Pitts / Fixed Wing Flight Model Review

aero4428 · January 22, 2019 at 5:43 PM

Thanks! The airbrake works pretty well. Slows down cruise slightly which I can adjust. Just requires some thinking about parameters to prevent each change from having unintended consequences on something else.

W.r.t. props, I tested lowering the governor minimum pitch to 0 which made the plane sink like a stone at idle albeit with the engine maintaining cruise RPM. So the negative alpha of the prop blades w.r.t incoming air in this case is properly working to slow the plane down rapidly. At least thats what i'm assuming is happening. I also tried increasing the lift coefficient of the blades thinking this would also increase negative lift in this negative alpha scenario. That also seemed to work. But, it would be useful if the lift coefficients for airfoils had optional separate values for positive and negative alpha. What I did also was influencing the lift in positive alpha cases.

My sense is part of the issue is related to what you mentioned -- the RPM is staying too high even at relatively low manifold pressure settings. I.e., the engine in real life has more of a braking effect than what is currently being modeled. I am going to see what RPM the actual plane drops to at 11 inches MP and 110 mph -- the conditions from abeam the numbers and base leg.

2. Related to my second question, I found that lowering the tailwheel spring constant to ~200 makes the ground handing fairly realistic at ~30-45mph, a little too squirrely below that, and a little too stable above that. So I thought it would be useful to have the spring constant a function of speed.

Jet-Pack (IPACS) · January 22, 2019 at 6:41 PM Official Post

Airfoils have a linear lift increase from the negative stall all the way to the positive stall, we can't deviate from the experimental results here, that would be the wrong direction.

ClAlpha is the gradient of the lift, that is lift coefficient per radiant of angle ot attack. 2 pi is the theoretical value for a thin airfoil, so around 6.28 is the correct physical value. Only for very short props you would see a decrease and only for insanely thick airfoils (e.g. towards the spinner) you would have a slightly higher value.

Cl0 is the lift at zero angle of attack.

Please check out the wiki for more details on the airfoil parameters: https://www.aerofly.com/dokuwiki/doku.…aft:tmd:airfoil

What you probably want to do is have the blades stall earlier in the negative region, for this you will need to change the AttachedCenter value, I think towards negative, so sth. like -0.07 for 4 degrees.

Overloaded · January 22, 2019 at 8:33 PM

An airflow driven flater pitch blade acts like an upside down thin cambered wing so the airflow on the rear/reverse side will be flowing against unstable, increasing then decreasing pressure unless the rpm and airspeed dependent coarse relative flow direction totally stalls the blade. The rotating blade will be generating significant ‘lift’ in the reverse direction unless the engine stops when the simple turbulent flow past the blades will produce less rearwards force.

Overloaded · January 22, 2019 at 8:57 PM

Quote from aero4428

My sense is part of the issue is related to what you mentioned -- the RPM is staying too high even at relatively low manifold pressure settings. I.e., the engine in real life has more of a braking effect than what is currently being modeled. I am going to see what RPM the actual plane drops to at 11 inches MP and 110 mph -- the conditions from abeam the numbers and base leg.

And the anti rotation drag from the ‘flicked angle’ airflow encountering what resembles the wing of the FS2 Sopwith Camel flying inverted.

KalEl · January 22, 2019 at 11:38 PM

Hi Jan (Jet-Pack) ! Thanks a lot for your forthcoming informations regarding the Pitts flight-model --> it matches to some observations I made on my rig at that time when I had both Aerofly versions installed: Aerofly FS (released 2012) and of course AFS2 from now.

Which leads to my question: May it be true, that the flight-model of the Pitts in Aerofly FS 2 has been "defused" for PC - Simmers (less rudder input) compared to the Pitts in Aerofly FS (2012) ?

Because I find the Pitts in version one more realistic regarding its flight-physics/model, more depending on correct rudder inputs. Especially during take-off it acts - from my perspective - more aggressively and realistically! ---> annotation: I am a strikt rudder-pedals "flyer".

Then as I tried the Pitts an AFS2 I even became some kind of disappointed (reg. too easy, less challenging to "fly").

Are these observations only the result of my imagination, or do I have a point there?

