The Physics Model

[RXP 2,338]

[I'm sorry for a 3rd post in a row, I couldn't merge the two previous because quoting from different pages]

@mrueedi whatever your opinion on RXP products, the main point of my post is not the RXP product but the product launch differences. And there is hardly any other way to think that should Asobo would have been in contact and working daily with Reality XP for the last 12 months, not only the RXP products would be available already for FS2020, but I'm certain we'd also have contributed in helping them spotting some of the most arcane internal flaws in the FSX SDK code base they are using and carried over into FS2020.

PS: Some of the flaws I know of and why, are the same preventing WT from making their Garmin gauges available for any aircraft, and are also why you can't have a F/D working without the A/P.

Edited April 6, 2021 by RXP

[mrueedi 702]

26 minutes ago, RXP said:

whatever your opinion on RXP products

Oh, I sure admire what you have created. 

[eslader 1,783]

1 hour ago, mrueedi said:

Therefore P3D is more similar to the MSFS launch than FSX

I wouldn't go that far. Until P3d went 64 bit it was really just warmed-over FSX. Many if not most FSX addons worked just fine in those early versions of P3d without the devs having to do anything. It was such a low-significance upgrade that many of us didn't even bother with P3d until it did do the 32/64 bit changeover, which was a significant upgrade over FSX.

 

 

Edited April 6, 2021 by eslader

[Dazzaboy 81]

5 hours ago, EvidencePlz said:

And I decided to do so. Here are some results:

P3D version 1 was released on November 2010.

Using Wayback Machine (WM) website, I found out it took A2A until 2014-15 to port the majority of their FSX products to P3D. By 2016-17 almost all of their stuff was available for P3D.

Using WM site and Avsim, I see that even in as late as 2016, neither RXP GTN 750 nor any of their product was ready for FSX: SE, and apparently, at that time they were merely wondering as to whether to port stuff to FSX:SE and P3D or not.  

PMDG released P3D's first addon the 737 in March 2015, and for XPlane 10 - which was released in 2012 - they released the DC-6 in 2016.

According to their Facebook and p3d's official forum, FSLabs released their first addon for P3D, the Concorde, on 27th January, 2016, and the A320-X, on 20th January 2017.

At this point I have to concede defeat and admit you are right...LOL.

I would like to issue an sincere apology to you and anyone else who might been offended by what I said above and anywhere else, and take back most of what I said. I still believe it would have been better if Microsoft partnered with top-end developers like PMDG, A2A, FSLabs, RXP etc earlier so that we could buy and use these addons sooner instead of waiting like how we are doing now. 

 

Full respect for acknowledging you were wrong . Just fly the sim you like and dont fly the one you dont . Makes for a far happier day . 

 

Dazza

[TrafficPilot 121]

On 4/2/2021 at 5:33 PM, snglecoil said:

I’m so sick of this debate. Pitch/power for speed has only one right answer! “It depends.” 🙃

Yup a bit of a club bar discussion this. It's power = pitch in my C42 but I know the swept wing jet boys treat it differently.

It can be pretty confusing to student pilots. That and trying to explain what "a stall" is. I read the classic Wolfgang Langewiesche book "Stick and Rudder" when I was younger and that helped me understand the "power=pitch" principle and also why you can stall at any speed. Great read for anyone thinking of learning to fly real aircraft. Despite being written in 1944 most of the information is still relevant today.

Sadly stalling and spinning doesn't appear to be simulated in this new "Microsoft Flight Simulator" unless I fly my FSX vintage SF-260 which DOES stall and spin in "legacy mode".

I was hoping that Microsoft/Asobo may fix the flight model one day but after reading the number of votes on the official forum from people wanting "animated trains" I think I'll give up and go back to FSX for any PC training. Someone needs to tell Microsoft that there is a reason aircraft manufacturers haven't replaced yokes and sticks with XBOX controllers. Can you imagine the carnage?

I've just uploaded "Microsoft Flight Simulator - Beauty Is Only Skin Deep. A review of the flight dynamics bugs in MSFS" to my YouTube channel if anyone is interested. Link below.

