FS9 Flight Dynamics, a complicated issue..

[Guest]

Just a few words if I may :-). I think that over the years of the Microsoft flight simulator franchise, there has always been a demand from the community for Microsoft to keep upping the ante on the accuracy of the flight model physics engine. Although Microsoft has created a very powerful simulation, there is a reason that the flight models will never be perfect. First and foremost, is the simple fact that modeled physics is not reality. This seems silly at first, but in fact a computer works in discrete time slices. It is not continuous as is reality and life. We cannot exactly apply pure Newtonian physics to a discrete situation, because of this lack of continuity. Now because we have to discretize the situation, error is introduced into the equation that will cause slight variations in calculations. While this is at the fundamental level, there is more reasoning at a much higher level. Large FAA certified simulations are modeled similarly to any flight simulation. The aircraft is broken down into its components, or discretized, where the moments of inertia, weights, fuel, instruments, luggage, seats ac units, etc. etc. are taken into account. These large aircraft simulation manufacturers have an advantage that Microsoft probably does not. They have all of the non-linear coefficients for each airfoil on the aircraft, as well as all of the engineering data needed. Microsoft most likely has to mimic this data, because usually aircraft manufacturers do not release the information for obvious economical reasons. Microsoft has to create these flight models by modifying known airfoil coefficients, or by modifying known linear coefficient data to match each aircraft that they offer. What this boils down to is a "reasonable approximation" to the flight characteristics of an aircraft. Secondly, the flight models in the Microsoft version are probably not as accurately discretized, thus producing rough approximations of the moments of inertia. As you can see, error is introduced at every angle. The only real way to remedy this situation is to have all airfoil coefficient data, and all aircraft schematics, including weights of everything, etc. readily available to the flight physics designers. This would clearly be a negative for Microsoft, who would undoubtedly have to pay a kings ransom for this information.This type of data is available however, in aerospace engineering texts by Roskam and others, etc., but only in select aircraft. If I recall, complete data sets were released for the 747, and the F-16A and a Cessna aircraft. Which explains the higher fidelity simulations in such popular titles as Aerowinx 747, Falcon4. These texts also provide linear approximations to many other aircraft, but again linear approximations to LDM coefficients cannot be 100% accurate. Now, I know we all want perfect flight models in our 50$ simulation, but this just isn't going to happen, ever. The engineers at Microsoft have done an outstanding job in my opinion, based on their available resources. The flight models in the aircraft are not perfect, nor will they be. The community has done a swell job tweaking the aircraft models to fly more "like reality", and isn't that what we are all after anyhow? If you want more realism, just get off yer duff and enroll in flight school dangit! Otherwise this really is as "real as it gets" for now :-)

[Guest SoarPics]

Interesting points. Can't say I agree with what you say about MS having to pay a kings ransom for airfoil polars, "schematics" (not sure what you meant here), and other data. This sort of info is readily available for any number of aircraft.I believe it comes down to offering what the masses will buy. The folks at forums like this are the discerning minority... and a very small minority, at that. The aftermarket gurus make the files we're interested in. And many of them are very good!Regards,

[Guest]

To properly model an aircraft requires more information that is available. Aircraft manufacturers do not just make public airfoil polars simply because a competitor may duplicate their design, or make a more efficient ranged version. This could result in profit loss to a company. There are hundreds, maybe thousands of airfoils available, including the NACA airfoils, but none for modern aircraft. Only linear coefficient data. The information available on the internet and in books is not accurate enough to provide a simulation with enough fidelity to function as larger simulators do, except in the example cases that were cited. Even then, the aircraft are not broken down with enough accuracy to completely model the airplane. The sources available in books, and the internet etc. do not give the detail needed to soft model every facet. Thus the actual aircraft design "schematics" would be needed, however, there are methods of approximating parameters that would yield inertial tensors that are "roughly" close to the real thing. Othewise, the engineers at boeing would have to release the exact inertia tensor etc. calculated during the design of the aircraft. I have yet to see anything but approximations of these values. Also, because the tensor matrix changes with payload, fuel and other things, there would be a need to have a way other than using a precalulated tensor to recaluclate these values for updating the flight model. Again, this requires much more than what is available, but approximation comes to the rescue!Oh, and in case I am wrong, send me a link so I can see this data for myself!! :-) I've been searching for a very long time. I have no pride to bruise, and I would welcome such information.Thanks!

