Jump to content

[1.x.x] On the particulars of Center of Lift and Center of Mass on winged craft


Val

Recommended Posts

8 hours ago, Boris-Barboris said:

It doesn't need to, it's CoL. It's builder's job to compare it's location to CoM. You have made errors while solving moment equation. And that sausage without engines would be stable with CoL right behind CoM,  because disregarded body lift is almost right on top of CoM (symmetrycal body).

I did some zero gravity experiments and  you are right once I put a nose cone on it to eliminate the source of drag at the front it was stable with COL just behind COM.  While doing this I noticed some very weird things about body lift.  Body lift is stronger on the back of the plane which is incorrect as body lift should be at the 1/4 chord point for a rod.  I also noticed that when pulling up I received no body lift  but when I pulled down I received lots of body lift with more coming from the rear.

VKiwVVe.png

fedgsUE.png

Would the fact that the back of the craft is rotating into the airstream and the front is rotatimg with the air stream make this big of a difference?  I would think the velocity as a part of rotation would be very very small.  Perhaps the devs are giving us a little something to make craft easier to make stable?

Link to comment
Share on other sites

8 hours ago, Nich said:

While doing this I noticed some very weird things about body lift.

It's not the body lift that is weird. It's just the overlay that is bugged. It's also displaying different lengths from left and right. Comparing length of different arrows have no meaning.

MK63Q7c.jpg

Edited by Val
Typo
Link to comment
Share on other sites

  • 6 months later...

     Very nicely written Val.  Lots of food for thought.  So in your experience, does the increased drag from increased main wing AoI (positive 1 - 6 deg) generally overcome the increased fuselage drag for low AoI?  Does a negative AoI on the tail-plane have a similar effect?  Does tweaking control surface AoI have much effect on drag?  

     I've only noticed terribly high fuselage AoA (completely outside the prograde bubble) at "not quite enough" speeds at very low and very high altitudes.  I wonder now if I could use control surfaces and action groups to correct AoA to compensate for fuel burned.  My Mk2 SSTO design flies, and makes orbit, but is awful to land because the stall speed is too high and the CoL shifts forward of CoM at end of mission.  If I can "set flaps," that might allow me to land it without losing pieces.  [ALT] Trimming is just too slow when you're coming up on the runway at 200 m/s to maintain stability and flipping at 150 m is a death sentence.

     I knew about wing AoI effect on lift from various childhood experiences (kites, paper airplanes, flying models), but didn't think to try it in-game.  Many real world aircraft (particularly fighter aircraft) have very significant AoI, and/or a noticeable difference in AoI between the canard, main wing, and stabilizers.  Dihedral also affects lift from tailfins and canards - which will move in-game CoL without affecting CoM or Aerodynamic Center - more dihedral gives less vertical lift for the same drag (horizontal lift increases, but is balanced using mirror symmetry).  I did tinker with dihedral for roll stability, but didn't notice much improvement.  I use mirror symmetry exclusively to add all fuel tanks, wings, engines, and control surfaces, but some designs just have a slow roll that increasing dihedral doesn't seem to correct.  Trying to trim it out just results in a slow roll to the other side.

     For new players, it might be more intuitive to picture moving AoA as moving a kite line fore and aft on the bridle (moving Center of Lift relative to Center of Mass).  Farther aft (increased AoA) makes it want to climb more (lift), but also pulls on the line in your hands harder (drag).  Too far either way is unstable.  The effect of physics warp on in-game wings is similar to a kite bending around its spine to increase sail angle in a strong wind (dihedral).  Drag stability from the vertical and horizontal stabilizers is similar to the kite's tail.  Putting the tail up front results in a flip (Aerodynamic Center ahead of Center of Mass).

http://www.kitesintheclassroom.com/parts-kite/

Link to comment
Share on other sites

20 hours ago, HalcyonSon said:

So in your experience, does the increased drag from increased main wing AoI (positive 1 - 6 deg) generally overcome the increased fuselage drag for low AoI?

Wing drag is not changed by adding AoI. At a given speed and altitude, the wings must always have the same AoA with regards to the airflow, to produce the same amount of lift.

The wings are carrying the fuselage through the air. Not the other way around. When you change the wings AoI you are only changing what angle the fuselage is carried at. (Ignoring any body lift the fuselage may contribute)

 

20 hours ago, HalcyonSon said:

Does a negative AoI on the tail-plane have a similar effect?

No. In KSP having 0° AoI on the main wing and negative AoI on tailplane only helps with the nose heaviness that craft would otherwise have from needing to have CoL behind CoM. By having negative tailplane AoI you can construct your craft so it flies straight unaided and needs much less compensation from SAS or trim.

