39 minutes ago, Glenn Fitzpatrick said:

From a meta perspective the approach taken in MSFS is going to give more consistent results long term than the Blade Element approximations done by X-Plane

With all due respect, do you have sources to back this up?  Because I have sources to show the opposite. 

31 minutes ago, Glenn Fitzpatrick said:

 

Totally disagree.

I did in fact fail my second year fluid flow and thermodynamics units 45 odd years ago at around the point I was dropping out of my engineering degree but mainly because I did not sit the exams, not lack of mathematical ability, and I did enough of it to know that the sort of math done by PC based algorithms is at best an approximation and guesswork and at worst just wrong.  I do still have my Mollier Enthalpy/Entropy chart I kept as a souvenir 😄

If nothing else turbulent flow is non linear and chaotic.

From a meta perspective the approach taken in MSFS is going to give more consistent results long term than the Blade Element approximations done by X-Plane but neither are going to give a real world approximation that will work alone without tables and adjustments based on "feel" .

 

 

The math done by algorithms ranges between totally wrong and perfectly on point. That turbulent flow is nonlinear and chaotic is what I wrote, more or less with the same words, so we do not disagree on that.

I think that in order to judge what approach is going to give more consistent results you need professional teams and a deeper knowledge of what is actually going on within the algorithms. Under this point of view, the sdk explanation, which seems pretty deep to an average simmer, is actually very interesting but insufficient for a professional to make judgements, The same goes with the other approach (which is not so dissimilar anyway, as Matt correctly pointed out), although many papers have been published in the last 25 - 30 years on the subject (but we don't know a lot about the actual implementation in X-Plane). And when I say "team", I mean that even a mathematician with a solid background in modeling, differential equations and numerical methods (this is my background) would not be adequate to the task: I would still need at least a physicist specialized in fluid dynamics and an aeronautical engineer with specific domain knowledge. 

In principle, a grid algo is finer and takes in consideration a better geometric description of the surfaces (which are, undoubtedly, bi-dimensional). But then there are the implementation details, assumptions, approximations, interpolations, and so on.

About fine adjustments based on "feel", that's what Matt and I were talking about: with table based models it's easier for the devs, but with more sophisticated algorithms, like that of MSFS, it's going to be harder. That's why I said that there are some recent developments in numerical methods that might come handy. So, even under this perspective, we agree :-)))

In all cases, my experience says that before you start "tweaking" to account for specific boundary conditions, you need a solid general model to start with. So I stand behind my proposition: the flight model algo is the most important part to create the illusion. Unfortunately, it is not sufficient by itself.

So, the fact that you prefer the model that gives you the right "feeling" is perfectly ok. But in order to get there, you need a solid scientific foundation, i.e. a great flight model. I like the approach the guys at Asobo have taken and I really hope that they will continue improving it, because just reading the sdk you see that they are passionate and competent about that.

A.

Question for "A.": I am impressed with the thought, and I assume expertise, that you bring to the discussion (I mean this sincerely). But all these lengthy posts must take considerable time to put together. My question is: how much do you actually fly MSFS? I ask this in part because I have become aware that I am spending far too much time in the MSFS forums and not enough using the sim itself.. For me, that needs to change. The forums can be useful, but flying this amazing simulator is much more satisfying!

 

 

Edited August 29, 20214 yr by cobalt

Yes it is easy to become a professional forum hound.  I don't have much to say so mainly troll and mark the site read  after scanning the topics.

Very habit forming if you are not careful.

I will do a quick view during a flight to kill time.

regards

bs

I honestly prefer - by far - the Combat flightsims, DCS, IL-2, when it comes to flight and overall physics modelling...

They're still 2nd to nothing else, with good aircraft / helicopters for X-Plane being also very interesting, but still not at the level of DCS Modules and IL-2 Battle of Stalingrad.

There's a lot that can / should be done about the modelling in MFS. There are some good aspects out of the equation, ranging from a limited representation of the lift generation surfaces / devices ( for instance no way to even define a bi-plane ... ), properly set the change in Cl / Cd / pitching moments due to lift augmentation devices, engine modelling ( turbine, turboprop, jet.... )...

Try the best Spitfire model available in MFS and compare it with either the DCS or IL-2 modules... It's so so so far off track 😕

Or simply use the default aerobatic taildraggers and find how crude ground physics is. Yes, they acknowledged that ( the non smooth transition between ground and flight physics, and it might become addressed, but... when ? )

There's probably also a lack of interaction with developers searching for the implementation of more detailed aircraft and systems, and that could partly account for not having, for instance, A2A on board, and yes I am aware of the Fénix A320 and look forward to see it "flying" in MFS.

