MSFS flight model

[NZ255 312]

9 hours ago, blingthinger said:

I'm talking fundamentals that explain why PMDG apparently has to inject significant motion adjustments to make it act the way it does. If that is the case, and I have every reason to believe it is, are they really using the FS2020 flight model anymore? Haven't they essentially written their own?

Yeah that’s a great point. We don’t know what flight model we’re comparing if that’s the case where devs either create their own, or inject data into the sim. We think it’s the base sim, but maybe it’s not.

That’s fine if it’s an improvement, but if done poorly, or not keep up to what the base sim offers, it will have negative implications for the entire sim. 

[lwt1971 7,456]

2 hours ago, robert young said:

Just a few days after MSFS was released I posted a long piece, and many subsequent ones, about the basic flaws in the flight model. A large number of members here said I was wrong and that the "new" flight model algorithms in MSFS were so much better than FSX and P3d and that the multiple lift points, so-called, were going to solve all the problems in previous editions of flight simulator. I pointed out many of the items that are now being discussed, and also wrote to Asobo about what I felt was a pretty disastrous core flight modelling methodology. I also pointed out here that many of the useful parameters in FSX which, though far from perfect, aided FDE developers in ironing out some flaws, had been removed.

I got a lot of flack, bordering on abuse, from certain members here. More recently, as this thread demonstrates, sim and real pilots have now made their views known and a lot of them concur with my originally expressed opinion.

For me the key flaw is that many behaviours, especially in lighter aircraft, are grossly exaggerated, particularly regarding apparent inertia values which present as being far too low (whether or not the inertia values in the flight_model.cfg appear to be mathematically correct). This includes way too sensitive pitch control, bouncing of pitch behaviour up and down at the slightest input, exaggerated yaw on the ground and crosswinds having a ridiculously exaggerated effect at certain speeds on the ground.

I managed to iron out a lot of these things in my Turbo Bonanza mod, but was not entirely successful at fixing ground behaviour because of the baked in extreme crosswind reaction.

These flaws show up much more in lighter aircraft than heavier ones, although even airliners suffer way too sensitive rudder control on the ground, as can be seen in even the best of third party airliners. I hope that the proposed solutions discussed by Asobo start to address all these issues because I agree with many views in this thread that it is a shame a sim with such beautiful scenery and visually satisfying features still has many core flight behaviours that are way off.

Just to be clear, whenever you've complained about aircrafts' flight models in MSFS, you were always speaking of default aircrafts from what I recall, where you could tweak flight_model.cfg values. Can't speak for any flack you received from whomever but like many have always said repeatedly the default aircrafts are poorly modelled in MSFS, and there's no arguing about that. In the hands of expert aircrafts developers however many good examples of properly flight modelled birds for MSFS are now available which can fly competently in the air (light aircraft examples of these include the Milviz C310, Sting S4, FSW C414, etc)

You speak of "the" flight model, but there is no global/single flight model in a sim though, as each aircraft as has its own flight model, and it all comes down to how each aircraft's FM is implemented and tuned using the core FDE (Matt Nischan explains it way better than I ever could: https://forums.flightsimulator.com/t/more-physics-more-real-winds/372656/256). Yes the ground handling issues are well known, and that's an example of issues in the core flight/ground dynamics engine itself, due to reasons like wheels being modelled as single points, which then has global impact across all aircrafts (though some 3rd party aircrafts have been able to reduce these issues through various tweaks/tunings/workarounds in their FMs)
 

Edited September 9, 2022 by lwt1971

[Glenn Fitzpatrick 3,799]

Up to FSX the MS flight sims were "tables all the way down" and any particular aftermarket planes FM was totally dependant on what tables those aftermarket devs created.

The claim there is no Global flight model is correct for FSX and earlier.  There is merely a system for creating a flight model.

 With MSFS the game now generates a basic FM from your 3D model (similar in some ways to how XPlane does it) that you then tweak with what are effectively tables. There is no single global universal FM, that much is true, but every MSFS aircraft starts from an initial FM generated by the game with any bugs or deficiencies from the algorithms hard baked in.  The issue with this is, when things are missing from the algorithms that generate the core FM from your 3D model, or are just plane incorrect, tweaking those issues is tricky, a bit arcane and invariably a compromise.

