All that is wrong with the MSFS Flight Model (Inertia, Stalling, Pitch Authority, Trim & Sensitivity)

User Support Hub Aircraft & Systems
, ,
62

Much of this was pointed out in the prior thread about inertia where someone collected data relative to roll rates…clearly nothing has changed.

I actually started looking into doing a freeware conversion of a cold war era jet from FS9, with permission from the original model maker…something free instead of people buying stuff like the Typhoon…but I gave up after I realized what a hacked together mess the MSFS flight model is.

I studied aeronautical engineering for four years and it boggles my mind what a mess they have made of this.

1 Like
63

It’s painfully obvious when we see things like the CoP in front of the CoG from OP’s post that Asobo have made serious errors or workarounds when it comes to aircraft handling and aerodynamics.
I would like to open a communication channel with Asobo to start fixing the flight model so we can all enjoy the sim as a proper simulator.
There’s are a few commercial pilots in this thread, myself included, and with the assistance of someone who’s studied aeronautical engineering I think we would have a good base to provide that input to help them.
The flaws at present have a significant impact because they are such fundamental basics like CoP/CoG and missing inertia but from what I’ve read about the complexity of the flight model it should be more than capable of providing a realistic response given the correct calibration. So maybe that’s what it needs, some more input from real qualified pilots and engineers.

Main topics I would bring to the table are

  1. Inertia:
  • This is quite a large subject for FS and I believe the basis of the majority of the flight model complaints, it needs to be re-evaluated completely. By far the biggest thing for me is the relationship between control deflection, AoA and resulting change in FPV. There is no sensation of mass dampening or accelerating the rate of change of direction, currently it’s more akin to how an RC plane flies. I’d like to find out why this doesn’t happen correctly given the current data the flight model is using. Other issues, lots of which OP has highlighted… Accelerated stalls basically impossible, spins require constant rudder input into the spin to maintain it, controls remain largely unaffected by dynamic pressure changes, no wing drop in uncoordinated stalls, control stop limits changed depending on airspeed, over effective controls in first few % of application, etc
  1. Sim changing controller inputs:
  • Interference being applied to control inputs. If the model is designed properly then this should not be needed. Inertia rework may solve the need for this to exist.
  1. Ground Handling:
  • Relating to issue 1, small rudder applications result in violent and uncontrolled yaw response without the inertia required to smooth it out. Rudder authority ineffective when a/c within demonstrated xwind limits. Nose wheel steering turning circle incorrect (too large) in most aircraft I’ve tried. Nosewheel steering self activates/deactivates at a certain unknown speed in GA’s. Tyres seem to have too much static friction, noticeable on xwind landings without decrabbing
6 Likes
64

I participated a bunch in that lengthy thread about inertia, which supposedly got forwarded on to the devs. Shortly after that they made an updated Feature Discovery video about the flight model to try to explain why it’s great. I guess they also promised to go make some measurements in actual aircraft, but that only gives measurement in the 1 aircraft used to get the data, where as what we need is a flight model that is correct for all aircraft.

The reality is that I’m a customer here not the one developing the sim. I have actual space hardware that I get paid to work on (I double majored in machanical and aeronautical engineering and ended up doing mechanical for my career), and a family, and other hobbies beyond simming where I can “control my own destiny” so to speak. So if MSFS isn’t up to par I just won’t buy any more paid addons, won’t recommend it to others, and will do other stuff. It’s annoying (having invested around $230 between the base sim+addons) but it’s not the end of the world. I’ve certainly spent more on things and had them not work out.

I feel like I’m banging my head against a wall trying to point stuff out that might not be right…especially given the way they have bolted things on top of parts of FSX with no clear outline of how it is all intended to work.

I don’t think all of this is on Asobo either. Microsoft set the requirements for the overall development effort and also required some level of compatibility with FSX (based on some comments in the SDK explanation of the flight model). Personally I would have not brought forward any of the old code, without having the team that wrote that code available. I would have started from the ground up from basic principles and gone from there. I got pretty caught up in the advertising for MSFS and thought that was what had actually been done - my heart sank when I realized how much of FSX was in there. (I actually think Bruce Artwick and the folks who were caretakers of the original codebase did a great job in the single-core-CPU era in which they operated, but trying to limp the thing forward further and further is a bit like trying to make a Model T match the performance of a Tesla)

6 Likes
65

That would have to be done on a case-by-case basis. There is no “one size fits all” flight model.

66

It needs to at least be able to do the correct numerical integration to handle inertia properly if the inertia of the aircraft is properly defined, without resorting to adding odd delays to the controls to try to create the “feel” of inertia.

67

Great work! The lack of improvements from the developers in that area and the focus on visuals just shows that the simulator is targeted mainly at gamers. I’m using it less and less.

68

I encourage you to continue your freeware conversion. I have no aeronautical engineering background, therefore I do not feel so much “pain” about Asobo/Microsoft doing it wrong. But I have my little successes. Compare for example my version of Pitts S2S to the Asobo version. See Pitts S-2S X flight model mod for Microsoft Flight Simulator | MSFS
I know that what I am doing is cheating and hacking. But in the end I don’t care about the numbers I enter into the cfg files, but about the “realistic feeling within the limitations of the simulation” I get out of the airplane.

69

After experimenting a bit with the flight model while I was trying to improve the one of the 747-8, I arrived to the following conclusion: the assumptions made by the game for the flight model are correct and working for straight and traditional wing profiles but are incorrect for large transport aircraft with swept wings and super critical aerofoils.

