Realistic Wings suggestion

[AeroGav]

So, on the one hand, you have the stock game, which treats all wing parts as equivalent, and only looks at wing area, dynamic pressure and AoA to calculate lift and drag.   In other words -  a straight wing off a Cessna 150 performs as well at mach 5 as a delta.   A highly swept wing or delta performs as well as a straight wing at low speed.

On the other hand you have FAR's highly sophisticated modelling, which uses voxels/ray tracing to calculate how highly swept the complete vehicle is and calculate an individualized rate of drag increase per mach number.  (that's a gross simplification i know, partly because i don't fully understand all the concepts and partly to keep this discussion simple enough for lay people)

As a middle ground,  maybe Squad could give individual wing sections different characteristics

eg. straight wings, develop the most lift but high wave drag.

swept wings , less lift , higher lift induced drag but lower wave drag.

delta wings , less lift, low wave drag but high lift induced drag.

Variable Geometry wings could also appear in game,  able to automatically adjust by mach number to offer straight wing characteristics at low speed and swept wing characteristics at high speed, for a weight penalty.

They could also give each wing section a proper attach node for the leading, trailing edges and wing tip, so you can build up compound wings easily.    The game can calculate additional drag from wing tip vortices, from open nodes on wing tip sections.  That way there is a reason to build high aspect ratio wings for better lift:drag ratio, same as you get in airliners and sailplanes.

 

[NathanKell]

This does not work because one makes various wing shapes out of the combinations of other wing parts.

[HebaruSan]

13 minutes ago, NathanKell said:

This does not work because one makes various wing shapes out of the combinations of other wing parts.

Are you saying you don't want to see giant supersonic biplanes built out of mosaics of delta wings??

[AeroGav]

1 hour ago, NathanKell said:

This does not work because one makes various wing shapes out of the combinations of other wing parts.

well, that is one example of where it could go wrong - a straight wing built out of deltas would have the aerodynamic characteristics of a delta, not the straight wing shape that it actually has.   I don't see that as a particular worry. 

More troubling is a large delta or swept wing built out of smaller parts - it is natural to put a rectangular section on the trailing edge, but that would mess up the dynamics of your high speed design.  I guess a delta flipped around on the trailing edge of a swept wing wouldn't be so bad.   You could probably work around it  by providing a selection of stock wing parts that fit together ok like this.

[Knaapie]

I'm actually starting to want to see how far does it.. Perhaps I should use it, as I should be able to make use of it.

These things are pretty complicated due to the fact that it matters a lot of what you attach to a wing node to influence the vortex. Perhaps it might be better to go for a mean wing aspect ratio.. Adding up all surfaces and placement around the center of lift, to make it look like 1 wing with 1 aspect ratio. Combine this with all the leading edge angles in comparison with the direction of the velocity and the velocity itself, to calculate a more precise yet simplified drag model. Yet this would create more than irl drag for "biplanes" where the wings are both in front and the rear of the plane...

edit: Perhaps this calculation could be done for every wing surface attached to another wing surface.... That kinda sounds getting close to irl, while keeping simplicity and creativity...

 

Could be possible, I guess

Edited May 7, 2016 by Knaapie

[AeroGav]

9 hours ago, Knaapie said:

I'm actually starting to want to see how far does it.. Perhaps I should use it, as I should be able to make use of it.

These things are pretty complicated due to the fact that it matters a lot of what you attach to a wing node to influence the vortex. Perhaps it might be better to go for a mean wing aspect ratio.. Adding up all surfaces and placement around the center of lift, to make it look like 1 wing with 1 aspect ratio. Combine this with all the leading edge angles in comparison with the direction of the velocity and the velocity itself, to calculate a more precise yet simplified drag model. Yet this would create more than irl drag for "biplanes" where the wings are both in front and the rear of the plane...

edit: Perhaps this calculation could be done for every wing surface attached to another wing surface.... That kinda sounds getting close to irl, while keeping simplicity and creativity...