Comments about what Pamela Booker (one of the recognised masters of FM work who has been contracted to do the flight models on many of the aftermarket planes we all fly)  thought about the core flight model ...

Quote

Meanwhile, Pamela Brooker took a deep dive into the new flight model, and while some aspects have been improved upon drastically in the new simulator, some other parts of the aerodynamics simulation seem to have been forgotten. The result is a flight model which needs extreme overdamping of the controls and still results in oversensitivity of rudder, elevator and ailerons.

 

Edited September 9, 2022 by Glenn Fitzpatrick

[lwt1971 7,456]

35 minutes ago, Glenn Fitzpatrick said:

The claim there is no Global flight model is correct for FSX and earlier.

Whether the flight model of an aircraft is initially auto-generated from provided geometry and then edited, or if it's hand-written from scratch, the point here is that ultimately the flight model is *per* aircraft.. and there is no global flight model that is always governing the FM of an aircraft when flying in the sim. I agree that tweaking an auto-generated FM with initial values can be trickier, but that could also be the case for a completely hand-written table-based FM. Again it all comes down to the aircraft developer, what aircraft it is and its geometry and other vital characteristics, how experienced they are on the core flight sim platform they happen to developing the aircraft for, what documentation and configuration/APIs/etc that the core platform provides, etc etc.

Edited September 9, 2022 by lwt1971

[Glenn Fitzpatrick 3,799]

My limited understanding is that the MSFS generated FM basically flies the aircraft and you are only fiddling with the inputs it receives to try and correct any errant behaviour -  meaning fixing one problem may introduce unwanted side effects, something which does not happen with a table based system.

However I am by no means a developer so my understanding of all this is limited..

[Alpine Scenery 1,609]

My thoughts, but people should not take too seriously, because just my opinion and I'm not a pilot...

Climb rates and fuel flow are easy to simulate, that's not really part of the flight model so much. Sure it is affected by it, but you can use overriding factors to control that and cheat the system anyhow.

The issues with the flight model are like all software issues, a lack of simplified encapsulated hierarchical layering with a good solid example. When providing 3rd party tools, each person has a different level of skills in order to tweak, and a different amount of time, etc... There should have been a few accurate default aircraft templates that people can use to understand. It's the same with all software, there are always people not layering complexity and organizing right, it's just almost impossible on a big project to always have organized developers. A lot of times developers just speed code to get stuff done and just don't even think about the implications of the way they organized stuff.

The issues with the Flight Model for me are twitchy lack of control on the ground, improper flap reactivity, questionable ground effect, strange reactions to wind, etc...

I don't doubt that the most skilled developers can tweak some of the issues out to some degree, but I'm sure it's not all that accurate when you hit extremes, even on the 310r. Like it might feel accurate anecdotally, but I bet if you compared the exact metrics of the reactionary fluid dynamics (or whatever) to the real aircraft, you'd still find certain things to be way off.

Edited September 9, 2022 by Alpine Scenery

[abrams_tank 7,454]

5 hours ago, Krakin said:

Listen, I don't care how passionate Austin and the folks at LR are, they are going up against competent people with access to a ton of resources. What took LR a decade to do can be accomplished in a lot less time if the resources are available. Let's be honest, guys, the leap from XP 11 to XP 12 looks about the same as Asobo doing 3 or 4 sim updates for free.

TBH, it’s not even 3 or 4 updates from The MSFS team, especially that the Sim Updates are bigger now because we only have 4 Sim Updates per year now in MSFS (I say the MSFS team because Working Title is also helping).

Looking at the feature list of XP 12, the MSFS team can do the same amount of changes/work in one or two Sim Updates.  That’s frightening, and IMO, why XP will never catch up to MSFS, MSFS will continually pull away from XP because MSFS is advancing at a much, much, faster rate, while it takes LR a lot of time to make a small set of changes.

[Bobsk8 12,836]

1 hour ago, Alpine Scenery said:

My thoughts, but people should not take too seriously, because just my opinion and I'm not a pilot...

