FS2004 Strange AI pattern/circuit behaviour - missing the turn to final etc

[Here for the Beer]

I just finished a model for the Druine Turbulent (for AI only) and I'm having some hassles with the FDE. The aircraft doesn't operate in a flight envelope I'm familiar with, lots of low speeds (stalls at 25mph!) and the dreaded taildragger ground handling. I've made the Jodel and RV4 previously and I'm still not much use at making these aspects work.

I can get the Turbulent to fly around at rational speeds (it gets a bit haywire sometimes), but its approach to a runway is, well, weird.

It barely descends at all, just floats down in its own sweet time. I have the lift set high so that the aircraft gets the speeds into the right area, but it seems to come in on final like a balloon more than something heavier than air. Altering the drag to make it work harder at descending just seems to throw the takeoff into disarray. The strangest part of all is that as often as not, the aircraft doesn't turn final! It sometimes does, but sometimes it flies through the centreline at about 45 degrees and lands somewhere other than the airport! Does anybody have an idea why? I was thinking it might be caused by the fact that the aircraft is flying at what FS probably considers to be a highish taxying speed (!).

The other hassle is the turning circle on the ground. With differential braking the aircraft should be able to turn around a mainwheel, but instead it takes such wide arcs that it departs the taxiway system and wanders around the countryside looking for a taxiway to latch on to. I have had the tailwheel steering angle set to 90 degrees and still it made little difference. I assume this might be something to do with the origin of the model (at least in the CFG). Would moving some datum around cause the aircraft to improve its behaviour?

Another strange thing I've noticed is the weird circuits the aircraft sometimes fly. I have had within the past half hour, an AI Turbulent fly a figure-of-eight pattern. Departure turn was to the left, it did a teardrop turn and crossed over the runway at circuit height to fly a teardrop late right-hand downwind and base from the opposite side of the runway. I know Turbulents are kind of maverick aircraft in some ways, but this is ridiculous!

Or perhaps I'm just flogging a dead horse here, forgetting that "you can't polish a Turb".

Edit: Just seen another one that flew base leg all the way to the ground (inside a hill) only to emerge a few minutes later, taxying across fields etc until it got a sniff of a runway link. They just don't seem to see the runway from base leg at all.

 

[Roy Holmes]

There is a discussion at http://www.fsdeveloper.com/forum/showthread.php?t=49157 that might help

I found that a lot of this sort of behaviour is due to the AI running out of aft elevator control which means it can not generate enough lift to turn to get aligned with the runway. As a first cut, you could increase the amount of elevator up movement in the .config file.

Most likely the CG is too far forwards, it should be at 1/4 chord. You could try that as well.

It is interesting how many aircraft fly OK when flyable in the sim, but all their bad habits show up when they are AI.

Tail draggers do tend to overshoot turns taxying as AI, their CG is aft of the main wheels and they are directionally handicapped on the ground.

Roy

 

[Here for the Beer]

Thanks, Roy. I had already improved the elevator lift as part of the process of getting the tail up early (and also changes to the CofG, although in this case I'd already been moving it backwards). The CofG going back was part of the attempt to get the ground turns to look better, because I was shifting the datum back too and the CofG needed to move back in tandem. The elevator authority was also fairly spectacular, the tightness of the turn onto crosswind had the aircraft losing a fair bit of speed and having to drop the nose quite a bit to maintain airspeed.

That doesn't mean you're off the mark, it is quite probable that some other value was negating all the elevator authority. I tried to find a CFG entry for "pilots_arm_beefiness" but there's not one there.

I dislike the process of getting one aspect right and then having to change something else that was affected, and which forces me back to the original situation to try to readjust it to satisfy both modes of the flight, so I approach these things straight away making changes on several fronts at once - mainly because it is what an FDE developer must end up doing anyway as they try to balance the different aspects.

For now - and probably for good - I've copped out and gone to the default FS9 Piper Cub for my FDE, although I'll increase its stability or decrease its aileron authority (or both) to stop it relentlessly yawing from side-to-side.

