More Physics, More Real Winds

Looking at the wind, I’ve just run some tests with the A320NX in Developer build. Take off and then climb circling the airport to cruise level.

With ‘live’ :thinking: weather and wind at 27107, the wind reported on the ND every 2000 ft up to cruise was fairly linear, starting at 7 on the runway and then ending up at 41 at 37000ft.

So I reset, changed the ground wind layer to 27120G30KT (with 5 gusts per minute), and did it again.

Reset again, and did it with 27130G45KT.

The main thing I noticed was that the gusts were there being displayed in the ND with corresponding movement of the aircraft all over the shop, and this was all the way up to cruising altitude. Between 21-29 kts for the first one and 32-45 kts for the second. I’ll come back to this in a second and make a thought re inertia.

Now I’m no meteorologist and the weather is a complex old beasty, but obviously wind gusts are in part caused by surface friction of the earth and other factors. The surface/planetary boundary layer of the atmosphere is the one nearest to earth, and where this takes place. It’s height changes based on a number of factors but it is not infinite, and has a cap.

I’ve read various heights mentioned dependent on conditions from just off ground level to 2-3000 metres. So say 10,000 feet as a maximum for the purpose of this.

Above the planetary boundary layer is ‘free atmosphere’, where the effects of surface friction are negligible, so it stands to reason that the gusts would disappear by then, if not before. So why are the gusts still there at 35000ft, and why does the strength of them remain constant from 1000 feet all the way to the top? Unless I’ve got something fundamentally wrong of course.

Onto inertia. When you setup wind with gusts and take off, notice how the wind speed changes in the ND. When a gust kicks in, the speed bounces back and forth rapidly between the actual wind speed, and the gust speed. In the 27130G45KT flight I did, it was rapidly and randomly showing on the ND between 32 and 45 knots. And by rapidly, I mean around 5 times a second. Sometimes it jumps straight from 45 to 32, then back to the 40’s, into the 30’s , then back up to 45 and so on. Then we have some calm as the gust subsides, then it’s rinse and repeat

So if the aircraft in the sim is responding to the effects of the rapidly changing wind speeds the weather model is creating, and that wind is ping ponging back and forth every 0.2 seconds, is that what is causing some of the excessive roll and yaw back and forth we’re seeing as the weather model keeps shifting the goalposts and the plane is simply reacting to the information it’s getting thrown at it?

1 Like

As a GA Pilot since 1982 and a Sim Pilot for only 3 months I find this thread quite daunting. Maybe some people are having a bit of a peeing contest?? For around $100 I think its a marvellous program and will only get better over time. Perhaps Sim Pilots have unrealistic expectations at this early stage of development?? It reminds me very much of Windows. In the early days it used to drive me nuts. Now Windows 10 is a very stable platform.


Exactly! That is the problem. If you want to simulate 30 kts gusting 40 kts you’ll need to punch in 60 gusting 80, which is nonsense. Above the friction layer it is normally less turbulent and gusty as those gusts are mostly produced by obstacles within the friction layer. What they should have done is work it the other way around. Take the ground wind you punch in, and modify it to simulate the wind above the friction layer and fade out the gusts. Also the 50% of the geostrophic wind thing at ground level might work fine for light to moderate winds, for strong winds I’m not so sure. Landing in 25 gusting 35 kts the wind isn’t 70 kts at 1000 ft, no way!

Might be, still in the real world there is quite some mass going down that flightpath, so rapidly changing winds especially when its pinging back and forward will probably not have any effect on the net flight path. In MSFS it doesn’t feel like there is any mass behind anything.

I’m complaining about this since day 1. It’s the typical RC plane behavior. I wonder why it is so hard to fix. It’s almost impossible to avoid a twitchy take-off roll.


No, I’m just an old, inactive GA pilot — I got my license in 1978. That’s why I don’t post much in this thread. There are some heavyweight aviation professionals trying to educate in here, and unlike BrayingBrioche, I enjoy hearing what they have to say.

As for my TBM impressions, that report only commented on the obvious changes that were made to the default flight behavior. Whether those changes were due to core code or to the TBM config files was left open.

I do have just over 50 hours in the Piper Arrow (II), and the TBM has always felt like a bigger, faster Arrow to me. I’ve got about 60 hours in the sim’s TBM. That influenced my evaluation.

1 Like

@hobanagerik if you have anything to add then let me know!

1 Like

I can confirm the elevator, aileron and rudders for say the C172 G1000/analog have been twitchy for me from day one. Matter of fact it’s the first thing I noticed from the first time I started using MSFS. A few updates before sim update 3 I was using some of the options devs introduced such as Reactivity, Extremity etc to come up with a bandaid solution. After sim update 3 I’m now using the “authority” slider for the rudders and I pull it all the way to the left before starting a flight. Now all the twitchiness is gone. May be you should try doing that and see how it feels?


