su27SM flight model improvement

[orsin 160]

After patching the 2008 Oct for WOX, it has become a quite apparent issue for virtually all agile jets that high AoA leads to speed bleeding at low. One of that jets affected is the outstanding Su27SM released lately.

 

In this post, I would like to discuss with anyone interested in resolving that issue. The eventually goal is to find a generic way to modify the FM of all birds and minimize the impact brought by the change of the 6DOF math model introduced by late-2008 patch.

 

As implied in the topic, SU27SM will be taken for experiment. By default, the result will be optimized for hard settings.

[orsin 160]

1. tools involved

AIDE was a very good tool to forecast the air performance. However after late-2008 patch, it has shown its shortage. My previous test flight under hard mode shows its inaccuracy nowadays. No offence to the author, it is just because TW added more parameters and changed the way that the game engine calculating the 6DOF math model.

 

As a result, AIDE is completely abandoned in this improvement. I will run thousands of test flight to achieve the data needed. It is no longer recommended to use the result of AIDE as guidelines for hard mode.

[+X RAY 83]

I'll take this post to add a line.

I've got woe on both configurations installed on my HDD,one is pre patched and one is post patched oct 08.

I've got Insky su-27SM,and 30Kn,plus some betas,on pre patched version sometimes the game stops,without crash,staying on the same frame,and I must restart the computer again,wich parameter cause it?Shadows?Or something to delete?

In Oct 08 patch this problem seem solved except some balck shadows that link the plane to the ground like a black wire,my system is updated to latest nvidia drivers.

About flight caratheristics i agree with Orsin about differencecies between pre and post patch versions

[orsin 160]

2. previous attempt

 

My previous attempt in that problem was through modifying

a) Drag vs Alpha for wings and control surfaces

b) Deflection angle of control surfaces

 

Drag vs Alpha, i.e. the CDLAlpha parameter, will control the multiple of the base drag parameter under different Alphas. UsingSmaller multiples bring less drag at angles. However there is a drawback - since the drag is lower, at altitude, there will always be an envelop where you can accelerate to over 1000km at maximum turn!

 

Another way is to limit the deflection angle of control surfaces, thus prevent high AoA to happen. Apparently that will worsen the performance at high speed.

 

As a result, these methods are not considered the best way and we will substitude with better methods.

[orsin 160]

3. the possible method

 

 

My impression for the issue as below

 

a) the drag&speed-bleeding problem appears more frequently at low.

b) The speed-bleeding issue does not appear at maximum turn all the time, when you do a maximum turn, the speed decrease slowly initially, till it reaches a "critical point", where all in a sudden the drag get a lot larger and you bleed 100km/h per second.

 

so the first thing that I planed was to test fly and verify my impression. If anyone can do a test flight and help verify it will be appreciated.

 

 

 

Modifying CDLAlpha turns to be a way sacrifice the realism of the aerodynamics, tweaking the deflection angle turns to be a way that weaken the performance. We want the speed-bleeding to be prevented at the small envelop where it happens, so the tweak will target at that envelop, while IAS is less that the "critical point"

 

The possible method is to modify the lift force of the control surfaces at IAS<crtical point

 

I will run a series of test flight at 1000m, or maybe smaller interval, to find out the rough speed of that "critical point"

 

We can do a plot of those critical points, link them together, and then estimate a IAS that is close to most of the plots.

Edited February 12, 2009 by orsin

[Fubar512 1,350]

3. the possible method

 

My impression for the issue as below

 

a) the drag&speed-bleeding problem appears more frequently at low.

b) The speed-bleeding issue does not appear at maximum turn all the time, when you do a maximum turn, the speed decrease slowly initially, till it reaches a "critical point", where all in a sudden the drag get a lot larger and you bleed 100km/h per second.

 

so the first thing that I planed was to test fly and verify my impression. If anyone can do a test flight and help verify it will be appreciated.

 

If my impression is correct, I will run a series of test flight at 1000m, or maybe smaller interval, to find out the rough speed of that "critical point"

 

We can do a plot of those critical points, link them together, and then estimate a IAS that is close to most of the plots.

 

(to be continued)

 

a) & b) sounds like a pretty accurate description of how an aircraft of that type might respond in RL. You must keep in mind that modern aircraft (such as the aforementioned Su-27) have an aerodynamic center that's located waaay aft of the center of gravity, than was customary with earlier fighter designs. This renders them quite sensitive in the pitch axis. In fact so sensitive, that they must rely on FBW systems to retain stability in level flight, as anything more than the slightest displacement in the pitch axis would otherwise result in a loss of control.

 

One problem is that this sim does not model FBW control, so one must employ a series of cheats to establish a workable compromise.

[orsin 160]

The possible method is to modify the lift force of the control surfaces at IAS<crtical point

 

Anyone has better idea?

[Fubar512 1,350]

The possible method is to modify the lift force of the control surfaces at IAS<crtical point

 

That would imply dumbing them down. What FM are you using, and who wrote it?

[orsin 160]

That would imply dumbing them down. What FM are you using, and who wrote it?

 

Fubar512 can you explain that a little bit? I'm using the FM come with su27sm

[Fubar512 1,350]

Well, someone wrote it...lol I'm just trying to determine whether or not it was current, or an older model, and what values are used. For example, the FM you're using may possibly have wildly inaccurate stall and control surface deflection & displacement rate values.

