Serious FDE bug in FS2004?

[Guest Tom Goodrick]

I have a simple solution to this crisis that works to my satisfaction. Assume you have a good model in FS02 that meets a cruise spec. 1. Fly it in FS02 and double check the exact cruise speed (use standard conditions, use the autopilot and a normal cruise power setting and altitude.) Call that true airspeed V1.2. Import the model into FS9 and make sure the weight is the same. Fly it FS9 in exactly the same conditions and note the cruise speed. Call it V2. [in both cases allow plenty of time for the cruise speed to become steady.]3. V2 will generally be less than V1. Divide V2 by V1 and then square the result. Put the result in for the parasite drag scalar in the FS9 aircraft.cfg file. When you fly the aircraft again in FS9, you will see close to the same speed as you saw in FS02. The aircraft will climb, cruise and glide the same way as it did in FS02.As an example, I fixed my Bonaza A36 using the relation (163/173)^2 = 0.888. I put 0.888 in for the parasite drag scalar in the FS9 aircraft.cfg file. It flew at 172 KTAS and was increasing slowly when I shut it down to write this.There are two things to note. After editing the aircraft.cfg file, you must select and load some other aircraft and then select the original aircraft and load it again. That reload key does nothing, at least in my case. This method works all the time for aircraft and panel .cfg edits as well as for .air file edits.You may ask "If the problem is improper induced drag, why does it help to decrease the parasite drag?" My answer is that, when you are cruising at the same speed under the same conditions in the same aircraft, lift matches weight and the lift coefficents must be identical. (If not, then somebody is violating the heck out of the laws of physics and aerodynamics.) If the lift coefficients are identical and the aircraft geometry is the same, then the induced drag must be the same so we "fixed" the problem.It is indeed comforting to see that it works.

[Guest]

>I have a simple solution to this crisis that works to my>satisfaction. Assume you have a good model in FS02 that meets>a cruise spec. > >As an example, I fixed my Bonaza A36 using the relation> (163/173)^2 = 0.888. I put 0.888 in for the parasite drag>scalar in the FS9 aircraft.cfg file. It flew at 172 KTAS and>was increasing slowly when I shut it down to write this.> >It is indeed comforting to see that it works.Well hello, Tom!Well how about that... I was busy tweaking the TB20GT to work in FS9 and noted the huge difference in cruise speed at 5,000' ASL (I use the ocean just west of KSEA for a test ground). In FS2k2 normal cruise is a nominal 163kts with 0/0 wind @ 29.92 (again, my standard test environment). In FS9 under the exact same conditions, cruise was 145 kts.Using my patented "hit-and-miss approach," I determined that reducing parasitic drag scalar to .80 yielded the expected cruise IAS. Using your furmula, that calculates to 0.7913356. My results would lend credibility to this process. Thank you for posting the aerodynamic reasoning to explain the empirical results I obtained, albeit very "unscientifically..." :)Now I just need to correct the +2

[Guest Ron Freimuth]

