Fundamentally wrong flight dynamics on swept wing aircraft

To continue where we left on the topic longitudinal (pitch) stability from the developer Q&A. It is a complex subject with many rabbit holes but I’ll try to keep it simple. Since Asobo announced they are gonna improve airliner experience, please help bringing this one to their attention.

The flight dynamics on the Airbus A320 & Boeing 747 (maybe also the Boeing 787) seem completely wrong when opening the Developer Mode. In short the Center of Pressure (lift) is far in front of the Center of Gravity (weight), even with the center of gravity on the forward CG limit. In real life this causes static pitch instability.

Looking at any random GA aircraft in MSFS we find the following distribution, lift in green is acting behind the weight vector in red with a subsequent blue downforce on the horizontal stabilizer.

When now comparing the above picture with that of the Boeing 747 below we can immediately see the weight and lift vector changed places with an up-force on the horizontal stabilizer.

Although it is not a direct requirement to have the lift acting behind the weight for static stability, the situation in MSFS is clearly wrong. Still the airliners don’t show static instability because there is something else fundamentally wrong:

On the real aircraft the Center of Pressure moves forward in increasing angle of attack causing a restoring nose down moment and vice versa. In MSFS the Center of Pressure is seen moving aft with an increase of angle of attack together with the reversed lift and weight vectors it creates stability but does create wrong flight dynamics:

  • Wrong pitch effect with slat / flap / spoiler movement.
  • Wrong drag affecting all sorts of flight dynamics
  • Wrong fuel economy as the stabilizer up-force is more efficient than the down-force in real life.
  • Wrong stall characteristics.

I did find a document supporting my case which seems to be official Airbus material:


Background information

I made an effort to make longitudinal static stability understandable below. Numbers, scale and distances are only for illustrative purposes and understand that this is a gross oversimplification of the actual subject but I hope its understandable.

Aerodynamic Center (AC)

When looking specifically at the wing, disregarding the horizontal stabilizer and other lift producing parts of the airframe:

The wing itself has a point, located at roughly 25% of the Mean Aerodynamic Chord (MAC) where the wing pitching moment (which is naturally nose-down) remains the same or is at least negligible with changes in angle of attack. To explain this effect, I have drawn two situations the green one represents straight and level flight, the red situation is following a disturbance increasing the angle of attack. Lets use the following numbers in this situation:

Picture1

  • Situation 1: lift force is 2 units, the moment arm from the CP to the AC is 5 units.
  • Situation 2: lift force is 5 units, the moment arm from the CP to the AC is 2 units.

Moment = arm x force.

  • Moment 1 = 2 x 5 = 10 units nose-down.
  • Moment 2 = 5 x 2 = 10 units nose-down.

As can be seen, taking the AC as reference point the moment remains the same as the increase in magnitude is compensated for by a shorter moment arm. If we now take a reference point in front of the AC, for example the leading edge of the Mean Aerodynamic Chord (Le MAC) which is located 1 unit in front of the AC the moment change as follows:

  • Moment 1 = 2 x 6 = 12 units nose- down.
  • Moment 2 = 5 x 3 = 15 units nose-down.

Conclusion, the nose-down moment has increased, creating a restoring pitch-down moment. When taking a reference point behind the AC it will be vice versa, the nose-down moment will reduce with increase of angle of attack.

Neutral Point (NP)

If we now add the horizontal stabilizer to the equation we are going down a rabbit hole as we have to take into account the wing downwash effect on the horizontal stabilizer, its hard to know what this effect exactly is. All the sources I have dived into writing this suggest the angle of attack on the tail becomes more negative with an increase in angle of attack on the wing due to an increased downwash angle from the wing. Although I used this in my explanation below, I believe this to be more correct for smaller aircraft opposed to large airliners. So to elaborate a little further on this.

  • If we completely disregard the effects of downwash from the wing on the horizontal tail (T-tail design), the angle of attack will increase on the horizontal stabilizer the same way as it does on the wing. The Neutral Point will be located behind the AC in this case as the reduced nose-down moment with increasing angle of attack will be compensated for by a reduced nose-up moment from the horizontal stabilizer.
  • If we assume the wing downwash compensates for the otherwise increase of angle of attack on the horizontal stabilizer with increase of angle of attack on the wing (stabilizer angle of attack remains constant), the neutral point will be located on the aerodynamic center (we can basically remove the horizontal stabilizer from the equation),
  • As in the situation below, if the downwash from the wing produces a more negative angle of attack on the horizontal stabilizer with an increase of angle of attack on the wing, the NP is located in front of the aerodynamic center.