Cheers and beers

Captain Moore

 

Edited April 6, 2021 by TrafficPilot

[eslader 1,783]

8 minutes ago, TrafficPilot said:

I know the swept wing jet boys treat it differently.

Mainly in how they refer to it. You might hear them say power for sinkrate rather than pitch (ex, landing a fighter on a carrier). But the bottom line is that you're doing the same thing - using the throttle to go up and down, and using the pitch to speed up and slow down. 

Interesting read here:

https://robertheffley.com/docs/HQs/NAVAIR_2002_71.pdf

page 23.

 

 

[TrafficPilot 121]

8 minutes ago, eslader said:

Mainly in how they refer to it. You might hear them say power for sinkrate rather than pitch (ex, landing a fighter on a carrier). But the bottom line is that you're doing the same thing - using the throttle to go up and down, and using the pitch to speed up and slow down. 

Interesting read here:

https://robertheffley.com/docs/HQs/NAVAIR_2002_71.pdf

page 23.

 

 

This?

"it shows how the back-side technique works everywhere on the thrust-required curve, while the “front-side” technique of controlling flightpath primarily with the longitudinal stick and airspeed primarily with thrust works over a limited range of the thrust-required curve"

Interesting article thank you!

[mrueedi 702]

7 hours ago, TrafficPilot said:

Sadly stalling and spinning doesn't appear to be simulated in this new "Microsoft Flight Simulator"

Are you using the right procedure? By doing it how its done in this video, I got in MSFS the same spins they got.

The ICON A5 on the other hand in real world can be flown almost normaly in stalled condition, something which I was able to accomplish very similarily also in MSFS. Even the sinkrate in stalled condition is very similar.

[Redglyph 4]

That's a good question by the OP. It's funny to see all the bickering and trolling, but there were a few very interesting posts among all that. If I understand correctly, MSFS's aircraft model is potentially almost as good as X-Plane's and better than P3D's but needs tuning regarding some of the moments to get a better stability (what several real pilots have confirmed by saying it was less convincing because it moved too much).

Has this been improved since this discussion last year?

Then of course that's just for the flight model but there's the question of the air it's moving in, and the aircraft system modelling.

[orchestra_nl 931]

34 minutes ago, Redglyph said:

 If I understand correctly, MSFS's aircraft model is potentially almost as good as X-Plane's and better than P3D's but needs tuning regarding some of the moments to get a better stability (what several real pilots have confirmed by saying it was less convincing because it moved too much).

Has this been improved since this discussion last year?

Then of course that's just for the flight model but there's the question of the air it's moving in, and the aircraft system modelling.

I'd say that MSFS's flightmodel is at least on par with X-plane's, and potentially better. IMO it depends mostly on the experience and dedication of a developer to get it as close to the real world aircraft. Even with lookup tables like in FSX one could get pretty close I think.

Wether a flightmodel is convincing enough or not depends on our expectations mostly. Most of us have little or no real world flying experience, at least not in the aircraft we fly in the sim. Yet we think we know exactly how an aicraft is supposed to behave.

Regarding the effect of the air we fly in on the aircraft, I think Asobo has laid the foundations and is expanding on those in the coming system updates.

 

[Tierborn 188]

On 3/28/2021 at 7:25 AM, orchestra_nl said:

IMO there are only two types of people who can know this: those who work at Asobo and possibly real world pilots who have been using both simulators. Everyone else is just using their gut feeling, with a dose of prejudice, expectations, hate and hope thrown in there as well.

gut feeling, or what article they read first

[jcomm 5,938]

2 hours ago, orchestra_nl said:

I'd say that MSFS's flightmodel is at least on par with X-plane's, and potentially better. IMO it depends mostly on the experience and dedication of a developer to get it as close to the real world aircraft. Even with lookup tables like in FSX one could get pretty close I think.

As a RW glider pilot for more than 4 decades one thing I can assure you - both, MFS and XP11, do a lousy job modelling gliders..., but so does FlightGear with the "best" JSBsim or YASim based model gliders.

I'd love to find gliders in these general purpose civil flight simulators modelled as close as possible to Silentwings and Condorsoaring, as well as Flight Unlimited 1 and 3...