>You touch on a lot of areas here. I'll keep my reply brief.>In my work where I did flight simulations for several>vehicals, I came to distrust the simplified linearizations>that are implied by the use of stability derivatives. in many>cases - peculiar craft and special flight problems - they are>insufficient. I think they were developed mainly to make>cohesive presentations in classrooms. These discussions are getting rather technical. Over the head of many. However, such threads also tend to draw in those who appreciate them. Normally I BS about this stuff at avhistory, however I've been writing most of my comments on the new FS9 deviations right here. I don't trust any formulas presented in engineering books until I have experience with them and know they work. I see more and more how theoretical assumptions are made and often nothing said about how real AC differ.--------------------------------- When I deveoloed a 6DOF>sim of gliding parachutes for the Army in the 1970's, I had to>compute the local airspeed vector and angle of attack at 8>points on the quarter chord line of a curved wing. From that I>computed three body components of force at those points. This>process enabled me to model not only the longitudinal flight>motions but steep turn effects and even spins where the>gliding parachute really gets wrapped around an oscillating>axis between the jumper and the canopy centers of mass. That>sim gave results that were verified by flight tests with>onboard data systems in highly dynamic conditions.>>I am simply suggesting that FS would do better if they>computed velocity and angle of attack at the median points at>the quarter chord on each half of the wing. Otherwise they>must play numerical games with linearized approximations of>highly non-linear phenomena. The same games most flight simulator alogorithm's play. True, a variation on 'standard SD simulations' might give improved results. But at the expense of a great deal of research and development. >You speak of incidence and twist as "cancelling". The Bonanza>36 has a root chord incidence of 4 degrees and a tip chord>incidence of 1 degree according to Jane's. I suppose that is a>twist of -3 degrees but what do you call the "incidence"?>Would it be 4? I'd say Incidence = 4.0 and twist = 1.0 - 4.0 = -3.0 degrees. Same as you. Note Wing Incidence can be calculated and set in a real AC. One takes the 'chord line' of the profile as zero incidence. The "3D" lift polar is what one would start with. One can set the 3D lift polar for TBL 404 if he knows the airfoil of the AC. I've often looked that up on a WEB site. The C172 uses NACA 2412, while the 23012 series is common for twins. The AC520, Baron 58 and C310 use that series; the last two numbers (t/c) vary from '12' to '16'. I just clipped this out of the header of my AC520 AIR file: "NACA 23012, 4 deg twist". From my C310 AIR file: "NACA 23018 root, 23015 tip" I massaged TBL 404 to approximate that combination. I also managed to get this data on the C310 published in one of Roskam's books:"CLo=0.288 (some in Wing Incid) 0.2858 setCL_alpha = 4.58? 4.817 set" Now that should apply to the actual C310, not just the wing. CL at zero Body AoA is given as 0.288. Note pitch is equal to Body AoA in level flight. And, the 'incidence_alpha' variable/parameter FS provides is Body AoA, not the wing AoA in TBL 404. By finding that value for CLo, and making incidence and twist consistent with my TBL 404 (actually, they canceled) I didn't have to adjust anything to get correct flight pitch. Since I based lift and lift slope on real C310 data. Incidently, the main variation in that NACA series is how hard they stall. However, the NACA polars are not for twisted airfoils. Thus, the place to incorporate the effect of twist is in the regions past CLmax in TBL 404. More twist softens the peak of CL. I tend to estimate that region and sometimes mess with it if stall dyanmics need improvement. Howver, TBL 473, Pitching Moment vs Body AoA is also important. It should force the nose down appropriately at stall. Cut and try is necessary since real data is almost non-existant. Again, in FS wing incidence should be set lower than what it is in the real AC. The real AC wing sees downwash (ahead of and along the wing) there is no reason to believe the FS model sees this effect. And, in FS9 we are stuck moving at least the center region of TBL 404 left or right to set effective wing incidence, in combination with the slope of CL vs AoA, that determines Lift at cruise, thus pitch. > I would use the median incidence on each wing Reasonable if the wing is not tapered. A tapered wing changes the effect of twist, but FS simply made twist have 0.50 the effect of incidence. Before FS9. ;) >as 2.5 degrees and compute angle of attack without considering>twist. The effect of twist is mainly in the lateral stall>progression. This can play a part in the stability derivatives>at high angles of attack. The problem identified in this>thread is very basic. It is how much lift should be generated>at a given angle of attack in the linear range of angle of>attack for a given aircraft attitude and central velocity>vector. Let us continue to address this central issue. One can calculate CLo if you know AC pitch at a given weight and q. q can be accurately calculated from 'TAS' and 'Atmospheric Density'. In fact, AFSD, by Herve' Sors gets the internal FS q through FSUIPC. AFSD calculates various coefficients by knowing S, W, and many other things not available in the regular FS variables. Herve' is working on a new AFSD for FS9. Many people have used it for FS2K2, though I bet many of the parameters displayed confuse the Hell out of them. ;) Note AFSD displays H Stab Lift and Elevator Lift. Further, they change with Mach number.>What is the simplest solution to make FS9 aircraft fly like>the corrected counterparts in FS2002? Calculate Incidence + Twist/2. If it is not zero TBL 404 will have to be slide sidways by the result of that sum (note twist is normally negative, so it subtracts). I HOPE the adjustment to TBL 404 works out exactly. Something I have yet to verify. I see the new TBL 404 in the C208 AIR files in FS9 have a higher slope (~7.5) between CL=0 to some small value than the slope in the main lift region of that table. That lift slope of 7.5 is impossible, I wonder if it came about due to more changes in how the Body AoA for Mininum Induced Drag" is set. We know that REC 1101 parameter is also dead in FS9. Actually, it isn't needed when twist and incidence aren't used. But, I wonder if it is set internally to what a correct setting for FS8 would be.RonPS: Don't be surprised if FS9 default AC have different top speeds than the same FS8 AC have. The two C208's had identical drag and speeds. Not at all correct, the float version has a lot more drag. This was changed in the FS9 air files, though I'm not sure how well. Further, the FS8 C172SP was slow by 10 kts or so. The engine couldn't hit red line, the prop was wrong. That has been improved in the FS9 172. I'd fixed the FS8 172 years ago and many people wanted it.