When including the effects of the horizontal stabilizer with the previous example, assuming the center of gravity is located on the AC and using an arm of 10 units and a down-force of 1 unit for the horizontal stabilzier in leve flight, increasing to 2 units following the disturbance:

  • Moment 1 wing = 10 units nose-down (previous example)

  • Moment 1 tail = 1 x 10 = 10 units nose-up

  • Moment 2 wing: 10 units nose-down (previous example)

  • Moment 2 tail: 2 x 10 = 20 units nose-up

As can be seen, following the disturbance the nose-down moment hasn’t changed while the nose-up moment has increased from 10 to 20 units. In conclusion the nose continues to rise and the aircraft is considered unstable.

To reach neutral stability the CG must be located 2 units in front of the AC, be aware that if we change CG position the whole equation changes. Its not really that important in this example so lets continue using the same arm of 10 units for the horizontal stabilizer for simplicity. Following the disturbance:

Moment = force x arm

  • Moment 2 wing = 5 x 4 = 20 units nose-down
  • Moment 2 tail = 2 x 20 = 20 units nose-up

A point of neutral stability has now been reached, the aft CG limit is situated in front of this neutral point to attain static stability. Take the drawing below with a pinch of salt as the scale and location of points is fictional. The CG limits are located within the MAC normally. I haven’t found any information about the A320 or Boeing 747 but I know that for the Boeing 737 the Aft CG limit is very close to the aerodynamic center so I assume if you add everything together the neutral point will be located slightly behind the AC.

Hope it makes sense.

This is why i have troubles to climb over 30.000ft with my 787 and perfect balanced cargo, because my plane always looks like on TO and the elevator trim is on the upper limit.

1 Like

Keep up the great work @Nijntje91. You’ve really done some amazing work explaining and highlighting these issues.

8 Likes

I thought I noticed something similar with the Boeing 747. I never fly that thing, I only flew it around in developer mode to see how it behaved but I needed a lot of trim for level flight.

1 Like

Good find!!
We must find a way to promote this thread…

2 Likes

You have my vote, this looks like a fundamental issue with airliner control and stability.

2 Likes

I’m afraid it is too technical for most people to understand or care about. In the Dev Q&A this topic did receive a lot of votes though and ended in the top somewhere but didn’t got answered by the devs unfortunately…

2 Likes

Thank you for putting this together. Very much appreciated.

On the Dev Q&A sessions, Jorg has said that they (Jorg, Seb, Martial) spend a lot of time reading the Forums. They don’t post or reply, but I would imagine they seek out this kind of detailed, highly technical thread. Hopefully, they put this to use in their development efforts.

Thanks again!!!

3 Likes

@Nijntje91
I did see this post back in November, but :slightly_smiling_face: it was too technical for me to understand or care about!

I re-read it a little more carefully yesterday in light of the discussions in our recent topics, and it dawned on me that there is a difference between those who do understand it, and the developers who maybe don’t understand it well and still have to make it work in simulator code.

If you take a look at any of the unencrypted flight model config files (I’m fond of the TBM), then you’ll see that there are aircraft geometry settings and flight model settings that have pretty sophisticated (complex?) parameters and tables for this. There are many tables that reference AOA, moments of inertia in all axes, coupling between axes, tail incidence angle effects, mach versus knots variations, etc.

So I think for FSX, the flight dynamics engine does model these, but ends up applying all those forces effects to a single point. The aircraft model is simply rotated around this point and the point is translated in reaction to the dynamic forces. I think “riding on rails” is describing this single point translation and model rotation, versus the new sim’s more localized reaction to dynamic forces.

Let’s say an updraft happens, and the effect should only affect the right wing out near the tip. The old model would translate the single model point upward and not produce a left bank. The new model – theoretically – will translate the right wingtip upwards. Here’s where it gets fuzzy.

In a very light plane, this should produce a little bit of rotation around the longitudinal axis, but also some significant upward translation of the model center and therefore left wing as well.

In a very heavy plane – and obviously dampened by the weight of the fuel in the wings – the right wingtip upward force would probably produce more of a model rotation, right? The right wing would go up and force the left wing down, while the model center would be less affected translationally.