All the very best,

sincerely Alex, from good old Germany

Jet-Pack (IPACS) · January 23, 2019 at 8:10 AM Official Post

Short answer: yes it was changed, not particularly to "defuse" it. I know especially the knife edge suffered a bit from the changes I made to get it to tumble...

Redtail · January 23, 2019 at 1:11 PM

Quote from aero4428

I am a private pilot (and fairly avid VR flight simulator user) and have been taking real-life instruction in a Pitts S-2C (and have some time in a Pitts S-2A) out of KSDL. I have been using the AeroFly Pitts S-2B as a training aid to supplement real-world flying. Site picture is perfect in VR, control response, and speeds are pretty close but there are some gaps that exist that I have identified going back and forth between real plane and simulator. These probably apply to other planes as well but I am limiting discussion just to the Pitts because that is all I have been using in Aerofly.
In the air:
Adverse yaw is very understated in AeroFly especially at slow speeds. Real Pitts has a lot of it.

P-factor doesn't seem to be modeled in Aerofly. During climb at 110mph in real Pitts, if you take your feet off rudder pedals, plane will yaw by ~5 degrees per second to the left requiring constant right rudder. In fact, in a left climbing turn, no left rudder is needed - you only need to reduce right rudder pressure to stay coordinated. In AeroFly you need no rudder input, and you still track perfectly straight. The converse of this is when pulling power (for example abeam the numbers on landing) in the real plane the nose will move to the right requiring momentary left rudder input due again to p-factor.

The model will not forward slip. Anything more than a mild aileron input and there isn’t enough rudder authority to prevent the aircraft from simply turning. The real Pitts forward slips incredibly well.

In flare, the Aerofly Pitts floats a bit long. Even with the two-blade prop it should decelerate a faster and touch down.

On the Ground – Landing Roll and High-Speed Taxis:
The real Pitts is very sensitive on the ground during rollout and is easy to overcontrol. To get students used to this, the instructor has you perform high speed taxis on the runway (with tail down) at about 30mph. Goal is to first track straight down centerline which is surprisingly hard the first few times. The plane is easy to overcontrol with rudder. So, a slight deviation will result in you putting too much rudder in (and/or leaving it in too long which is probably the bigger issue) giving you a slightly bigger deviation the other direction, and so on. By swerve 3 you are on your way to running off the runway. Anyway, you eventually get the feel of this. Then he will have you move left of centerline 10 feet and then right of centerline, etc. Until you can do it without getting into any swerves. This whole procedure is performed with the stick completely back. In Aerofly, with the stick completely back, it is as if the plane is on rails. It is perfectly stable and the tailwheel is basically locked to the ground. This isn’t accurate. I have found you can sort of recreate the real instability in Aerofly by keeping the stick 3/4 of the way forward of neutral and keeping the speed low. In Aerofly, this will sort of recreate the swerve tendency that exists in real life. However, when you start picking up any speed in Aerofly, the stability increases regardless of stick position (for example, when you are still moving fast on landing roll). In real life, this swerving tendency also exists at higher speeds especially when decelerating on rollout. (The plane is obviously easier to control on takeoff with full power pulling the nose forward giving the plane more stability.)
On the Ground – Regular Taxi
In a real tailwheel plane when doing s-turns for example, once you input rudder and get the turn started, the plane continues to travel through the arc even after you neutralize rudder. In Aerofly, the turn stops as soon as you neutralize rudder.
On the Ground – Takeoff
Real-plane – Smooth full power with stick initially fully back combined with very gentle rudder inputs to keep tracking straight. Requires some constant right rudder presumably due to p-factor of the plane sitting at a high angle of attack in three point attitude and engine torque putting more weight and friction on the left tire. About the point you get to full power, release back pressure on stick slowly. The stick will move forward on its own through about the midpoint between full aft and neutral which will require a gentle forward pressure to push through this. If you push too hard through this (which I have done), you get a very pronounced and startling swerve to the left due to gyroscopic precession of the tail coming up which you inevitably overcorrect for giving you a swerve to the right! Anyway, get the plane in slightly nose high attitude and then keep tracking straight until it flies off. (When first learning, there is a very strong urge to just lift the plane off too early. This is because once the wheels get light, they sort of start hopping and skipping across the runway giving you the urge to just get off the ground and into the safety of the air!). This is obviously wrong, but I had a lot of mental trouble with this.
In Aerofly, from the start there is still too much stability in the takeoff roll. Also, in Aerofly if you push the nose forward too quickly it doesn’t show the exaggerated gyroscopic forces. In fact, a while ago Aerofly actually had the force going in the wrong direction. When you would push the tail up, the plane would yaw to the right. I had submitted a bug and it seems like it was sort of fixed. But either way, the stability on takeoff in the sim is too much and the gyro force is understated.

