The not so great flight model

[Kaan KASIM 54]

It's clear from SDK and ingame dev mode, you can't have proper biplane.

 

As you can see there is only one wing. I believe there can only be one vstab too.

[EmaRacing 302]

ouch

[Chock 17,769]

Well although the Pitts S2 has two aerofoils, which differ for the upper and lower wings: NACA 63A015 (on the upper) NACA 0012 (for the lower). They are both symmetrical ones, and they are designed to work cumulatively, so if you can design one aerofoil which averages things out, it will essentially be the same.

After all, on every biplane, the wings are meant to work together - their main function is the increased lift of two wings and the box bracing they engender for high G resistance -  and it's only on something like a Nieuport Bebe or an Albatros DVA sesquiplane, where one wing is meant to act significantly different from the other, and look how that worked out. Both of them had to be redesigned so the wings didn't fall off. 🤣

Edited August 18, 2020 by Chock

[MattNischan 3,778]

The flight model is actually quite neat compared to other sims, from playing with the SDK a bit. Virtual wind tunnel, great data, flight surface sampling, very very cool stuff.

But yes, there are currently limitations in the available flight surfaces. You need one wing, a vertical tail, and a horizontal stabilizer. There is not currently the ability to create arbitrary flight surfaces. For bi-planes, a single wing in the average of both is a fairly close aerodynamic approximation, although won't produce exactly the same stall or spin characteristics as the real deal (but it will be close, as there are rarely flight envelopes where one wing is acting a ton differently than the other).

X-Plane is a bit ahead in this regard, as the inputs to the flight model can include any number of flight surfaces anywhere along the aircraft. This presents its own challenges, though, in designing aircraft that fly by the numbers. They're both reasonable approaches.

[Slides 1,552]

So where's that Frenchman who said he would resurface on the 18th and expose Asobo's subterfuge on reusing the FSX flight model.

[MattNischan 3,778]

1 minute ago, Slides said:

So where's that Frenchman who said he would resurface on the 18th and expose Asobo's subterfuge on reusing the FSX flight model.

He'll be nowhere to be found, for sure. Maybe he saw that the aircraft flight cfg files had many of the same named parameters and made an assumption. That's the nice take.

More likely, someone who just wanted to stir up some rubbish.

[Paraffin 2,197]

5 minutes ago, MattNischan said:

X-Plane is a bit ahead in this regard, as the inputs to the flight model can include any number of flight surfaces anywhere along the aircraft. This presents its own challenges, though, in designing aircraft that fly by the numbers. They're both reasonable approaches.

I wouldn't say both are reasonable because one is far more limiting in available aircraft types. You couldn't model a canard aircraft design this way, and rotary wing aircraft need their own specialized treatment of airfoils. It can't just be a "flying disk" for helicopters; it requires modeling each rotor blade independently, so you can model things like retreating blade stall. 

Anyway, it's early days yet. Let's assume this will be expanded to include what's needed to model all aircraft types realistically.

[Slides 1,552]

3 minutes ago, Paraffin said:

I wouldn't say both are reasonable because one is far more limiting in available aircraft types. You couldn't model a canard aircraft design this way, and rotary wing aircraft need their own specialized treatment of airfoils. It can't just be a "flying disk" for helicopters; it requires modeling each rotor blade independently, so you can model things like retreating blade stall. 

Anyway, it's early days yet. Let's assume this will be expanded to include what's needed to model all aircraft types realistically.

In most cases you end up getting more accurate results using lookup tables, parameters and simplified equations than you do with modeling every surface. Look at how much work XP needed before it become believable.

And other sims like IL2, DCS and F2 use a combination of methods and have great flight models. Not everything needs to be BET. Especially edge cases.

[MattNischan 3,778]

27 minutes ago, Paraffin said:

I wouldn't say both are reasonable because one is far more limiting in available aircraft types.

We have different definitions of reasonable here.

Canard craft are not in the majority of planes that an aviator is likely to encounter. Helicopters are great fun to sim, but even in aviation heli pilots are a small fraction of aviators. I agree that not being able to currently define arbitrary flight surfaces makes the model less extensible, but I think the flight model approach is still quite reasonable, especially if the goal is to get planes that people enjoy that can be flown by the POH. Just fly the DA40 NG in the sim. Aside from the simplified engine setup (no simmed ECU), it's within small percentages of the POH on the numbers, and maybe just a touch underpowered. That's pretty darn neat for a default plane in a sim.

But yes, early days yet, all that considered still.

Edited August 18, 2020 by MattNischan
Typo

[ilya1502 42]

Has anyone found yet where exactly this FM is different from the FSX one? I'm still into downloading the SDK, but looking into the cfg-files I can't see anything new except that good old tables from air-files were moved there.

[ilya1502 42]

-del-

 

Edited August 18, 2020 by ilya1502