Some of the posts in this thread really feel like posts from "Lovers" 🙂  Everything is "Perfect" when you're in love  🙂

 

Edited August 29, 20214 yr by jcomm

2 hours ago, cobalt said:

 My question is: how much do you actually fly MSFS? 

I wanted to fly a lot during my vacation (6 - 22 Aug, not going anywhere because people are crazy here and Covid is no joke). Unfortunately since the 27th I had a series of ctds, so I started writing on the Forum to see if there was a solution. Also, what I saw between ctds was pretty underwhelming compared to what we had before. So no flying, and a lot of free time. Including yesterday and today, for the server issue. I admit: I fly the steam gauges 172 for the visual impact. If I want to fly an airliner it's PSX 747. Only occasionally the A320 from Rome to Olbia, because it's quick (35 minutes) and with an interesting approach that I can hand-fly.

Thank you for your kind words about my expertise, but I haven't really written anything new, valuable or original in this thread, just putting some ideas together and  trying to remind that even an expert is not enough in a multidisciplinary context like this. 

Don't worry, as soon as MSFS is back to what it used to be I'll be flying my Cessna somewhere in Italy.

BTW: I have more free days this week, before WU7, so you know what to expect.

A.

Bit late to the train, but I recently found this thead, and  was pleasently surprised by very interesting discussions on Sim Comparion, without ending up in a battle of throwing stones at each other, a Simulation record!! Sorry to bring some old posts up.

 

 

On 8/27/2021 at 6:04 PM, MattNischan said:

.. The reality is that within the features that MSFS supports, it has objectively, from a mathematical and implementation perspective, the highest resolution and most capable flight model of any home flight simulator, by an order of magnitude. And all the comparisons to Level D sims or FAA training approval, are downright silly. Level D sims use old school table lookups with single CoG point force application and absolutely zero air mass simulation, FS9 era mathematics. The dynamic 4D air mass simulation alone is totally a complete first in the simulation space. The _default_ 172 in MSFS flies so incredibly close to measured data that it's down to that last extra degree of adverse yaw here, that one or two knots of extra deceleration in a slip there..

-Matt

This comparison doesn't feel adequate to me Matt.

What is the end goal for a Simulator, getting maximum points of resolution, the most dynamic experience possible, despite innacuracies in major ares of expected flight behavior? I don't think that's what those Level D Simulator Companies are trying to achieve at all. I agree that you probably won't have the air instability generated from a steamy Lake before the Runway making your flight bumpy on short final. But, is that more important than simulating a realistic behavior of the aircraft when you press on the Rudder Pedals X Speed? I really don't think so. These Level D Simulators doesn't need to bother with Virtual WindTunnel Simulation in Real time, with 1000s points accross the fuselage, because they literally have real gathered data points for every point in the curve of the flight chracteristic of the Modeled Aircraft. It's so well populated, that they can guarantee in basically any normal envelope situation, the LevelD Simulator will respost within very strick tolerace of accuracy to the real plane. You don't need real time calculation when you have such valuable and real data to populate your Simulator. The outcome will just be better. 

So, while this is a cool feature in MSFS2020, it has very little significance compared to way more important checkpoints that a simulator must be able to replicate within tolerances. These arguments of Dynamic and Real Time is always better, so good Real Manufacturers use out product to design their real planes, reminds me of a certain individual that is on other side of the competition, and I really hope the Asobo team doesn't go down that path of believing what they have done is so brilliantt we are all flying real multimillion dollars  Nasa Simulators in our Desktops.

 

On 8/28/2021 at 9:59 AM, ADamiani said:

All in all, the only desktop simulator I have tested in which the flightmodel is really good is PSX (the 747-400 sim). I tested it many times with a retired 747 captain and he found it absolutely excellent. But that's a single aircraft sim, so it's in another league. And it has virtually no visuals, so it's in another league for that reason also. 

This is a nice example, I'm lucky enough to own PSX 747, and I'm 99% sure it's all done using Table Based Flight Model, and I can guarantee it's more fluid and realistic than anything I Have ever flown in other sims, be it X-Plane, FSX, P3D, or MSFS2020. This just shows my point above. 