Conclusion
I am by no means an expert in the Pitts. I am a student learning. But I have had the unique perspective of going back and forth between the real airplane and the simulator and trying to closely document the differences. I saw there is a professional flight model for the new R-22 helicopter. I was hoping there might be a professional flight model for fixed wing aircraft and also hope that this post may be useful. I would be happy to provide any more input and perhaps can take some GoPro video of the next flight to demonstrate some of the things I have mentioned in this post. Aerofly is a really great sim and I would love for it to get to an even more advanced place.
One final note, the most accurate tail wheel sim I have used so far is the DCS P-51 Mustang. I obviously have never flown a real Mustang but prior to getting my tailwheel endorsement, I practiced heavily in the DCS Mustang – takeoffs and landings (normal and crosswind), high speed taxis, and wheel landings. I flew in a Super Decathlon for my actual tailwheel endorsement and found it very easy and intuitive. To me that is proof that the DCS flight model is done very well. It is mainly a military sim though. If Aerofly could recreate that on the civilian side it would be fantastic. For reference, my hardware consists of Warthog flight stick and throttle with 10cm extender on the flight stick to match real plane dimensions better. I have slaw device rudder pedals which was an upgrade from MFG Crosswinds. All of the tailwheel stuff I mentioned requires good rudder pedals to get value out of a good flight model.

Display More

I got my tailwheel endorsement and basic aerobatic course in the Super Decathlon as well. It's the only TW airplane I've flown, but boy the most fun plane as well!

Forward slips are very impressive in it, compared to doing them in the Cessnas and Pipers I've flown. A lot of fun.

Took a little while to nail the wheel landings though!

"I did a lot of them in the real Super Decathlon and the behavior was very similar to doing them in the DCS Mustang. Was actually easier in the real plane for me which again is why is see so much value in simulation even for stick and rudder training."

Yep, definitely easier in the real plane than the DCS Mustang! But very challenging at first. Had a few PIO moments, in which I promptly powered up and went around.

Man, I'd love to fly a Pitts or Extra one day!

Really enjoy reading this thread.

Thanks.

Trespassers · January 23, 2019 at 2:50 PM

Quote from aero4428

Thanks! The airbrake works pretty well. Slows down cruise slightly which I can adjust. Just requires some thinking about parameters to prevent each change from having unintended consequences on something else.
(...)

When you get a satisfying improved behaviour, please make sure to share your tmd's as a package on https://flight-sim.org/

Thanks for your inputs

Cheers

Antoine

KalEl · January 23, 2019 at 3:54 PM

Quote from Jet-Pack

Short answer: yes it was changed, not particularly to "defuse" it. I know especially the knife edge suffered a bit from the changes I made to get it to tumble...

Thank you for your straight answer!

Well, which lines do I have to edit to get back the "old" flying behaviour? Is it an possible or even easy task?

Jet-Pack (IPACS) · January 23, 2019 at 6:26 PM Official Post

Quote from KalEl

Thank you for your straight answer!
Well, which lines do I have to edit to get back the "old" flying behaviour? Is it an possible or even easy task?

Pretty much all of them

It can be done but I'd rather not put back old faulty values, rather fix the current pitts

Overloaded · January 23, 2019 at 11:07 PM

The NASA Foilsim program has unfortunately died with Java but a simplified version is working off their web site. The pressure distribution across the chord isn't available but lift and drag for various aerofoils can be demonstrated. I tried a 3% thickness 2% camber foil to represent a propellor blade. The relative airflow direction hitting the blade depends on the airspeed and propellor rotation speed, imagine the airflow coming from say a nominal 45 degrees.