 

20 hours ago, HalcyonSon said:

Does tweaking control surface AoI have much effect on drag?

No, it does not have much effect. if you're just tweaking them enough to make the craft fly straight unaided, then you're doing exactly the same thing that the SAS or trim would need to do to keep the craft flying straight, if you hadn't done it.

 

20 hours ago, HalcyonSon said:

I use mirror symmetry exclusively to add all fuel tanks, wings, engines, and control surfaces, but some designs just have a slow roll that increasing dihedral doesn't seem to correct.  Trying to trim it out just results in a slow roll to the other side.

There's supposed to be a small bug with joint rigidity in symmetry, which causes, especially mirror symmetry, craft to always deform slightly asymmetrically with load. Engine thrust, gravity and lift is enough. It's a small asymmetry, so it's easily compensated for by SAS and other Autopilot functions. But just big enough that it requires quite a large dihedral or very low wing load to compensate.

Link to comment
Share on other sites

4 hours ago, Val said:

Wing drag is not changed by adding AoI. At a given speed and altitude, the wings must always have the same AoA with regards to the airflow, to produce the same amount of lift.

The wings are carrying the fuselage through the air. Not the other way around. When you change the wings AoI you are only changing what angle the fuselage is carried at. (Ignoring any body lift the fuselage may contribute)

 

No. In KSP having 0° AoI on the main wing and negative AoI on tailplane only helps with the nose heaviness that craft would otherwise have from needing to have CoL behind CoM. By having negative tailplane AoI you can construct your craft so it flies straight unaided and needs much less compensation from SAS or trim.

 

No, it does not have much effect. if you're just tweaking them enough to make the craft fly straight unaided, then you're doing exactly the same thing that the SAS or trim would need to do to keep the craft flying straight, if you hadn't done it.

 

There's supposed to be a small bug with joint rigidity in symmetry, which causes, especially mirror symmetry, craft to always deform slightly asymmetrically with load. Engine thrust, gravity and lift is enough. It's a small asymmetry, so it's easily compensated for by SAS and other Autopilot functions. But just big enough that it requires quite a large dihedral or very low wing load to compensate.

Wow...  this is the most comprehensive and knowledgeable discussion of in-game aeronautics I've ever seen.  Thank you!  Probably have to take my most of my designs back to the drawing board now LOL

It would really be nice to have a mod similar to RCS Build Aid that gives a more detailed view of Aero forces while building a plane.  Lift/drag/control force arrows can be shown in flight, so I wonder if there's a way to feed those functions just enough info to get values in the SPH.  Maybe have sliders for user airspeed, altitude, and angle entry.  Output could then be visible arrows and a chart of values... Now I wonder if the tables aren't already available in KER or MechJeb...  Doesn't look like KER has it.  Does MechJeb?

Link to comment
Share on other sites

  • 2 months later...
On 9/28/2016 at 0:34 PM, HalcyonSon said:

Wow...  this is the most comprehensive and knowledgeable discussion of in-game aeronautics I've ever seen.  Thank you!  Probably have to take my most of my designs back to the drawing board now LOL

It would really be nice to have a mod similar to RCS Build Aid that gives a more detailed view of Aero forces while building a plane.  Lift/drag/control force arrows can be shown in flight, so I wonder if there's a way to feed those functions just enough info to get values in the SPH.  Maybe have sliders for user airspeed, altitude, and angle entry.  Output could then be visible arrows and a chart of values... Now I wonder if the tables aren't already available in KER or MechJeb...  Doesn't look like KER has it.  Does MechJeb?

I apologize for reviving an old thread but it's a great one (@Val, seriously, this is amazing). @HalcyonSon, FAR has some of what you're looking for where you can see stability for velocity sweeps and AoA sweeps. Val should be able to answer this better than me but this guide might not apply to FAR? The concept of CoP, CoL, AoI, etc. should all apply regardless of which aerodynamics model you're using (stock or FAR) but I don't know if the wings generate lift and drag the same way. Something that works in one might not work the same way in the other.

It would be nice to have some kind of modeling tools to use in the SPH like RCS Build Aid except for aerodynamics.

Link to comment
Share on other sites

14 hours ago, Racescort666 said:

I apologize for reviving an old thread but it's a great one (@Val, seriously, this is amazing).

It's still relevant, so no worries, and thank you for the praise. :D

Quote

Val should be able to answer this better than me but this guide might not apply to FAR? The concept of CoP, CoL, AoI, etc. should all apply regardless of which aerodynamics model you're using (stock or FAR) but I don't know if the wings generate lift and drag the same way. Something that works in one might not work the same way in the other.

The basic premises of this thread is also valid in FAR.