We have associated all limitations of Table Based Simulation, to those of the FSX and Pre limitations. If FSX didn't model correctly various factors, it's not right to blame it all on the method, but how the Microsoft team implemented it within the Simulator back in those times. If it was that bad and unrealistic as everone says, we should all be afraid all real pilots in the world who have all learned to fly 100+ Passengers Airplanes in Table Based Simulations.

On 8/29/2021 at 11:36 AM, ADamiani said:

...I think that in order to judge what approach is going to give more consistent results you need professional teams and a deeper knowledge of what is actually going on within the algorithms. Under this point of view, the sdk explanation, which seems pretty deep to an average simmer, is actually very interesting but insufficient for a professional to make judgements, The same goes with the other approach (which is not so dissimilar anyway, as Matt correctly pointed out), although many papers have been published in the last 25 - 30 years on the subject (but we don't know a lot about the actual implementation in X-Plane). And when I say "team", I mean that even a mathematician with a solid background in modeling, differential equations and numerical methods (this is my background) would not be adequate to the task: I would still need at least a physicist specialized in fluid dynamics and an aeronautical engineer with specific domain knowledge. ...

...In all cases, my experience says that before you start "tweaking" to account for specific boundary conditions, you need a solid general model to start with. So I stand behind my proposition: the flight model algo is the most important part to create the illusion. Unfortunately, it is not sufficient by itself...

With your deep background with Math and Modeling, if you allow me to make a question, which I would like to know your opinion.

You say even such an expert in this area wouldn't be able to know for sure whats going on in the Simulation, without going deep into the code. I agree 100%. What I also want to point out, in my opinion is that even those who created it, should "benchmark" it, entering airplane data into itm which they have the real data outcome that should be outputted, and then compare how off that simulation is, and if needs to me further worked on, or is it considered within satisfactory performance. I don't think that was ever done in the MSFS2020 Flight Model Engine. If you enter as close as possible real values for an aircraft, and compare the data it puts out. it doesn't check nowhere even close to what those dimensions and data should be outputting. Then, you say, that's why developers need to tweak.

But, if you have such real time Simulation, isn't it exactly for this reason you use a Realtime FlightModel Simulation? To reduce how much tweaking it needs to be done, generating realistic results with accurate data, is why I think such approach is chosen in a Simulator. If you need to tweak it, and in the case of MSFS2020, the tweaking needed is of such high order, in certain areas of the Flight Behaviors, that you end up loosing control of your outcome, generating unexpected noisy innacurate results in other areas that you didn't want to influence, but end up happening anyway. You loose all the strenghts of such model, and the Table Based Model would be better suited for this kind of heavy tweaking, where you can control exactly where you need to finetune, with much finer precision.

Would be a pleasure to hear your opinion!  

 

 

 

Edited August 30, 20214 yr by Alec

Superbly put Alexis !  👍

Was bout to add. among other, the present limitations regarding simulating the effects of dynamic pressure on the controls.

The "elasticity" tables can do their job, but I just noticed that visually your control surfaces still move full speed / deflection even if your aerodynamics cycle sees it as being reduced according to the associated elasticity table.

In such an "advanced" simulation with the 1000 points that are used to "average" the aerodynamic coefficients of your aircraft model you still can't properly cope with basic sutf like the effect of the deflected wing downwash as you use stability augmentation systems, like simple flaps. You also don't have detailed effects for each of the stages of deflection of these surfaces, at least in it's present state of affairs...

Aircraft have their control surfaces gaining efficiency way earlier than IRL. You can check that in the rather twitchy effects of rudder input during takeoff. Also, you can notice that if you input trim corrections, your control surfaces immediately deflect, irrespective of there being any significative dynamic pressure...

It's nice, it's promising for sure, specially from the scenery pov, and hopefully they'll be able to get the best from their Meteoblue feed, but let's not over-magnify it's present state of implementation...

Can it evolve in the right way ? We all hope so - the Team @ ASOBO is rather dynamic. For instance regarding the ground physics modelling they already acknowledged being still tied to the Legacy ground physics, and there being a blend between their Modern FM and this legacy ground physics that isn't perfect. I guess they intend to further develop the ground physics. There's also mention of gliders and soaring, which is something that interests me in special. Meanwhile I enjoy some of the default aircraft, as well as the PMDG dc-6, and the FBW A320N 🙂 and of course the EXCELLENT addons made available by teams like that of Matt himself. 

Dedication, dynamism, can probably become strong points in the MFS development cycle, like never before with other civil flightsims, and that's very promising !