[Guest christian]

I think the point that has been made in the past by a number of airfile editors is that the basic engine is still heavily based on FS2 or whatever version. And there seem to be a number of variables that can't be controlled in the air file (eg turbo props if I remember rightly). I can only give an example from the panel side. In FS2000 it wasn't possible to have a seperate DME channel (I believe in FS2002 it's still not fixed). However, a simple Bedix/King DME has a separate channel. In effect, you can't model a 100% accurate Bendix/King DME because of that limitaton. Similarily, I suspect there are things missing on the physics side. I have spent a bit of time with panel/gauge programming in FS2000 and there really was a whole heap of basic functionality missing. FS2002 has improved that situation somewhat, but there is still room for improvement. I'm quite eager to see the new GPS system, I hope that we will be able to access it once an SDK comes out.If it is really true that the physics engine base is quite ancient, then a rewrite could be in order simply based on the fact that current processors should be able to handle a much more complex physics engine (and more physicly relavant items could be modelled). On the other hand I'm not really able to judge how good the current engine really is, since I'm neither an airfile wizard, nor have any real flying experience (one trial flight barely counts, or does it :) ). From reading some posts and talking to people who know both real and sim, I have the impression that some effects are not modelled in the sim. It's not about making the physics 100% real, but pushing the envelope further. In that sense the demand to make the physics engine more realistic seems fair enough to me. On the other hand I'm quite happy with the announced features for FS9. The announced dynamic weather system can be seen as a major improvement of the underlying flight physics. It doesn't make any sense to model the aircraft physics 100% without paying any attention to the surrounding athmospheric conditions, so I guess we'll be in for a nice surprise there. Further, the current scenery engine still tolls the CPU by quite a bit. I really would expect that FS10 will make heavy use of DX9 vertex and fragment programs, and all the graphical tasks will really be offloaded onto the graphics card. That would free up the CPU to do more physics modeling, so I would expect more in that area in FS10.Just my thoughts,Christian

[Guest SoarPics]

crgse,You believe this to be an enginering issue... I believe it to be a business issue.MS could easily attain the engineering data if they wished to. But at what cost in implementing it into the physics model. And for whom? A few discerning simmers, whom they seem to listen to much of the time, but make up such a small market share that it would be economic suicide for them to do so.They aren't getting rich off this product. They have to weigh the balance between what would sell and what makes the sim better. For the masses.As Christian says, the physics model could certainly use some work, but we won't see anything close to what you (and I) hope for. That type of work will be left to gurus within the discerning community.Cheers,

[Guest]

I believe it is an engineering issue simply because it is a business issue. Microsoft could spend a huge sum of money on the flight models, but they dont have to because they can create linear approximations essentially for free, increasing their return on investment. The budget they are willing to give to the simulation development cycle does not allow for the flight models to be 100% accurate. I have no doubts that if Microsoft spent the money to purchase the data, that the egineering issues would dissolve, as would flight model complaints. They can't do that economically, for multiple types of aircraft (and still charge 50 bucks), therefore they have to design within the limitations of their given budget. Not to mention the fact that such data may not be able to be purchased, except for specific types of contractors developing larger simulations. Anyhow, I think we have the same viewpoint, but come from different sides of the field.

