A More Classical Departure

[Caesar 305]

Hey all,

 

I was wondering if anyone here has gotten their plane to depart from controlled flight without having the plane go to negative airspeed. I mean, I've gotten planes to go irrecoverable by going straight up till I run out of speed and it goes negative; then the thing's damn tough to get out of; but has anyone departed under more normal operations, like too much AoA or too high yaw-rate? I find most planes to be highly resiliant, and began to wonder about this when I voluntarily put an F-16 into a "flat spin," by selecting full burner, rudder, and opposite lateral stick. Thing is, I was able to recover from said spin with opposite rudder extremely quickly. I've tried the same with the F-14 but couldn't get a rate fast enough to be considered a flat spin, or any other departure, save for the straight up-run-out-of-airspeed circumstance, so I was just wondering if anyone's actually gotten their plane to depart, or if it's even possible for that matter.

 

-"Caesar"

[Heck 496]

Hey all,

 

I was wondering if anyone here has gotten their plane to depart from controlled flight without having the plane go to negative airspeed. I mean, I've gotten planes to go irrecoverable by going straight up till I run out of speed and it goes negative; then the thing's damn tough to get out of; but has anyone departed under more normal operations, like too much AoA or too high yaw-rate? I find most planes to be highly resiliant, and began to wonder about this when I voluntarily put an F-16 into a "flat spin," by selecting full burner, rudder, and opposite lateral stick. Thing is, I was able to recover from said spin with opposite rudder extremely quickly. I've tried the same with the F-14 but couldn't get a rate fast enough to be considered a flat spin, or any other departure, save for the straight up-run-out-of-airspeed circumstance, so I was just wondering if anyone's actually gotten their plane to depart, or if it's even possible for that matter.

 

-"Caesar"

 

Seems to depend a lot on the individual plane, Caesar. The stock aircraft were modelled very benign, if I remember other threads correctly. This was a purposeful choice on the designer's part. Mod aircraft seem to behave very differently from each other, depending upon how "ruthless" about the flight model the builder wanted to be. I've had Zur's F-86's depart on me in turns, trying to follow Mig 15's in sustained turns at low speeds (read 100 knots, or lower), but have never had problems with recovery once I learned to yank the throttle back to idle, then pour it back on as I pushed the stick forward. Then again, I've never had a Mig 15 depart on me. When it comes to WOV and SF aircraft, I don't recall ever having any departures in turns, so I must be the greatest flyer in the world, or it's not in those aircraft ini's to depart. I don't know what part of the data ini controls it, but I'm sure some part does, because some do it, some don't.

[Caesar 305]

Ah, understood. Now that you mention it, I have gotten an F-86 and an F-106 to depart under such conditions as too fast turn rate, or the like, but the more modern models seem very resiliant. Thanks for the input.

 

-"Caesar"

[fireengineer 1]

The latest GR3 Harrier is a real handfull in conventional flight. Lost her in a tight banking manuver and she went to a crazy 3 axis spin. The AI planes do not like the harrier either you can loose half your flight to crashes in any given mission. Especially so in low level low speed patterns the ai cant handle the sudden departure of the harrier and soon you can hear lots of exploding ground impacts as your flight get to grips with mother earth!!!

[+Dave 2,322]

The latest GR3 Harrier is a real handfull in conventional flight. Lost her in a tight banking manuver and she went to a crazy 3 axis spin. The AI planes do not like the harrier either you can loose half your flight to crashes in any given mission. Especially so in low level low speed patterns the ai cant handle the sudden departure of the harrier and soon you can hear lots of exploding ground impacts as your flight get to grips with mother earth!!!

 

Play the GR.3 FM on normal and they are fine. Depsite what some people say I think any thing on Hard FM messes with the AI more. Since I use it normal I do not have that issue.

[+baffmeister 1,205]

Hey all,

 

I was wondering if anyone here has gotten their plane to depart from controlled flight without having the plane go to negative airspeed. I mean, I've gotten planes to go irrecoverable by going straight up till I run out of speed and it goes negative; then the thing's damn tough to get out of; but has anyone departed under more normal operations, like too much AoA or too high yaw-rate? I find most planes to be highly resiliant, and began to wonder about this when I voluntarily put an F-16 into a "flat spin," by selecting full burner, rudder, and opposite lateral stick. Thing is, I was able to recover from said spin with opposite rudder extremely quickly. I've tried the same with the F-14 but couldn't get a rate fast enough to be considered a flat spin, or any other departure, save for the straight up-run-out-of-airspeed circumstance, so I was just wondering if anyone's actually gotten their plane to depart, or if it's even possible for that matter.