The process had been fun until the FDE, but then it often is.

Once this has been sorted out, I might make a Jodel Bebe over the Christmas period. This one should have taken 3 days to make, now it'll be four, a 33% project overrun, tsk. Cheap 'n cheerful, just like the original.

 

[Roy Holmes]

One interesting point which is buried in that long thread, is what appears to dictate AI approach speed. I built an "AI chaser" Jet Provost T-4 and gave it the MFD's from an F-15E I had built earlier. On the left MFD I could select the AI I wanted to examine and on the right one I has speed height heading throttle position etc, all I needed to see what the bugger was up to.

I found that all AI I looked at adopted a precise 2 degree nose up pitch angle on the approach and then adjusted power until they were at 5 degrees angle of attack, which combination gave them a 3 degree approach slope. Their speed was what they got at 5 AOA. You can change that speed with different flap settings etc. The 5 degrees was not rock solid, it depended whether they were on an exact 3 degree slope. Crossing the threshold, throttle was cut and a flare made just above the runway. I could record the percent of stick movement and see it max out with some planes. Those ones tended to land short of the runway. Brakes seemed to come on at about 50% of the approach speed. If the contact points were not well set up and the wheels were not in good contact with the runway, braking was really bad.

In the JP I could chase about any plane around a circuit and see flaps going up and down depending on AOA and lots of other interesting things. For tail draggers I had a B-17, DH Rapide and a Piston Provost. They all performed well even when taxying. The biggest problem I had was with a F-104 and I can not remember it ever making it to the runway!

Just a few thoughts
Roy

 

[Here for the Beer]

In theory, then, I could use a pitching moment of the flaps to achieve something too. I'd been standing back from altering that value, yet it might actually be the kind of thing that could change the way several other values interacted and almost solve the problem by itself.

Either way, I have stuck with using the default Cub FDE since it sort-of works. I had to make the ailerons and rudder far more sensitive/effective and ladle in a large dollop of stability in roll and yaw to get it to behave. Now the FDE is almost unflyable by humans - very twitchy, but the AI pilots love it. So for humans the Turb now overspeeds, is capable of flick rolls (sometimes accidentally) and utterly unTurblike unless the AI engine is flying it, when it stays unperTurbed. While I was testing those changes I LODed the thing, so all I need to do now is figure out which ones are still flying amd make some flightplans for it otherwise, frankly, it'd sit on the shelf because people wouldn't make the effort to get it airborne.

It's been a useful exercise trying this low-speed stuff, though, educational and has given me a "stock" FDE for any low-speed AI I make in future. Nothing is ever lost if we decide to give it value.

+++

On a discursive note - I'm using my more flyable FDE to take G-ARBZ from Cannes (where I tested) up to Headcorn (where the real one is based), via Geneva, Chatillon-sur-Seine, Le Touquet (some of my good old days IRL were flying there) and via Shoreham to clear customs, then back into Headcorn - some more fun real-life flying there too. Crossing the western Alps has been great fun (something to do while researching the active Turbs in the world). The beast won't fly much over 8000ft and the mixture gets leaned off to the point where you wonder whether the engine can manage on so little juice (no EGT gauge either, so you listen to the revs). Found a couple of valleys that were dead ends to a Turb with no more climb left in it, and threaded my way through to Geneva via a couple of detours, IMC at one point and praying that the GPS was right about the terrain. Flying with almost no available power is a real treat when there's no mortality involved. 25kt winds from the NE meant that on some legs the ground speed was halved.

Kind of on-topic if you count digressions (!), I can remember watching a 15 minute aerobatics routine performed by Anne Welch at Old Warden one remembered summer, fifteen minutes of impeccable energy management - in a glider, after an aerotow to 2000' AAL. Pure genius.

I will probably include the flyable FDE in there for people that want to put themselves up against the weather and the mountains (I might have to cut some ailerons, rudder and elevators, though... after all).