  • The CoL marker is not the same as Aerodynamic Center/Center of Pressure/Drag and can therefore be misleading if used as a measure of craft stability.
  • A small Angle of Incidence can always reduce drag, compared to no Angle of Incidence.

But it's been too long since I used FAR, so I don't even know if it has a CoL marker still?

 

Quote

It would be nice to have some kind of modeling tools to use in the SPH like RCS Build Aid except for aerodynamics.

I can recommend @Boris-Barboris's CorrectCoL mod for stock aerodynamics.

 

Link to comment
Share on other sites

On 27/09/2016 at 4:57 PM, HalcyonSon said:

  .  My Mk2 SSTO design flies, and makes orbit, but is awful to land because the stall speed is too high and the CoL shifts forward of CoM at end of mission.  If I can "set flaps," that might allow me to land it without losing pieces.  [ALT] Trimming is just too slow when you're coming up on the runway at 200 m/s to maintain stability and flipping at 150 m is a death sentence.

http://www.kitesintheclassroom.com/parts-kite/

I think the best thing here is not to have the CoM shift so much in the first place.  The cargo bay should be on the plane's CoM, so empty or full does not move CoM.   There needs to be an equal amount of fuel ahead and behind CG so that it doesn't shift as the fuel burns off.    The mass of the engine, at the back, needs to be balanced by the cockpit at the front.  If you rely on fuel or cargo for this, then you will have a problem when the fuel and cargo are gone.

The trouble is on larger designs.the engines are too heavy to be balanced by the cockpit alone, so you need them mounted either side of the main fuselage close to the CoM.

 

Link to comment
Share on other sites

@Val

I did some experiments with incidence/AoA.

I attached various wing segments to a fuselage at 1 degree increments, then took it for a flight with SAS set to Prograde.

From the debug menu, I enabled drag data in right click menus, and made a note of lift and drag on the wing sections at different angles of incidence.

Mach 1.7                
                 
Incidence Angle 1 2 3 4 5 6 7 8
Lift 19.5 39.7 59.9 7.63 9.15 10.68 7.32 8.14
Drag 7.8 8.29 9.1 0.88 1 1.13 0.77 0.86
L/D ratio 2.5 4.788902 6.582418 8.670455 9.15 9.451327 9.506494 9.465116
                 
                 
200 M/S                
                 
Incidence Angle 1 2 3 4 5 6 7 8
Lift 8 16.1 24.21 35.98 42.73 49.48 59.93 66.72
Drag 0.55 0.58 0.64 0.73 0.83 0.94 1.24 1.39
L/D ratio 14.54545 27.75862 37.82813 49.28767 51.48193 52.6383 48.33065 48

 

Before we start -

At the same AoA, speed and altitude, all wing and control surfaces have exactly the same L/D ratio.   The only difference is that pure control surfaces like an all moving tailplane/canard or elevon  have twice as much mass for the same wing area (ie lift rating) as a wing.  Also the basic swept wing unlocked by the Aviation tech has a lift rating/mass ratio that's half that of all other wing parts.

Disclaimer -

There is some possibility for error in the above data,  because at the time of this experiment, I thought Prograde SAS was perfect and always held the body at 0 degrees AoA exactly.   Observing with Aero Data GUI  I now know this is not the case.     If you have an airplane that settles into 8 degrees of body AoA when flown "hands off",  prograde assist will reduce this error to about 2 degrees nose up,  but will not eliminate it entirely.  

If you are trying to make something with angled wings fly on prograde assist with near zero body angle, you need to adjust the incidence of your tailplane or canard in the SPH so the plane flies at less than 3 degrees nose up "hands off".  Then on prograde assist it will fly with a body angle of only half a degree or so, and have good lift/drag ratio.

Disclaimer 2

I did some tests at mach 5.5 at very high altitude and got pretty identical numbers to mach 1.7.

 

Conclusion 1 - optimum AoA

7 degrees looks like the sweet spot, but 6 and 8 are so close it hardly matters. You could argue that carrying extra wing mass beyond LKO decreases your delta V, so going with a smaller wing angled at a higher AoA will get you more delta V for only a small aerodynamic penalty.   

Conclusion 2 - Body drag is still huge even at Prograde

This aircraft had a lower wing loading than even my spaceplane designs, and a skinny mk1 fuselage.  Even so, the whole vehicle Lift drag ratio numbers as reported by the Aero Data GUI  were a lot worse than that for the wing panels themselves.     The very best supersonic L/D seen was 4.5 to 1, just under half of what the wing itself was getting.  This means the body must be generating more drag than all the wings put together.

At subsonic speeds, the whole airplane was getting about 6 to 1 ,  which is over 8 times worse than the wings themselves.

So, body drag is massive, but only increases moderately when supersonic.  Wing drag starts off very small but increases by a lot more when supersonic, even so most drag still comes from the body in all reasonable designs.