I support IL-2 in the hopes of one day getting some Normandy glider add-on, which looks like a very remote chance by now 😕

I look fwd to test gliders in MFS post SU10, and XP12... So far I use them just to "sightseeing" 🙂

Edited August 19, 2022 by jcomm

[lwt1971 7,416]

Ah, an oldie-but-goodie topic 🙂

Recently posted on the MSFS forum a collection of insights and observations re: the MSFS flight dynamics engine and framework, how flight models all come down to how aircrafts implement on top of this framework, default aircrafts vs high fidelity aircrafts, etc: https://forums.flightsimulator.com/t/msfs-flight-models-aerodynamics-etc/526532

Matt Nischan's insights on this topic are very informative, and believe he posted these after this thread ended, so will point a couple out here below... keep in mind these were all before the prop physics and CFD advancements came along (https://docs.flightsimulator.com/flighting/html/Developer_Mode/Aircraft_Editor/Debug/Debug_Aircraft_CFD.htm), and in SU11 the CFD stuff will be extended to atmospheric airflow and interaction with the aircraft.. so the core flight dynamics/physics engine is always evolving and advancing. Hopefully Matt can distill more info on the current FDE in MSFS if he gets the time 🙂


https://www.avsim.com/forums/topic/601526-msfs-has-the-most-advanced-flight-model/page/12/?tab=comments#comment-4549236
"Additionally, MSFS categorically _does not_ use Blade Element Theory. Blade element theory is the idea that you can slice an airfoil up into cross sections, evaluate those cross sections, and then come up with a single lift and drag component for each cross section. XP does this slicing across the defined lifting surfaces to generate a limited number of lift points. It is relatively coarse and doesn't generate different values across each individual surface cross-section, but nonetheless it is used to great effect and the work done with it is quite good, as I've said before.

MSFS also starts with a base geometrically defined lifting surface, but then goes a completely different direction and discretizes the lifting surface into a large number (comparatively) of grid samples. Each individual grid sample receives its own airflow simulation that gets input from the airflow model in true 3d space: i.e. the atmospheric model is also 3d and thus the air itself is not a just a single scalar contribution but instead a varying 3d contribution across each grid sample where the atmospheric model and grid intersect. This means that each grid sample on any lifting surface contributes its forces individually and is also affected by a 3d atmospheric model individually.

Whether or not one believes the current aircraft flight model configurations use this well or whether enough parameters are exposed, the base grid sampling of the MSFS flight model is of a much higher resolution and the atmospheric contribution in 3d is a consumer sim first (to my knowledge, anyway). It also has the benefit of generating different lift values across the surface from front to back, which can be critical value differences at the flight envelope edges."


https://forums.flightsimulator.com/t/more-physics-more-real-winds/372656/256
"There’s no conclusive observation that can be made by looking at one specific aircraft flight model configuration and then applying that conclusion to the entire flight simulation. How well a particular aircraft meets book values is entirely dependent on how well the flight model author adjusted the values to make the book values possible.

This is exactly the same in both MSFS and X-Plane. X-Plane only uses geometry to the same extent MSFS does, for the most part. All the complex study level flight models developed in XP heavily use datarefs to adjust various tables and scalars to modulate the output of the simulation, because all simulations are imperfect.

If the flight model designer has not input the correct parameters into the model, then you get a ■■■■■■ simulation, both in MSFS and XP. It’s why the default 172 in XP flies like it has no idea what longitudinal stability is, while payware offerings are much better: that doesn’t mean XPs flight model overall is garbage, just that the configuration of it may be for a given airplane. Similarly, taking the default 787 which doesn’t match book and claiming it means something about the core of the MSFS flight engine is just misguided.

In the right hands, the MSFS modern engine is going to produce some seriously accurate aircraft. How do I know that? Because our Working Title CJ4 does actually hit those book values at all regimes, with correct N1s, fuel flow, climb rates, over various altitudes and ambient pressures. Not only that but we have stall speeds within a knot of two of book, proper approach angles, correct bank rates, etc.

