[SU11 Beta] - Variance in handling AI injected plans vs SU10

I have a simple example piece of code injecting 3 plan files via
AICreateEnrouteATCAircraft using pure C++ simmconnect. The plans place
aircraft at 3 points on an approach route to KLAX and are identical except for
the starting point being slightly further out for each. In SU10 all 3 planes
will land. On SU11 the first one will land, the 2nd one will eventually
overfly the airport while still climbing past 10,000ft, the 3rd one turns and
heads in the opposite direction from the airport never to return. I’m looking
for confirmation of changes that would have impacted this behaviour and
indication if it will be looked as a bug. I do note that the altitude in SU10
is inserted incorrectly but in SU11 it is more accurate (related to live
traffic altitude changes perhaps). Maybe this then leads to the odd behaviour
in SU11 approach. But ultimately SU11 leads to lots of aircraft overlyflying
airports and circling before landing and some planes never trying to land at
all - different from SU10. This link is a video of the aircraft moition in
little nav map in SU11.
https://drive.google.com/file/d/1210NrpnBUCUFVDkE_ntpCDSj1L0PegNg/view?usp=sharing
This link is a video of the aircraft moition in little nav map in SU10 - exact
same code and plans.
https://drive.google.com/file/d/1HYZTcwCKTIY1syi-0sxEg7xXuF7hdn9F/view?usp=sharing
Update: After leaving the SU11 test running for an extended time - Plane 3
finally landed after about an hour and lots of ciruits above the airport.
After an hour Plane 2 is still doing circuits above the airport at 6000ft.

Hello @whakamolenz We suspect those aircraft
are waiting for crowded runways to be available. There has been some changes
with traffic density and behavior with SU11 and we are aware of some issues
with AIs not using all available runways and causing some traffic jam. Can you
check if this is the case here? (by disabling base traffic) Regards, Sylvain

Appreciate the response but when this test is running no other ai is enabled.
These are the only 3 aircraft in the sim over and above the user aircraft.

Hey, I will run some Stress-Test on the AI System next weekend and provide you
with detailed Feedback.

Ok, thank you for the feedback. We are investigating this.

If you want the basic code that creates these examples i can package it up.
Ultimately load testing in FSLTL shows that this behaviour has a significant
impact on the amount of landing traffic. But i thought a simple example would
be easier for replication purposes.

Further to this - observing arrivals at NZAA - could it be possible they are
now all being forced to some STAR that isn’t correct for their approach.
The Red line shows how they
should approach coming from airports south of NZAA for 5R typically DAVEE7D.
But instead all the jets in this image flew past NZAA and then turn to
approach from the north. The plan they fly is quite similar to APABO 3A STAR -
which happens to be the first star for NZAA if you take them alphabetically -
could it be just selecting the first STAR in the list for all arrivals if one
isn’t in the plan? Looking at the waypoints being followed in the simobject
view - this does seem to be exactly what is happening. Update: Our test team
is confirming this behaviour at all other airports tested and also seems to
apply to departures (first SID in the list alphabetically). For example at
EGLL we see STAR ALESH1H being used for all arrivals.

@FlyingRaccoon I was able to replicate this issue with our Traffic Controller
as well. (Should be replicatedable with the version you have as well). It
seems that AI that is spawned near (40-90nm) around the airport is climbing
back to CRZ FL while flying the approch waypoints and starts way to late the
decend torwards the airport. On some approches it seems it will start a
holding circle and slowly decend (way to slow btw), while on others it will
overfly the runway at around 20k Feet, start a GoAround (with out worjaround)
and try a new approch, sometimes this results in a landing (sometimes in a
second go around). So far I did not oberserve any AI with our app that needed
more then 2 GoArounds to finally land.

@FlyingRaccoon - is there a liklihood of
improvements to this issue in the final release build of SU11. Currently this
is going to cause significant issues for us if released in the current state,
with no obvious workarounds.

Hey @FlyingRaccoon have you guys made any progress on this issue? Can you
comment on if we can expect a fix for the final SU11 release next week, or
shoud we start thinking about some workarounds we can add in our tools to
reduce the impact of this bug? Kai

Hello @whakamolenz
@Kaiii3 No, we haven’t addressed those issues in
SU11. We are still in the process of reproducing and identifying the problem.
Regards, Sylvain

Hey Sylvain, thanks for the feedbacl. Do you need anything from us to help
reproducing the issue?

@FlyingRaccoon as i have a very simple injection program setup that created
the videos provided but either way I assume we must expect SU11 to be released
as it is. Can you provide any guidance on workarounds for this? In particular
a way to specify in the flight plan file for AI or via a simconnect API call
that approach procedures shouldn’t be used and vectors should apply?

no such call to avoid STARs, but I am thinking about some other
“manipulation”. Will report back when we have somthing that might work (at
least to some degree). Not sure how complex the FSLTL injector is regarding
knowing/reading stuff from the Sim, but this idea I have would require some
data to be read on runtime.

We read (and set) a reasonable amount of data in the sim including the new
facilities methods. So far my findings are that when the sim has assigned an
approach, i’m unable to manipulate those waypoints in anyway using the
existing AI Waypoint simvars. If you find something that works - great!

Hello @whakamolenz @Kaiii3 We are using one of your PLN file and tracking the
issues you mentioned. Am I right to assume your traffic is spawned airborne
because you face issues when they start from the ground? I know we’re also
tracking a problem with AI not taxiing and wanted to know if that’s related.
Regards, Sylvain

AI is spawned enroute

The AI not starting taxiing seems related to Ground Vehicles set to 0%

will work on a design this week and test the idea during the weekend - if it
works I will post it here, so you can implement it as well.

Yes same for us - airbourne. If very close to the airport (less than 50km) may
still land - otherwise above behaviour occurs.