 

Could be possible, I guess

FAR is good to play with,  but i've found it just introduces too many new ways to kill Kerbals for the game to be fun any more.     Landing speeds double or triple,  so forced landings/crash landings are inevitably fatal.   Using procedural wings (not been updated for KSP 1.1+?)  I had a few designs with good transonic drag despite the large wing area, landing speeds weren't as good as hoped but 60-70m/s is fair.  Unfortunately I either had zero roll control below 100 m/s or adequate roll control at takeoff/landing  and far too much roll control at higher speed.    Gave up at this point, though perhaps i should have tried a stack of reaction wheels instead of aerodynamic surfaces.      

I like spaceplanes that are more than just simple orbit taxis - i want mine to be capable of getting a refuel from an IRSU base then landing on Laythe or Duna.     Guess that requires variable geometry wings to get a sufficiently wide flight envelope.   In the stock game, the wings all behave aerodynamically like VG.   The animation is simply missing  !

[Blaf]

On 6. 5. 2016 at 8:49 PM, NathanKell said:

This does not work because one makes various wing shapes out of the combinations of other wing parts.

Btw, why doesn't KSP use FAR approach to aerodynamics? To me it seems more intuitive (and more realistic as well)...

[cantab]

Because Squad didn't hire Ferram.

And yes, making a plane that can land at a nice friendly speed on rough terrain, and also blast it at Mach whatever, is hard in FAR. It's hard in real life too.

Edited May 11, 2016 by cantab

[thereaverofdarkness2]

22 hours ago, Blaf said:

Btw, why doesn't KSP use FAR approach to aerodynamics? To me it seems more intuitive (and more realistic as well)...

Because FAR is very difficult to use. NEAR was much more realistic than the old aerodynamic model yet it was easy to use, so Squad adopted and adapted it into the game.

[Streetwind]

49 minutes ago, thereaverofdarkness2 said:

Because FAR is very difficult to use. NEAR was much more realistic than the old aerodynamic model yet it was easy to use, so Squad adopted and adapted it into the game.

Both Squad and Ferram would likely disagree with you on this

NEAR was basically FAR with the UI for aerodynamic simulations removed, and aerodynamic failures switched off. Everything else was exactly the same. Ferram stated that he did this intentionally to prove just how much of the "difficutly" of FAR is solely in people's heads - NEAR is widely regarded as "easier", when it was in fact the exact same aerodynamics. The only thing he did was take tools away from the user, tools which were originally meant to help the player out. Turns out, as soon as the player is no longer confronted with with the means to do stability analysis, he instantly starts thinking that his planes are automatically stable...

NEAR has also little relation to the system that Squad ended up implementing into stock. They made the conscious decision at the time not to use Ferram's way of doing things (although Ferram offered), and instead used something which actually is simpler, under the hood. That's also the reason we don't have NEAR anymore these days, because there's no need for something like it - whether played straight, or tongue-in-cheek.

And of course, keep in mind that the FAR of today is vastly different from the FAR/NEAR of the .90 days when these decisions were made. It's so different as to be a completely new mod altogether.

[ferram4]

...even now seeing people saying that NEAR was easier to use than FAR makes me laugh.  It was so much easier to get into flatspins with NEAR than with FAR, but no one complained just because more difficult with less information available somehow translated into easier.  That was pretty much the only change other than no UI or options.

Anyway.  There is a very, very easy quickfix to KSP wings that could be made that don't have to worry about overall geometry at all, which is that the wing lift curve (ignoring dynamic pressure, just handling Mach effects) as a function of Mach number is completely wrong.  It should follow approximately 1/sqrt(1 - M^2) at M < 1 and 1/sqrt(M^2 - 1) at M > 1, which of course breaks noticeably near Mach 1 so some simple fairing between the curves from Mach ~0.8  to Mach ~1.3-1.4 fixes it.  This is similar to what FAR does.  Stock instead has a curve that decreases in silly ways that makes lift far too low at high speeds, as shown below.

Spoiler

Note: comparison is not perfect, stock curve does not account for tangent effects at points, FAR curve is highly approximate near Mach 1

Note that the huge drop in lift at such low Mach numbers means that stock has to apply a very high base multiplier to get planes to fly reasonably at around Mach 0.5 - 0.7 (170 - 238 m/s) where most fighters are intended to fly.  The end result is that you have multi-ton shuttles and fighters being able to float along at 30 m/s when in reality 70 - 80 m/s is the lowest speed they should be able to manage with full flaps.  Before even attempting to handle any kind of shape effects, this curve needs to be fixed and the base multiplier needs to be changed; whether it's set to assume that everything is U-2 wings, shuttle wings, or some compromise doesn't matter so long as lift as a function of Mach number is right.  That'll fix a lot of the "things feeling wrong" effects right away.

