The Physics Model

[sd_flyer 6,042]

Tthere are no superiority. Just certain areas of one sim handles better than another. World "realism", "authentic", "study level" applies to a degree of relative comparison. For example, Sublogic ATP sim was the most realistic in early 90th, we can't say it's now.

Edited March 28, 2021 by sd_flyer

[Chock 17,769]

8 minutes ago, mtaxp said:

https://github.com/nasa/XPlaneConnect

'This research tool has been used to visualize flight paths, test control algorithms, simulate an active airspace, or generate out-the-window visuals for in-house flight simulation software'

So, not for actual flight modeling then; which is what we're on about here.

Seriously, you need to be aware that when Laminar bang on about this aspect of XPlane and Blade Element Theory etc, it's PR, it's the same as when PMDG go on about that 'within five percent' stuff and everyone just swallows it all up without looking into it in detail. Blade Element Theory was developed for testing underwater drive screws (propellers) on ships back when paddle steamers were still common; it's not some new magic flight modeling system.

Edited March 28, 2021 by Chock

[Biology 1,347]

No, P3D's flight dynamics engine is inferior to MSFS' modern flight dynamics engine. P3D's flight dynamics engine does not even model local effects like induced roll caused by rudder input. Add-ons like FSLabs, PMDG, A2A and Majestic etc. modify or completely override default flight dynamics engine, which is how they achieve more realistic behavior.

P3D uses data tables for global aerodynamic coefficients (Cl vs AoA, Cd vs AoA, Cm vs AoA etc.), which determines the behavior of an entire aircraft. As these coefficients are empirically accurate, the aircraft’s flight performance can be accurately reproduced. However, local aerodynamic effects like induced yaw, induced roll due to rudder input, stalls etc. can not be reproduced accurately, as they depend on geometrical properties of an aircraft. For instance, a yaw motion causes one wing to have more lift, causing a roll motion.

Reading the SDK documentation, this is how I think MSFS flight dynamics engine works:

MSFS works by dividing aircraft into several surface elements, giving each surface element its own set of local aerodynamic coefficients, then integrating element forces to determine aircraft behavior. But how exactly are these local aerodynamic coefficients determined, given that they don’t have empirical data unlike global aerodynamic coefficients? This is where MSFS' normalization algorithm comes into play.

MSFS' normalization algorithm has a single task - finding the correct set of local aerodynamic coefficients which will match global aerodynamic coefficients in all flight conditions. In order to accomplish this, MSFS starts by distributing global aerodynamic coefficients to surface elements proportional to their surface area. After that, an iterative optimization algorithm will further tune local aerodynamic coefficients so that when all element forces are integrated (when a “zero-order” solution is reached) resulting global aerodynamic coefficients will match the global aerodynamic coefficients entered in tables. I am not sure why they try to estimate local aerodynamic coefficients for wings though, as they are already available in the form of airfoils, which is what X-Plane uses. I assume this is for backwards compatibility purposes, so that FSX tables still can be used the same way as before.

I did lots of experiments and it looks like MSFS almost perfectly matches Cl vs AoA and Cd vs AoA tables for almost all flight conditions. However, Cm vs AoA tables and other moment & stability related data seem to be rather off - maybe this is the reason MSFS aircraft feels twitchy / inertialess: Unlike other local aerodynamic coefficients, Cm and other moment related coefficients have a worse fit. I hope Asobo addresses this and allows for a more complex fit with more data points. Several days ago I've noticed someone else reporting the exact same thing - local Cl and Cd fitting the aircraft really well while Cm having a worse fit, so I think this is reproducible too.

In short, MSFS flight dynamics engine maps global aerodynamic coefficients in FSX tables to local (surface) aerodynamic coefficients using a regression algorithm, which allows keeping empirical accuracy of data tables while being able to represent local effects. Once they address Cm related fit, it will be a really good flight dynamics engine. While I find X-Plane's physical approach for wings better (as MSFS tries to estimate local coefficients for wing elements, which are already empirically available in the form of airfoils), MSFS is not too far behind, unlike P3D, which needs massive improvements.

Also, using default aircraft is not a good way to test a flight dynamics engine. Don't believe me? Here's an example: Default Cessna 172 in X-Plane doesn't even spin properly, because it uses wrong airfoils. With just a few tweaks, it performs so much more realistic. Garbage in, garbage out - I've looked at flight models of a few default MSFS aircraft and surface areas of many control surfaces were higher than real-life, which might explain abnormal surface authority. Values in tables don't seem to be matching the real aircraft either. While the Cm fit issue I've mentioned is not just a garbage in, garbage out situation, the rest clearly is.

Edited March 28, 2021 by BiologicalNanobot

[tweekz 2,165]

12 minutes ago, Chock said:

So, not for actual flight modeling then; which is what we're on about here.

Of course not. XP can never replace a proper CFD software. That's a marketing trick. Actually, with all its flaws, MSFS is closer to a CFD approach than XP. 😉

Edited March 28, 2021 by tweekz

[mtaxp 666]

16 minutes ago, Chock said:

'This research tool has been used to visualize flight paths, test control algorithms, simulate an active airspace, or generate out-the-window visuals for in-house flight simulation software'

So, not for actual flight modeling then; which is what we're on about here.