The pics represent the lift and drag of a prop in cruise, feathered and windmilling at low rpm with the relative airflow approaching the blade stall. The numbers are arbitrary and the relative wind is simplified but it gives an idea of how disasterous it is to have a failed feather in the event of an engine failure. The windmilling prop has an enormous rearwards thrust and the drag keeps the rotation speed low maintaining the coarse negative angle of attack on the blade.

Overloaded · January 24, 2019 at 5:04 AM

Quote from Redtail

I got my tailwheel endorsement and basic aerobatic course in the Super Decathlon as well. It's the only TW airplane I've flown, but boy the most fun plane as well!

My first hands on aerobatics was in the earlier Citabria version, ‘airbatic’ spelt backwards. It was my first time starting a fuel injected engine, I had to, the instructor in the back seat didn’t have the full engine controls so I had to fuel pump prime her and start her initially in idle cut off. It seemed so strange, sophisticated and high tech’ at the time. We did nearly all the flying inverted, out of Moorabbin in Melbourne, Australia

There was talk of FS2 getting a Ju 52/3m, possibly from outside.

Jet-Pack (IPACS) · January 24, 2019 at 8:46 AM Official Post

Quote from Overloaded

The NASA Foilsim program has unfortunately died with Java ...

I've experimented with a lot of different tools like this, but it doesn't help much. Sure, if I know the airfoil and have the coordinates then I can plot the lift , drag and momentum vs angle of attack and that helps for the normal flying behavior but Aerofly also needs to simulate the angle of attack region beyond 20 degrees and below -10 degrees, where these tools start to fall apart. We need a full 360 degree chart and we need one for the fuselage as well, good luck with that

That's why these tools are good for a first guess, which I have already done with the new pitts. Anything beyond that has to be checked manually.

aero4428 · January 24, 2019 at 8:49 AM

Thanks for the comments everyone. Foilsim looks pretty neat too. So, I found in the real plane the rpm doesn't drop much when power is reduced to 11 inches. More the deceleration that is noticeable I guess. Flying a pattern I have 21 inches on downwind, ~120mph, 2,700 rpm. Abeam the numbers, 11 inches and RPM is still at 2,700 but plane is decelerating fast at this point. RPM stays pretty much between 2,500-2,700 while pitching for 105-110 until power is reduced to idle in which case rpm drops to 2,100 at ~100 mph. This is in the S-2C which has the 3-bladed Hartzell Claw. I believe the Extra 330 has the same prop.

Regarding using the airbrake to fine tune the prop, seems like it would be useful if the drag from the airbrake could be set proportional to the magnitude of propeller alpha for negative alphas but then zero for positive alphas. I think it would help dial in the deceleration without affecting normal flight and also be more true to what is actually happening, albeit still a hack.

I should have time this weekend to play around a lot with the sim parameters and will report back with any questions / findings.

aero4428 · January 25, 2019 at 4:40 PM

Had a couple initial questions regarding yaw in the Aerofly model:

There seems to be excess stability in the yaw axis in the air. The plane will react to a change in rudder position but then decay back to neutral rudder even if rudder is still pressed. I tested moving the y position of some things (like airbrake and prop) to create yaw. Then tried correcting for this yaw with the rudder. The two things that seemed apparent is that it is sort of hard to create sustained yaw and it is also sort of hard to correct for any sustained yaw with the rudder. Is there some sort of yaw stability augmentation or auto-trim currently in place? If so, is there a way to reduce?
On a related note, is there a way to significantly increase adverse yaw? The documentation says induced drag is modeled entirely inside the aerowing class. So was thinking there isn't an easy way to increase it. As an alternative, I had considered putting two airbrakes (one out on each wing) and having their input tie to aileron.output. Is this doable and can you have more than one airbrake per TMD file? Also, I see there is an AeroDrag class but it isn't documented. Curious if that could be used.

Thank you!