[Guest Ron Freimuth]

>Just a few words if I may :-). I think that over the years >of the Microsoft flight simulator franchise, there has >always been a demand from the community for Microsoft to >keep upping the ante on the accuracy of the flight model >physics engine........>>Large FAA certified simulations are modeled similarly to any >flight simulation. The aircraft is broken down into its >components, or discretized, where the moments of inertia, >weights, fuel, instruments, luggage, seats ac units, etc. >etc. are taken into account. The same thing can be done in MSFS. Some of my station settings in aircraft.cfg include the weight and location of engine oil. In flight variation of CG, mass, and MoI's due to fuel burn are taken into account in MSFS.>This type of data is available however, in aerospace >engineering texts by Roskam and others, etc., but only in >select aircraft. As a matter of fact I set Roskam's 727 Stability Derivatives, including variations due to Mach, in my B727. However, I had to increase some dampings so the lousy FS2K2 autopilot would be stable at higher FL's. While I'm modeling the turbines from data recorded from a FM simulator and also by reverse engineering FM tables. FM tables aso give information needed to model both lift dependent and non lift dependent drag sources that change with Mach number.>Now, I know we all want perfect flight models in our 50$ >simulation, but this just isn't going to happen, ever. The >engineers at Microsoft have done an outstanding job in my >opinion, based on their available resources. As far as I'm concerned, the flight model people at MS are incompentent and doesn't even understand the flight model code MS purchased. The more they work on this code, the more they mess it up. Further, the default AIR files provide plenty of evidence that some of the most basic things are not understood by the MS guys. Zero Lift Mach drag that decreases at Mach 0.80 (747), Torque curves no real engine produces (most of the reciprocating engine AC). Inaccurate prop tables, dampings 8X real in larger AC. Overall: Messed up LG dynamics in FS2K2, fake 'Yaw Damper', Lousy 'autopilot'. No prop torque in FS2K2, no real P-Factor, 'prop effect on pitch' lost. Useless FSEdit, poor and misleading documentation. It goes on and on. Also, most of the time default and inaccurate non-linearity tables are used (same appears to apply to the turbine tables). CFS2 has more custom curves, but spins were grossely overmodeled. > The flight >models in the aircraft are not perfect, nor will they be. >The community has done a swell job tweaking the aircraft >models to fly more "like reality", and isn't that what we >are all after anyhow? If you want more realism, just get >off yer duff and enroll in flight school dangit! Otherwise >this really is as "real as it gets" for now :-) Flying is very expensive, no one has a large range of AC available for his own use. And, very few are qualified to pilot the full range of AC. Further, GA is dangerous, much more than driving. -RAF

[Guest]

I doubt RealAir people purchased the data you're mentioning but their airfile for the Siai-Marchetti SF-260 beats everything else in accuracy. Including the a/c for which MS would have bought the data. That only goes to show that 1- decent simulation of actual a/c behavior is possible without heavy $$$ investment, 2- Working around MS FD engine takes thousands of hours in itself, 3- Grey matter and dedication always makes the difference.

[Guest]

Well, RealAir could not make use of the data even if they had it. The physics model is built into the simulation code, and is by default unalterable. I agree with your point about working around the basic FDM, but we still only get a rough approximation of the flight characteristics of the real aircraft. However, thats ok by me :-) My point is that if we want to fly by the numbers, we have 2 choices. One is to fly the real aircraft, or do a complete model as in Aerowinx, or any full size training simulator. These are too costly when dealing with multiple aircraft, thus we have a comprimise that is our inaccurate, but tweakable and many times fun generic linear flight sim.Later!