Edited August 30, 20214 yr by jcomm

18 hours ago, ADamiani said:

I wanted to fly a lot during my vacation (6 - 22 Aug, not going anywhere because people are crazy here and Covid is no joke). Unfortunately since the 27th I had a series of ctds, so I started writing on the Forum to see if there was a solution. Also, what I saw between ctds was pretty underwhelming compared to what we had before. So no flying, and a lot of free time. Including yesterday and today, for the server issue. I admit: I fly the steam gauges 172 for the visual impact. If I want to fly an airliner it's PSX 747. Only occasionally the A320 from Rome to Olbia, because it's quick (35 minutes) and with an interesting approach that I can hand-fly.

Thank you for your kind words about my expertise, but I haven't really written anything new, valuable or original in this thread, just putting some ideas together and  trying to remind that even an expert is not enough in a multidisciplinary context like this. 

Don't worry, as soon as MSFS is back to what it used to be I'll be flying my Cessna somewhere in Italy.

BTW: I have more free days this week, before WU7, so you know what to expect.

A.

Thanks. I guess the only difference in our experience is that, for me, since the SU5 update MSFS Is, by far, the best it has ever been, with great performance (50-60 FPS practically everywhere) and excellent graphics. Just reporting.

 Alienware Aurora R11, 32 GB ram, Intel i7-10700F, GeForce RTX 2080 Super, Ultra graphics settings

1 hour ago, Alec said:

It's so well populated, that they can guarantee in basically any normal envelope situation

 

1 hour ago, Alec said:

I really hope the Asobo team doesn't go down that path of believing what they have done is so brilliantt we are all flying real multimillion dollars  Nasa Simulators

But this is exactly the rub that most folks don't really understand. There are two absolutely enormous caveats that you yourself state here. One is that there is so much data, and the other is that tiny little point about "normal envelope". Most (nearly all?) big training simulators are focused on normal procedures, and the multimillions part is generally not the software side, it's the ludicrous amount of custom hardware required. And there's zero way for normal consumer space aircraft developers to get data like that. Huge expensive simulators still generally punt on flight characteristics out of the envelope, because they can't really get great test data and it's super hard to model lookup tables for it. They'll have some scripted behavior in that area that is somewhat fixed, because generally the idea is to train people not to get into those flight regimes, and if there is any stall recovery training at all, they want you to follow the procedures so the scripting will be geared towards that. All this is from talking from folks who have been in that sector a bit in the past.

Level D simulators are not the bastion of mathematical flight modeling, despite misconceptions. They simply have more available table points to use with their lookup based flight models that are from measurement. This _can_ make them more accurate for normal procedures, which is what the training is heavily geared towards: making sure the pilots who fly the big iron have the procedures drilled deep into their heads and available with easy muscle memory. Do also note, though, that the high praise of Level D sims generally come from the consumer sim fanbase either who hasn't flown them or hasn't flow the actual plane to compare to; I know quite a few pilots who have opinions ranging from "yeah, felt just like the real thing" to "it doesn't act like that at all". And that's a big issue: a ton of what consumer simulator people say is realistic is based on their non-experience-based beliefs only. I understand that you find the PSX 747 to be quite realistic, but what is that in comparison to? The actual plane? The Level D sim of the plane?

All that a consumer simulator developer can really do here is make sure the underlying mathematical model is sound and provides correct emergent behavior from the model given certain conditions. The rest is up to the individual flight model config designer.

2 hours ago, Alec said:

If it was that bad and unrealistic as everone says, we should all be afraid all real pilots in the world who have all learned to fly 100+ Passengers Airplanes in Table Based Simulations.

No, we shouldn't, because nobody learns to actually fly a plane in those simulations. They already know how to fly a plane 1000s of hours before they step a single foot into them. Those simulations prepare the pilot for airplane specific procedural tasks, and give them a general sense of performance differences between what they know and what they're being trained on. Nobody learns how to fly a 737 on a 737 sim. They learn how to pilot one.

-Matt

Guys, this is getting interesting! 

I only have a few minutes before resuming work (not much sleep tonight I'm afraid, submission deadline tomorrow), so forgive the approximations.

Welcome to mathematical modeling 101.

The first recommendation to my RW students is: modeling is a multidisciplinary art, it needs a team, domain experts, knowledge, experience, time, patience and other three or four things:

Data, data, data.

More data.