We know that inertia – both overall and in respect to conservation of momentum in any axis – is incorrectly modeled as of this moment. Is that the “root of all evil” in the core flight behavior? The planes yaw inappropriately in reaction to crosswind, torque, and nose steering forces. The planes “porpoise” continually while on AP. The plane’s wings rock continuously while trying to fly level. The planes are nearly impossible to trim and reach a realistic stability. Nose wheel steering jerks planes around like the nose wheel had infinite traction and the plane had no weight at all.

I think that’s my new theory. In general terms, lack of mass and inertia, and in specific terms, lack of the conservation of momentum that correct mass and inertia physics would provide. Controls seem overly sensitive because the control surfaces do not have to overcome realistic inertial momentum.

All these individual complaints about porpoising, wing rocking, uncontrollable steering and inappropriate yaw effects – they can’t be tuned out. They are probably reacting appropriately if dampened by appropriate inertial forces.

B

3 Likes

I’m not sure how the MSFS flight model works. All I know is that the depiction of forces and pitching moments in the developer mode do not correspond to real life flight dynamics. And longitudinal is a pretty difficult subject, you need to draw it a couple of times probably to really understand how it works. There is also plenty of wrong information on the internet or confusing at least, not taking certain effects into account. It still explains longitudinal stability but does not reflect the location of forces and pitching moments correctly. Asobo said in the Q&A that the coming updates are focused on airliners, if it is still a mess after the next couple of updates its time to complain again I guess :joy:.

4 Likes

I did this landing challenge with the C152 in strong winds, the turbulence behavior is laughable really. You can completely ignore the rocking of the wings due to turbulence, the net flight path is straight without any pilot input :sweat_smile:. Feels very unrealistic and on rails to me.

That sounds like the flight model is simply doing model rotations around a central point, giving the visual appearance of turbulence affecting it, but without translating the model (moving the flight path). Maybe for that challenge they internally turn on Legacy mode or something equivalent so that it looks hard, but really isn’t.

Oh, but that would actually make it harder for real pilots who are correcting all over the place for the perceived and expected deviations.

Interesting story – on topic, I swear. I learned to fly in south Florida (KFLL) starting in March – very, very windy and gusty. This was in 1978, and my instructor was one of the early female Delta co-pilots at the time. That woman had magical stick and rudder skills – better than any instructor or pilot I’ve seen since.

Back then, we generally flew a 152 and used 9L for training. That meant that on short final, you came in the wide concrete expanse of I-95, then a pretty wide canal, then the grass of the airport just before the threshold. In the afternoon, it was generally pretty hot, so you not only had very gusty winds, but also significant thermals from the terrain. On short final, your wings would be rocking back and forth unexpectedly, and you would audibly grunt as you bumped up over I-95, then nearly banged your (6’4") head on the ceiling as you absolutely dropped over the cooler canal downdraft, the bumped back up over the grass, and again over the runway threshold.

Talk about a demonstration of flight dynamics. Most of my early landings were like narrowly avoided crashes. But Virginia?

It was just like you described where the plane was moving all over the place but the flight path was as smooth as glass. I would look over and she would be using her right hand to move the yoke in sharp, jerky motions left and right, and back and forth. Her left hand would be subtly jockeying the throttle while I could swear she was tap dancing as I watched her feet on the rudder. The transition from crab to slip was something to behold.

One thing she told me that has always stuck with me: “Fly the plane – don’t let the plane fly you!” To this day I get annoyed at myself and pilots on YouTube if they’re not aligned with the centerline on approach. I miss those days, and the sim is the closest I come to reliving them.

I hope they get the flight models and physics worked out soon.

4 Likes

Just like you I remember and treasure those flight school days. Please remember also that Asobo very talented game developers are unlikely to address your concerns. It’s up to 3rd party partners to see if this platform can be used for the kind of simulation you (we) are looking for. Right now it’s an open question.

Actually Flyguy, from reading posts in here, they are actually quite concerned with them.
To the point where they have put a priority to fixing them.
Indications seem to point that this will be fairly soon, rather than later.
I read in one post from a source that should know, that this could be in this upcoming update.

1 Like

Not likely, they said the Aerodynamics showcase is delayed that was meant to be coming with the next update.

1 Like

Regarding your description of your short final:

This is something that has always missed in flightsim, no matter what version. My early flight training was in a ASK13, wooden sailplane, 480 kg, 16m wing span, glide number 28.
Runway 28 was slightly rising terrain, brown fields. Wind most of the time 5-10ish kts from west there were some thermals over the fields giving you a good lift (1-2m/s) on final, often we had to push the nose down and extend the spoilers pretty far.
RW10 on the other hand was slightly offset to the north and led in over a small forest where we always had a sinking of 2ish m/s, so the spoilers were almost completely in for a 100km/h final until we cleared the trees, then the downdrafts suddenly stopped and if we had anything but a headwind we would have made a looong landing if we hadn‘t been aware.