 

  

 

Link to comment
Share on other sites

36 minutes ago, AeroGav said:

From the debug menu, I enabled drag data in right click menus, and made a note of lift and drag on the wing sections at different angles of incidence.

Angle of Incidence reduces drag on the fuselage, not on the wings. So if you want to test the effect of AoI, you need to compare the crafts total drag at different incidence angles.

Remember, for level flight at a given speed and altitude a wing will always have the same AoA.

The wings carry the fuselage through the air, which means changing AoI changes what angle the fuselage is carried at.

 

Link to comment
Share on other sites

5 minutes ago, Val said:

Angle of Incidence reduces drag on the fuselage, not on the wings. So if you want to test the effect of AoI, you need to compare the crafts total drag at different incidence angles.

Remember, for level flight at a given speed and altitude a wing will always have the same AoA.

The wings carry the fuselage through the air, which means changing AoI changes what angle the fuselage is carried at.

 

The purpose of that experiment was to find the ideal AoA for wings.

We know the ideal AoA for the fuselage is zero.

The next part is to work out how much wing we need to get our plane to climb at the correct rate with both wing and body at ideal angles.   I am writing another post on this problem right now!

Link to comment
Share on other sites

27 minutes ago, AeroGav said:

The purpose of that experiment was to find the ideal AoA for wings.

Well, in that case, your findings also seems to agree with real world pretty well. :)

Quote

Wings are typically mounted at a small positive angle of incidence, to allow the fuselage to have a low angle with the airflow in cruising flight. Angles of incidence of about 6° are common on most general aviation designs.

Source: Wikipedia

And since real world is generally using cambered wing profiles, the extra degree of Incidence Angle you found makes sense, as that would compensate for the lack of cambered profiles on KSP wings.

Edited by Val
Fix underscoring
Link to comment
Share on other sites

10 hours ago, AeroGav said:

@Val.

Mach 1.7                
                 
Incidence Angle 1 2 3 4 5 6 7 8
Lift 19.5 39.7 59.9 7.63 9.15 10.68 7.32 8.14
Drag 7.8 8.29 9.1 0.88 1 1.13 0.77 0.86
L/D ratio 2.5 4.788902 6.582418 8.670455 9.15 9.451327 9.506494 9.465116
                 
                 
200 M/S                
                 
Incidence Angle 1 2 3 4 5 6 7 8
Lift 8 16.1 24.21 35.98 42.73 49.48 59.93 66.72
Drag 0.55 0.58 0.64 0.73 0.83 0.94 1.24 1.39
L/D ratio 14.54545 27.75862 37.82813 49.28767 51.48193 52.6383 48.33065 48

 

 

@Val

I've been doing a lot of space plane launches flying at constant AoA recently.

We know the ideal angle for wings is 7 degrees, so I used editor extensions to create a custom angle snap at 7 degrees and attach  the wings to the body at that angle.  

We also know that the ideal body angle is zero, so I fly almost all of the ascent with SAS set to prograde to keep the body as near to zero angle as possible.

The next question is how much wing should we fit? 

Let's say we are flying at 240 m/s at sea level.   The wings will produce more lift than our weight under these conditions, so the aircraft will want to climb.  It will stop climbing when the air becomes thin enough that lift is no longer greater than weight.

In practice of course, a spaceplane does not cruise at constant speed.  It will continue to accelerate if it can.

However, a low wing loading design will tend to be at higher altitude for any given airspeed than a high wing loading one during the ascent.

Lift drag ratio does not appear to be affected by altitude, only mach number and AoA.

Since we need the same amount of lift, and  have the same AoA and mach number, then  a low wing loading airplane will have the same amount of drag from its wings as a higher loaded one  flying at a  lower altitude for the same speed.

However, the body of the lower wing loading design will be at a higher altitude for the same airspeed, therefore it will produce less drag.  

In stock aerodynamics,  the body creates more drag than the wings on any airplane that doesn't look like a glider,  so for the airplane as a whole, the low wing loading design will have better lift drag ratio.   

However, you also need to bear in mind that the more rapid climb of the lower wing loading airplane will make it harder for the engines to make power.

400px-CR-7_R.A.P.I.E.R._Engine_atmospher


Fortunately the RAPIER is quite forgiving.  When the atmosphere has declined to only 0.08 times sea level pressure, it still makes 0.3 times its sea level power.  So up to about 14km or so drag falls off faster than thrust.

  
Using CorrectCoL mod to plan wing size

20161201090501_1_zpslgtt4rps.jpg


This is a low tech Panther/Terrier SSTO. The Panther's thrust (in afterburner) also declines more slowly than atmospheric pressure up to 14km.  Thrust also climbs with increasing speed to mach 2.5 (750m/s), but has a very sharp decline if you go faster than that.

