Fixed Wing Flight Model (dev branch)

[teabagginpeople 398]

4 hours ago, jone_kone said:

Any new findings on the flight model? Not able to test myself but would be interesting to see how much changes for 1.72

Was testing on rc 1.72

 

Alot less of the hanging in air stalling out.  I mean alot. Which is really good.

[jone_kone 158]

20 hours ago, teabagginpeople said:

Was testing on rc 1.72

 

Alot less of the hanging in air stalling out.  I mean alot. Which is really good.

 

Sounds good! :) How about rolling and turning with and without rudder input? 

[dragon01 902]

Works, coordinated turns are quite possible with analog stick and rudder. The FM is not quite perfect from a flightsimmer's perspective, but it seems to get most of the basics right.

[x3kj 1247]

On 15.6.2017 at 4:55 PM, teabagginpeople said:

Was testing on rc 1.72

 

Alot less of the hanging in air stalling out.  I mean alot. Which is really good.

increased thrust, thats all apparently.

[dragon01 902]

First of all, I have to congratulate you guys on finally getting it right. :) The FM on DLC jets is perfect. It may not be DCS, but what matter is that the planes simply feel right. The Buzzard is one exception, though. It seems like it's either too heavy or not getting enough lift. I noticed that its AoA is always very high, like it's riding the edge of stall just to stay aloft. As a low speed, straight-winged aircraft, its behavior should exactly opposite.

 

Second, there's the matter of AoA indicators. On Black Wasp, the indexer works fine, and if it wasn't for Malden airbase's glideslope literally going through a mountain, I'd have pulled of a "nominal" landing just like in a sim. Two other jets are simply missing theirs, though. Buzzard does have an AoA indicator, but it seems to indicate an AoA that puts me at about 500km/h with full flaps, which is way too much for a safe landing. This might be related to the problem above, but it's just as likely that the indicator itself is not working properly. Speaking of HUDs, it would also be nice if alpha (AoA), G-force and mach number were displayed on the HUD (usually, they're where you put pitch, roll and numeric climb rate).

 

Thirdly, I wanted to mention the G-load limiters and/or effects. Real planes are usually limited to either 9 or 10G positive and 4G negative acceleration. The latter, in particular, can kill the pilot very quickly over that limit (and even -4G can be very dangrous, if sustained). This is where "corner velocity" comes from - the faster you go, the more agile the aircraft becomes (due to control surfaces being able to exert more force), but it also increases the G-forces acting on the pilot. Corner velocity is, as such, the velocity at which peak G during the turn just reaches the limit. Non-fighters (or particularly heavily loaded fighters) can have the limits set much lower, because of structural concerns. I don't think there's a need to implement G-based airframe damage, but software limiters themselves would be a great idea and blackouts/redouts would be a nice touch, as well (thought it could be good to make them difficulty-based). 

[oukej 2911]

Thanks a lot for the feedback. Will take a look at Buzzard (apart from also finishing the old CAS planes).

 

2 hours ago, dragon01 said:

Thirdly, I wanted to mention the G-load limiters and/or effects. Real planes are usually limited to either 9 or 10G positive and 4G negative acceleration. The latter, in particular, can kill the pilot very quickly over that limit (and even -4G can be very dangrous, if sustained). This is where "corner velocity" comes from - the faster you go, the more agile the aircraft becomes (due to control surfaces being able to exert more force), but it also increases the G-forces acting on the pilot. Corner velocity is, as such, the velocity at which peak G during the turn just reaches the limit. Non-fighters (or particularly heavily loaded fighters) can have the limits set much lower, because of structural concerns. I don't think there's a need to implement G-based airframe damage, but software limiters themselves would be a great idea and blackouts/redouts would be a nice touch, as well (thought it could be good to make them difficulty-based). 

You're right, it's possible to achieve quite unrealistic negative Gs. Atm can't be avoided without hampering maneuverability. Agreed on black/redouts.
Otherwise the airplanes have been configured in a way to let them pull:
* Shikra - 10G turns
* Black Wasp II and Gryphon - 9G turns
* Buzzard - 8G turns
And mainly so in the Mach 0.4-0.8 region. Becoming slightly more 'sluggish' once they move into transsonic and supersonic speeds (usually coming to something like 6-7G).