[orsin 160]

Well, someone wrote it...lol I'm just trying to determine whether or not it was current, or an older model, and what values are used. For example, the FM you're using may possibly have wildly inaccurate stall and control surface deflection & displacement rate values.

 

my believe, that model is a combination of Marcfighters original su27sk, ErwinHans and Lindr2's data based on reference, and some imagination for the parameters not mentioned in the reference.

[orsin 160]

That would imply dumbing them down. What FM are you using, and who wrote it?

 

it might be a reasonable tweak as I find pulling the control surfaces to the maximum deflection angle does not bring significant improvement at turn rate. I didn;t measure it but my experience is about 2 degrees/second more, while the quick speed bleeding make the turn unsustainable. It appears as the bird is turning the same rate and losing its speed. Tell me if you observed anything different.

[Fubar512 1,350]

it might be a reasonable tweak as I find pulling the control surfaces to the maximum deflection angle does not bring significant improvement at turn rate. I didn;t measure it but my experience is about 2 degrees/second more, while the quick speed bleeding make the turn unsustainable. It appears as the bird is turning the same rate and losing its speed. Tell me if you observed anything different.

 

I'll bet the values have been all buggered up. If it was Kreelin's original FM, it was fine to begin with, until all the "cooks" showed up and put their respective spins on it.

[column5 63]

I'll bet the values have been all buggered up. If it was Kreelin's original FM, it was fine to begin with, until all the "cooks" showed up and put their respective spins on it.

 

Fubar speaks the truth.

[+76.IAP-Blackbird 3,557]

I thought this is a thread to improving and debugging it with the help of those people who knows something about it.............

[column5 63]

I thought this is a thread to improving and debugging it with the help of those people who knows something about it.............

 

According to orsin it is neither improved nor debugged--sounds like quite the opposite is true, in fact. Ruined and buggered up.

[+Gunrunner 314]

Ahem, not to play the carebear, but turning that into "us experienced modders" against "them amateurs" was not the purpose of the thread and nothing good would come out of it.

Comparing the FMs and explaining why the changes are bad, though time consuming, is a little better and may help the "cooks" to avoid the same mistake later on (of course it would have been better if they started by asking those who know better first).

[orsin 160]

According to orsin it is neither improved nor debugged--sounds like quite the opposite is true, in fact. Ruined and buggered up.

 

It is apparent that the FM needs fix, whoever made the orignal FM does not justify the accuracy at all. In fact, when the game is patched all the so called accuracy have gone IMO. Should we try oversee the most obvious problem nowadays? The bird just doesn't feel right now.

Edited February 12, 2009 by orsin

[column5 63]

Ahem, not to play the carebear, but turning that into "us experienced modders" against "them amateurs" was not the purpose of the thread and nothing good would come out of it.

 

You misinterperet completely. This isn't about experienced versus amateur modders. ErwinHans and his team are quite experienced and do excellent work. The problem is that people take these flight models and introduce error after error, compounding problems, and do it under the guise of "improvement" when there is no clearly defined problem. This is running rampant in the CA downloads and its practically impossible to download something and have any confidence in its performance relative to other aircraft.

[Fubar512 1,350]

The point of my post was not to pit experienced modders against new ones, but to lament that someone may have taken something that worked well, was balanced, and pretty accurate out of the box, and changed it, without really improving it.

[column5 63]

It is apparent that the FM needs fix, whoever made the orignal FM does not justify the accuracy at all. In fact, when the game is patched all the so called accuracy have gone IMO. Should we try oversee the most obvious problem nowadays? The bird just doesn't feel right now.

 

I'd suggest going back to Kreelin's original flight model for the Su-27 and using it as a starting point. Its very good, and was based on actual data and had the benefit of Kreelin's many years of experience. You can then start to identify individual problems that it may have under the new patch, and address them one by one.

[+76.IAP-Blackbird 3,557]

I must say that I don`t knwo the ini difference between hard FM and normal FM, some of my favorite birds seems to be buggy after the last patch in hard FM so I stepped down to normal FM. All work ok until now. The prob is that we have only a few very good people around here, kreelin, fubar, column5 I htink I forgot someone but we can count them on one hand. Making a right FM is for me something like try and error.

 

To start with a good base FM is always a good point to start. And the SM isn`t final it`s version 1.0 but still some stuff to work on ;-). Maybe someone with a good experiance in FM modeling can help..

 

best regards

 

Martin

[+kreelin 2]

I've donwloaded and checked the SU27SM's FM. It seems that it was based on very first beta by Starfighter2 of Marcfighters's SU27. So Orsin, as C5 stated, I suggest you to test my latest SU-27 FM (almost one year old now). If you are happy with mine, use it as a base for your 27SM

 

You can grab my FM here http://kreelin.free.fr

 

@76.IAP-Blackbird

 

you forgot Streakeagle and Starfighter2

Edited February 12, 2009 by kreelin

[+76.IAP-Blackbird 3,557]

As I said you can count them on one hand ;-) thanks for the link

[+X RAY 83]

Not to bore anybody up,but is someone that can answer me about my previous reply?