[Guest Tom Goodrick]

Yes, I got that plaque. It is a real nice one. Thanks, Bill.This morning I was doing slowfing around the airport in the Learjet 45. At 140 KIAS it has such a very high pitch angle it is very awkward. I do wish we could simply adjust the incidence a degree or 2. I used to have some contact with a guy who flies Learjet 45's for a living. But I have lost contact with him. I'd pose the question to him of how real or unreal the attitude is. He first contacted me when he could not fly the default FS2000 Lear. I fixed it up a bit and we both enjoyed it. He gave me a copy of the weight and balance sheet he uses daily on his notebook computer in the plane. I still refer to it.There is one thing about being a pro jet pilot. You are seldom home as you "live with the aircraft."

[Guest Ron Freimuth]

I never set the lift and drag scalars in aircraft.cfg. I set REAL Cdo, Induced Drag, etc. in the AIR file (as far as possible). I also set realistic power, thrust, prop efficiency, etc. When real engineering values are used, realistic performance will be obtained as long as the FS 'flight model code' is correct. I can calculate the effect of changes if I feel like it. One can always adjust one drag compenent to get 'real' performance at one specific AC state. However, don't expect the AC to perform correctly at any other condition! Assuming FS9 sets 'Body AoA for minimum Induced Drag' to CL=0 in TBL 404, then then minimum drag AoA with TBL 404 unchanged will be in the wrong location. 1101:50 no longer sets this location. Note Cdi is NOT proportional to CL^2 in FS8 and earlier versions. Here is the equation I put in Aired.ini for 1101:50. An equation that took thousands of hours of flight testing and research by others to get to: "CD = Cdo + {(AoA +AoAmd)*(dCL/dAoA)}^2 * IDK" Where IDK = 1/Pi*AR*e and AoAmd was to be set corrected in 1101:50. Now dCl/dAoA is calculated some way by FS, it is approximately the lift slope at AoAwing = 0 radians. A typical value would be 5.5, the theoretical value for a 2D airfoil is 2Pi. However, dCl/dAoA also changes due to GE (TBL 400) and Mach Number (TBL 401). Those effects result in AoA decreasing for a constant CL. And, CL will be constant in non-acclerated flight at a constant weight. CL stays the same, AoA decreases. With the FS2K2 and previous flight model code this caused Cdi to decrease since AoA decreased. This models both reduced induced drag in GE and reduced drag due to increased lift slope as Mach increases (though, TBL 401 will eventully drop below 1.0). I imported my current Concorde FM into FS9 and it appeared to fly the same. Induced drag and Mach variations are very important in this AC, if drag is a bit low one can climb when he shouldn't be able to. As with many of my other AC, Incidence and Twist exactly canceled (Twist = -2*Incidence). Looks like I've been moving for some time to setting those two variables so AoA in TBL 404 is the same as Body AoA. This also makes setting the AoA related tables easier. I also imported a 727 into FS9. It will need a small adjustment in TBL 404. It also appeared reasonable. But, as I've said before, I have yet to verify that drags are exactly the same as in previous versions of FS when TBL 404 is correctly adjusted. It it should turn out FS9 calculates Cdi as K*CL^2 then many things will be damaged. For one thing, drag will decrease after stall, since CL drops in a stall. In real AC drag continues to increase post stall, even more than the AoA^2 variation FS2K2, etc. set it to. Ron