Unless you are flying in Gatwick…

Oddly enough during storm Dennis last year I had almost exactly that - down to around 1500’ was around 220ish/70 and reported at threshold was 190/ something gusting 35.

I would certainly say it isn’t typical though!

1 Like

You can add about 30 knots to those speeds. And note that it is possible to exceed the limit load factor in an abrupt pitch-up maneuver at Va without stalling the airplane.

1 Like

I just looked into the developer’s discord for the A32NX where in the FAQ (frequently asked questions) section it says: “The Developer build is updated daily and is a constant work in progress and although we test thoroughly each update, minor issues may occur from time to time. If you find this to be the case, you can report these issues in #help (channel)”.

If you wanna give them feedback about your observations, whether positive or negative, you are more than welcome to do so. 93 real world Airbus pilots and more than 60 aircraft engineers are doing exactly that in their spare time.

As you can see, it’s a work in progress, and so are the other versions (stable and experimental). The same has been the case with expensive study-level addons for Prepar3D and XP as well, and it took those developers years to get to the stage where they are now. It took FSLabs more than 6 years to get to the level of details and accuracy they have today, yet I can replicate everything in your video using those highly expensive “study level” aircraft.

FlyByWire team need to prioritise things as well. Is it important to dive deep down into the flight model and abnormal stuff first (things that 99 percent of simmers on Vatsim or PilotEdge will never do or use, such as intentionally stalling or doing loop-the-loop manoeuvres like you did in the video?) for the first six months? Or is it more important to work on the features that are more important to most simmers (such as accurate systems, navigational features such as holds, missed approaches, VNAV, ILS, VOR/DME etc) so that most of us can take it to VATSIM, IVAO etc and enjoy a nice and realistic point A to B flight?

It’s the developers who decide what to concentrate on first. Regardless, I highly believe your feedback as a real world pilot should be of great importance to them. If you are interested, you can go to their discord and tell them about your observations:

Here’s where I believe you have some confusion that needs to be cleared up. You have claimed earlier that you are not a developer and that’s why you probably don’t quite get it. You are not sure about the flight model of what? The base, core sim? or the flight model data that’s taken from either the manufacturers or real world pilots which is then added to the base sim?

For example, A2A Simulations is known for their excellent flight model for home desktops. They develop exclusively for Prepar3D and the core sim’s flight model is not good enough. That’s why they developed their own system that totally bypasses the core sim’s flight model, and injects their own flight model (it’s called Accu-sim) into the sim.

There are two parts in most Microsoft ESP-based home desktop flight simulators: 1) the base sim’s flight model, and 2) unique flight model data for the aircraft you are developing. You can’t just draw a plane and throw it into the base sim expecting that the plane will magically start acting like the real thing. You have to take the flight model data for the plane, translate it into a language (C++ or whatever) that the computer understands, and then feed it to the plane.

The overall accuracy of the flight model will be dependant on the quality of the data and how good the developers are at converting it using a computer language. You can get this data for the plane either directly from the manufacturers themselves (which will cost an insane amount of money, and that’s one of the reasons why those big motion airliner simulators cost millions) or you can ask real world pilots. The latter option will never be very accurate because you are trying to make approximations based on hearsay, and different pilots disagree on a lot of things too.

The default aircraft in MSFS 2020 are meant to be used by a broad range of people: kids, young adults, adults, people who never flew a plane, people who may be flew a plane for some hundred hours. The default flight model is “highly realistic” (as it says in Microsoft’s adverts) for a lot of people (more than 90 percent of flight simmers are not pilots, according to the latest Navigraph survey). Microsoft never claimed that any of the default aircraft and their systems and flight model will be study-level “highly realistic”. “Realistic” means a lot of things to a lot of people. Asobo developers and Microsoft representatives have been extremely honest and straight-forward about this from the very beginning, and explicitly defined the “realism” of their base default aircraft even way before release. I can show you ample evidence of this if you ask me. If we are asking for study-level level-D types of “realistic physics for each aircraft” which no home desktop simulator ever had straight out of the box, this simulator would have been released in 2050, and cost thousands and thousands of dollars per copy, and definitely not $60. The end result would be yet another stagnating, dying version of FSX, Prepar3D and XP that nobody cares about.

This is how it’s always been in the past for any cheap home desktop flight simulator, and will always be. You buy it for $60, and get some default aircraft with basic systems and flight model that is accessible, useable and would be considered “highly realistic” to a certain degree by a broad category of people, whether it be for entertainment or basic cockpit familiarization. But there will also be some specialist third parties who are both pilots and software developers, and they will acquire data for the unique study level third party planes they want to develop, and they will be highly skilled in translating those physics and maths into a format that the computer can understand and simulate at more than 30 fps. They will take years to do this and will sell you the highly accurate and feature-heavy end result for upwards of $60 on average, on top of the $60 you already paid for the core simulator. Just an expansion of a 90 dollar PMDG DLC costs 77 dollars now, and that’s for Prepar3D which alone costs $60.

