up as some cartoony simulation of reality (maybe as it was more in 0.90).

Cheers,

~Claw

Edited October 23, 2015 by Claw

[NathanKell]

I find it particularly humorous that, prior to KSP 1.0, FAR also used the same per-part approach, with the same downsides. The main difference between 1.0 stock Aero and pre-1.0 FAR is regarding wings, where FAR does try to ascertain total wing geometry and calculate AR and sweep effects, where stock does not. That's not to minimize how awesome old-FAR was, of course; I adopted FAR in my first month of playing (June 2013) and never looked back, but we also shouldn't minimize the advances that ferram brought with post-KSP-1.0 FAR.