[Guest]

Seems like the initial, quick approach would be:1. Determine net incidence.2. Adjust table 404 as needed.3. Check cruise pitch and speed4. Adjust zero lift drag so that your FS9 cruise speed (or top speed or whatever benchmark you're using) is correct at optimum altitude.I agree that adjusting the scalars is a Band-Aid approach and should be avoided.This assumes everything else has already been done and these are the final adjustments!!!

[Guest]

>I agree that adjusting the scalars is a Band-Aid approach and>should be avoided.Were that always true, there'd be no point in even having them available at all, would there?It makes little difference in the long run whether you multiply by 0.8232345 in the .cfg or elsewhere, the end result is the same... :)Sometimes you only need a "bandaid" and not 4-O silk sutures and a body cast...BillAVSIM OmbudsmanFounder and Director,Creative Recycling of Aircraft Partshttp://catholic-hymns.com/frbill/FS2002/images/fartslogo.jpg

[Guest]

Attn: FS Developers.Regarding these discussions and FS2002/2004 "compatibility". What happens now? How can we ensure authentic compatibility?Many aircraft will simply be loaded across to the new sim without accounting for these changes. This begs the question, what does "FS2004 compatible" mean? When a developer says their add-on product suits both FS2002 and FS2004, does that mean they've updated the model to account for the changes?It would be very easy for any developer to ignore the changes, as long as their aircraft still flew in the new sim, and at the same time badging the product as FS2004 compatible. There are already a number of products out there that say, "for FS2002/FS2004".There needs to be a way that users can know that any add-on aircraft that we purchase will behave authentically under the new design rules. "FS2004 compatible" is just too ambiguous. There is one way this can be done. Regular industries have a way of dealing with this by citing a new rule and then stating that the new or updated product is "Rule 5988 compliant". Perhaps such a system should be used here.Andrew

[Guest]

>It makes little difference in the long run whether you>multiply by 0.8232345 in the .cfg or elsewhere, the end result>is the same... :)>>Sometimes you only need a "bandaid" and not 4-O silk sutures>and a body cast...>>Bill>AVSIM Ombudsman>Founder and Director,>Creative Recycling of Aircraft Parts>http://catholic-hymns.com/frbill/FS2002/images/fartslogo.jpgYes, it does make a difference.As Ron correctly pointed out, the most accurate and effective way to adjust flight dynamics parameters are within the .air file itself. The scalars are placed there as a way for those who are not familiar with the particulars of the .air file to make adjustments without having to know this information.Needing to use the scalar to adjust the performance is an indication that something is wrong somewhere else in the flight model and needs to be addressed.I realize that this may seem like picking nits, but if the goal is to create the most accurate flight model possible, the scalars should be ignored completely and the proper parameters noted and adjusted within the .air file. Every aspect of the aircraft's performance can be adjusted this way.

[Guest]

I have not come across a single instance where the same flight model worked equally well in both simulators. In fact, many of them are fine for the sim for which they were intended and bomb in the other sim.The difference may in some cases be small, but I would not at this time envision being able to have the same flight model for aircraft that are to be used in both simulators. Each sim needs its own flight model. Thus, if an add-on is offered for both sims, separate flight models will need to be developed for each.Clearly, if it concerns a payware product, the customer should contact the seller and confirm that separate flight models are included for both simulators. This issue is really no different than what we had for CFS2 and FS2002, although there is a much greater level of compatibility between those two sims.