Since you ask, here’s what I think: when you say “you are not sure about the flight model”, you are looking at it from a pilot’s perspective and not a developer’s perspective. That’s why I’m inviting you to get in touch with the developers of the A320NX and give them your feedback, if you care about the accuracy of the flight model of that particular model at all, and if you think there are things in it that should be fixed. Asobo have released the core flight model and sold it to us for $60 which needs to be extended and added to over time by the specialist third party pilots and software engineers in order to add the extremely accurate study level aircraft to the core sim. Think of it as Windows 10, which you buy for $100. Microsoft Paint which is pre-installed with your OS is a simple drawing application that we get for free. It can do basic things, but for highly advanced stuff such as professional animation and visualizations, you need to pay upwards of $3000 dollars for 3D Studio Max or Maya.

I don’t want to see your energy and enthusiasm for highly accurate flight model go to waste. It’s not the business of neither Microsoft or Asobo to aim for and deliver the kind of accuracy you and a loud minority in this forum and elsewhere are looking for. Same is the case with Lockheed Martin and Laminar Research. Show me one study level aircraft with highly realistic flight model that either Lockheed or Laminar ever created and sold it along with the base flight simulator software for $60. You won’t find it. It doesn’t exist. It’s not their business. Microsoft and Asobo have said exactly this several times in public, in numerous interviews, in both text and video formats. Deliver your wisdom, time and energy to those who can and are supposed to make great use out of them.

Here’s what I think: Next time when you install Windows 10 or buy a new PC with Windows 10 preinstalled, do not expect 3D Studio Max, Maya and Adobe Premier to be pre-installed for free. Microsoft is a for-profit business and not a charitable organization.


Do you know the ages of everyone on the forum? What is the right age to comprehend the risk? This sentence does not add to your argument and makes unclear assumptions.

Some of us here are pilots, and from what I’ve read some of the pilots are ATP rated and have time in jets from CRJs on up. I think that again you’re making baseless assumptions, this line of argument is specious and borderline ad-hominem. Personally, I’m a lowly private pilot with 137hrs which is admittedly peanuts.

What exactly makes you so sure? This is more wild speculation.


Thanks, I was looking at the aircraft creator but couldn’t find it, you’ve made my day!

Do you also happen to know if it’s possible to turn on atmospheric/terrain wind vectors as we’ve seen in some of the Asobo videos?

I’ve not found a way to do that sadly.

Actually the in sim data points really are pretty bad, I was expecting to be able to see a lot more after their expose. I’m trying to ‘heavily’ modify an aircraft and can’t see certain vectors like the propeller forces for example. Can’t even see where my tailwheel position is exactly with it, just some offset box which represents not a lot really…

Thanks for your extensive post!

I don’t think the FBW team changed anything to the base flight model and instead have modified the fly-by-wire itself to correct for the misbehaving flight model. Also the FBW A320 has this weird and wrong dynamics, same as the base A320, the the CP being in front of the CG, up-force on the horizontal stabilizer and CP moving backwards with an increase in AOA. I asked one of the FBW devs about this once somewhere here on the forum and he answered that I should ask Asobo about it.

Agree, and again the A32NX flies marvellous with the flight computers (fly-by-wire) on, so the FBW did a great job. I never intended to do those loops until I noticed the aircraft didn’t stall at all during my design maneuvering speed tests, so I thought, lets see what happens when you keep pulling the side stick full back…

Thats the point, I can’t figure out if it is the individual flight model or the core sim because I have no knowledge about these things. The only thing I notice is that the aircraft has no inertia, I’m not familiar with the A320, haven’t flown it in the real world, I do fly the Embraer 190/195 and have loads of hours on a 737-400 level D sim, they are build essentially the same way so I don’t expect their behaviour without fly-by-wire to be much different. I don’t know where my theory stands but it goes something like this to explain this odd behaviour:

  1. There is no inertia, in real life there is a lot of mass following that flightpath, the aircraft does not suddenly change direction and / or speed when apply control inputs or when the wind is gusty / turbulent. If you fly straight and level in cruise and suddenly decide to yank the yoke full back, the nose pitches up while the aircraft continues down the original flightpath for a moment. The aircraft flight path will be lacking behind the nose-up pitching moment and therefore AOA will be exceeded. I don’t see these effects on any aircraft, not the C172 and not the Boeing 747 either. The aircraft instantly responds and the flightpath does not lack behind the nose-up pitching moment enough to enter a accelerated stall.