Climb rates and fuel flow are easy to simulate, that's not really part of the flight model so much. Sure it is affected by it, but you can use overriding factors to control that and cheat the system anyhow.

The issues with the flight model are like all software issues, a lack of simplified encapsulated hierarchical layering with a good solid example. When providing 3rd party tools, each person has a different level of skills in order to tweak, and a different amount of time, etc... There should have been a few accurate default aircraft templates that people can use to understand. It's the same with all software, there are always people not layering complexity and organizing right, it's just almost impossible on a big project to always have organized developers. A lot of times developers just speed code to get stuff done and just don't even think about the implications of the way they organized stuff.

The issues with the Flight Model for me are twitchy lack of control on the ground, improper flap reactivity, questionable ground effect, strange reactions to wind, etc...

I don't doubt that the most skilled developers can tweak some of the issues out to some degree, but I'm sure it's not all that accurate when you hit extremes, even on the 310r. Like it might feel accurate anecdotally, but I bet if you compared the exact metrics of the reactionary fluid dynamics (or whatever) to the real aircraft, you'd still find certain things to be way off.

I got a great idea, why don't you save up some money, go down to a local flight school in your area and take some lessons up to the point where you solo. Then tell us how a sim should feel.

Edited September 9, 2022 by Bobsk8

[Alpine Scenery 1,609]

40 minutes ago, Bobsk8 said:

I got a great idea, why don't you save up some money, go down to a local flight school in your area and take some lessons up to the point where you solo. Then tell us how a sim should feel.

Point was more directed at their software style more than the model itself. It could be the best model in the world, but even a much worse model that the average developer hits a lower delta error is generally a better model. The elitist route tends to make it to where only the most skilled people can have a lower error. The 310r is definitely A LOT better than the other planes I've flown, like worlds better in the handling, but A2A and Milviz were always good at tweaking that stuff.

I mean you could sell a piece of furniture to where only 5% of people can assemble it correctly, but is that the best way...

Edited September 9, 2022 by Alpine Scenery

[robert young 2,897]

3 hours ago, lwt1971 said:

You speak of "the" flight model, but there is no global/single flight model in a sim though, as each aircraft as has its own flight model.....
 

Indeed so, and I have repeatedly said the same thing many times. However, there is also a core set of parameters which offer the ability to "tune" each individual aircraft, and very early on I stated that Asobo had removed or truncated many of the quite useful parameters available in FSX, but they insisted that the "new" flight modelling algorithms in MSFS were much better and afforded better control. I have always been unconvinced by this claim. There are many examples but to give three: Asobo has always claimed that each wing is divided into discrete sections or lift points, but I have yet to see anything like this working in practice. Another example is the way a stalling aircraft becomes unstable in roll. The parameters that controlled this appear to have been reduced to just one as far as I can see, and it is just a virtual copy of the same parameter in FSX, but the FSX method had more sub-routines in order to fine tune. Last example is pitch control, where FSX offered several different ways of controlling pitch sensitivity, whereas the same controls are very limited in MSFS.

[MrBitstFlyer 3,892]

1 hour ago, abrams_tank said:

That’s frightening, and IMO, why XP will never catch up to MSFS

I've had XP12 for two days. In the first hour with a default aircraft I had none of the flight model problems being talked about here - no twitchiness in the rudder and fine pitch and roll control.  The settings for controller sensitivity curves are so easy to use too.  As somebody else said, if the flight model moved forward enough, MSFS would be hard to beat.  

[blingthinger 649]

12 hours ago, mrueedi said:

Your are talking about this?

Can you be more precise, what stuff does fully explain, that the model would feel OK at best?

That's the one. Some of my head scratches:

- The summed up force and moment values on all those force elements everyone is raving about, get squashed down such that their sums equal the FSX tables. I get the backwards compatibility aspect, but this is no badge of honor. And like @robert young says, they got rid of a lot of stuff that is useful to make this work when table models are still involved.