Incidentally, à propos of nothing in particular, I used to know a Roy Holmes when I was working for an avionics company in SE England. Your penchant for English hardware does lead me to wonder... Your mention of JPs makes me recall being in a hot air balloon in solid cloud over the midlands somewhere when a couple of JPs heaved out of the clag and passed us by about 100ft horizontal with no vertical separation. It was quite nerve-wracking because of course in a balloon you can hear everthing, and a rising jet sound is, how can I put it? Unnerving. Especially when it seems to be all around you.

 

[Roy Holmes]

Could have been me, not too many Roy Holmes out there or, at least I have never met another one. I worked for MEASL/Marconi Avionics/GEC Avionics in Airborne Display Division from 1977 to 1988. Lived in Tenterden down the road from Headcorn and I used to fly Robins from Headcorn and Rochester. I flew the Piston Provost at Cranwell and the JP4 on a so-called scientific test in Singapore and Malaysia. Rest of my flying was in Hunters and F-4M. Spent a year at Upavon. My first ever flight was in a DH Rapide from Dublin to Liverpool and my FS Rapide is in Aer Lingus disguise. Moved to the US in 1988, had the operation and became a US citizen.

Roy

 

[Here for the Beer]

Ah, you're the RH I met a few times, then, we shared a company and indeed divisions (ADD). I moved out of military work - just lost my appetite both for it and GEC - and eventually out of Kent to the Cotswolds to work for the emergency services as a contractor, where ironically the first job was for Kent Police. Extremely rewarding work, though. Ran my own company for many years doing systems design and business analysis before turning my hand to writing (Roanoke got a mention in my first book too, which is a strange coincidence).

I had lived in the US prior to GEC, working for a telecomms company in Maryland for a while (getting shipped around the world fixing bugs). I never "had the op", although I came very close to staying on and taking a commercial pilots license with my accrued salary (I was on a per diem, so I'd never touched my salary). I did a lot of flying there because it was dirt cheap, but decided I didn't want to be paid for it (a pivotal choice I'm certain was right).

Flying out of Rochester was good fun, although Headcorn was a bit too hands-off for me; still a fun airfield, but when everybody is acting "at your own discretion" it depends on everybody having some level of discretion! At least at Rochester I managed to get my grubby mitts back on biplanes, flying "gashes", Hotel Sierra, quite a bit. Years before I'd flown the BAe Tiger Moth up at Hawarden and always pined for old biplanes after that. These days it is more DR400s, R2160s and the occasional Cherokee familiarisation flight if I'm going to be using one - I know a Cherokee owner in St Lucia and have occasionally rented his N-reg a/c for island hopping. I have a US licence as well as JAA, so no problems there.

Also, for laughs, I did fly in a Phantom formation once upon a time (and after a fashion). Well, I was flying DR400 PH-SRG from Seppe to Rotterdam and two WGAF F4s formated on the aircraft, gear down, flaps down and wallowing, but they managed it for a few moments before tucking everything away and vanishing quickly in a roar of reheat before we could bring our guns to bear. Good times.

 

[Roy Holmes]

Small world!
I flew GASHS only four times, flew really well, but was a pig to taxi, the brakes did not work.
Roy

 

[Here for the Beer]

I tended to treat 'HS like a Tiger Moth and assume there were no brakes at all. Fortunately the rudder on the Stampe is really effective. Even so, on one occasion when I was threatened with a groundloop on takeoff from Rochester r/w 16 in a westerly wind that was coming over the industrial estate (meaning the crosswind varied rapidly all along the runway - something you really notice in a tailwheel a/c) the rudder couldn't counteract it so I hauled the aircraft off as it was turning and departed over the top of the warehouses instead of along the extended centreline. We went off-piste for about 6 feet into the tall grass before it lifted, though, which was a very anxious moment.