So, we want lowest possible drag at these "top speed" conditions so we can minimize the number of panthers required.   That means having enough wing that our overall AoA is close to zero under these conditions.

The blue vertical line on the upper graph shows how much AoA this airplane will need to get enough lift to maintain 1g flight in these conditions.   On the right you can see the boxes where i specified the conditions to simulate - 750 m/s and 14km.   And you can see that after adding and removing wings, i got the design so it needs just 0.5 aoa to fly in this regime.   The wings themselves are angled at 7 degrees with respect to the body.

 

Other optimisation strategies

The above optimisation strategy might be called a "speedrun" optimisation, since it's about getting the best airbreathing top speed possible from the plane.
400px-CR-7_R.A.P.I.E.R._Engine_velocity_


On a RAPIER engine you can see that once you get supersonic, thrust grows so much most designs will be limited by heat.  

Also, power falls off so rapidly after mach 4.5 (more than halves  between mach 4.5 and 5.5, then falls to zero thrust by mach 6) that trying too hard to optimize for a good speed here will encounter diminishing returns , making it not worth the price you are paying elsewhere.

So, another design point you could optimise for would be the sound barrier.   Due to the RAPIERs low thrust at low speed, the limit on how much you can lift with one rapier is set by the ability to pass the sound barrier.   Looking at the engine velocity curve,  any design with enough thrust to pass mach 1 should have enough to hit mach 5.5 in theory, given that the thrust multiplier in both conditions is the same.

The problem is that optimising for mach 1 at 14km results in a very low wing loading, that's not optimal for the speedrun.  It also results in a very long and tedious subsonic climb phase.        

Finally one can optimise for the closed cycle flight to orbit.   I like my interplanetary NERV powered space planes, but these engines are heavy and we want to minimise drag here to keep them to a minimum.  Here the rule simply seems to be "more wing = better" since these engines don't loose thrust as you get higher, and it minimises body drag.  The only limiting factor is the dry mass of the wings themselves impacting your delta V.  But wings are a lot lighter than NERVs.


Squaring the circle

In view of such contradictory requirements, how do you optimise?    I'm still learning this part.    The best compromise is going to lie somewhere between a mach 1 optimised design (mach 1 at 14km, huge wings) and a speedrun optimised one (mach 5+ at around 21km, tiny wings).

Notice that although the best altitude for speed on the RAPIER is 14km, based off the fact that thrust starts to decline faster than drag above this point,  you'd only use 14km for the mach 1 case, since going 5.5 mach at 14km would melt your ship.  

In my experience 2400k parts would need to be above 22km to not explode at such speeds.  In turn, such high altitudes reduce power, so something that can only just get through mach 1 is probably limited to about mach 5 realistically (which then means you can fly a bit lower on your speedrun to get more thrust and not blow up, but not by much).  


Phugoids - Welcome to Jeb's Roller Coaster aka Vomit Comet

https://en.wikipedia.org/wiki/Phugoid

Flying on prograde lock at constant AoA can set up quite a phugoid tendency.  This seems especially bad when the amount of thrust or wing is excessive for the conditions 

ie. at low altitude

or when the power available or airspeed have undergone a rapid change

eg. when you've just taken off, or have just gone supersonic and the rapiers are coming into their power band.

You can attempt to dampen these by locking the nose angle with stability assist when the nose is rising or falling through the climb angle that results in the plane neither gaining nor loosing speed.      Then monitor AoA , which will deviate from 0 as SAS tries to hold a constant flight angle.  

If you uses stability hold to suppress a climb, the AoA will go negative until the air starts to get thinner and lift declines.  Go back to prograde lock when AoA is getting close to 0 again.  

Same thing with arresting a dive , only this time AoA will go above 0 until the plane is able to find more lift, return to Prograde once AoA is nearing 0 again.

However, mild to moderate Phugoids don't actually seem to hurt performance much if at all, so i mostly just let the airplane do what it wants.

note - the wings on this one are only angled at 3 degrees, possibly not optimal.

 

Edited by AeroGav
Link to comment
Share on other sites

Well, I found out that the whole CoM/CoL relation thing is actually more of a swing, and the individual lifting surfaces (fuselage included) behave actually like static RCS.
Not only the angling is important (by angling wings or gear), but also the distance they are placed away from center of mass (while not affecting CoM/CoL spheres). The higher - the more RCS effect they give.

Thus, this plane is far more stable and maneuverable:
JpGldde.jpg

Than this one:

G6Ehlai.jpg

Thus the designs where majority of wing is placed right underneath of the CoM will produce good drag to help lift the mass up, but will not contribute to pitch control efficiency.

