AI aircraft oscillation on final approach

[Guest]

I downloaded and installed AFCAD and it works great. I am in the process of writing a traffic program for FS2002 simulating all arriving and departing flights from Philadelphia International during Feb 1986. (I came upon an old airline schedule and was inspired) I am about half way completed. Everything looks to be fine when I run FS2002 and go to PHL, except one thing. The aircraft that are inbound on the ILS starting from about 6 miles out grossly oscillate up and down. They dive then climb, dive then climb then end up slamming hard onto the ground. The departing traffic looks to be okay. Is there anything I can do about this? Also, is there a program out that will allow these AI aircraft to fly actual STAR and SID routes to and from PHL rather then great circle routes? appreciate any help. Jason Long

[wsieffert 0]

When you start fs2002, the AI are spawned around an approx 70nm envelope of the airport or situation you started. Because of this some aircraft landing are not at the correct altitude to make the correct landing to your airport. Another problem may be the choice of aircraft with poor aerodynamics. Several aircraft in my hangar do the up/down trying to feel for the correct slope to land. A lot of the aircraft land too far down the runway. Runways shorter than 8000ft have a lot of aircraft calling missed approach because they didn't land in the landing zone.

[betelgeuse 41]

>>Several aircraft in my hangar do the up/down trying to feel>for the correct slope to land. I've also run into this porpoising problem on approach. It seems to be worst at airports which have long flat approaches and which have default ATC. PADQ (Kodiak, AK) and PAFA (Fairbanks, AK) fall into this category and the porpoising on approach is really bad. On the other hand, MT29 (Sunday Creek Airpark, Montana) has no default ATC. After installing ATC etc there I can get an AI Dash 7 in no problem and no porpoising! I've tried setting the flightplans for the problem airports in VFR but still get porpoising. Does the ILS play a part in the AI landing engine? Where does this 'correct slope' come from and can it be modified?

[Guest]

You can use one of the default planes' .air and .cfg files. Just copy and paste the .air file and rename it what it was before, than copy the CFG file but remember to copy the contact points & texture part from the old cfg file.

[betelgeuse 41]

Yep, I know this. But if I do that I'll get a STOL aircraft that flies like a Baron or something - urrggh! What I'd like to find out more about is this 'correct slope' to which W.Sieffert refers in his post. Clearly, many of the non-default planes have problems picking this up. That's why they porpoise along during the early approach. The default planes have no problem identifying and flying along this slope. So what is it? Can it be vertically widened or tweaked in other ways?

[Guest]

I think the AI machine cannot make aircraft with complex FDEs land well.

[betelgeuse 41]

You may well be right. The question is why?

[Guest]

Well, this is Microsoft's first try. Maybe they never thought us simmers could crack the AI code and just programmed the AI program to read the default planes' FDE and the FDEs in par or lower with that.

[Holger 0]

Howdy all:if I may hazard a guess regarding this issue - it's fixable but would require a fair bit of work and knowledge about tweaking .cfg and .air files. See these posts for resources: http://www.simviation.com/cgi-bin/yabb/YaB...;num=1052161672http://forums.avsim.com/dcboard.php?az=sho..._id=9844&page=2 The thing is that AI flight dynamics in FS2002 use the same complex flight modelling procedure as any aircraft you're flying yourself. Thus, small changes to or incorrect settings of one or a few parameters can have a huge impact on the underlying differential equations. In my experience, porpoising is usually related to incorrect settings for the moments of inertia and/or control surfaces. Tom Goodrick's Flight Dynamics Handbook is an excellent resource for improving AI flight behavior.Cheers, Holger

[betelgeuse 41]

Hello Holger! I'm working with John Hubbard on the AI_D7 so good to hear from you. Thanks for the tips and the links. I've discovered some intersting things about the AI engine. At PAFA (Fairbanks, AK) there are 6 runways i.e. three strips. The longest (IL/19R) is over 11,000ft and the shortest (1R/19L) just over 3000ft. If I set up the D7 (and this might apply to all STOL aircraft) to land on the 11,000 ft strip it porpoises from about 12 miles out until the last 4 miles before touchdown. Then it setttles down and lands well. In total there are about 12 cycles, pitching up and down maybe 500 ft per cycle.If I close off this long runway with AFCAD and force the aircraft on to the middle length North/South gravel runway (around 4000ft) the approach is much more stable - maybe two cycles of porpoising at about 6 miles out and then a stable approach on to the runway. I then tried shortening the 11,000ft strip to 4000ft but got no reduction in porpoising. So it seems that this STOL AI aircraft works best on short runways - the opposite to the default planes.

[Guest frankcubillos]

For grins you might want to try our PAI V.5 FDE's, not perfect (as mentioned earier V1 MS AI is a hard one to crack, hopefully FS04's 2nd Gen AI will be better) but it's better than the default FDE's, our guys have been working hard on this well known MS AI issue.Cheers :-wave---Banners? We don't need no stinkin' banners!---Visualize FSGateway/ProjectAI sig banner here

[betelgeuse 41]

Hi FrankI've checked out the PAI v.5 FDEs - 12 varieties of 737 and a Jetstream was what I found - couldnt help wondering if anyone had worked on a turboprop/STOL. Maybe I've missed something obvious.