We typically learn to solve equations at school: that's the problem of discovering the cause (x) knowing both the algorithm (model: f(x)) and the outcome (y).

f(x) can be easy: 3x+6 = 0  means  x = -2

f(x) can be nasty, so you have numerical methods.

Modeling, instead, is about discovering the model f(x) when you have x and y.

Data are typically noisy, so you need statistical learning.

There can be a lot of complexity, so you resort to machine learning (like deep learning)

But with more complexity, you need more data.

There are situations in which Physics comes to the rescue by giving you the structure of the model. Most times this happens in an implicit, differential way (PDE: partial differential equations, like those in the MSFS sdk document - except if I remember correctly those are componentwise ODE - ordinary differential equations). Physics also tells you under what hypotheses those equations are accurate (the envelope) and Engineering can add some information about acceptable real-world approximations. Mathematics has methods to solve numerically that stuff. Finding a close form solution (the actual functions) is often impossible, so you have to represent the solutions via interpolating functions most times. It's an approximation, but you can also decide to compile adaptive tables (finer where variations are more intense, coarser elsewhere) and then interpolate at runtime to obtain a good approximation of the desired result.

At the envelope boundary it's a big mess. Physics tells you "I would not bet my reputation on that model", you know that you are approximating, odds are that you will fail to represent something acceptable. You cannot expect regularity where Physics tells you there is chaos.

So what do you do? Model chaos? Feasible, with the help of random components, in controlled configurations with reduced dimensionality. In aerodynamics? I'm not sure. How many stalls will we need in a real aircraft before we have the data to attempt modeling? You don't want to know.

The last bunch of data is, how correctly pointed out above, for validation.

You take your model, run it against A LOT of real world data and check if the results obtained by the model are the same as the real ones. Rinse and repeat.

What's the best model? The one consistently closest to validation data, regardless of how it was designed: the sampling method, the modeling technique, wild guesses, approximations, prayers. 

I also saw some quite interesting considerations about dynamic pressure on controls. Interesting and worth some comments from an aeronautical engineer: control surface deflection as an equilibrium of balancing aerodynamics and mechanical forces is really fascinating, but it needs an appropriate framework to be understood.

A.

Edited August 30, 20214 yr by ADamiani

6 hours ago, Alec said:

Bit late to the train, but I recently found this thead, and  was pleasently surprised by very interesting discussions on Sim Comparion, without ending up in a battle of throwing stones at each other, a Simulation record!! 

You have seen nothing. Try and visit a soccer team fans forum in Italy :-))))))

I'm talking of fans of THE SAME TEAM :-)))))

A.

I don't trust the interface between cold hard rational logic, and human emotion. So I'll stay anecdotal 🙂  I bought MSFS on day 1, launched it, and within 4 hours and seen enough for me to quite comfortably delete it from my SSD.    I tried again after SU2 and the experience was almost the same.  

I left it a while and installed MSFS literally 1 week before the infamous SU5 dropped.  Something had changed.  A lot had changed.  I was still enjoying it 3 or 4 days after the re-install and poor X-Plane 11 was sulking in the corner of the SSD, with jealous pangs.   SU5 was a game changer for me.  With an older i7 4700 3.8 CPU and a GTX 1080, I now had fluidity and FPS that I had no right to be getting with these specs! 

Still, there are a lot of areas of frustration for me within MSFS.  I hate the flight planner beyond words. Both it's UI but mainly it's instability.  I utterly detest the "logbook pop up" when you cut the avionics or engine, more than I hate cabbage - and I truly hate cabbage.   There are many other issues too.   But there is also tons of joy and I've already seen exponential signs of improvement.   One of the things that made me delete MSFS on Day 1, was the inability to make a simple "Direct To" in either the G1000 or GNS530/430, without them then deleting the flight plan (as FSX stock avionics also used to).   Now here we are with the wonderful G1000 NXi, with perfect DTOs and a heck of a lot more.   That's made a staggering difference to my enjoyment. 

I will also say on balance though, that after a few unstable days on MSFS (due to the server issues), I loaded up and took a flight in the Aerobask DA62 in X-Plane.   I hugely missed the quality of the visuals from MSFS, but almost everything else was better.   I'm not saying "XP11 is better than MSFS", I'm just saying that it's so much more mature and the 'basics' as they subjectively are for me (not bothered about ATC!) are all so well covered for my preferences, in Mr Meyer's sim.   