Link to comment
Share on other sites

@Kerbal101

That's why a plane's elevators and rudder are typically on the tail.

The wing's control surfaces are flaps And ailerons (for rolling. They are on the far end of the wing so are already far from CoM)

In short, if you want to play with weird four wing designs, that's great, but you don't have to do that to build a stable airplane.

Link to comment
Share on other sites

@Starwaster I can't comprehend the sense of your post. You started with generalization and rephrasing and ended with "four wings" being special class? Well, rockets also typically fly to the moon. Its not even remotely about four wings. Its about most players putting the wings right behind CoM - the place with lowest effect on stability, keeping the eye on perfect CoM/CoL balance placement, while ignoring the distance-to-swing-center effect and then wondering why their aircraft has difficulty taking off or flips over afterwards.

Speaking of "weird designs", how much N wing design is this one?

Link to comment
Share on other sites

6 hours ago, Kerbal101 said:

@Starwaster I can't comprehend the sense of your post. You started with generalization and rephrasing and ended with "four wings" being special class? Well, rockets also typically fly to the moon. Its not even remotely about four wings. Its about most players putting the wings right behind CoM - the place with lowest effect on stability, keeping the eye on perfect CoM/CoL balance placement, while ignoring the distance-to-swing-center effect and then wondering why their aircraft has difficulty taking off or flips over afterwards.

Speaking of "weird designs", how much N wing design is this one?

I don't think you really are unable to comprehend my post. You presented a four wing design as being more stable; I replied that you can have stable aircraft without resorting  to that kind of design. Pretty straightforward really.

Link to comment
Share on other sites

10 hours ago, Kerbal101 said:

Speaking of "weird designs", how much N wing design is this one?

Heh. That crazy old design. Needs a lot of redesigning, especially the landing gear, to work a gain in current KSP.

The reason it has so many wings is that the fuselage would simply bend so much that it was uncontrollable, if it only had one big wing in the middle.

Link to comment
Share on other sites

  • 2 weeks later...

Ok  @Val, here's a conundrum.

I got into a debate with Slashy over the canard vs tailplane thing.   His position is that canard designs are inherently less stable and that they should be avoided by newcomers.   I was arguing the opposite, that CoL and CoM are the only factors to influence handling.   We were trying to help a guy whose plane kept flipping on re-entry.    I was advising him to move his engines further forward,  but i think i lost the debate and he's going to redesign his ship as a tailed craft.

We both attempted an experiment to prove our positions, and got different results.

http://pastebin.com/74XUmWv3

20161216200113_1_zpsgmtsfbto.jpg

http://pastebin.com/73rZcd88

20161216200416_1_zpsljaesed6.jpg
These test craft are not space-capable without infinite fuel cheat.  To save time,  I used "Set Orbit" to get them in space, then put them on a re-entry trajectory and attempted to hold radial out (90degrees nose up) for as long as possible (infinite electricity).  

Spoiler

 

Both aircraft verge on excessively stable, not allowing more than 10 degrees AoA in the atmosphere.        For pictures, look at the OP -

To summarise - both the canard and tailed versions of this craft became unable to hold the radial out position at the same altitude - 52km.  On both aircraft, the nose began to fall steadily despite maximum nose up input, after this point.   On both aircraft, I was unable to keep the nose above stalling angle at 37km.

 

Slashy took one of his existing SSTOs,  and modified i into a canard , which according the the indicators in SPH should be just as stable as his tailed version - 

SSTOC1_zps3aqksway.jpg
 

Spoiler

 

However, on re-entry you can see that SAS is already applying a large nose down input to stop AoA rising further

SSTOC5_zpsxsxaueqj.jpg

Nose down input is maxed out, the plane is about to start flipping

SSTOC6_zpsgcfvhwid.jpg

He manages to recover at 500m,  thanks to the rapier engine gimbal.  Coming in to land, you can see that SAS is applying masses of nose down trim to stop the 3 -4 degree nose up AoA from turning into another flat spin.

SSTOC13_zpsbdclgnfl.jpg

 

So why the different result?  I do have CorrectCoL installed.  Also he does not appear to be using RCS build aid nor using infinite fuel cheat, so i am not sure if it's a dry CoM issue.    But, does this not prove his point, if you're unable to build one that flies correctly with the info provided to you by the unmodded game? 

Quote

Static stability is still the same as the conventional design, yet *dynamically* it becomes uncontrollable during reentry

 

Quote

You will have a much easier job if you put the elevators behind your wings instead of in front of them. Putting them in front is still within your ability as a new spaceplane designer, but it will make your ship behave in ways that you won't expect at first. If you put them in back, it will make it more difficult to hold the nose up, but it will be easier for you to control as a new pilot.
 Not saying that you should avoid a canard design, just saying that you should be aware of how such designs behave during reentry and why they act that way.

