Flight Model artefact ? -

I started doing some new turn performance tests with the new flight model and first let me say that on the whole, performance charts look better now, more what I would expect than before (pretty much constant G until corner speed).

However, while testing sustained turns, I came across something that seems really weird : at 10 000 ft (configuration mentionned below) there is a very sharp peak in sustailable load factor at M0.92 or 520 KCAS, up to 9 Gs. I wonder if this is intended, but it looks like it might be an artifact from the flight model to me.

On my end at least, it is very reproducable (note: I do my tests with unlimited fuel and GLOC off, to guarantee accurate results - supposing these do not modify the flight model behaviour in any way)

Tests were made in DCS World 1.5.8.12823

Tests were made in DCS World 1.5.5.12823

I'm assuming you mean 1.5.8.12823, latest stable version at the time of posting?

Affirm, thanks for noticing the typo

Hi, nice finding.

By curiosity, how do you test and extract the STR values ?

No problem. Here's my test procedure for sustained turn rate measurements

Test flight

- be at the correct altitude (I am using ME altitude instead of altimeter altitude)

- start on the high speed end

- start a high-g turn, pushing the throttle to maximum

- adjust roll and stick force to maintain calibrated airspeed

- keep the airspeed and altitude stable for about 5 seconds

- reduce speed and take another measurement

Data computing

- open Tacview Pro

- open graph view, in relative time for the X-axis

- export TAS, CAS in a file, export Mach, G in another file (since only 2 values can be exported at the same time)

- look at the graph (in M-G, or CAS-G), spot the sustained turns

- zoom and choose a sweet spot where airspeed and G look stable enough

- go ahead and copy TAS, CAS, Mach and G for this time (Tacview measurements are made every .25 seconds)

- to check for variability it is possible to take several measurements at different points in the same sustained turn**

- finally, calculate turn rate and turn radius from aerodynamic formulas (using TAS and G)

*: sometimes it can be hard to be perfectly level, so I accept a slight descent or climb, as long as everything is smooth

**: I have had pretty close results doing that, so unless I am unconvinced by a specific turn, I now take only one measurement

Note: it would be possible to extract turn rate and turn radius from the Tacview directly, but I'd rather calculate them from TAS and G, as I am afraid Tacview might introduce deviations (I think Tacview uses actual heading change, which, if you are not wings level, may overestimate your turn rate - just picture a loop, with in instantaneous 180° heading change). Now I may be wrong about that, but it shouldn't matter anyway if flights are performed up to standards, and the tests I did make when I started this had the Tacview turn rate measurements virtually identical to the calculated ones.

It is a rather tedious process, although it isn't that long with a little practice. Count maybe 2-3 hours for one altitude and configuration. High speeds are easier than slow speeds, speeds around the best STR can be tricky because of the very big effect any change in roll will have.

OK, I was doing pretty much the same, except I use TacView turn rate.

I thought you had a better way of doing it because you have a lot of data points for STR :thumbup:

I didn't noticed your pick but I was testing M1.0, M0.95, M0.9, M0.85, M0.8.

I check FPM on horizon and VVI for STR level flight.

FPM and VVI, yep. As you can see, the sudden increase in turn rate is very localized, and may be missed at first. After my first test flight I had only one point that was affected by it, although it was not that higher than the others ; still I thought it was weird, so I ran a second (and then a third) series of tests around this speed range to see what happens exactly.

Normally I try to take a reading every 25 knots.

It is a problem indeed! If it would be right, then that peak in greater STR should be spread across a higher range of speeds, not just for a particular speed. Otherwise said, in terms of alpha (which governs the lift to drag ratio, or the aerodynamic polar), the better L/D ratio which would translate in higher STR in the end should be found over a higher alpha range (2 to 4 degrees I'd say). Moreover, simply the fact that at some point the STR becomes greater than the ITR=))LMAO tells how professionally this flight model was worked out! The way it looks right now definitely confirms that for an exact alpha (doesn't show up on your graphs though) the L/D suddenly becomes very high. Slightly above or below that alpha, the L/D gets back to the original polar's function and not outside it. So, this problem is either due to the fact that the lift might suddenly jump higher at that particular alpha or the suddenly drops.

Anyway it might be, this must be fixed along with a lot more flight model problems that I personally found regarding: lift slope, critical AoA, drag curve to AoA, drag at 90 AoA, pitching moment vs AoA and engine thrust which reflects into a thrust to weight ratio which is almost 20% higher than that of the F-15C which is total rubbish.

