@ sign in AI flightplans

[Deleted]

Hi,

In most AI flightplans arrival times are preceded by an @ sign. As I suppose adding this sign should fix arrival times to what it matchs more or less to real schedules. However especially for long distance flights the arrival times aren't those at which AI land effectively. Here an example from traffic 1962:


Leg with @ sign

AC#678,HB-IDA,25%,WEEK,IFR,0/01:58:08,@0/09:54:19,310,F,0861,LSZH,1/08:43:50,@1/11:13:57,260,F,0816,LPPT,1/11:58:15,@1/20:23:58,320,F,0816,KJFK

Leg without sign

AC#678,HB-IDA,25%,WEEK,IFR,0/01:58:08,0/09:21:23,310,F,0861,LSZH,1/08:43:50,1/10:55:48,260,F,0816,LPPT,1/11:58:15,1/18:20:06,320,F,0816,KJFK


In the first case the AI isn't shown up at 20:13 in Traffic Toolbox Explorer, nor at 18:10. There are some other enroute AI not shown. So I don't see any sense to put @ signs in flightplans. Maybe someone could explain a good reason of doing it anyway.

Cheers,
Bernard

[Tom/CalClassic]

There is indeed some controversy about the @ symbol. Some say it works well for them, other say it doesn't. In my *personal* experience I get more arrivals if I use it. So I do.

For the details, here is an excerpt from Don's AIFP manual:

"4.5 The @ Symbol (User-Specified Arrival Times) – TTools allowed estimated times of arrival to be specified by prefixing such times with a “@” symbol. Alternately, the arrival time could be left blank, in which case it would be computed by the compiler based on the distance to be travelled and the specified cruise speed of the aircraft used.

AI Flight Planner does not use these concepts. The ETA is always calculated by AIFP but, nonetheless, may be overridden by the user.

The "@" symbol was a (useful) TTools "invention" that, unfortunately was not “understood” by the AI engine. Thus, its use in TTools-format flight plan files involved some “trickery”. When arrival time was preceded by the “@” symbol, the TTools compiler subtracted 15 minutes - a nominal amount to allow for approach, landing and taxiing - from the specified arrival time and recalculated the aircraft cruise speed for that leg of the flight plan and used that revised speed to determine sector arrival times. Unfortunately (as will be seen in the next section), there is no place to save this revised cruise speed in the traffic file. (Changing the specified cruise speed of the AI aircraft itself would affect all flight plans using that particular aircraft.)

4.6 The “37-Minute Problem” when Using @ – The MSFS AI/ATC engine doesn’t “like” tardy AI. If an AI aircraft is delayed on departure due to, for example, runway congestion or a long queue at the hold-short point such that the aircraft has not taken-off within about twenty minutes of the scheduled departure time, it simply disappears from the taxiway. Likewise, if an AI is forced to perform several missed approaches making it very late for landing, it disappears.

There seems to be a similar condition applied to enroute AI. In particular, if an AI is scheduled to be activated (i.e., enter the sector occupied by the user aircraft or one of the surrounding eight sectors) more than about twenty minutes later than as calculated by the AI/ATC engine using the distance travelled from departure to destination at the cruise speed specified for the aircraft in the traffic file, that AI will not materialize for landing. Instead, it will spawn in a parking spot at the destination airport some time later in preparation for the next leg.

When the sector entry time is based on the cruise speed specified in the traffic file, AI activation will always occur “on-time” (i.e., at the time calculated by the AI/ATC engine) and all is well. But, if the arrival time was specified using “@”, the cruise speed used by the compiler to calculate the sector entry time may have been significantly different from that specified for the aircraft in the traffic file. When “@” is used to specify a later arrival time (as it usually is), the AI will be activated “late” relative to the time calculated by the AI/ATC engine. If it is more than about 20 minutes late, it is discarded. From extensive testing (by others), it has been determined that the critical time difference is 22 minutes. Given the standard 15 minute allowance for approach, landing and taxiing, the 22 minute interval equates to 37 minutes later that a user-specified arrival time. Hence, the name “37-minute problem”.

The 37-minute problem is most likely to occur when simulating scheduled airline long-haul operations where the user-specified arrival time is often substantially later than the simply-calculated (distance/speed) next-to-final-sector arrival time plus the fifteen minute approach and landing allowance.

Apparently, the AI engine doesn’t “care” about AI activating arriving early. So, to overcome the 37-minute problem, some suppliers of freeware AI flight plan data, such as WoAI, MAIW and, until recently, AIG Alpha-India Group, specify a cruise speed in aircraft text files in the order of 200 kts for all jet aircraft. This ensures that AI activation time for any AI having a reasonable user-specified arrival will not be “late” - since the cruise speed then specified in the traffic file will be less than the cruise speed calculated by the compiler to accommodate the user-specified arrival time."

Don uses a similar technique in AIFP, but he doesn't use the @ symbol to specify arrival times:

"AI Flight Planner uses a similar (but subtly-different) approach. It halves aircraft cruise speeds before saving them in the traffic file. But, when it decompiles a traffic file that it has previously compiled, it restores the cruise speed seen in the Aircraft List by doubling the cruise speeds saved in the traffic file. Thus, this “evasive action” is invisible to the user – unless he/she happens to notice that a cruise speed originally being an odd number is returned after de-compilation one knot lower. Other than this possible slight change, there are no known side-effects of this scheme.

Should you decompile an AI Flight Planner-generated traffic file other than with AI Flight Planner, you will find an unusually-named airport as the first entry in the airports.txt file. This airport is a “flag” used by AI Flight Planner to indicate that special measures to address the “37-minute problem” have been taken and that cruise speeds derived from the traffic file should be doubled."

Hope this helps,

[Deleted]

Hi Tom,

Thanks for your ample explaination. Now I remembered to have red once about the "37-Minute-Problem". The fact is, that the above mentioned flight to Idlewild only can be catched with Traffic Toolbox Explorer far afield over the Atlantic. There is no way to see it somewhere between Newfoundland and Long Island again. Anyway, I do a try without @ symbol for a while.

Cheers,
Bernard

[Tom/CalClassic]

Yes, you will find AI planes that you cannot "capture". I found it better overall with the @ symbol, but as I said others like it without.

[Deleted]

Jul 28, 2014 17:40:25 GMT -5 Tom/CalClassic said:

Apparently, the AI engine doesn’t “care” about AI activating arriving early. So, to overcome the 37-minute problem, some suppliers of freeware AI flight plan data, such as WoAI, MAIW and, until recently, AIG Alpha-India Group, specify a cruise speed in aircraft text files in the order of 200 kts for all jet aircraft.

Hi Tom,

Is that the case for CalClassic traffic too?

Cheers,
Bernard

[Tom/CalClassic]

Yes it is. The aircraft speeds listed are about half of the real values.

[Deleted]

Sorry, but now I'm quite confused. While I was intended to restore the initial Traffic 1962, I noted that for the following both Swissair jets speed has been set to 477 mph

AC#677,477,"Caravelle Swissair"
AC#678,477,"DC-8 Swissair"

For all other jets speed is about 200 mph. So my questions:

- why just these two models have other speeds?

- should the @ symbol not be deleted in their flightplans?

- could that be the reason why the DC-8 doesn't appear at approachs?

BTW, with the actual speed the schedule is approximatively respected, even if curiously westbound the flights are faster.

Thanks for your explanation.

Cheers,

Bernard

 

[Tom/CalClassic]

They are errors, of which there are many...

They should be around 200 kts.