[Guest]

From a minority member:The last three versions of MSFS have had a common and very irritating error. When extending spoilers in the FS Lear (or any FS jet, for that matter), speed drops at too great a rate and the aircraft tends to pitch up. That is not the way a real Lear flies. As a matter of fact, a pitchup of only a few degrees could be disastrous with the spoilers extended. In reality, the nose safely drops several degrees because the center of lift shifts rearward when inboard-mounted spoilers are extended.Also, with FS2002, the aircraft cannot be backed up on the ground with reverse thrust and jets slow down at an unrealistic rate at flight idle. I suspect it's because the FS braking and zero throttle settings are set for C182 dynamics. I'm not certain because I'm no programmer. But I do have a few thousand hours of flight time in GA aircraft, mostly piston, with enough turbine experience to remember how they feel.AllynRetired ZTL ARTCC ControllerATP-LRJETCFI-II-MEL

[LAdamson 240]

> As far as I'm concerned, the flight model people at MS are >incompentent and doesn't even understand the flight model >code MS purchased. The more they work on this code, the >more they mess it up. >snipped>No prop torque in FS2K2, no >real P-Factor, 'prop effect on pitch' lost. Useless FSEdit, >poor and misleading documentation. It goes on and on. >Well Ron, as you know me, sometimes I agree with you, and sometimes I don't.....When it comes to the actual effects of "feeling" the left drift caused by P-Factor, slip stream, and torque with single engine (CCW rotating props), it's my opinion that FS2002 models or slight alterations are the best in the business ---- along with IL-2 Sturmovik ------------------ when using rudder pedals!!!!Without rudder pedals, there is NO comparison. I feel they're essential in providing a feel of pushing against a force, which FS2002 does quite well. Other sims, seem to imply NO force, or simply start drifting the other way. I first noticed this improved "feel" with Microsoft's CFS2. It was overdone in FS2K, but tamed in FS2002.edited: I'm even going father out on a limb here.....I fly from spot view quite often because I can, due to many years of R/C flight. It's the MS flight sim's (CFS2 included) along with IL-2 Sturmovik (once again) that seem to have the best dampening movements, a simulated feel of weight, and reasonable power versus weight movements, done much better than most other flight sims I've run into. As an example, X-Plane lacks almost all feel of weight between large & small aircraft. FLY doesn't simulate dampening and movement near as well as MS. We all know that third parties can either enhance the built in properties of MSFS, or even degrade them.I think you're way off the mark, in your thought's regarding MS programmers. It's been the simplicity of the default MSFS 172 using the VC and your modified airfile for dampening qualities, that's been one of my "finer" moments while landing at my 3rd party Sedona, Arizona airport. This airport has the look of an aircraft carrier approach. With the VC for peripheral visual cues, this particular flight seemed very real, both in look and even a sense of "feel"! If the programming is near disaster, as you imply, this sense of delight would never have occured within FS2002I've never felt it's the MS programmers problem to program everything in total correctness to the real aircraft. It involves many hours and appeals to "few". This is where the 3rd parties come in and fill the gaps, and make a few bucks by hanging on the coattails of the original sim. L.Adamson

[Guest Ron Freimuth]