The Mirage 2000C must enter a complete flight model overhaul before things get messy with it!

Anyway it might be, this must be fixed along with a lot more flight model problems that I personally found regarding: lift slope, critical AoA, drag curve to AoA, drag at 90 AoA, pitching moment vs AoA and engine thrust which reflects into a thrust to weight ratio which is almost 20% higher than that of the F-15C which is total rubbish.

 

The Mirage 2000C must enter a complete flight model overhaul before things get messy with it!

You don't seem to have flown the last update, did you?

Looks like your opinion reflects the state of the FM before december 2017 overhaul.

Regards,

Az'

You don't seem to have flown the last update, did you?

Looks like your opinion reflects the state of the FM before december 2017 overhaul.

 

Regards,

Az'

You might be right! But I didn't do the analysis performed by the one who showed the STR and ITR charts, so I took his word for granted and didn't re-check this on my own.

Otherwise, although I didn't yet install Tacview I've done other tests, like CL max determination (through stall speed test), CD 90 (drag coefficient at alpha 90) and post stall pitching moment behavior and what I found proves that the plane needs other aerodynamic data fixes.

If I could have a word with the ones at RAZBAM who control or can modify the aerodynamics of this plane, I will sure be useful.

Right now, the plane's maximum lift coefficient and also lift slope is way too high for this plane's wings, of course, taking into account not only the wing shape but also the slats and generated vortexes (which can be neglected for not being too strong).

I would like to put things into detail with someone from RAZBAM involved in the flight model.

Regards!

Regarding the T/W ratio, it seems that the Mirage has it right (after I've made some further tests) and it only seems that ED (somehow) didn't correctly input the engine thrust performance into all the Flaming Cliffs 3 aircraft (MIG-29, Su-27, Su-33), because for a simple instance, we know what a thrust to (weight+drag) ratio is and how it affects the plane's X axis acceleration.

For some reason, all the FC3 planes might be plagued by a relatively low acceleration from what they should have, but this is another discussion and I will address it in it's appropriate thread.

So I now take back my words when I said it's "rubbish" the fact that the Mirage out accelerates the F-15 (which normally should be the best in this area) in DCS, cause it seems this is not a DCS Mirage 2000C's T/W ratio problem.

You might be right! But I didn't do the analysis performed by the one who showed the STR and ITR charts, so I took his word for granted and didn't re-check this on my own.

 

Otherwise, although I didn't yet install Tacview I've done other tests, like CL max determination (through stall speed test), CD 90 (drag coefficient at alpha 90) and post stall pitching moment behavior and what I found still proves that the plane needs another overhaul.

 

If I could have a word with the ones at RAZBAM who control or can modify the aerodynamics of this plane, I will sure be useful.

 

Right now, the plane's maximum lift coefficient and also lift slope is way too high for this plane's wings, of course, taking into account not only the wing shape but also the slats and generated vortexes (which can be neglected for not being too strong).

 

I would like to put things into detail with a the man from RAZBAM involved in the flight model.

 

Regards!

This is CptSmiley who is coding the FM.

https://forums.eagle.ru/showthread.php?t=197059

For instance, have a look at this:

While holding full aft stick during the glide (no engine thrust for pure aerodynamic lift), the AoA settled at around 28. The CAS settled at 155km/h. The plane's weight at the time was 9300kgf (20500lbf). With a reference wing area (41m^2), the lift coefficient resulted as 1.96. Tremendously high for this type of wing. Rather said, almost twice as high. From my experience and analysis, this wing, with full slats, at Mach 0.2 produces no more than a 1.3 CL at alpha 30. How come this gets 1.96 in actual simulation? I've opened the file called "M-2000C.lua" in CoreMods folder and at the aerodynamic data section, I do indeed see 1.14 at Mach 0.05 and (surprisingly / abnormally) the maximum CL drops continuously until reaching Mach 0.3. From my personal experience (being an ex-aerodynamicist), the CL max increases up to Mach 0.3 and only afterwards it decreases (especially at shock stall Mach), so I find this a bit strange too. Still, it baffles me to find that during simulation the max CL (Cymax) gets to 1.96.

Also during deep stalls at around 90 AoA (achievable through pitch limiter disengage), the fall speed was only 125km/h CAS. This determined a CD (drag coefficient) at alpha 90 to be 2.7 o.O!! Tests and experience show that the majority of planes share a CD at 90 AoA around 1.15, very similar to that of a flat plate in 3D flow, not 2D tests. So both the lift and the drag functions are exaggerated. Can someone explain this?