[Guest]

Thanks for that info. Agreed, the purchaser should contact the vendor. But very often the purchaser only learns that there is an issue after the fact. (Not that I have an answer for that either.) I've seen several a/c that are offered as "FS2002/FS2004". Makes me wonder.

[Guest Tom Goodrick]

Ron, your statement:"Note Cdi is NOT proportional to CL^2 in FS8 and earlier versions. Here is the equation I put in Aired.ini for 1101:50. An equation that took thousands of hours of flight testing and research by others to get to:""CD = Cdo + {(AoA +AoAmd)*(dCL/dAoA)}^2 * IDK"is curious because that IS the proper equation showing CDi is equal to CL squared times IDK. There is no need to look CL up in a table when operating in the linear range as we are in all normal aircraft operations.(AoA +AoAmd)*(dCL/dAoA) = CL

[Guest Tom Goodrick]

I strongly disagree that there is something wrong with using the scalars to tune the performance of an aircraft and that one should only insert absolutely correct values of known coefficients. First such absolutely correct values of known coefficients are seldom available unless you have access to wind tunnel data on the aircraft. But in the final analysis, the only thing that truly counts for anything is the observed performance of the aircraft. In most cases the proper performance can only be observed after a long period of trial-and-error adjustment of parameters. Adjusting a scalar is a perfectly fine way of achieving this goal. In FS02, I have used the aircraft.cfg file whenever possible rather than using the .air file editors, all of which have some uncertainties. By that I mean no .air file editor is perfect. Many of the terms used are non-standard even in the aeronautical engineering profession. But by simply editing the text in the aircraft.cfg file, you can be sure of effecting the intended change. At least, you could be sure until MS did this change in FS9 which I consider a bad mistake.

[Guest]

>It would be very easy for any developer to ignore the changes,>as long as their aircraft still flew in the new sim, and at>the same time badging the product as FS2004 compatible. There>are already a number of products out there that say, "for>FS2002/FS2004".I should think that a statement "for FS9 only" would be a good indicator... :)Remember, there are a lot more differences in the package than just the flight dynamics! The new .mdl file allows for a large number of "animation tags" that simply don't exist in FS2k2 or earlier versions. Using an FS9 model in an earlier sim will have unpredictable results...More problematic are changes to the gauge coding that render some FS2k2 gauges unusable in FS9, and vice-versa. This will become even more of a problem after the new SDK is released and gauge programmers begin exploiting the new variables and control events. In many cases already noted, "tricks" that worked well in FS2k2 will simply cause FS9 to crash to the desktop without warning.EaglesoftDG has already made the decision to market new products with two clearly labeled versions: one FS9 only and the other FS2k2 only, at least for the next twelve months or so.BillAVSIM OmbudsmanFounder and Director,Creative Recycling of Aircraft Partshttp://catholic-hymns.com/frbill/FS2002/images/fartslogo.jpg

[Guest Ron Freimuth]