I did get nagged about sideslipping too enthusiastically on finals and told that the rudder post was only mortal, and I could ease off and give it a longer life. Old habits, again down to the Tiger Moth I had a relationship with, whose engine occasionally decided to stop before the aircraft had landed, particularly if it was idled for too long. Consequently I'd approach the airfield at 2000' as though for a typical overhead join, no matter what. If I was cleared direct to base leg I'd still come in high and slip the height off or do a few s-turns to get down. I'm normally good with crosswind landings and departures, but with the TM I never felt secure above a 7kt crosswind component and, at base, they'd not release the a/c for flight if the x-wind was over that threshold (fortunately they had a large grass area that the TM could depart from in almost any direction (wheels off at 40mph, initial climb at 55mph and throttle back to 1950rpm when reaching 300ft to be kind to the engine - something that can be done in a very short run).

I did try sideslipping a DR400 at Dunkeswell one time when I did the same high approach. I'd been cleared to the overhead and they changed their minds very late and cleared me to base leg, so I gave it a try - the DR400 was not great at slipping, which is a pity given those vestigial flaps it has.

Happy days...

Apologies for the delay in coming back, when I'm not modelling something I don't drop by as often.

 

[Roy Holmes]

I also have time on the Tiger Moth and the big difference from the ground handling aspect was it had a tail skid, not a tail wheel like the Stampe. If you pulled the stick back and gave a blast of throttle that skid would try to get stuck in the grass and it really helped straighten the beast out. Mr Stampe had a castoring tailwheel that gave it the directional stability of a supermarket shopping trolley. It was awful when taxying down wind.

However, the Stampe's inverted flight fuel system was a big plus and it was nice to be able to do a slow roll without the engine cutting half way round the roll like it did in the TM. That engine would also die in a vertical stall turn.

Loved your story about the cross-wind take-off! Tall grass is nothing compared to a take-off ground loop.

I was flying from somewhere to Lasham in a TM one day. Weather was OK on take off but Lasham is on a sort of hill and as we approached, the cloud base dropped. All I could see was the tops of trees and I made a 180 with them as reference. Bit of a knee trembler. Had to stay in contact with the ground because there were no radios or nav aids. No inter cockpit comms either, the TM one consisted of a tube that you shouted into and hoped the other pilot could hear. It was not working and we, myself and my good friend Ewan Perreaux (later killed when flying with the Red Arrows) had decided that "You have control" would be signified by a strong aileron waggle. He was in command on the flight and took over after my 180 and very strong aileron waggle. Since the weather was getting worse he landed in a field. We walked to the nearest house and had a few cups of tea until the weather cleared and he swung the prop, I turned on the mag switches and we returned to wherever it was we started from. Never got to Lasham.

On the subject of hand swinging the prop, it is a scary, but well tried practice, especially if you have someone else in the cockpit to control the throttle.

If not, this can happen

http://www.dailymail.co.uk/news/art...horror-biplane-takes-pilot-crashes-trees.html

Needless to say it happened to a Stampe with no brakes!
It and the TM were fun to fly, as you said, happy days

Roy

 

[Here for the Beer]

Returning to this thread after how long?

I have an answer to my rather old and outdated problem now.

What happens with low speed AI is that the sim (FS9, but others might do the same) flags the aircraft as having landed even if it is somewhere on base, or joinging a fake ILS for routing, when the sim drops the airspeed to final approach speeds. If the a/c is slow on approach the sim tells itself that the a/c has landed and that it needs to taxy in.

I'm working on integrating helicopters into my FS9 setup before stopping any development efforts and just becoming a user. Needless to say I'm dealing with glacial approach speeds and have tweaked FDEs to get one working, and have integrated it into all of the helicopters I'm using. I now longer have the problem of "taxying before landing" in the sim. So, my guess at the time of the original posting was in fact a good one. Get an AI aircraft down to about 30 knots and FS9 assumes it is taxying. If it has to reach that speed it needs to do so pretty much as it touches down.

I never used Traffic Toolbox for diagnosis back then, and the flags for the phase of flight are visible in there and changes are easy to follow.

I re-read the flying tales as well, and it reminded me that the Grumman AA5 range was similarly a pig to taxy in strongish winds because the nosewheel was castoring. Staying straight on a taxiway meant riding one of the toebrakes and pretty much leaving the other alone.

This taxying behaviour might be a feature of later simulators, so if you see this sort of thing happening at least this thread might provide a better clue.