>> As far as I'm concerned, the flight model people at MS are >>incompentent and doesn't even understand the flight model >>code MS purchased. The more they work on this code, the >>more they mess it up. >>>snipped >>>No prop torque in FS2K2, no >>real P-Factor, 'prop effect on pitch' lost. Useless FSEdit, >>poor and misleading documentation. It goes on and on. >>>>Well Ron, as you know me, sometimes I agree with you, and >sometimes I don't..... >>When it comes to the actual effects of "feeling" the left >drift caused by P-Factor, slip stream, and torque with >single engine (CCW rotating props), it's my opinion that >FS2002 models or slight alterations are the best in the >business ---- along with IL-2 Sturmovik ------------------ >when using rudder pedals!!!! As I said, there is NO "P-Factor", "Torque (engine)", or "Prop effect on Pitch" in FS2K2. The so called "P-Factor" slider only adjusts Prop Helical Effect. It did include real P-Factor in FS2K. The Prop Torque slider HAS NO EFFECT. I compared one AC in CFS2 and FS2K2 and was able to see and adjust the effect in the AIR file running in CFS2. The same file(s) running in FS2K2 displayed no 'torque'. Same for "Prop Effect on Pitch". It died in FS2K2. Increasing the power will not result in the tail coming up on a taildragger. I think this error also propagated to CFS2. This damages the C208 Amphibian since it's hard to pull the AC 'up on the step'. Fortunetely, I was able to adjust the "Prop Effect on Rudder" to give some steering from the prop blast in this AC. >Without rudder pedals, there is NO comparison. I feel >they're essential in providing a feel of pushing against a force.. You don't need rudder pedals or even a JS to see the effects. All should appear when an AC is trimmed for slow, climbing flight. Incidently, I can see the "Prop Gyro Effect" on left and right turns, the nose in higher when the AC is turning one direction than in the other. >force, which FS2002 does quite well. Other sims, seem to >imply NO force, or simply start drifting the other way. I >first noticed this improved "feel" with Microsoft's CFS2. It >was overdone in FS2K, but tamed in FS2002. Tamed by killing P-Factor. The pull to the left on TO could have been fixed in FS2K2 AIR files by simply reducing "Prop effect on Yaw" in the AIR file. Did MS say anything about that in the FS2K2 beta test? I don't think so. They would not mention ANYTHING having to do with the AIR file parameters. >edited: I'm even going father out on a limb here..... >>I fly from spot view quite often because I can, due to many >years of R/C flight. It's the MS flight sim's (CFS2 >included) along with IL-2 Sturmovik (once again) that seem >to have the best dampening movements, a simulated feel of >weight, and reasonable power versus weight movements, done >much better than most other flight sims I've run into. May well be true. I know X-Plane has numerous flaws. Every month a new X-Plane beta attempts to fix them. ;) I flew a 10 lb RC AC 30 years ago. Got tired of gluing it back together and never flew it again after a battery wire broke and it dived into the ground. > As an >example, X-Plane lacks almost all feel of weight between >large & small aircraft. FLY doesn't simulate dampening and >movement near as well as MS. We all know that third parties >can either enhance the built in properties of MSFS, or even >degrade them. But, can't fix things the MS guys ruin (or, can't fix). A large amount of time is spent trying to work around things we can. Currently, this includes a realistic EPR reading for a certain B727 in the works. >I think you're way off the mark, in your thought's regarding >MS programmers. It's been the simplicity of the default MSFS >172 using the VC and your modified airfile for dampening >qualities, that's been one of my "finer" moments while >landing at my 3rd party Sedona, Arizona airport. This >airport has the look of an aircraft carrier approach. With >the VC for peripheral visual cues, this particular flight >seemed very real, both in look and even a sense of "feel"! >If the programming is near disaster, as you imply, this >sense of delight would never have occured within FS2002 All my prop AC lack the effects I mentioned at the top when flown in FS2K2. They are also much worse on autopilot control. Mainly the jets, which fly at high altitudes where air density drops to 1/2 that at sea level.>I've never felt it's the MS programmers problem to program >everything in total correctness to the real aircraft.>L.Adamson They didn't. I think MS contracted out for the flight model code, and don't understand it well enough to either design good AC for it or to enhance it for the few limitations it has. I don't expect 'the best possible AC' from them, but there are many simple gross errors in the default AC. The C182 series is about the best, but lost some realism moving from FS98 to FS2K2. Part of the problem is MSFS managment. Every version ships on schedule: READY or NOT. Then, they won't even fix the many problems found in FS2K2. In fact, I doubt they are aware of many of the flight model problems they introduced. Ron

[LAdamson 240]

