Serious FDE bug in FS2004?

[Guest Ron Freimuth]

>Curious George, you said,>>"The aircraft works as a collection of interacting systems>and it is the behavior of the entire aircraft as a unit that>we are concerned with here. It's a flight simulator, meant to>simulate flight in aircraft already designed and developed,>rather than a wing design simulator that considers only the>wing as a single component. To be even more brief, the flight>simulator is intended to simulate the experience of flying and>airplane, not designing one from the ground up">>I'd like to tackle you on this and I ask others to also advise>because I don't know if the 2002 flight model covers what I'm>asking.>>If I can, by example:>An aircraft is turning at a particular rate of turn in a heavy>wind. The lift from the wing on the inside of the turn is>reduced, whereas the lift on the wing on the outside of the>turn is increased. Add to this, the action of each aileron>changes the differential lift between the two wings. Add to>this, the moving direction of the wind, relative to each wing,>significantly alters the effect on each wing. Further add, the>angle of sweep (forgotten what it's called) on each wing,>relative to AoA. Given conditions, the wind could actually>produce a situation of more lift on the inside wing than the>outside wing. Can a "whole aircraft" approach calculate an>attitude for this?>Andrew Yes. -- Well, mostly. I'm not sure the standard elements of the FS flight model cover all important effects of 'uneven lift' Most flight simulation approaches use the same form MSFS does. There are standard "Stability and Control Derivatives" which describle how the six degrees of freedom of a moving body are coupled in order to describe Vx, Vy, Vz; Roll, Pitch, and Yaw. Also their rates (derivatives). Actually, I think Roll, Pitch, Yaw and their Rates are the six basic 'degrees of freedom'. Velocites, which integrate into distances come in at a different level. They were named 'Stability Derivatives' when AC analysis was firt done to figure out how stable an AC would be; only small changes in the effects the parameters set were considered; one can assume they are linear for small changes. "Dihedral Effect" relates to Roll Moment due to Sideslip. That effect is more complicated than often considered, the fuselage itself has a significant effect. Regardless, one can add up all the omponents to get one number for 'Dihederal Effect'. Then, adjust the value to make the flight model display appropirate Dihedral Effect. A Roll Moment twists the AC (wing) CW or CCW. So, it amounts to an additional up force on one wing and a down force on the other. Ailerons are the main way to generate roll. However, you are thinking about a tight circular path, where the outer wing has more lift than the inner wing. That results in a 'Yaw Rate'. The last of the Roll Moment parameters in REC 1101 is "Roll Moment - Yaw Rate". "Yaw Rate" is high in a tight turn, so this effect twists the wing. Causing a Roll. It does not change net Lift, but does have the effect of reducing lift on one wing and increasing it on the other. Further, the 'Stability Deriviatives' are usually in a normalized form. One has to multiply Roll Moments by q*S*b. q is dynamic pressure, S is wing area, and b is wing span. So wing span is seen to be a factor in Roll Moments! In fact, that is the only way Wing Span has an effect, one could use span/2 and double the roll coefficients and get the same end result. Tip to tip span is easy to measure. Whether the wing is tapered or not, that effect is in the Stability Derivatives that generate the twisting and turning. ---------------------- An engineer might want to calculate if an AC is stable in roll. Do the wings tend to level by themselves? Or, does a roll develop into a spin unless a pilot over-rides the natural tendency? "Control Derivatives" apply to things such as "Elevator Moment due to Elevator Deflection". Larger deflections (~20 deg) don't produce as large a moment as expencted, the effect is non-linear. But, in normal flight, the effect of moving a control surface is proportional to how far it is moved. Most of the standard Stability Derivatives are in REC 1101 of the AIR file. Others are implicit in TBL 404 CL vs AoA (dCL/dAoA) is the slope, and that is one component of "static pitch stability", Another component is dCm/dAoA, which is the slope in TBL 478 - "Pitching Moment vs AoA" Things get more complicated as speed increases. At high enough speeds an aileron many twist and not generate the roll moment expected. Further, the pilot has limited strenght, and can't move the control stick when it takes too much force. The above effects are modeled by tables which reduce elevator, aileron, and rudder control as dynamic pressure, q, increases. That reduction at higher speeds also makes an FS AC easier to fly, since control movments become more critical. FS/CFS also have tables that let one make the elevator stall at some AoA. Reduce Aileron Control, change pitch damping. Then, there are tables that modify the Dihedral Effect, Roll Damping, Yaw Damping, etc. vs AoA. Most all the data I see on the WEB, and in engineering books involve the same parameters and variations one can set in the AIR file. However, information is lacking on more extreme attitudes, such as in stalls. I have to visualze air flow over the tail to estimate how to adjust some tables. Then, see if my changes have a reasonable effect on the AC. Fortunatly, we can edit the AIR file, and reload it in flight to immediatley see the effect. FSEdit attempted to calculate these 'stability derivatives' from the new lines in aircraft.cfg. However, it messed many things up. At best, only an approximate set of flight model parameters could be calculated from the data in the air file. More sophisticated programs require more detail of the airframe. And, still are only good to come up with an initial approximation of the paramters involved. It would take a supercomputer hours to calculate the transonic effects for the wing; further the exact form of the wing would need to be entered to get correct results. Spins are very difficult, even FM sims tend not to model them. Sure, one can make an FS AC spin, but that doesn't mean it spins as the real AC would. -RAF