 

-"Caesar"

 

It's possible to get the F-100 to do an "interesting" departure. Look Here for information. I havn't really done much more experimenting with this but the F-100 departure is following general aerodynamic principles at least for the first few seconds. Example: in a climbing turn the aircraft departs opposite the turn direction and in a descending turn the aircraft departs into the turn direction. I suspect Third Wires flight engine would support departure modelling to some degree, but how to do it properly is the question. Also raises some interesting questions for flight modellers. Using the F-16 as an example, would the fly by wire computers even allow an F-16 to depart controlled flight? Another example is the F-104G. It was equipped with a stick pusher that would push the control stick forward [ I think around 60 to 100lb force] if the aircraft was approaching a critical angle of attack. So, would it even be realistic trying to model departures for those aircraft? For the F-104G it might be more realistic to try and model a sudden nose drop when the aircraft reaches a critical AOA, to simulate the stick pusher. Interesting topic. :)

[Fubar512 1,350]

Post-stall behavior (ie; departures) have not worked properly in SF since the 3.X service packs. Streakeagle had it down pat with his F-4B flight model, until a change was made in how the sim interprets FM data, and it has been porked since then.

 

For example:

 

A.) stall induced drag is no-longer linear, it arrives all at once. The work-around is to "stagger" stall values for the effected surfaces.

 

B.) Any positive value in the stall moment variable will now result in an unnatural pitching moment.

 

That being said, here's some snippets from the man himself (TK), that may shed some light on stall modeling.

 

 

"StallMoment=

 

StallMoment is the additional pitching moment you get when the wing section starts to stall. positive pitch (+) makes it so it pitches up as you stall (making it more unstable), negative pitch (-) makes it so it pitches down when you stall (making it more stable). The default value, if you don't specify a value, I think, is -0.02, which I think would make for gentle mushing forward type stall I was looking for...

 

There are a couple of other stall/depart variables added, all of them have default values so if you don't specify, you just get the default behaviour you see...

 

PostStallCma=

 

determine the Cma (pitching moment due to Alpha) past stall. Pre-stall, the engine uses lift value and xac value to calculate the pitching moment. Once the wing stalls and departs, additional pitching moment of StallMoment + the PostStallCma * (AoA - AlphaDepart) is added. I think this defauls to -0.2 or something.

 

StallHysteresis= (should be from 0.0 to 1.0)

 

determine how far the AoA has to be reduced in order for the airflow to reattach itself. I think 0.0 means the wing section will be unstalled as soon as AoA goes back under the stall angle, 1.0 means wing will unstall only when AoA goes back to 0 deg. I think the default is 0.4 or something.

 

PostStallZeroLiftAlpha= this is the angle at which the lift goes down to zero. Post AlphaDepart, the lift is linearly reduced down to zero at this angle. I think this defaults to 90-deg..."

[Caesar 305]

Ah, interesting stuff. I didn't know the latest patches had affected a/c departure. I'm still off finding the full envelopes of different 3rd party a/c...idk why really, maybe I just like seeing how far I can push it? I've inadvertantly lost an AV-8 in a hard turn, like fireengineer but in an older variation (the AV-8C). Of course, this may have to do with the fact that I fly all things on Hard, so that probably affects it also, as USAFMTL pointed out.

 

As for the F-16's flat spin and if it can be done, I have no idea as to whether or not the computers would let the F-16 enter into a departure envelope, but my flat spin wasn't really much of a departure; really a highly controlled spin that was quickly recoverable. I still haven't lost a Tomcat to a spin due to excessive yaw-rates, but the velocity at which I've been trying may still be too high; roughly 300KIAS, sometimes 280-250. Maybe I'll make graphs of the data I find when I'm done...

 

Well, back to work!

 

Vale,

-"Caesar"

[+baffmeister 1,205]

I'm still off finding the full envelopes of different 3rd party a/c..

 

Caesar, if you find any good websites that have detailed performance data for specific aircraft, please make a link to them. That kind of detailed information seems to be hard to find on the internet. Also, thanks to Fubar for the stall info.

Cheers. :)

[Caesar 305]

Caesar, if you find any good websites that have detailed performance data for specific aircraft, please make a link to them. That kind of detailed information seems to be hard to find on the internet. Also, thanks to Fubar for the stall info.