- The algorithm for said element-squashing and scaling process starts by calculating the force and moment coefficients at 20 pre-defined control surface positions and making a big interpolation volume. This happens once at aircraft loadup. During the flight if the user has the controls at a position that is between those points (which is MOST OF THE TIME), the forces and moments generated are averages of the nearest points in the volume. This essentially means that the "personality" traits of an airframe (as defined by the FSX table model) get smeared out. 

I think these 2 are the big reasons why the dynamic responses of even the newest planes are still a bit off. I struggle to see even the likes of PMDG fully overcoming them fully even with their own flight model tweaks. Also note this largely influences the dynamics. Getting steady-state stuff like a cruise climb and associated fuel consumption are going to be just fine from the FSX tables.

A couple others:

- No fuselage effect in the lift equation? Are the fuselage force elements only generating sideways drag?

- No drag equation.

- No roll moment equation.

- At each element, they're generating force and moment coefficients independently and simultaneously. I'm not talking about neighbor elements affecting each other. I'm talking about the fact that it appears that they've decoupled forces and moments. This makes zero sense to me given that the forces are distributed in 2 dimensions across the wing, fuselage, and tail surfaces. Moments are calculated FROM forces in a scenario like this. Not simultaneously. The chances of them both being correct at the same time are astronomical. Is this done just to match the FSX equations? May be wrong, but I can see this one being important when precision movements are really noticed: e.g. anywhere near the runway in takeoff roll or landing flare.

[Sky_Pilot071 581]

9 minutes ago, MrBitstFlyer said:

I've had XP12 for two days. In the first hour with a default aircraft I had none of the flight model problems being talked about here - no twitchiness in the rudder and fine pitch and roll control.  The settings for controller sensitivity curves are so easy to use too.  As somebody else said, if the flight model moved forward enough, MSFS would be hard to beat.  

I haven't seen these issues either in MSFS.

sp

[lwt1971 7,456]

Great discussion at end here.. despite the current SDK docs being far better than they were at initial state, would be great to have some tutorials or further insights out by the likes of the Milviz or Sting S4 developers, Matt Nischan, iniBuilds, Fenix, etc re: how they implement their aircrafts using the MSFS platform, what tweaks/workaround they have to do, positives/negatives of the current FDE & SDK vs previous sims, etc.. without giving their trade secrets of course. Personally, as someone who comes from a software development background who's also interested in flight simming these deep dives from aircraft developers would be very interesting to read/learn from.

https://devsupport.flightsimulator.com/index.html, https://www.fsdeveloper.com/forum/ and https://www.fsdeveloper.com/forum/forums/flight-dynamics.65/ are great resources  but as per usual with these Q&A type forums the info is scattered everywhere.

Edited September 9, 2022 by lwt1971

[jcomm 5,975]

Sometimes they - ASOBO - write stuff that really makes me think they don't really have a sound idea of what they're doing... Like in this answer given to a developer ( MrTommymxr ) by what I believe is one of the ASOBO team members ( FlyingRacoon ), regarding the total lack of Coanda effect modelling in MFS:

https://devsupport.flightsimulator.com/questions/10265/cfd-broken-body-not-affecting-airflow-coanda-effec.html

Rob Young, one reference in terms of "non-external" flight modelling for MS FS, has been referring to various omissions / truncations effective with the "Modern FM". I believe some of those parameters are supposedly no longer required since they should be "inferred" from their "CFD" approach.

There is a collection of new aerodynamic and engine ( mostly for prop aircraft ) parameters that allow for fine tunning of their MFS's CFD approach, but sometimes I think this kind of approach poses limitations to the fine tunning of highly detailled aircraft models like those that Real AIr and A2A made available for fsx and p3d, just as in X-Plane's BeT approach this same fine tunning is also not easy to achieve, although X-Plane offers a worth of additional design options for aircraft and their various aerodynamic components ( generators of lift and drag or thrust and / or moments ) that we simply do not have in MFS's Modern FM.

Yet MFS is amazing on many aspects, although I confess I'd rather prefer not to have the great scenery and weather, ATC and AI, and could now simply jump back to my good old X-plane...

 

 

Edited September 9, 2022 by jcomm