The Flanker's speed at a held alpha of 30 is 183km/h CAS with maneuvering flaps down. While weighing 22300kgf (49100lbf), it's lift coefficient proved to be 2.2. So the Mirage is almost as good as the Flanker in terms of aerodynamic lift? Keep in mind that the Mirage doesn't benefit from high energy vortexes generated by the LERX which usually increase the maximum lift coefficient by at least 50%. The Mirage doesn't have flaps (can't have) which also increase the maximum lift coefficient (even though the critical alpha becomes lower, the CL with flaps is higher). This should be a good enough proof that the DCS Mirage has too high lifting performances as compared to the real plane.

Kind regards!

This is CptSmiley who is coding the FM.

https://forums.eagle.ru/showthread.php?t=197059

Thank you Jojo! I'll try to talk with them and share some thoughts!

Regards!

Thanks you Jojo! I'll try to talk with them and share some thoughts!

 

Regards!

thanks for your contribution to the FM Maverick Su-35S!!, interesting stuff you posted, really appreciated. Hope it can make the FM better.:thumbup:

For instance, have a look at this:

 

[ATTACH]176891[/ATTACH]

 

[ATTACH]176892[/ATTACH]

 

While holding full aft stick during the glide (no engine thrust for pure aerodynamic lift), the AoA settled at around 28. The CAS settled at 155km/h. The plane's weight at the time was 9300kgf (20500lbf). With a reference wing area (41m^2), the lift coefficient resulted as 1.96. Tremendously high for this type of wing. Rather said, almost twice as high. From my experience, this wing, with full slats, at Mach 0.2 produces no more than a 1.2 CL at alpha 30. How come this gets 1.96 in actual simulation?

 

I've opened the file called "M-2000C.lua" in CoreMods folder and at the aerodynamic data section, I do indeed see 1.14 at Mach 0.05 and (surprisingly / abnormally) the maximum CL drops continuously until reaching Mach 0.3. From my personal experience (being an ex-aerodynamicist), the CL max increases up to Mach 0.3 and only afterwards it decreases (especially at shock stall Mach), so I find this a bit strange too. Still, it baffles me to find that during simulation the max CL (Cymax) gets to 1.96.

 

Also during deep stalls at around 90 AoA (achievable through pitch limiter disengage), the fall speed was only 125km/h CAS. This determined a CD (drag coefficient) at alpha 90 to be 2.7 o.O!! Tests and experience show that the majority of planes share a CD at 90 AoA around 1.15, very similar to that of a flat plate in 3D flow, not 2D tests. So both the lift and the drag functions are exaggerated. Can someone explain this?

 

The Flanker's speed at a held alpha of 30 is 183km/h CAS with maneuvering flaps down. While weighing 22300kgf (49100lbf), it's lift coefficient proved to be 2.2. So the Mirage is almost as good as the Flanker in terms of aerodynamic lift? Keep in mind that the Mirage doesn't benefit from high energy vortexes generated by the LERX which usually increase the maximum lift coefficient by at least 50%. The Mirage doesn't have flaps (can't have) which also increase the maximum lift coefficient (even though the critical alpha becomes lower, the CL with flaps is higher). This should be a good enough proof that the DCS Mirage has too high lifting performances as compared to the real plane.

 

 

Kind regards!

- The filed you opened is for AI planes. Player's FM is different and located in mod folder and you probably can't open it.

- Mirage has strakes on air intake sides generating vortexes over the wing roots, and the sealed SLATs are generating vortexes for the rest of the wing.

thanks for your contribution to the FM Maverick Su-35S!!, interesting stuff you posted, really appreciated. Hope it can make the FM better.:thumbup:

Let's hope, but now because a lot of fanboys in DCS have found a pot of gold in how much thrust to weight ratio this plane develops in contrast to the DCS F-15 by beating the F-15 in vertical climb (climbs higher than the F-15C in vertical climb between 500 to 3000m, depending on climbout conditions). Right now the Mirage 2000C beats every aircraft in DCS in vertical climb and horizontal acceleration. Test please! Although this might happen (from what I concluded through calculations) because ED Flaming Cliffs 3's fighter planes engines (of F-15, Su-27/33 and MIG-29) develop lower maximum thrust than the real engines do or the Mirage 2000C's engine develops higher maximum engine thrust than the real engine can do, so it's a bit blurry to say the Mirage has too much output thrust or the Eagle has too low as compared to reality. I believe that a thorough investigation must be done by both producers to find the cause for this simulated aberration.