[dragon01 902]

As far as G limits go, this would be best implemented as simple, "artificial" limits, where the plane would have been able to turn faster, but the software doesn't let it. Note that corner velocity is about turn rate, not G. In a real plane, you would be able to pull 9-10G at supersonic speeds, too (unless your plane is not designed for those speeds and loses control surface efficiency, but this is not the case with the aircraft we have) - it's just that the actual rate in degrees per second would be lower, due to greater speed. The faster you move, the slower you need to turn to produce the same G.

 

One way of doing this that would not involve messing with physics is changing control surface deflection limits based on speed. It would probably still require a code change, but you could scale positive and negative input axes separately. I think that the scaling is basically linear with speed, with a cutoff point at corner velocity. This is, as far as I know, how the actual G limiters in aircraft work. An additional use of this feature would be making it harder to break off the rotor in advanced helos.

 

Another way would be not implementing any limiters, but implementing mandatory blackouts and redouts instead. This is potentially easier to do (since you're not touching existing systems), but could make it difficult to manage Gs for KB+mouse players due to lower control precision. It would also be somewhat harder on all players, and you may get complaints about it being easier to black/redout than IRL. In that case, a G indicator on the HUD would be a must.

[anilbu 3]

Latest FM is really good. I can't help myself to stop flying over Malden with some AI fellows. Thank you guys.

I just wonder if the modded planes also get those improvements automatically or do they need to be configured by their creators?

[xxgetbuck123 945]

9 hours ago, anilbu said:

Latest FM is really good. I can't help myself to stop flying over Malden with some AI fellows. Thank you guys.

I just wonder if the modded planes also get those improvements automatically or do they need to be configured by their creators?

 

If they're inheriting off of a BI jet with no changes to the flight model, then yes they'll have that aircrafts flight model. 

 

Obviously modders can adjust these values themselves, however we don't have any documentation or anything on the new config stuff for flight models that I presume BI have added. I'm sure there are some people out there who have no idea what certain values do etc..  @oukej @reyhard

[oukej 2911]

1 hour ago, xxgetbuck123 said:

 

If they're inheriting off of a BI jet with no changes to the flight model, then yes they'll have that aircrafts flight model. 

 

Obviously modders can adjust these values themselves, however we don't have any documentation or anything on the new config stuff for flight models that I presume BI have added. I'm sure there are some people out there who have no idea what certain values do etc..  @oukej @reyhard

The documentation is here https://community.bistudio.com/wiki/A3_CfgVehicles_Plane_class_config_reference

The recent FM configuration has used slightly different approach from the "apex one". Once that's validated by players being happy :) I'll update the wiki with some more details.

[jerminhu 25]

Did the new FM go into the 1.72 update? If yes, the fighter jets still bleed speed way too fast.

[MrFies 20]

4 hours ago, oukej said:

The documentation is here https://community.bistudio.com/wiki/A3_CfgVehicles_Plane_class_config_reference

The recent FM configuration has used slightly different approach from the "apex one". Once that's validated by players being happy :) I'll update the wiki with some more details.

the flight model is really bad the 5th generation jets are turning like old cruise chips... thats hillarious

[Beagle 684]

Version 1.72 did not change the bad behavior of fixed wing planes regarding the turn rate in a standard turn for Stick users. While aspects like nose drop are better now, a standard turn just by banking is stil not possible leadign to unatural need for banking and yanking istead of smooth controls for directional changes. THis behaviour contadicts everythign ive ever seen in a real plane or even all Simulation that come to my mind. The Control more is still more like TIE Fighter vs. X-Wing.

 

There is still no lift factor modeled, planes can go 5000kmh on the ground without even the slightest tendency to lift off without stick input. The same problem produces hard landings since you have to fly the plane into the ground instead of letting it settle by itself at 10° nose up angle in landings.

 

Look as the simplistic but working full real FM in Warthunder a free to play game. Thats what I would exspect from ArmA III at least.