Quote

 That's why conventional designs tend to push the nose forward while canard designs tend to make it want to fly backwards. One approach is inherently stable while the other is inherently unstable. It's always better to have a design that's inherently stable. That way if you lose control, it automatically reverts to a flyable state

.

Quote

Canard deltas can be perilous and frustrating for new designers, so it is helpful to know the mechanism by which they have been known to bite you in the butt.

 

Edited by AeroGav
Link to comment
Share on other sites

On 17/12/2016 at 7:17 AM, AeroGav said:

Ok  @Val, here's a conundrum.

 

Alright, I think I have an explanation and there is no voodoo involved..

SSTOC1_zps3aqksway.jpg

The main wings have incidence but the canards do not.

The canards are angled 5 degrees lower than the main wing in flight.  That's a large nose-down trim.  It causes the CoL indicator to move backwards,  but this does not give a true picture  - you actually have more lifting surface area ahead of CG than behind it.  

At zero AoA, pointing directly prograde, the craft appears balanced.  But as AoA increases, the canard's lift rises faster than the main wing's, because it is starting from a lower value.  Eventually the main (rear) wing stalls while the front is still generating lift , and the airplane flips.

unstable_zpsje5rumlu.jpg

 

To fix this - angle the canards up by the same amount as the main wing. This causes the CoL to move forward to its true position.  You will have to adjust your wings back some more to get the CoL touching the back of the yellow ball again.   This will be stable, but it still will not fly the same as your original craft.

Why? Because I bet the original SSTO has angled main wings and a tail plane that is not angled.  Tail plane angled 5 degrees below the main wing is a substantial UP trim.   It moves the CoL forwards forcing you to slide the wings further back to compensate, leading to an airplane with a very aft CoL.   And of course, the main (front) wings will always be at a lower AoA than the tailplane, and will stall first.

tailplane_zpspif7bpzh.jpg

To create a canard that really flies similar to the above tailplane craft, you'd need to adjust the canard to be at 10 degrees of incidence with the main wing angled at 5.  That way, in both aircraft, the front lifting surface stalls at 5 degree lower angle than the one behind. It has the same built in nose UP trim as the original tailed craft, which forces us to design a very aft CoL to compensate.

uptrim10deg_zpspxlpa1e5.jpg

CoM moves aft as you pitch up.   

Conclusion 

Canards are OK

Adding incidence is OK

Canards + Incidence whilst forgetting to angle the canards too = flip happy

IF you are going to use incidence, lifting surfaces ahead of CoM should have incidence => incidence of surfaces behind CoM.

 

Edited by AeroGav
Link to comment
Share on other sites

  • 1 month later...

The OP is good, and I like it, but I read it several times and the aerodynamic center still made no sense to me (not an engineer, solely basing this off of my hazy memories of high school physics). So then I read Wikipedia and some "basics of aerodynamics" texts by NASA and others, as well as old forum posts where @Boris-Barboris argued with people about the difference between center of pressure and the aerodynamic center, and it makes a little more sense now, maybe? And I find the best way of actually improving your understanding of something is to try to explain it to someone else, so, I'll give it a shot.

When our aircraft is in flight, the aerodynamic forces acting on it are usually separated into two component forces: our familiar friends lift and drag, where the lift is perpendicular to the direction of the airflow and drag is opposite to the direction of motion through the airflow (although, in KSP I think lift might actually be perpendicular to the lifting surface instead, which probably has a whole bunch of implications that I do not fully understand the extent of). When you consider the sum of these aerodynamic forces, you find that they produce a torque on the aircraft. For the aircraft to maintain a constant attitude, the control surfaces need to be arranged such that this torque is zero (or, well, we would also also need to consider any torque caused by the engine thrust not being aligned with the center of mass, but let's disregard that here). What we're interested in here for the purposes of this discussion on "CoL behind CoM", or in fancy terms "static longitudinal stability" is obviously only the longitudinal torque, or the pitching moment.

Now, what we must avoid in order to achieve static longitudinal stability is a situation where an increase of the AoA leads to a positive pitching moment (that is, in the nose-up direction, or more strictly speaking in the "increasing AoA" direction), because this causes a positive feedback loop where the pitching moment causes a further increase of the AoA, which causes a further increase of the pitching moment, and so on. Going by the rule of thumb, we know that ordinarily the greater deal of the aerodynamic forces is lift (because our lift to drag ratio is usually greater than one), that lift is directed roughly upwards-ish, and that lift tends to increase with AoA, so going by the seat of our pants it seems obvious that if we place the center of lift behind the center of mass (around which the aircraft rotates), we'll be home safe and the resulting torque will be negative (acting to decrease AoA) and grow with increasing AoA, more the further the CoL is from the CoM. Remember, the magnitude of the torque is the force multiplied by the length of the lever arm, and this is why people say that aircraft with the CoL closer to the CoM are more maneuverable - the counter-torque caused by any AoA that differs from the equilibrium is smaller, so you need less force, that is smaller control surface deflections, to maintain the attitude you want. And all that is true - it's just that lift is only one component of the aerodynamic force, so considering only the center of lift doesn't tell us the whole story.