Is the MSFS simulation completely perfect and without limitations or quirks? No, but neither is XPs, by a long shot. These strange questions and tests are apples and parsnips."

 

 

Edited August 19, 2022 by lwt1971

[Redglyph 4]

Thanks for your detailed reply @lwt1971 . Yes, exactly, there's an API that can be used or bypassed by each vendor, so giving a personal feeling about one aircraft isn't really relevant.

I was wondering about this earlier observation (which seem to agree with what you said and quoted) and the comment made on moment stability because it seems from what I saw and read that many aircraft were very sensitive to what I assume are thermals or wind variation (it has nothing to do with hardware controllers since it's visible without any input from the player). Or at least they showed an exaggerated oscillating behaviour that could be due to this. I must say that I haven't tried the sim yet to experiment with this, but I don't have any deep knowledge of aerodynamic modelling anyway, those courses were long ago. 😉

On the other hand, it's a good sign that the model isn't ultra-simplified and behaves like a perfect level flight or glide like it used to with P3D's default aircraft. So I expect it's mainly a matter of adjustment and perhaps it's already been done.

Edited August 20, 2022 by Redglyph

[blingthinger 649]

On 8/19/2022 at 11:23 AM, lwt1971 said:

It is relatively coarse and doesn't generate different values across each individual surface cross-section

This is incorrect. If more spanwise resolution was needed in XP, the dev can simply break the wing into 2 halves and then suddenly have twice the number of elements. The forces calculated at each element do change with span (less force moving towards tip) and forces obviously do respond to control surface movement (chordwise direction). 

In fact, the spatial resolution Asobo has in place is overkill but isn't hurting anything either. Given that the forces at each element for all aircraft models are calculated based on a generic straight-hersey-bar wing and fuselage and then scaled to be equivalent to the FSX tables, that kind of spanwise resolution (80 elements) is unnecessary.

The force scaling being done is also likely a big reason why the moments are off. At a fundamental level, moments are derived from the sum of all those force elements. If you force the forces into a specific 3D distribution on a) a geometry that doesn't represent 3D reality and b) such that they become the FSX values, the moments will not be right.

 

On 8/19/2022 at 11:23 AM, lwt1971 said:

It also has the benefit of generating different lift values across the surface from front to back, which can be critical value differences at the flight envelope edges.

That this is a benefit is highly debatable. The only time it might remotely be useful is near stall where flow separation is appearing along the chord ("envelope edges"). However, 5 elements along the chord is woefully inadequate to predict something like this. For example, by default, the 2D airfoil flow solver Xfoil sets up nearly 100 points on the suction and pressure surfaces (each). Asobo doesn't provide any details on their stall model other than acknowledging that they flip a switch to treat the elements as falling flat plates. Given that the models do stall reasonably well, there are certainly a lot of "fudge factors" in place.

The "fudge factors" required to get the models to behave are the biggest differences between the flight models, as far as developers are concerned.  Examples are varied and range from XP devs having to add extra wing surfaces to approximate fuselage drag or MSFS devs having to tweak a CFD viscosity term that makes absolutely zero physical sense. Maybe the CFD_AirViscosity term is actually more of a scale factor? I hope so, because if the default's a dimensional number, they're running the CFD with motor oil!

And that's just the tip of the iceberg in both cases. It would be easier to compare the tricks the devs have to play if LR's dataref documentation was more descriptive, but I'll just have to keep dreaming. And I'd be surprised if the big name devs on both sides weren't getting addition information that doesn't get shared.

 

On 8/19/2022 at 11:23 AM, lwt1971 said:

the atmospheric contribution in 3d is a consumer sim first

 

This means that if you have one glider wing in a thermal (3rd party) or part of a microburst column, you'll get a rolling moment. I have yet to see that in action. They've talked about it, but has it been demonstrated in any way?

 

On 8/19/2022 at 11:23 AM, lwt1971 said:

MSFS categorically _does not_ use Blade Element Theory

Funny.... 2 yrs ago Asobo claimed they were. Maybe that was just the M$ marketing department winning an internal battle...

Edited August 20, 2022 by blingthinger