[dragon01 902]

I can tell you that this was fixed on (fairly recent) devbranch. Flight model is good. Banked turns are possible, landings work great and the planes are really agile (at least up to their G limits). 

 

As for not lifting off without stick input, this is realistic behavior. Jets don't have much of AoI, and their wings are thin and symmetric, for efficient supersonic and transsonic flight. The result is that they will generally require a bit of a pull to take off.

[Beagle 684]

1 hour ago, dragon01 said:

I can tell you that this was fixed on (fairly recent) devbranch. Flight model is good. Banked turns are possible, landings work great and the planes are really agile (at least up to their G limits). 

 

As for not lifting off without stick input, this is realistic behavior. Jets don't have much of AoI, and their wings are thin and symmetric, for efficient supersonic and transsonic flight. The result is that they will generally require a bit of a pull to take off.

Well im talkign about the flight model i the released 1.72, the currently newest version. While banked turns are now possible, standard and half turn rate can onyl be achieved when using rudder into the bank. I would exspect that the plane would do that by itself, since we are talking about modern Jets with FCS at least as modern like a 90's F-16 or F-15 which have rudder auto coordination. Without rudder input the Gryphon and F-181 and Shikra need a bit more than 4 minutes for a 360° circle at 25° bank and 450 km/h Thats worthy of an airliner like an Boing 737. Power settign is also a bit wierd since i can keep that 450km/h at 35% throttle. Another strange behaviour is that thr turn rate does not really increase significantly at 40° bank.

[MrFies 20]

yes they are more like a airliner right now in stable 1.72 ;D

[x3kj 1247]

10 hours ago, dragon01 said:

As for not lifting off without stick input, this is realistic behavior. Jets don't have much of AoI, and their wings are thin and symmetric, for efficient supersonic and transsonic flight. The result is that they will generally require a bit of a pull to take off.

when he says landing is bad he means that they have too much lift compared to the drag at given flight parameters. Because having too much lift vs. drag leads to not beeing able to make a proper landing approach. If you slow down for landing normally you have to increase AoA quite a bit to increase lift, which should cause more lift induced drag. This makes controlled decend easy. When the plane has too much lift and not enough lift induced drag at slow speed, you have to dive straight on into the runway (otherwise you gain alt) and level of at just the right momemt to not explode. Or make an extremely shallow approach.

 

Also, seeing the weird "in goo" behaviour of the airplanes (observable when launching a plane into the air via force command) i'm not surprised if this would reduce sinkrate significantly.

[Strike_NOR 898]

Just wanted to chip in my few cents and say that after latest patch I think the jets feel more like jets. I am really liking the improvements and tweaks so far :)

[dragon01 902]

On 6/23/2017 at 11:47 AM, Beagle said:

Well im talkign about the flight model i the released 1.72, the currently newest version. While banked turns are now possible, standard and half turn rate can onyl be achieved when using rudder into the bank. I would exspect that the plane would do that by itself, since we are talking about modern Jets with FCS at least as modern like a 90's F-16 or F-15 which have rudder auto coordination. Without rudder input the Gryphon and F-181 and Shikra need a bit more than 4 minutes for a 360° circle at 25° bank and 450 km/h Thats worthy of an airliner like an Boing 737. Power settign is also a bit wierd since i can keep that 450km/h at 35% throttle. Another strange behaviour is that thr turn rate does not really increase significantly at 40° bank.

The throttle thing is unsurprising, remember that those engines are quite powerful. I agree that automatic rudder coordination would be useful (though it should be enabled on a per-plane basis, so that it can be disabled on the likes of Caesar BTT). The jets do feel somewhat "airliner-ish", but I think this is still a great improvement from when they used not to feel like airplanes at all.

 

On 6/23/2017 at 5:53 PM, x3kj said:

when he says landing is bad he means that they have too much lift compared to the drag at given flight parameters. Because having too much lift vs. drag leads to not beeing able to make a proper landing approach. If you slow down for landing normally you have to increase AoA quite a bit to increase lift, which should cause more lift induced drag. This makes controlled decend easy. When the plane has too much lift and not enough lift induced drag at slow speed, you have to dive straight on into the runway (otherwise you gain alt) and level of at just the right momemt to not explode. Or make an extremely shallow approach.