Cheers. :)

 

For performance data, it's usually best to try to find aircraft manuals, like (for Navy A/C) NAVAIR NATOPS operations ones if you can. It is true what you say, finding that data online is not easy. Also, in terms of books you can find some real gems too; but specific data for stall speed, max AoA and whatnot is real tough. Even the official data is not 100% accurate; or is modified for certain circumstances.

 

The Tomcat (A-model) for example, is given a +6.5/-2.4G envelope in the latest public release of the NAVAIR 01-F14AAA-1 manual (1998). This is a precaution to preserve the old airframes, as through the 1970's and 1980's, the Tomcat was a 9G-rated plane. During its initial flight testing against an F-4, the Tomcat started with a +7.4G, forcing the F-4 outside the Tomcat's turn, reversed, and ended up on the Phantom's tail. The Tomcat pulled +8G with 6 AIM-54A Phoenix missiles, 2 267 gallon drop tanks and 2 AIM-9G sidewinders onboard at Mach .9, 20,000 feet altitude. (that little bit o' info from a 1975 release of Grumman F-14 by J.P. Stevenson, pg. 68) At M1.2, 20,000 feet the plane pulled 9G. With the glove veins extended the 'Cat pulls over 7G at M2.04 at 50,000feet. Below 275 knots, probably clean or with 4 AIM-7's, the 'Cat can pull over 6.5G. Under controlled circumstances, the plane can hit -5.5G on the opposite end.

 

On AoA, the 'Cat in the NAVAIR manuals is not supposed to exceed 30 degrees AoA; but the Tomcat has hit nearly 90-deg. AoA. It can roll 360 degrees at 50 knots, 25 degrees AoA (again, J.P. Stevenson's book, pp.68) something that the NAVAIR manual would normally prohibit. So, to get an idea of the a/c true operational envelope, you have to look to more than one source, and finding test data, where they push the aircraft to find those limits is the best you can come by, but also extrememly rare.

 

For now, I'm trying to find the limits of the a/c within the SF game; or, the limits that SF puts on the a/c.

 

Vale,

-"Caesar"

Edited March 29, 2006 by Caesar

[+baffmeister 1,205]

Tomcat =

[Fubar512 1,350]

On AoA, the 'Cat in the NAVAIR manuals is not supposed to exceed 30 degrees AoA; but the Tomcat has hit nearly 90-deg. AoA. It can roll 360 degrees at 50 knots, 25 degrees AoA

 

For now, I'm trying to find the limits of the a/c within the SF game; or, the limits that SF puts on the a/c.

 

Vale,

-"Caesar"

 

Those figures are not in degrees, they are instead measured as"Units" of AoA. General rule-of-thumb, 20 units of AoA = 16 degrees. To get the proper figures for AoA limits, you really need a dash-1 manual for each of the USAF birds, and a NATOPs for the Marine and Navy A/C.

 

The MiG-21/23/29 and SU-27 are easy to figure, the limits are plainly marked on their alpha-meters

 

Simply knowing the stall limit (AlphaMax) is only part of the story. You also need to find out where the stall buffet (AlphaStall) begins to set in. That data that us usually not listed in any manual, you'll need to obtain "anecdotal data" for that, preferably from someone who's flown the aircraft type that you're modeling.

 

Also, a variable-sweep design such as the Tomcat is impossible to model accurately, as the SF engine currently supports only one set of values, whereas you'd need values for each wing-sweep position. The obvious work-around is to model the values as an average, ie., "mid-sweep".

[Caesar 305]

Ah, I see; my mistake. As for the flight models, I'm completely satasfied with the Tomcat and every other 3rd party plane I've flown; they're real kickass designs. The given data up there's just some of the test limits and earlier limits of the plane to compare it before the Navy had to put restrictions on it due to airframe age and those damn TF-30's. God I'm gonna miss that bird. Thanks for bringing more to light, Fubar.

 

Vale,

-"Caesar"

[+streakeagle 871]

To keep CPU workload down, the A.I. does not fly by the same rules as players. They get a slightly simplified model. When a player selects autopilot, his aircraft is actually flying under the AI model. This can be observed using debug mode where you will see that an aircraft's weight is constant when flying on autopilot.

 

Even with its slightly dumbed down FM, the A.I. cannot handle realistic FMs. They pull too high of AoA, enter deep stalls, won't use enough power, and won't try to recover when they stall/depart. I only recommend using hyper-realistic FMs for multiplayer dogfight mode where there is no AI, or just for personal enjoyment of flying since the AI otherwise ruins the single player experience with difficult FMs.