[Guest]

Thanks Ron,Very helpful and well explained. :)Andrew

[Guest Ron Freimuth]

>Thanks Ron,>Very helpful and well explained. :)>Andrew Actually, I don't understand all these things fully. They only slowly sink in. Further, things like the "vertical wing distance' (relative to the vertical CG have an effect on 'wing leveling'. I don't know if published values for 'Dihederal Effect' include that or not. Anyway, it gives me an excuse for setting 'Dihederal Effect' as required in the AIR file to get nominal wing leveling. I now suspect the 'dihederal_angle' in aircraft.cfg moves the vertical offset of the wing to its 'average'. So, I've been setting it to zero, and using AircraftContainerManager to estimate the height of a line that amounts to the wing's 'waterline'. FSEdit would also probably use 'dihederal_angle' (and other lines that otherwise make no difference' to set up new 'stabilility derivatives' in the AIR file. Unfortunately, it messed up too many things. ;( Ron

[Bob Scott 7,721]

So with all this having been discovered and said, does anyone besides me find it utterly dismaying that both PSS and PMDG have released their latest pricey payware offerings with a one-size-(doesn't)fits-all FDE for both FS2002 and FS2004? The PSS A340 is abysmal in FS2004...they've really lost the bubble on flight dynamics this time.CheersBob ScottATP IMEL Gulfstream II-III-IV-V L-300

[Guest Ron Freimuth]

>So with all this having been discovered and said, does anyone>besides me find it utterly dismaying that both PSS and PMDG>have released their latest pricey payware offerings with a>one-size-(doesn't)fits-all FDE for both FS2002 and FS2004? >Bob Scott When I modify my FD files for FS9, I can copy them back to FS2K2 and they fly virtually the same. In fact, they will fly the same in FS2K, but for the fuel consumption. There are variations in how the autopilot works, it's by far the best in FS2 (with properly adjusted REC 1199). Ron

[Bob Scott 7,721]

>When I modify my FD files for FS9, I can copy >them back to FS2K2 and they fly virtually the >same. In fact, they will fly the same in >FS2K, but for the fuel consumption. If you've made the table 404 corrections, then zeroed-out the wing incidence and twist, and restored any modified drag scalars back to unity, I tend to agree that an FS9 FD config would exhibit approximate backward-equivalence.But in the PSS and PMDG examples, we have the same airfile and config file, which have non-zero settings in Table 1101 x50h, wing incidence and twist, (and in PSS A330/340 some rather odd drag scalar settings)...such a pair of airfiles cannot possibly act equivalently across the two versions. Not that I want to see the PSS A330 doing approach stalls at charted Vapp in *BOTH* versions or anything!I'm just dismayed at the flight from quality exhibited by the latest supposedly "high-end" prayware offerings. The last thing I want to do when loading up a sim for which I chunked out $50-75 is re-write the FD. Feels like it's becoming the norm lately.CheersBob ScottATP IMEL Gulfstream II-III-IV-V L-300