>Ron, your statement:>>"Note Cdi is NOT proportional to CL^2 in FS8 and earlier>versions. Here is the equation I put in Aired.ini for 1101:50.>An equation that took thousands of hours of flight testing and>research by others to get to:">>"CD = Cdo + {(AoA +AoAmd)*(dCL/dAoA)}^2 * IDK">>is curious because that IS the proper equation showing CDi is>equal to CL squared times IDK. There is no need to look CL up>in a table when operating in the linear range as we are in all>normal aircraft operations.>>(AoA +AoAmd)*(dCL/dAoA) = CL Yes, the 'FS way' to set Cdi is equivalent to "IDK*CL^2" as long as one is in the linear, low AoA range of the lift polar. Though, the REC 1101 'Offset' parameter had to be set correctly to put the parabola in the right place. At high wing AoA's, where the CL vs AoA curve starts sagging, the 'FS way' keeps increasing Cdi as AoA^2, not with CL^2. One advantage of this is drag continues to increase post stall. One could move the minimum of the drag parabola to the right, to try to emulate a 'laminar flow' drag bucket. However, I was never able to get climb and cruise performance to match PoH tables when I tried that. So, I always set 1101:50 so the drag parabola hit minimum where CL hit zero. This had to include the effects of wing Incidence and Twist! Though they tend to cancel. ------------------------------------- I made more flight tests to compare FS8 with FS9. First, I flew my C208 Amphibian in FS2K2 at 85 and 150 kts IAS. Noting TAS, Pitch, and PPH. Then, I flew the "modifed for FS9" AIR and Aircraft.cfg files in FS9 at the same conditons. I didn't have AFSD to show details in FS9, but could check that the flight weight was nearly the same by going to the FS9 Fuel/Payload menu. I could read PPH digitaly from the tooltip when I placed the mouse pointer over that panel gauge. I have SPD hold set in aircraft.cfg, to make it easier to hold IAS constant. I could see pitch with 1 deg resolution from the tool tip and TAS on the GPS. Prop RPM, Condition Lever, etc. where set the same in FS8 an FS9. Results:FSVer IAS PPH Pitch Weight9.0 85 132 8-9 deg 7060 lb " ~140 282 ~2 "8.0 85.0 133 8.0 deg 7140 lb " 140 282 1.5 7070 To some degree, I set PPH and let speed stabilize. Note PPH (a measure of drag) is *virtually the same in FS8 and FS9* at both low and high speeds. I couldn't read pitch as closely from the tooltip in FS9 as I could with AFSD in FS8, but they appeared to be within 0.3 degrees of each other. ------------------------------------- Now the FS8 AIR file used the original MS TBL 404 and incidence/twist settings. Howver, I replaced TBL 404 in the C208 AIR file I copied to FS9 with TBL 404 in the FS9 AIR file. It appears to have been appropriately changed to give appropriate results. I left the FS8 incidence, twist, and 1101:50 settings in aircraft.cfg and the AIR file. Knowing they have no effect in FS9. -------------------------------- To make one set of aircraft.cfg and AIR files work the same in both FS9 and FS8 I will either set incidence and twist to 0.0 in aircraft.cfg, or set Twist = -2.0*Incidence. I will also have to have the new TBL 404 in the FS8/FS9 AIR files and set 1101:50 for the appropriate value, since it does have an effect in FS8. The correct value will now be 1101:50 = - 2.00 deg. Since the new TBL 404 shows -0.0350 radians at CL=0. This is clipped right out of aired: "Fuselage AoA at Min Induced Drag=-2.813". I just changed that to "-2.00". No effect in FS9, but makes the AIR file compatable with FS8! I also changed REC 1204: Incidence = 0.852. I left "Twist=-1.692". However, aircraft.cfg is where this really counts. So, I'll change Twist in it to be the same as above. BTW, I think these two FD files will also work in FS2K. Where Incidence and Twist have to be in the AIR file. FS2K will ignore most of the aircraft.cfg lines, so the corresponding parameters in the AIR file have to be correct. While I think the C208 AIR file will give the correct FD perfomance in in FS2K, that sim doesn't support all the FS2K2 things, such as floats and 'water rudders'. Nor does it support the .mdl. ;) Note I changed Incidence so it is just -Twist/2. Thus the two will cancel in FS8. After all, TBL 404 is now shifted left, giving more lift for the same wing AoA. So, less root incidence is needed. ------------------------------ Aircraft.cfg and the AIR file can be compatable between FS2K2 and FS2K4. One just has to understand the details. ;) Incidently, the DF C310 was 'ready for FS2K4' when first released. It had been tested under a developer copy of FS2K4 (not by me) and click points on the panel, etc. were already added. However, the installation exe did install slightly different cfg files, depending on the FS version. -------------------------- I checked the files in FS9 AIRCRAFTtemplates. They appear to have been 'fixed' by MS to work in FS9. While incidence and twist are set in aircraft.cfg (they have to be something), the AIR files appear to have new TBL 404's (when appropriate). They are just the same old default AC, other than for this change. Ron