>> Part of the problem is MSFS managment. Every version >ships on schedule: READY or NOT. Then, they won't even fix >the many problems found in FS2K2. In fact, I doubt they are >aware of many of the flight model problems they introduced. >>Ron Hi again.....Been thinking about this today. It probably all comes down to "time" and the most important issues that appeal to the potential customer base . As most of us know, Steve Small does his wonders with multi-engine aircraft. You do yours with Dreamfleet, & Rob Young spent an enormous amount of time getting the RealAir SF260 to simulate spinning.Rob also stated that it isn't an easy process to just get every aircraft spinnable. Far too much code changes are involved. In that vain, I don't expect to see spinning in any Dreamfleet or FSD models anytime soon. Perhaps not even in certain future RealAir products. We now, all know that it's possible, but far too time consuming..... considering spins may be out of the normal everyday flight envelope & won't appeal to everyone. Therefore, I won't and shouldn't get upset if any "spinnable" real life plane doesn't do it in MSFS. And I won't get all upset & claim that particular models are inaccurate because they don't spin either. I believe the same should apply to Microsoft. They do some things average, well, or poorly. But I don't dismiss the whole flight dynamics file as garbage as some do!BUT.....................I consider RUDDER PEDALS as ESSENTIAL for takeoff roll and climbeout in the single engine aircraft. Over the years, I've experienced both ways. It's just too much of a difference, and as I've said before--------- both MSFS & IL-2 Sturmovik just have that certain rudder pedal feel that other sims don't! MS really does "feel" and maintain a quite realistic climbout on the runway centerline when using pedals------------no matter what the coding is! :)L.Adamson

[Guest Douglas K]

>It's the MS flight sim's (CFS2 included) along with IL-2 Sturmovik (once again) that seem to have the best dampening movements, a simulated feel of weight, and reasonable power versus weight< I hope you mean the best amoung flight sims, because comparing these simulated flight dynamics to real aircraft is inaccurate and misleading, in many respects they do not even come close! As an example, the ME109 in IL-2 swings first to the right and then almost uncontrollably left when you SLOWLY open the throttle (and I lock the tailwheel), the acceleration is pretty fierce on the takeoff run and it should be if you are going for realism, but then magically, after liftoff and even with 25% fuel on board it takes forever to accelerate past 190kmph, and if you raise the flaps to lessen the drag it will fall out from under you and dump you back on the runway if you don't catch it in time! Great fun on a computer, but real world takeoffs would be a pretty grim "going to the gallows" white knuckle type of affair. This is an aircraft that in real life will climb at over 3000 ft per minute! That implies a high rate of acceleration. Landings are another "sweating bullets" type of experience where the possibility of flying the aircraft on a stable approach, or smacking into Mother Earth hard enough to leave a smoking hole seems greatly in favor of the smoking hole outcome as the drag seems to increase in inverse proportion to the altitude AGL. Be ready on that throttle! An entertaining game certainly, and I love it, but not quite accurate enough when it comes to flight dynamics. While I have read that 5% of all ME109's were destroyed in takeoff and landing accidents, evidently it was the narrow gear and poor ground handling characteristics that accounted for these losses. In reality an aircraft that behaved as in IL-2 would have been like an allied secret weapon, killing scores of German pilots - and I'm sure the young men tasked with flying them would have been more afraid of takeoffs and landings than being shot at in combat. Having said that, I should also say that I do not have any flight time in a ME109, but I did own a T-6 for several years, and I've been lucky enough to ride "back seat" in a civilian P-51 many years ago, (no problem accelerating in THAT aircraft let me tell you!) and I've managed to accumulate over 9000 hours of PIC in a wide variety of single and multi engine aircraft during the past three decades so I believe I have a pretty good yardstick by which to compare the computer aircraft behavior. >I think you're way off the mark, in your thought's regarding MS programmers.I've never felt it's the MS programmers problem to program everything in total correctness to the real aircraft. It involves many hours and appeals to "few".

[LAdamson 240]

Doug....Hope your using rudder pedals..edit ---- 172 nose "bobbing wildly" from PIO? Not even my defaults do that with my hardware setup. I hope after 9000 hrs. that your's don't either! Maybe it's my less than 9000 hrs in real life, but close to that in R/C with "little" control sticks that make the difference.And BTW---- regarding IL-2, it's only takeoff & climbout I'm concerned with. Not exacting specifications for every aircraft.L.Adamson