 

Also, seeing the weird "in goo" behaviour of the airplanes (observable when launching a plane into the air via force command) i'm not surprised if this would reduce sinkrate significantly.

I see what you mean, but I think that if anything, its them not having enough drag, especially at high AoA. Lift is OK, but induced drag is somewhat low. However, I was able to make a controlled, 3-degree approach with the Black Wasp using the ILS and AoA indexer just fine. Control speed with throttle (in such low ranges it's pretty easy), align with stick. It's a bit though on Malden because the airbase glideslope goes through a mountain, but it does work even there.

[Beagle 684]

1 hour ago, dragon01 said:

The throttle thing is unsurprising, remember that those engines are quite powerful. I agree that automatic rudder coordination would be useful (though it should be enabled on a per-plane basis, so that it can be disabled on the likes of Caesar BTT). The jets do feel somewhat "airliner-ish", but I think this is still a great improvement from when they used not to feel like airplanes at all.

 

On the contrary. When I use DCS:W su-25 as a reference, it is impossible to accelerate the plane with rpm less than 60%. Even for taxi at 40km/h at least 40% throttle is used. Our ArmA III airplane feel weightless and power is instant...you start acellerating at 1% and fuselage drag seems not to exist...only flap, brake and gear drag.

 

No matter what Simulation I use for reference, a normal landing in a jet airplane would always involve an AoA of 5-10 degrees and a power setting around 60-70% due to the induced drag.

in ArmA III you need to deploy airbrakes and basically glide in at 0%, otherwise you start to accelerate again.

[dragon01 902]

Are you talking RPM percent, or throttle percent? They're not quite the same thing. ArmA displays a "throttle setting", which goes from 100% to 0% in normal operation. RPM, on the other hand, goes from 100% to about 25% after engine is spooled up. What ArmA displays as "0%" is more like DCS' 15-25% (depending on aircraft) idle setting. While this idle thrust isn't simulated in ArmA (doesn't blow things around or suck people into intakes, which is mostly what idle setting does IRL), I think it's a reasonable approximation.

 

Drag is too low, I agree, but I found that taxiing feels more or less OK, at least in DLC jets. Su-25 is an old-ish, heavily armored CAS aircraft, while DLC jets are light/medium weight (especially since we don't simulate either armament or fuel weight), high performance fighters. Of course, the vanilla CAS planes should be significantly less powerful and more draggy, but they're not updated yet.

[jone_kone 158]

Is it just me or does the black wasp feel like it's loosing speed really fast when turning independent of aoa? It also feels a bit underpowered.

 

However the improved fm is much better now (while not perfect). Never expected 100 % realism so im pretty ok with the current model. Landing on the carrier is now more consistent and in general flying feels more intuitive now.

[Strike_NOR 898]

13 hours ago, dragon01 said:

Are you talking RPM percent, or throttle percent? They're not quite the same thing. ArmA displays a "throttle setting", which goes from 100% to 0% in normal operation. RPM, on the other hand, goes from 100% to about 25% after engine is spooled up. What ArmA displays as "0%" is more like DCS' 15-25% (depending on aircraft) idle setting. While this idle thrust isn't simulated in ArmA (doesn't blow things around or suck people into intakes, which is mostly what idle setting does IRL), I think it's a reasonable approximation.

 

What @dragon01 says regarding engine power setting and rpm is true and they must not be mistaken as the same thing. The 0-100 indication in ArmA is generally just a power-setting feedback to the player. 0 = throttle at minimum, and 100 = throttle at maximum. You can read it as both a feedback as to where your throttle lever is positioned, but also as "what am I telling the aircraft to do" in terms of power.

 

Like @dragon01 mentions, 0% throttle would normally be your idle RPM, and 100% would be your maximum RPM or augmentation (afterburner), depending on engine type. Some manufacturers like to list power in PLA (Power Lever Angle) where it relates to a physical angle of the throttle lever on the engine fuel control. More modern engines only receive digital information which is generated by a potentiometer, hall sensor or equivalent in the throttle quadrant.