[Guest Ron Freimuth]

"If you've made the table 404 corrections, then zeroed-out the wing incidence and twist, and restored any modified drag scalars back to unity, I tend to agree that an FS9 FD config would exhibit approximate backward-equivalence. .." If incidence + twist/2 = 0 they cancel one another. Many of my past FD files were already set that way and didn't require a change for FS9. I'd think the drag scalars would work the same. However, in FS2K2, changing 'induced_drag_scalar' or 'zero_aoa_lift' will reqire a change for the REC 1101 'AoA for min Induced Drag at CL=0' to keep induced drag the same after the change. I don't think that is a problem in FS9 since that AoA must be automatically set in it. So, an incorrect setup for FS2K2 won't come out the same in FS9 even with TBL 404 shifted. Ron

[Bob Scott 7,721]

Sorry...I think we're talking past each other here.My methodology for conversion of an FD set from FS2K2 -> FS2K4 is to shift table 404 to build in the incidence + twist/2 (not necessary, as you observe, if net is zero), then modify induced drag scalar to correct for FS9 not taking table 1101 rec x50h into account, per the procedure described eariler in this thread by Tom Goodrick. Even with zero net delta resulting from incidence and twist, the post-conversion FD set will not generally be backwards-equivalent if used in FS2002 because of the drag scalars present in the converted file set.The PSS A340 FD, for example, has wing incidence and twist values that do not cancel out. If one assumes that the FD have been validated in FS9, then using same FD set in FS2002 will yield considerably different performance due to presence of wing incidence and twist data in conjunction with an already-modified table 404, plus the effects of the drag scalars that were presumably used to better fit the FD to FS2004.CheersBob ScottATP IMEL Gulfstream II-III-IV V L-300

[Guest Ron Freimuth]

>Sorry...I think we're talking past each other here.>>My methodology for conversion of an FD set from FS2K2 -> FS2K4>is to shift table 404 to build in the incidence + twist/2 (not>necessary, as you observe, if net is zero), then modify>induced drag scalar to correct for FS9 not taking table 1101>rec x50h into account, per the procedure described eariler in>this thread by Tom Goodrick. I didn't see that 'procedure', but FS9 must set an internal value equivalent to 1101:050h. I've moved lots of AC into FS9. Making "incidence -twist/2 = 0.0" and correcting TBL 404 if necessary - then changing 1101:50h so it's still correct for FS2K2. I have NO REASON to think induced drag is any different if that is done. In fact, I may copy the two FD files from FS9 BACK to FS2K2 so I have the identical files in each FS version. FS2K2 ignores the new lines. > Even with zero net delta>resulting from incidence and twist, the post-conversion FD set>will not generally be backwards-equivalent if used in FS2002>because of the drag scalars present in the converted file>set. I always set 'drag scalars' to 1.0. I don't know just what components of 'zero lift' and 'lift dependent' drags they affect. I set only simple Induced Drag effect with the Oswald Efficiency'. Which is equivalent to including it the Induced Drag Factor in REC 1204 for FS2K. >The PSS A340 FD, for example, has wing incidence and twist>values that do not cancel out. If one assumes that the FD>have been validated in FS9, then using same FD set in FS2002>will yield considerably different performance due to presence>of wing incidence and twist data in conjunction with an>already-modified table 404, plus the effects of the drag>scalars that were presumably used to better fit the FD to>FS2004.>Bob Scott Guess they didn't read the messages in this thread. ;) BTW, I just found/verified 'wing_root_chord' has NO EFFECT (unless FSEDit is used). This was mentioned in the CFS3 SDK and just verified in FS2K2. FS/CFS must calculate MAC from wing_area and span. Regardless, the CoL is 25% aft of the calculated MAC (I assume FS calculates it the same way I do) of where one sets the wing's LE. Unless set differently in the AIR file record that lets one move the CoL in three directions. The Concorde.AIR file is an example. 'Twist' may still be used by any new FSEdit for FS9. However, probably not in an appropriate way. ;) Ron