 

Reducing the FM from Hard to Normal removes some aspects of the FM engine. It makes planes fly better by removing or overriding some of the laws of physics. We are all playing this sim to have fun. If putting the FM on Normal makes you happier, please do so. Personally, I want every aircraft in the game to fly as realistically as possible and refuse to fly with anything less than Hard for all of my settings. Most aircraft fly just fine on Hard settings. Some of the FM makers have joined me on the dark side in trying to produce very high-fidelity flight models. While I love this, the flight engine and AI limitations make it a pain-in-the-butt for the average player. In the future, I intend to produce pairs of realistic FMs: one made as realistic as possible, the other cleaned up to make it AI flyable. Of course, I intended to finish and release a much improved version of AIDE along with a complete set of revised F-4 FMs as well... but real life takes priority over modding sims, so you never know what I will be able to do or when I will be able to do it.

[+baffmeister 1,205]

Even with its slightly dumbed down FM, the A.I. cannot handle realistic FMs.

 

I had a modified A.I. that worked quite well with your F4B. [flight setting on hard] The A.I. mod has reduced pitch imputs/rates as well as a higher safe altitude. Also, all the aircraft have their cruise speeds set at 550kts which might help them carry more energy into the dogfights. It hasn't been tested much yet but I'm back at it and will be keeping track of kill ratios and crashes to see how it works out. Lot's of fun testing this stuff.

Regards.

[Fubar512 1,350]

Raising the cruise speed will not fix the issue, as the AI is not "cruising" during a dogfight. Raising the corner speed by 10% or so, and bumping the stall speed up to at least 70 m/s helps to a degree. The real trick is limiting the AI's tendency to exceed alpha stall, or, better yet, having the developer add an AI max alpha variable that can be defined in each aircraft's data.ini. I've discussed the latter with TK, and he seemed receptive to the idea.

 

Another approach is to edit the "AIData section of the aircraftobject.ini to limit the AI's predilection to-wards radical turns, excessive pitch inputs, and to force it to throttle up at the slightest drop in airspeed.

[Caesar 305]

Hi again,

 

Just to share some of my experiences with the AI:

 

I actually find my AI pilots to handle their a/c pretty well; not too many AI losses on my machine due to stalls or awkward maneuvers (this could be because I usually fly the more modern a/c like the F-14, -15, -16, and -18 moreso than planes like the A-4, F-4 or -106.) Now if there was something I wish the wingmen wouldn't do is close in when they have 6 remaining AAM's to try to gun a bomber with its own gun in the tail that usually accounts for more losses than their handling the plane too roughly.

 

That said, I have had several occasions with the older planes where they loose it, and after following the last 3 posts or so, I guess I now know why...of course, considering that I, the human pilot have done the same thing on more than one occasion (including that backflip incident) with cold-war era planes allows me to forgive them.

 

Vale

-"Caesar"

[+baffmeister 1,205]

Raising the cruise speed will not fix the issue, as the AI is not "cruising" during a dogfight. Raising the corner speed by 10% or so, and bumping the stall speed up to at least 70 m/s helps to a degree. The real trick is limiting the AI's tendency to exceed alpha stall, or, better yet, having the developer add an AI max alpha variable that can be defined in each aircraft's data.ini. I've discussed the latter with TK, and he seemed receptive to the idea.

 

Another approach is to edit the "AIData section of the aircraftobject.ini to limit the AI's predilection to-wards radical turns, excessive pitch inputs, and to force it to throttle up at the slightest drop in airspeed.

 

 

The increased cruise speed I'm using is to make it more difficult to chase down enemy aircraft that are in " Navigation Mode". I think it helps the initial stages of the dog fight as the aircraft have decent speed at the start. All my Third Wire aircraft have the stall speed raised 20%. I raise the stall speed for add-on aircraft on a case by case basis. Streakeagles F4-B stall speed remains unchanged. Havn't done anything to the corner speeds yet. I have edited the A.I. data section of the aircraftobject.ini [ got some inspiration from your Uber-A.I. mod but the current one is built from scratch.] Ran three quick air to air missions last night against Mig-19s and Mig-21s and the F4-B was doing quite well with a better than 2 to 1 kill ratio [including 2 F4B crashes] but then on the fourth mission there were 5 F4-B crashes out of 16 aircraft. They may have been chasing Migs that were in the landing pattern but not sure. Made a few tweaks to the A.I. and will try again.

Edited April 2, 2006 by baffmeister