Interestingly enough, the CJ4 can be modelled accurately with the standard flight model and assumptions due to its straight wing and traditional profile (the MMO is low at 0.77 compared to the 747-8 MMO of 0.90).

For supercritical aerofoils, the centre of pressure is significantly more aft than for a traditional aerofoil, with the pressure distribution over the wing being more spread than with a traditional aerofoil, to reduce the drag generated by the shock wave forming over the wing when flying above Mcrit.

Right now, it is possible to change the wrong position of the centre of pressure from forward of the centre of gravity to aft of the centre of gravity by changing the empty CG position in the config file and in the game when loading the aircraft to be in front of 25% MAC, which might be more restrictive than what the real aircraft can do, for example the 747-8 can be loaded between 11% MAC to 33% MAC IRL (I’m aware the mass and balance chart is more complex than this, just trying to illustrate my point here).

I have also realised that the SDK is not correct on what the wing_pos_apex_long and the aero_center_lift parameters do: if you have the line compute_aero_center = 1 in the config file, then the aero centre position can be changed by changing the value of the wing_pos_apex_long parameter, not the aero_center_lift one as indicated by the SDK. With the computer_aero_center parameter enable, you can also modify the lift_coef_aoa_table and the pitch_moment_aoa_table to change the calculated aero centre position. When the compute_aero_center parameter is set to 0, then the aero centre position is determined by the value of the aero_center_lift parameter in the config file. Changing this value will move the centre of pressure position, but will maintain the same distance between it and the centre of gravity, moving the centre of gravity in the process.

Finally, it is possible to improve a bit on the standard autopilot behaviour by writing pitch_use_trim = 0 in the autopilot section of the systems.cfg file, emulating an elevator control based autopilot as they are IRL. I say emulating because the trim won’t be used to control the aircraft in pitch, but neither will be the elevator, the plane pitch will just change and the aircraft will trim itself, allowing the AP to be disengage with the aircraft in trim.

Hopefully, this reaches Asobo and they can take these things into account to improve our simulator.

Thank you for reading me.

9 Likes
70

This paragraph describes the most annoying issues i have with MSFS… I don’t get near the limits of the plane to encounter issues with stalls and spins, but landing is inevitable. And it’s always the same. You land, plane seems to go straight down the runway, but then it starts veering off to the side (weathervaning in the wind, i imagine, but it’s incredibly exaggerated). You apply a tiny bit of rudder to correct it and the nose completely swings over to the other side, like you gave it full rudder input. From there it turns into an unstable mess as you’re trying to manage an exaggerated weathervaning tendency, with an overly sensitive rudder (i have mine set at -66% sensitivity), and braking.

Takeoffs are pretty similar. The transition between different ground speed regimes is very jarring. Which is interesting as there shouldn’t be any such transition. You have the aerodynamic torque generated by the rudder and the friction torque generated by the wheels. Those don’t suddenly go away when you get above a set speed.

For me the rest of the flight model is passable. In normal flight it behaves better than the old FSX one. But as bad as the old model was, it didn’t have jarring moments like this, where everything would suddenly break down and send you skidding sideways down the runway at 70kts.

5 Likes
71

Why wouldn’t there be? Physics apply universally to all planes, in the same way. The aerodynamic properties that keep a Piper Cub in the air are the same that keep a 747 flying, and they apply in the same manner.

6 Likes
72

Exactly, if the core model is incorrect then no aircraft will ever be correct no matter how many adjustments or work arounds are made in the cfg.

3 Likes
73

Dynamics are not entirely the same when comparing straight versus swept wings, super critical wing profiles etc. flying at transonic speeds…

74

If the basics are correct, it doesn’t matter if it’s a straight or swept wing, sub or supersonic aircraft etc.
E.g. In FSX, P3D and x-plane designing a WWI biplane and a Mach 2 fighter works with the same aerodynamic ‘base’ simulation.

1 Like
75

They may work somewhat differently, but the same rules apply. And while i would understand having one model for subsonic flight and another for hypersonic flight, i don’t see any reason to have separate models for GAs or for airliners.

1 Like
76

Agree, although for hypersonic flight (> Mach 5) you’ll need to download orbiter 2016 :upside_down_face:. From the developer mode the dynamics on most GA aircraft seem ok, airlines however (maybe the longitude as well, in other words, everything with swept wings) is completely off. In the developer mode you can visualize the pitch moments, you can see the CP in in front of the CG through most of the flight and CG envelope, CP moving aft with an increase of AOA etc. I don’t know where this comes from, maybe they did this to overcome some shortcoming in the core flight model?

1 Like
77

Ok, fair enough. “Supersonic” was the word i was looking for.

As for the rest… yeah. That’s the problem. Looks like they made a relatively restricted model that applies (barely) to GA planes, and then they tried shoehorning airliners in it. I don’t think that’s how things should be done.

1 Like
78

I’m pretty sure that Asobo will be able to provide a basic one-size-fits-all FDM in the future.

Just tested one of my old P3D FDEs with the flight model set to modern in MSFS and it worked pretty nice without a single change.
This is IMO an impressive feat, but it will take considerable time to fix all the present FDM problems.

1 Like
79

That is in line with what i’m hearing from people working with Asobo. At no point does it seem to me that this is a money grabbing operation. Both PMDG and A2A have said that Asobo are genuinely committed to delivering the best possible experience with this sim, and that they’re working hard on it. So yeah, we just need patience, i suppose.

2 Likes
80

Microsoft made the decision to release when they did. Asobo just does what Microsoft says.

2 Likes
81

Which was a good idea. I’m glad that I don’t have to wait another year…or two…