[Guest ChanRak]

>I had some time this morning to investigate some of FS>Aviator's findings concerning the wing AOI and twist, and>Fuselage AOA at Min Induced Drag settings being inop in FS9,>and after some testing I produced nearly the same results as>he did - until I restarted the flight sim per Ron F's>observation.>I'm sorry to say that the "reload user aircraft" function>doesn't work for me as far as changes to wing AOI and twist,>or Fuselage AOA at Min Induced Drag are concerned, I can only>see the results of any changes I make after shutting FS9 down>and starting it again. >I'm using Windows ME which has a really interesting way of>managing RAM, and suffice to say that after two of these>shutdown/startup sequences following .air or .cfg file changes>FS9 runs only a little faster than grass grows!>I must then use FreeMem to allocate and free up some RAM,>adding an additional step to the long sequence of this>festival of good fun. >I'm not sure if FS Aviator managed to choose the only three>flight dynamics parameters to which this applies, and as>obligations to the business of survival prevent me from>pursuing this investigation full time, I'm unable to offer an>in-depth analysis just now.>I can say that as a quick and simple check I made a change to>CM_de Pitch Moment-Elevator in the airfile and this change was>recognized immediately after reloading the aircraft by both>the keyboard "reload user aircraft" event and by re-selecting>it from the menu so the mystery deepens.>>As far as this is concerned:>>>2) Current air file editors are not loading or displaying new>unknown sections of the FS2004 air file to which some>aerodynamic variables have migrated, or<>>This seems unlikely, the FS9 airfiles have smaller file>sizes.>>For questions 1 & 3 it would appear that the changes are>recognized after the flight sim is restarted, so the bug isn't>show stopping just really, really annoying, and as for the>notion that "MS have now split the FDE into three files one of>which is now secret and hidden" at this point that also seems>unlikely, although by now nothing about this latest version of>FS would surprise me.>>>There is potential for some interaction in FS2004 with the>new payload altering module. Note however that the new v1.5>RCS B-25J is very nose heavy at default load (due to another>'problem' in FS2004) and that in FS2004 all my FDE have CoG>well forward at exactly zero wing chord (I know why, but that>is a separate FS2004 issue/bug). For the moment note that>these false NOSE HEAVY default CoG conditions in FS2004>(compared to correctly processed default CoG in FS2002) cannot>be causing the observed NOSE UP condition in FS2004.<>>I've noticed this also, can you elaborate on the cause of this>forward C.G. phenomenon, and what -if anything- can be done>about it? >>>>Has anyone even bothered to check out the new aircraft.cfg>entries?>>Yes Bill, I've looked into them, and besides the entries>already mentioned regarding the autopilot, we now have these:>>>New entry added to the GeneralEngineData section>>min_throttle_limit=0.0 //Minimum>percent throttle. Generally negative for turbine reverser>>>And these new sections>>[brakes]>parking_brake=1 //Parking brake available>toe_brakes_scale=0.68 //Brake scalar>>[hydraulic_system]>normal_pressure=0.0 //PSI>>[turn_indicators]>//Type: 0=None, 1=Electric Gyro, 2= Vacuum Gyro>turn_indicator.0=1,1 //Turn AND Bank>>[stall_warning]>type=1>>>Also, the .mdl files have some new content, possibly related>to gauges or the "clickable" VC's. The vintage aircraft .mdl>file sizes are huge, the Jenny's .mdl file is 2.55 MB!>>>Edited to format message in plain textWas in contact with Microsoft about this problem. I had a suspicion of out of a hunch, (Microsoft is well known to change the names around in their files) and what was going on in 2002, now is not in 2004 because they have another name to identify it. Strange that they would do this, but I do suspect this is the case.Microsoft will not tell me what name it is under now.So I guess we got to find the link that affects the mention as above. The game looks as I was told at three files at same time. It must correlate.I am not the expert but just passing on some information to you.Good luckDaveYo :-wave:-wave