 

However I'd like to elaborate more on engine thrust, which is what I believe is somewhat missing in ArmA 3, without going too far into detail. I'll use some info from the Pratt & Whitney F-100-PW-220E engine which is commonly used in the F-16 and F-15, and share the same design philosophy as our in-game counterparts.

 

The F-100 idles at about 65-70% RPM. At this RPM, the engine produces enough thrust to actually move an F-16 on flat surface, hence the pilot must ALWAYS apply brakes. (This was actually a even bigger problem in the early F-100 days because it caused un-necessary brake wear during taxiing and the engine was later modified). The engine reaches 90-100% RPM during military power setting (max), and stays at the same RPM during afterburner (remember that afterburner fuel is injected AFTER the core engine turbines, and as such does not interfere with RPM). 

 

The F-100 is actually controlled by something called Engine Power Ratio demand. Let's just quickly simplify EPR: It is the engine power potential and translates to how many times the atmospheric pressure is multiplied across the engine. If an engine has an EPR of 25:1, it means that in the engine exhaust, the pressure is 25 times higher than the engine intake. This effectively means you have a lot of potential thrust. As we all know, thrust in jet engines comes from the mass of air pushed backwards at ridiculously high speeds, and newtons laws state that every action has an equal and opposite reaction. So force of mass moving rearwards = force of thrust forwards. 

So when the pilot selects a throttle setting (0-100 in ArmA), he is actually requesting that the engine produces a specific EPR, which will correspond to a specific thrust. It is now the engine controllers job to make the RPM increase, to match the desired thrust regardless of atmospheric temperature and pressure. 

 

Older jet engines were regulated based on RPM, but since air mass changes with altitude and temperature, this would mean that a 80% power setting at sea level, is a lot more thrust than 80% power setting at 30.000 ft. By using EPR as regulation, the RPM will adjust itself automatically to maintain the desired thrust.

 

Military jets, like in ArmA, also have variable nozzles. These convert high pressure gases, into velocity gases (or high static pressure, to high dynamic pressure) which increase or decrease thrust. The smaller the jet area cross-section, the higher the velocity. The larger the jet area cross section is, the lower the velocity. To prevent brake wear on the F-16, the F-100 engine was modified to open the Nozzle to 100% open as long as the aircraft was on the ground. This meant reduced thrust and less brake wear.

 

So not only RPM, but also nozzle cross section determines the amount of thrust an engine produces.

 

So what about ArmA?

 

Well.... I have not sufficiently tested the "post-patch low speed characteristics" yet, but I very often used to get the notion that the aircraft really do not operate on a proper thrust vector. Rather that the power setting, combined with angle of attack, determines aircraft speed. If you are stalling in other words, the thrust just isn't there. Even if you are theoretically using engines that, on 100% power, produce more than the aircraft's own weight of thrust, you still slow down. It's very weird, but I get the sensation that if you are in the ArmA 3 stall speed, the only sure way to regain speed is to point the nose against the direction you are moving. The speed will come very fast, as if a brake was released. But if you stay in the high angle of attack area, it can almost "freeze" your acceleration. No matter the angle of the aircraft. As long as your nose is off course in relation to travel direction, the aircraft grinds to a stop, even if you in theory have more than enough thrust to "push through" the stall.

 

If the thrust was an actual force or vector that pushed the object through the air, it would behave much differently, and more realistically. In risk of being too bold, it should work like this:

 

Imagine a plane falling straight down, but with the fuselage aligned to the horizon. It would essentially be falling like a brick, while in a normal flight attitude.

 

There are two ways out of this:

 

1. Point nose down and build airspeed across wings, regain control and pull out of dive (this seems to be the ONLY option in ArmA 3).

 

2. Theoretically if you can hold the plane horizontally, you should be able to give max thrust. Eventually the aircraft will start moving forwards, even though it's still falling there's a constant thrust vector being applied horizontally. Eventually, this vector has accelerated the jet sufficiently that air is now flowing over the wing at an acceptable angle of attack to produce lift. This lift, will cancel out the downward movement and finally the jet will be back to level flight.