Jet-Pack (IPACS) · January 25, 2019 at 5:21 PM Official Post

1. What you can do in the sim that you can't do in real life is adding an infinite force to move the rudder. IRL you can't instantly deflect the rudder which you sort of can in the sims. This creates side effects that seem unrealistic but actually come from unrealistic inputs. But what you are probably experiencing is another effect, yaw inertia. It might still be too high, because, as mentioned earlier, I just guessed the masses of the wings. And the heavier the wings the more inertia you have.

2. I'm not actually sure how we could increase it. Yes you can add as many airbrakes as you want, I have at least 7 in the airbus (for each gear and for ground and flight spoilers on each side). But this is also not the correct way of doing it unless you know for sure that the ailerons cause an asymmetric drag due to flow separation, e.g. the nose of the ailerons deflects downwards and causes significant drag.

AeroDrag would be used to model drag of wires or hot air ballons, etc. We usually account for this in the Cd0 of the fuselage and therefor never used it in FS 2. Same code as the airbrake just without any inputs.

What I plan to do once I get time to change the pitts is to add aerodynamics for the vertical wing struts and gear. Both of these would destabilize the yaw and help to maintain a constant slip. Also the aerofuselage Cm could be higher to significantly destabilize any yaw and pitch.

Last but not least there are parameters for the prop (drag and force coefficients for when the incoming airflow for the propeller is off the rotation axis)

RoyPettit · January 25, 2019 at 9:46 PM

I mentioned about a year ago that the AFS2 Pitts will not hold knife-edge. IPACS contacted a Pitt's pilot who said that's the way it really is. What has been your experience? I've seen Pitt's doing knife-edges in air shows. What are they doing to make that happen?

Jet-Pack (IPACS) · January 26, 2019 at 9:00 AM Official Post

I've never seen an actual knife edge of a Pitts in any video, most of them were high bank angle side slips, none were really 90°. All of them cheated by tilting the wing slightly.

aero4428 · February 2, 2019 at 5:32 PM

So I've made some good progress and have learned a lot about the physics model after experimenting with many settings. The ground handling in my opinion is pretty much as good as you could ask for and that was simply done by turning the tailwheel spring constant down. I increased adverse yaw by reducing the lift coefficient of the vertical stabilizer. This generally destabilizes the yaw axis which also feels a bit more real. Has the additional benefit of making the high speed landing and takeoff roll slightly more realistic. I also slightly reduced the lift coefficient of the rudder (flap) to balance with the decreased vertical stabilizer effectiveness. I increased propwashrotation slightly on the vertical stabilizer which gives a slight left turning tendency on the ground. However, nothing I have done is really giving a sustained yaw in the air. Even a fairly massive increase to propWashRotation will make ground handling uncontrollable but once in the air everything flies pretty much the same. Same goes for crosswind.

I also found this in the aerowing documentation:

"... P-Factor which can also be modeled with the AeroPropeller class using its .LateralDragCoefficient, .LateralForceCoefficient, .SideThrust and .DownThrust attributes."

Would be curious as to who wrote this and if its accurate, if an example of modeling p-factor could be provided? I do not see a way to do it if I understand these settings correctly but hoping I am wrong ; )

Other things I am trying to figure out are: 1) getting prop speeds and airspeeds closer. 2) forward slip. 3) getting the bounce on landing slightly more realistic. If you touch down slightly on the main, the oscillation is too high a frequency. No questions on these latter parts yet as I haven't experimented enough.

RoyPettit I have done transition training into the Pitts but not anything other than basic aerobatics so don't want to comment on the knifeedge as it would be just speculation. That's why I have limited my Aerofly efforts to basic ground handling, adverse yaw, etc. Having said that, most of the low and slow knifeedges you see (Skip Stewart) are in planes that have much more horsepower and also not at 90 degrees as JetPack mentioned. Four point rolls at speed seem reasonable in Aerofly. But again, don't take my word for it!

Mr. Bean · February 2, 2019 at 5:52 PM

In proper knife edge, the fuselage is providing virtually all of the lift. Flat sided fuselage is best. The Pitts has a tiny rounded fuse.

…in theory, with sufficient fuselage side area (fuselage is now the lift device), lots of thrust, and a powerful rudder (to control the fuselage angle of attack), you could fly all day long in knife edge.

Tags