The Mirage also turns better than the F-15C. Yes, it proves relatively the same ITR (instantaneous) and STR (sustained turn rate) performance as the Su-27, which is utterly absurd and it doesn't seem to be due to an apparent higher engine thrust, but due to incredibly higher simulated lift for any given AoA and airspeed.

Here are some tests I did right now. The 3 aircraft are having 50% fuel, turning at around 1000m MSL while holding around 30 AoA (Mirage has around 27-28). The results speak for themselves and everyone can test them.

Regards!

- The filed you opened is for AI planes. Player's FM is different and located in mod folder and you probably can't open it.

- Mirage has strakes on air intake sides generating vortexes over the wing roots, and the sealed SLATs are generating vortexes for the rest of the wing.

OK. Thank you! So that .lua file is for the AI and not for a player flown aircraft. Then if that lift coefficient of 1.28 (which seems to be maximum in the tables) is for AI, I wonder what's the value simulated for the player's aircraft and why would it differ if this is a realistic simulator O.o?

Regards!

OK. Thank you! So that .lua file is for the AI and not for a player flown aircraft. Then if that lift coefficient of 1.28 (which seems to be maximum in the tables) is for AI, I wonder what's the value simulated for the player's aircraft and why would it differ if this is a realistic simulator O.o?

 

Regards!

Probably because:

- the player's FM is still under work and has the priority.

- the coding of player's FM, is different from AI (SFM), you can't copy/ paste.

My guess is that AI will be adjusted when player's FM is assessed as final.

Right now the Mirage 2000C beats every aircraft in DCS in vertical climb and horizontal acceleration. Test please!

I did. At least for level acceleration from M 0.9 at FL350 MAX AB.

Before december FM update: the 2000 beats the F-15. Wrong.

After december FM update: the other way around. Good.

Also I heard from several former and actual pilots on the aircraft saying "good job".

So they may be not test pilots nor have they perhaps tested the whole flight enveloppe, but still... ;)

Also I didn't test AI aircraft, only player-controlled aircraft (those have the AFM/PFM).

Finally, I'm not (at all) an aerodynamicist so can't comment on some values you quoted; lacking the proper background unfortunately.

Right now the Mirage 2000C beats every aircraft in DCS in vertical climb and horizontal acceleration.

Regards!

I really appreciate the work you are doing as a former aerodynamic expert but I can`t help to get rid of some comments:

1. When hoping to be taken serious by the developers I would renounce from calling people fanboys who are happy about the latest FM changes

2. Stop from repeating false allegations (see your quote above). This was already proved false after your post in the F-15 forum. If you say you tested it, be sure you use the latest patch/FM iteration. The current M2000 does not out accelerate the F-15. Tested by at least ten people.

3. Be sure you know how the FM structure of AI/Player aircraft works and why before comparing those (not everybody has to know but people who critizize the FM should imo)

If you take at least one of this points into account I'm sure we will see a lot of constructive critics from you. Would be a pity if your knowledge is ignored because you used the wrong words or forgot to patch....

Regards

I did. At least for level acceleration from M 0.9 at FL350 MAX AB.

...

You did from FL350 and Mach 0.9 only? Why didn't you try testing from higher altitudes and different speeds also? Try testing from above 11000m or FL360+ and see what happens. Try accelerating in level flight from 400km/h IAS to 700km/h IAS or from any lower Mach ranges and see the difference! Also, test both planes (Mirage and F-15) starting from whatever fuel quantity (but both the same percentage) in vertical climb at FULL AB and from various altitudes and climbout G-loads (between 2 and 4). You will surprisingly find out that the Mirage climbs higher than the F-15 every time. Depending on initial conditions (initial altitude and TAS) the maximum difference of altitudes reached may range between 0.5km to 3km. From what can be rapidly concluded is that the Mirage's engine doesn't loose maximum thrust with increasing altitude as any other fighter in the sim does, even the MIG-29 which has pure turbojets (very low bypass engines) loses thrust with altitude faster. Is it really that way? Again, it's either the Mirage's engine modeled better than anything else in DCS and ED has a problem on all their FC3 planes (not impossible), we shall investigate that if what we are doing here is in the name of realism or the Mirage's engine thrust tables need a revision if the engine's thrust doesn't decay with altitude as fast as in reality.