So, what do we do? You'd think that it would be useful to try to find the center of pressure instead, because just like we can think of the gravitational force as a single force acting through the center of mass, we can think of the total aerodynamic force as acting through the center of pressure. The problem with this though is that when the AoA changes, not only does the total aerodynamic force change considerably (both in magnitude and direction), the center of pressure also moves (it can in fact move outside the aircraft in certain conditions). So, while it is certainly possible to analyze stability based on the center of pressure and the total aerodynamic force, it is much more complex than "place center of lift behind center of mass". There is an easier way, though. Remember, what we're interested in is the pitching moment, or torque, and how it changes with AoA. It turns out that if we disregard the center of pressure and play around with where the total aerodynamic force is applied, considering it as a force which varies in direction and magnitude with AoA plus a pitching moment which varies in magnitude with that force and the length of the lever arm (that is, the distance between the point through which the force acts and the point where we choose to apply that force) it is possible to find a fixed, AoA-independent point such that if we apply the aerodynamic force there, the pitching moment about that point is constant (well, almost) regardless of AoA. This point is called the aerodynamic center, and that is what we want behind the center of mass. It accounts for the total aerodynamic force regardless of AoA and as long as it's behind the center of mass the torque about it acts to push the nose in the direction we want.

You can't actually see this point in KSP, of course, but the center of lift is reasonable substitute in many cases, at least for eyeballing purposes. We've talked about why already, but additionally contributing to why this is the case is that the pitching moment produced by the lifting component of the aerodynamic force tends to be much greater than the pitching moment produced by the drag, since the lift is perpendicular to the direction of motion while the drag is parallel to it. You can of course construct craft where this assumption goes right out the window by giving the drag a very long lever arm, especially one that isn't parallel to the longitudinal axis of the craft, but if you do that, well, I hope you know what you're doing. Adding angle of incidence to the wings does shift the aerodynamic center away from the center of lift, since the wings and the body now have different angles of attack and thus differing aerodynamic torques, but for the most part the difference is pretty small.

If you just want to make stable craft in KSP without worrying about all of this, just install the Correct CoL mod and if it says your plane is stable, it almost always is.

Does that make any sense at all? By all means, please point out where I'm wrong here. I'm sure there's mistakes - again, just working with hazy memories of high school physics, so terminology might be wrong, I may have misunderstood concepts completely etc. I think I understand what's going on better, but philosophizing all by yourself can lead you into weird mental sinkholes.

Edited by renhanxue
Link to comment
Share on other sites

Thank you, @renhanxue. That is a very well written explanation and does a great job of explaining what I'm trying to say with this thread.

 

2 hours ago, renhanxue said:

... in KSP I think lift might actually be perpendicular to the lifting surface instead.

This got me thinking. If you have Correct CoL installed and pitch your craft up 90° in the SPH, so it's pointing straight up, the lift vector and drag vector become parallel. All pointing towards the back of the SPH.

How close are the various aerodynamic centers in this situation? (CoL, CoP, AC)

Is there a difference between lift and drag? Is it even still a CoL?

Can CoL can be used as a true indicator of stability with the craft oriented in this way?

I'll have to test this.

Link to comment
Share on other sites

9 minutes ago, Val said:

This got me thinking. If you have Correct CoL installed and pitch your craft up 90° in the SPH, so it's pointing straight up, the lift vector and drag vector become parallel. All pointing towards the back of the SPH.

Then it's not lift and drag.

12 minutes ago, Val said:

How close are the various aerodynamic centers in this situation? (CoL, CoP, AC)

AC is undefined (on infinity) and has no useful meaning since assumptions that made it static on small AoA are gone.

CoP somewhere on the craft chassis, quite informative.

AC is CoL by definition in my vocabulary. There is another way to call things though - let AC be AC and CoL be blue dot in KSP. Then CoL is approximation of AC under long list of assumptions.

23 minutes ago, Val said:

Is there a difference between lift and drag? Is it even still a CoL?

Two orthogonal components of the same vector. Game class property names and tutorial pictures aren't exactly a good terminology reference.

Link to comment
Share on other sites

This thread is quite old. Please consider starting a new thread rather than reviving this one.

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...