[Jarin]

13 hours ago, Streetwind said:

Both Squad and Ferram would likely disagree with you on this

NEAR was basically FAR with the UI for aerodynamic simulations removed, and aerodynamic failures switched off. Everything else was exactly the same.

Bolded seems like a pretty important difference. My biggest problem with FAR was always with my wings snapping off under stress, because I don't fly with a joystick and ham-finger the keys sometimes.

Edited May 12, 2016 by Jarin

[Streetwind]

8 hours ago, Jarin said:

Bolded seems like a pretty important difference. My biggest problem with FAR was always with my wings snapping off under stress, because I don't fly with a joystick and ham-finger the keys sometimes.

Well, nowadays you can turn them off (or configure their magnitude, if you prefer that) in FAR.

[ferram4]

Even back before NEAR existed you could turn those off.  The option was available the very second that the feature was implemented into FAR, because I knew that people wouldn't like it and would want to turn it off.

But back to the topic; anyone have any arguments for/against lift scaling with Mach number as I've suggested?

[AeroGav]

The lift : drag ratio numbers in the stock game seem too low,  8:1 at 0.4 mach decreasing to 2.5 to 1 above mach 1 - is this because of the curve you mentioned Ferram?  Real world jet airliners manage nearly 20 to 1 at 0.84 mach.          Of course, the crazy low orbit speeds and unrealistic jet engine performance still allow us to build space planes.

From what I remember, in FAR  L/D tapers off more gradually, but is still really poor by mach 4 or 5 no matter how much you sweep back or try to optimise wave drag.     For me this is the most critical phase of flight as i might only have NERV engines powering the craft at this point.

Seems we could do with something that models the hypersonic regime, where rules are different again - apparently you can get up to 10 to 1 at mach 6 using compression lift ?

[cantab]

Well, I think FAR should basically be stock, so yeah ... correct lift with Mach please.

I do find a few things difficult aerodynamics-wise. One is handling "Mach tuck" and other effects that shift CoL with speed. But now that we can pin right-click menus, I wonder if it won't be easier to use fuel transfer to trim that out. And in-flight views of the CoM/L/T would help a lot too I feel.

Another is that flaps seem nearly useless. 10 m/s slower landing if I'm lucky. Either I'm doing something wrong, FAR makes them too weak, or they're actually nearly useless in real life.

A third is just that everything seems to glide really well in FAR, and shedding speed is a pain in the backside. That may be real of course.

[AeroGav]

2 hours ago, cantab said:

 

Another is that flaps seem nearly useless. 10 m/s slower landing if I'm lucky. Either I'm doing something wrong, FAR makes them too weak, or they're actually nearly useless in real life.

 

Real flaps increase the lift produced by the entire airfoil, but it appears that in both FAR and stock, all they do is increase the lift produced by that control surface itself.   This is small in relation to total lift, furthermore, they do so by increasing the angle of that control surface relative to airflow so you get more lift if the whole aircraft is at 10 degrees nose up, because your "flaps 15" surface is now at a total 25 degree AoA, but if you pitch the whole aircraft beyond 15 AoA then this panel stalls out and stops making lift altogether.

The last aircraft I made under FAR had leading edge high lift devices to offset the pitch-down caused by the flaps.  However, if you have them droop downwards like real-world slats, they act like spoilers and make negative lift.  You actually have to angle them upwards to generate lift.   With decent sized, full span leading and trailing edge devices the speed you'd have at 10 degree AoA did come down some 25%.   Unfortunately roll control went south at low speed.   I guess when you compute the angle of attack of the flaperons =  ship AoA + flap droop angle + roll command input , it was going over 30 degrees, meaning that the harder i tried to pick a wing up , the less lift that surface made causing a roll in the opposite direction to intended.  Even if the flaperon is not going over stalling angle, it will be at very high angle of attack so will cause a lot of adverse yaw for a small amount of roll.   Reaction wheel spam, i'm thinking.