Seriously, you need to be aware that when Laminar bang on about this aspect of XPlane and Blade Element Theory etc, it's PR, it's the same as when PMDG go on about that 'within five percent' stuff and everyone just swallows it all up without looking into it in detail. Blade Element Theory was developed for testing underwater drive screws (propellers) on ships back when paddle steamers were still common; it's not some new magic flight modeling system.

I have no idea why are you to deny the fact that people designed and simulated real aircrafts before their first flight, on x-plane. instead trying to nitpick.

Latest known one is evtol by BETA. Ask them, not me.

[Biology 1,347]

3 minutes ago, mtaxp said:

I have no idea why are you to deny the fact that people designed and simulated real aircrafts before their first flight, on x-plane. instead trying to nitpick.

Latest known one is evtol by BETA. Ask them, not me.

Sure, X-Plane's flight model is good, but I don't understand why are you discussing X-Plane's flight model when the topic is about MSFS' flight model, compared to P3D.

Edited March 28, 2021 by BiologicalNanobot

[fogboundturtle 2,708]

The best physics is the flight sim where I invested the most money.

[mtaxp 666]

33 minutes ago, tweekz said:

I remember someone who claimed we'd hear something about XP12 by X-mas... so, don't go too hard on him. 😉

Although not an official "x-plane 12 is coming", couple months later though, we already got a tease for a new native HUD (no new features are coming for v11..), they are still silent though on graphics stuff probably kept for the major announcement (my bet is june in fsexpo).

[tweekz 2,165]

1 minute ago, mtaxp said:

Although not an official "x-plane 12 is coming", couple months later though, we already got a tease for a new native HUD (no new features are coming for v11..), they are still silent though on graphics stuff probably kept for the major announcement (my bet is june in fsexpo).

wow

[mtaxp 666]

2 minutes ago, BiologicalNanobot said:

Sure, X-Plane's flight model is good, but I don't understand why are you discussing X-Plane's flight model when the topic is about MSFS' flight model, compared to P3D.

I responded to a quote that compared to x-plane.

[Biology 1,347]

4 minutes ago, mtaxp said:

I responded to a quote that compared to x-plane.

It doesn't even come to a conclusion, it just says that MSFS uses a similar approach to X-Plane (using local coefficients and integrating surface element forces), I can't see why this caused you to think the post says X-Plane's flight model is inferior.

Edited March 28, 2021 by BiologicalNanobot

[Chock 17,769]

12 minutes ago, mtaxp said:

I have no idea why are you to deny the fact that people designed and simulated real aircrafts before their first flight, on x-plane. instead trying to nitpick.

Latest known one is evtol by BETA. Ask them, not me.

I'm not denying anything, I'm pointing out to you that designing something and analysing for the basics and not even necessarily in terms of aerodynamics, before you go to creating wind tunnel models, is not done using the fifty quid copy of XPlane you get off Steam, it's done using the seven hundred and fifty Dollar version, plus a load of additional software and hardware. None of this means your 15 quid Cessna in XPlane flies rings around one in P3D, FSX, MSFS, AeroFly etc. You are conflating the fact that Laminar tout some aspects of what XPlane's professional version is occasionally used for, with this meaning the version on Steam is some kind of super-sim.

P3D has similarly professional customised versions, which some of Lockheed Martin's clients use for real world training - including the Royal Navy using it for some of its F-35 simulators. But does not mean that your home version of P3D will make you capable of mounting a naval assault on North Korea.

Edited March 28, 2021 by Chock

[Biology 1,347]

3 minutes ago, Chock said:

But does not mean that your home version of P3D will make you capable of mounting a naval assault on North Korea.

That's... oddly specific 🤔

Edited March 28, 2021 by BiologicalNanobot

[abrams_tank 7,419]

1 hour ago, mtaxp said:

You really need to get into facts rather thab quoting other and then trying to twist the bottom line.

Long story short; x-plane is used in a pro manner to develop REAL aircrafts, test how they will fly, analyze performance, or in other words it is reliable enough for cessna, the us military, NASA...so if somedbody looking for aerodynamics research, x-plane is the best answer currently.

LOL, I don't think NASA is using the home version of XPlane for aerodynamics research. I trust that person's opinion, who at least did the math, over yours.

Edited March 28, 2021 by abrams_tank

[mtaxp 666]

2 minutes ago, Chock said:

I'm not denying anything, I'm pointing out to you that designing something and analysing for the basics and not even necessarily in terms of aerodynamics, before you go to creating wind tunnel models, is not done using the fifty quid copy of XPlane you get off Steam, it's done using the seven hundred and fifty Dollar version, plus a load of additional software and hardware. None of this means your 15 quid Cessna in XPlane flies rings around one in P3D, FSX, MSFS, AeroFly etc. You are conflating the fact that Laminar tout some aspects of what XPlane's professional version is occasionally used for, with this meaning the version on Steam is some kind of super-sim.

I will not talk more about x-plane since people are a bit sensetive here. But yes, aerodynamics are a part of it and there are specifc powerful tools that come even with the home version of x-plane, but also, getting a flight simulator flexibile and open enough to support this kind of developement is something really to no underwhelm, whenever I see MSFS (including their marketing team that claims it is superior) it kinda cringes me due to reality literally proves them wrong.

All I did is support my claims compared to a quote that vaguly quoted another, there is no reason to get mad (no directly to you).

Now i'll go..:-)