This is what results as X axis acceleration at any altitude higher than FL360:

Some people's attitude looks just like a "sneak around the corners" quite often, not to accept what's true the first time!

If the overall thrust (neglecting those abrupt variations with speed and altitude) might rather be a problem with ED's FC3 fighter planes engine performance instead of Mirage's engine performance which seems correct in terms of T/W ratio compared the the real Mirage, the aerodynamic polars (lift vs drag vs AoA) really are a problem. The drag to AoA's function exponent is too high, reason why the plane falls on it's belly (90 AoA) at around 125km/h IAS (this tells the drag coefficient is of a wooping 2.7 at alpha 90) instead of at least 190km/h (which corresponds to a realistic CD of 1.15 at 90 AoA) as EVERY plane does in reality.

The resultant lift force is also exaggerated! No Mirage 2000C flies at critical AoA (30) at just 94KIAS while weighing 9300kgf (20500lbf) as it happens in the sim right now. I have read somewhere about a demo flight in 1978 where a Mirage 2000C prototype (much lighter than today's combat fighter) managed to fly at a minimum speed of 100KIAS and lower weight than 20500lbf (perhaps maximum 19000lbf), thus the CL was way lower than it is in DCS. I'm not trying to make things up, it wouldn't be my in advantage after all, but I just want to look at things from a more realistic point of view. About those strakes near the inlets which people think can produce tremendous lift, from my experience as an aerodynamicist they can't produce more than a couple of percents more lift or higher critical AoA (5%, mostly 10% more, idk), firstly due to their position relative to the wing, secondly due to their size. I don't know their exact reason for implementation yet, but it might be to enhance the directional stability (better/cleaner airflow wash on the vertical stab) as I know other planes have needed before, but I strongly disagree that above the crude wing's maximum lift capability with slats out, these elements would increase the lift as dramatic as I see it in the sim.

Regards!

1. When hoping to be taken serious by the developers I would renounce from calling people fanboys who are happy about the latest FM changes

I might apologize for using that word, but it's the way these "people" comment about the Mirage (as it's some sort of a miracle for them) during play for how quickly the plane turns and climbs and does everything else against all other fighters in DCS and don't understand that this aircraft's simulated performances need more attention. They like it a lot, of course they do, cause they beat both the Flanker and F-15 in turns and climbs, so, how could they be "disappointed" of the new product? Some of them have spotted that something's odd about how this plane performs against all other, some don't want to pay much attention which degrades the ability to spot the things that don't add up.

2. Stop from repeating false allegations (see your quote above). This was already proved false after your post in the F-15 forum. If you say you tested it, be sure you use the latest patch/FM iteration. The current M2000 does not out accelerate the F-15. Tested by at least ten people.

They must test a little bit more, cause I have done just a few tests (not to comprehensive) and it seems that my allegations are facts. Check the latest message.

3. Be sure you know how the FM structure of AI/Player aircraft works and why before comparing those (not everybody has to know but people who critizize the FM should imo)

I agree, I didn't even believe there would be a difference in aerodynamic and engine thrust values if both the AI and player aircraft would fly with the same level of realism. This is strange and I hope this won't remain the same forever, but of course, this is ED's problem not a third party's problem.

If you take at least one of this points into account I'm sure we will see a lot of constructive critics from you. Would be a pity if your knowledge is ignored because you used the wrong words or forgot to patch....

 

Regards

I totally agree, so let's be constructive and make more tests and gather more real data!;)

Cheers and kind regards!

You did from FL350 and Mach 0.9 only? Why didn't you try testing from higher altitudes and different speeds also?

Lack of time + other reasons :)

About those strakes near the inlets which people think can produce tremendous lift, from my experience as an aerodynamicist they can't produce more than a couple of percents more lift or higher critical AoA (5%, mostly 10% more, idk)

I do happen to disagree, with solid grounds, on that particular point. I'm sorrry I can't elaborate why; and I certainly can't enter an aerodynamicists debate (I'm simply not qualified for this).

For the rest, it will probably take me several days to process all this (or more likely, it will require better suited people than me), in particular because no DCS for me right now.

I certainly find your comparizon vs the Eagle interesting, and will try to replicate.

Regards,

Az'

This discussion has gone quite off-topic (although I do appreciate any effort to test and improve the FM). I doubt that the weird peak that I highlighted in the OP is related to other problems (real or supposed) in the FM. I think this spin-off discussion would be better suited in a separate thread, so as to keep from muddling the issues.