In fairness though, I am mainly comparing there; a heavily addon-utilized X-Plane 11, with a majority vanilla MSFS.   That's a real advantage for MSFS.  I can fly one of the default aicraft, in default scenery in MSFS and thoroughly enjoy it.   I could never do that in vanilla X-Plane; I'd never fly the default aircraft - other than the showcase C172 - and I would literally never fly in X-Plane's default scenery! (Only fly in Orbx TrueEarth Scenery).

So for me, it's all about maturity.  MSFS is not yet adequately mature but I am confident that it will get there.  In the meantime, I am happier than I've ever been in the sim hobby, sharing my time almost equally between MSFS and XP11. 🙂

Edited August 30, 20214 yr by JYW

Sorry for the time for replying your contributions to my Post! This takes my Mind in top shape to keep up!

On 8/30/2021 at 2:56 PM, jcomm said:

Superbly put Alexis !  👍

Was bout to add. among other, the present limitations regarding simulating the effects of dynamic pressure on the controls.

Thanks Jose!! 

This topic really deserves an entire Month of discussion, and we still wouldn't be able to achieve a good all around solution 🙂

On 8/30/2021 at 4:11 PM, MattNischan said:

But this is exactly the rub that most folks don't really understand. There are two absolutely enormous caveats that you yourself state here. One is that there is so much data, and the other is that tiny little point about "normal envelope". Most (nearly all?) big training simulators are focused on normal procedures, and the multimillions part is generally not the software side, it's the ludicrous amount of custom hardware required. And there's zero way for normal consumer space aircraft developers to get data like that.

All that a consumer simulator developer can really do here is make sure the underlying mathematical model is sound and provides correct emergent behavior from the model given certain conditions. The rest is up to the individual flight model config designer.

Let me first agree with you here on one point. I never said Table Based Simulation is the best Approach for a Flight Simulator made in 2020, to Simulate such a diverse range of Unique Airplane types and Sizes. I agree, a Dynamic and Real Time Model for Aerodynamics is a progress in the Flight Simulation Business, compared to the Old Sims of the past, with limited processing power to do such thing in real time. Also, Table Based Modeling needs way too much data, as you correctly pointed out, those basically impossible to get on your hands for 98% of Airplanes in the world, unless you are a lucky employee of such manufacturer. But my point is, you can't say it's better, just because you enter all the data you can imagine and it does it's thing in real time. You need to care of what is coming out, if it's within what you define as valid for your entered dimensions of an Airplane, or you end up with worse outcome than those old tables, where you entered the few data you had, and interpolate the rest with what you think should be reasonable for that plane characteristic, and even if i's like 30% off the real behavior, it's still better than what the 2020's state of the art Code is interpreting from your data.

23 hours ago, ADamiani said:

The last bunch of data is, how correctly pointed out above, for validation.

You take your model, run it against A LOT of real world data and check if the results obtained by the model are the same as the real ones. Rinse and repeat.

What's the best model? The one consistently closest to validation data, regardless of how it was designed: the sampling method, the modeling technique, wild guesses, approximations, prayers. 

A.

 

Damiani, First thank you very much for the excellent and deep reply to my Post!

It's an area I find most interesting, and it's certainly very good to learn from someone who is deep into the inner workings of this industry.

What I wrote, is based on the few materials I have read, and common sense. What you described, is exactly my point. WIthout Validation, your 1 million lines code real time simulation is as good as a Casio calculator, for representing real time aerodynamics. You need to check it for Best and Worst Case scenarios, and which areas needs further work , to give something close enough to the realism you are expecting from the Simulation. I don't even mean advanced Turbulent Flow Simulation, Post Stalled Airfoil Characteristics, Transonic Effect, Propwash areas of influences, nothing like that. I mean a Cessna 172 cruising in calm air, and what you get when you move your controls. I sense this was a step that was skipped during the development of Microsoft Flight Simulator, and they went forward with their current Model, and it was never checked since if that is giving an expected output from the data entered in the Sim.

They just add more and more scalars, to attend the demand of 3rd party developers complaining that the expected behavior for a specific plane is impossible to achieve, without ever "auditing" the source of all deviations. It's like Adding Sugar to a Sauce that turned out too Salty. 

Edited August 31, 20214 yr by Alec

26 minutes ago, Alec said:

It's like Adding Sugar to a Sauce that turned out too Salty. 

And yet, the current C172 flies quite well... so they must have figured out how to correct the earlier mistakes made..

I agree that the early flight model was pretty bad, but for the C172 at least, that is not the case any more. 🙂