 

This is only possible if you have flight control systems to keep jet horizontal (and not tip out of balance and stall horribly). Arma 3 jets supposedly have flight control systems that are capable. And secondly you need very high thrust from your engines. Which arma 3 jets also supposedly have.

But if the engine does not simulate thrust as a vector that works together with the three remaining vectors, lift, drag and weight, then it is simply not possible to simulate realistic flight. 

 

Now if any of the devs read this, don't get me wrong. I am very happy to see work being put into the flight models, and I am amazed at how far ArmA flight has come (both since OFP 2001 and since ArmA 3 release). I am just sharing some thought as to what I "feel" during ArmA 3 flight.

 

It is my subjective view, and please correct the heck out of me if I am wrong. But I feel thrust is a "synthesized" feature in ArmA, and is only really influencing aircraft acceleration in parallell to the angle of attack, instead of fuselage orientation.

 

Thanks for reading :)

 

 

[dragon01 902]

Actually, most real FBW systems won't keep the jet horizontal when "falling like a brick". Indeed, for most part, they will do all they can to prevent such a situation from arising in first place. There's a reason that most fancy aerobatics are done with limiters off. Also, jet thrust increases with airspeed (more air is rammed into the intakes), and if you're falling, then depending on intake design, you could end up with them in aerodynamic shadow. All in all, coming out of stall by increasing thrust is not the correct technique. IRL, stall speed varies with the weight of the aircraft, which is influenced by both its mass and G. Thus, by putting your nose down you actually put the aircraft in a low-G condition, which lowers stall speed (if you can manage 0G, to zero) and makes it easier to recover.

 

In fact, aerodynamic stability prevents normal planes from ending up in situation where they fall straight down, since the drag distribution will reorient the plane nose down. When it doesn't happen, it's called a deep stall (on fighters, at least. A T-tail deep stall is a different thing). On Viper, at least, no amount of throttle is gonna get you out of it, though I suppose powering out of such a situation is theoretically possible, if not with F-16's engine.

 

All I can say, the jets feel good enough over most of their envelope. Extreme conditions, stalling out and supermaneuverability would be nice, but ultimately, what's the use? ArmA is not a flight sim and as such, it primarily needs to maintain fidelity in the part of the envelope players most often find themselves in. Accurate stall modeling is nice, but how many times is it going to get noticed and appreciated?

[Strike_NOR 898]

1 hour ago, dragon01 said:

Actually, most real FBW systems won't keep the jet horizontal when "falling like a brick". Indeed, for most part, they will do all they can to prevent such a situation from arising in first place. There's a reason that most fancy aerobatics are done with limiters off. Also, jet thrust increases with airspeed (more air is rammed into the intakes), and if you're falling, then depending on intake design, you could end up with them in aerodynamic shadow. All in all, coming out of stall by increasing thrust is not the correct technique.

 

While all of this is 100% true, I was indeed addressing the issue of "lacking thrust vectors" on a theoretical level, not real FBW jets. Point was that if "falling brick vector" is joined by horizontal velocity vector due to thrust, then eventually this should bring the sum of vectors on a 45 degree downward slope. Continuing to increase the velocity will eventually bring it to a shallower dive, and then, as the effect of lift comes into play, a horizontal vector.

 

As you state, coming out of a stall is more about getting your nose towards the velocity vector, rather than trying to use power as this may worsen the situation. Stall spins need opposite rudder and yada yada... *Aerodynamics*.

 

In all honesty I owe to the developers to thoroughly test the new aircraft physics in all flight envelopes, takeoffs and landings to get a broader and better opinion on what can be improved upon or what works just right :) Was just trying to convey that Arma Jets have traditionally lacked that sense of inertia and flown more like styrofoam RC planes :)

 

As for suggestions. I would really like to see some headshake for the jets. Like when any form of drag is applied to the jet, such as gear down or flaps full down, speedbrake etc.. it would be nice to see some mild vibration. It's also good gameplay feedback to the player to allow him to know something is slowing him down (forgotten your gear down eh?).