There is absolutely NO such behaviour in MSFS. These things were advertised massively, with air animations moving over mountains, falling over forests and water etc. But if you fly across a mountain hang you will see nothing but wildly rocking wings. The variomater remains exactly where it is.

My guess is that a lot is still planned regarding flight dynamics, physics engine, etc but they completely miscalculated the time required for that to accomplish. And to make the whole thing feel alive they put in some strange effects to move the plane around but unfortunately they are completely wrong.

That sounds about right. All was good intentions and GREAT marketing, but I think they lost about a year of anticipated development time due to COVID effects and trying to deliver on the VR and Xbox timeline.

I hate to be so optimistic all the time – when I’m not whacking trolls and stamping out misinformation – but I truly feel that all this apparent chaos is really still within the framework of a master plan. I do think they underestimated a lot of stuff regarding the overall flight model, physics model, and systems model. But they admitted that is the case and are working on all those things. It’s just going to take more time – like a year or so, not a week or month.

The great marketing and fuzzy road maps are making people set themselves up for disappointment with unreasonable expectations. That creates a lot of understandable negativity on the forum, but then, so would Asobo discouraging the community to kill the excitement and wish list.

As for gliders, I’m a real pilot, too, and us powered aircraft types frown on parachutists, “Why would anyone jump out of a perfectly fine plane unless they had to?” The thinking on gliders is similar but different. It is a plane, after all. And when you don’t have an engine, it certainly makes you pay attention to the subtleties of stick and rudder and the world around you. I find that attractive.

I never did it in real life – jumping out of a plane or gliding – but I wish I had now. I tried gliders a couple of times in FSX and X-Plane, but the terrain effects and scenery were just boring. But when they add them to this sim, especially if the flight model and environmental effects are realistic – I’m all in! It will be as awesome, challenging, and fun as the real-life experience you described.

Wow. Real flying is SO much fun, and I feel really bad for the sim pilots who have never experienced it. Especially those who somehow think airliners flying high and fast is the ultimate flight experience – no offense, airline pilots. The biggest thrill I get in the sim is from relatively short flights in beautiful VFR conditions, and from takeoffs and landings in IMC. At times, I literally forget I’m not actually in a plane, and when I do come back to reality, it’s always with a big smile on my face!

4 Likes

Wouldn’t they have to fix the aerodynamics before they could create all the pretty videos and stuff? That would mean we can – if we’re optimistic – expect the fix sometime before the showcase, right? So if the showcase was originally scheduled close in, and has now been delayed a little, maybe the fix is due really soon!

2 Likes

I somehow understand them, at least I try to. You need money for development to progress, the whole sim is a give and take. They had sold 2 Million copies in December and even if several hundred thousands of them were part of that gamepass it’s still a huge amount of money to spend. I hope they’ll do it right.

I usually fly WT-9 with my girlfriend and sometimes, if I find someone to pay the hour, C172, mostly couples who want to do a scenic flight around the Swabian Alb or Black Forest in southern Germany. But also I fly Skyvan dropping those who we’ll never understand :smiley: I jumped out of a Do-27 once and it was NO fun at all for me. I’m extremely scared of altitude (sounds ridiculous for a pilot) but as soon as there is nothing around me the world is about to end. A simple ladder to clean the car port from leaves is too much, “she” has to do that ^^ But flying? I could strap an SG-38 on and would be fine probably. lol. Flying is the greatest thing I could ever have started and while it’s a challenge in the sim to manage a 747 as correctly as possible for our purposes I haven’t even installed the airliners in MSFS and only do low level VF without R ;D with some of the planes, Tommy’s DA40 mod is my current favourite together with the CriCri ^^

1 Like

I am not really sure. The language used is a bit confusing, and maybe lost in translation. I bolded the parts.

Our Feature Discovery Series: Aerodynamics, will be postponed until further notice as we are still making a few edits to the video to ensure we hit the same quality bar for information that we have in the past. We are aiming to have this released next week but will update you should anything change. January is already in full swing and so is the team.

I guess by video they mean graphics and are is supposed to be be were unless it was meant to come before the update next week.