  2. Something which might have an influence on this is the effectiveness of the horizontal stabilizer and elevator, I think it became aerodynamically limited after some update. I’m speculating that this is the “improvement” they implemented to have the tailplane stall at large elevator deflections, I think they might have overdone this, you can see the aircraft pitching up more sharply in the initial moments after applying full up elevator, after which elevator effectiveness seem to reduce. The horizontal stabilizer is having a negative AOA during straight and level flight and while the AOA increases (becomes more negative) while moving the elevator towards the up position, the angle of attack decreases as the tail starts to move down (nose moving up), I doubt the horizontal stabiliser will stall before the wing does while performing these sharp pitch-up manoevers.

I found this video on Youtube, Airbus A320 in either normal or alternate law? Doing unusual attitude recoveries. You see how difficult it is to recover without entering an accelerated stall, you can hear and see the wing buffeting during recovery:

There are more videos like this on Youtube, you can see how easy it is to exceed the critical AOA when being too aggressive on the pitch. I don’t know if these things can be modified in the individual aircraft flight model or if it is a problem with the core flight model?

The only experience I have with this in real life on Transport Category aircraft is the Upset Prevention Recovery Training (UPRT) in level D sims, after unusual attitude recovery I find it very easy to enter an accelerated stall on the aircraft types I have experience with. If you end up nose low at high speed, you need to very gently pitch up as not to enter an accelerated stall, yanking the stick (full) back as in MSFS it would immediately stall and drop a wing instead of looping end over end.

I agree with you, but this approaches more the accuracy level of GTA V in some respects. We should at least expect the default aircraft to represent the basic effects, not accurately maybe but at least to some extend, slipstream, adverse yaw, P-factor, propeller drag, inertia, ailerons affecting angle of attack and leading to wing drop. Inner wing stalling during base to final turn etc. I don’t care a thing about needing a little more right rudder in real life, or in real life it feels a little more like this or that, I’m sitting stationary behind a computer with a joystick lacking feedback from the flight control system, with a different range of motion etc. It is never going to be the same as the real deal. The basics should be right, and they will at some point get it right I assume.

On the topic of unusual attitude recovery, it’s an old video but this guy does a fantastic job explaining it:

The term “Corner speed” he is using means design maneuver speed in this context (Va).


Will certainly try that. The question that comes to mind is: Does that solve the core problem of twitchiness, or does it just limit the deflection of surfaces - which is more of a workaround than a solution.

MSFS main problem appears to be a too direct calculation of what forces on an aircraft should do. While it is more accurate, it feels like in GTA.

We reached a mile stone people, 1000 posts in this thread (mostly useless arguing). Congratulations everybody :+1: :smiley:

See my post above.


Just out of interest they are in direct law there, not something we generally do in our sims - our own UPRT is mostly done in alternate law.

It is indeed quite easy to achieve a secondary accelerated stall in such manoeuvres if you aren’t paying attention, but of course one must remember in no small part that’s likely because of the lack of g-force feeling in the simulator. In real life you would certainly feel how hard you are pulling back during the recovery when over 2g is making you feel like you had a few too many pies for Christmas. In the sim the only real g-force feedback you have is the tiny number in the bottom of the SD page - you can hear the PM actually calling out this number for situational awareness (again something we don’t personally do but not necessarily a bad idea) but obviously not much of a replacement for the real thing.

Fully agree with everything you say, I started this discussion because it feels like there is very little mass behind anything in MSFS, gusts, turbulences etc. throw even the Boeing 747 of path directly, controls (especially the rudder) are very twitchy. I’m aware you normally won’t fly in alternate law or even direct law mode, same on the Embraer (although less advanced it has a “normal” mode with AOA protection). The intend is to demonstrate the lack of inertia, pulling fully back on the side stick in direct law mode, at high speed should immediately trigger an accelerated stall, not a loop :sweat_smile: :joy:.

1 Like

This is the old problem of estimation. In air traffic control we have the problem of estimation of position in three coordinates and estimation of velocity in three coordinates. The position estimation is always better then the velocity estimation, because velocity is the derivative of position. The same is true for moments.

I agree. The Xplane fan boys think different, but they are the Xplane fan boys.

Well, I made flight model mods to fill the gap. Actual there are 3189 downloads of my TBM930 and 1295 downloads of my C172

My problem: Bad documentation. Some parameters have no documentation. I believe that some parameters are not used - for example the Aircraft Editor, Debug, Legacy parameters. See for my up-to-now knowledge.

1 Like

The relative lack of inertia is pretty evident from the very start where aircraft in general are far too happy to get rolling quickly with idle thrust.

Btw having scrolled through this thread I must commend you on your patience, it’s admirable.