Auto Constraint Deletion

[Owl]

Hi Hardy,

I don't know if this has ever been brought up before, but there is an interesting behaviour in the FMC which isn't modelled in PSX:

Upon satisfying an "At-or-above" constraint in the CLB phase, the associated constraint is deleted automatically by the FMC, as it should have no further relevance.

What Honeywell missed was that when a constraint is deleted in this way, it also erroneously deletes any associated speed restriction. If you are flying a normal VNAV departure with the speed window closed this can cause all kinds of fun and games if your SID has terrain clearance that is predicated on a max speed and you miss the bug moving up!

A bulletin (Effectivity: All 747 Airplanes) was issued around 2008 or so to alert crews to the anomaly, however remains in effect to this day (They say changes are planned for the -8 so pretty much acknowledging they will never bother to fix it for the -400).

Bulletin contents:

Boeing has confirmed operator reports of a Honeywell FMC anomaly that incorrectly deletes a speed constraint. Some SIDs are designed to limit turn radius to maintain clearance with other traffic or restricted airspace. Some of these procedures also have an AT-OR-ABOVE altitude restriction inconjunction with the speed constraint. Typically, the airplane will be required to limit speed until passing the respective waypoint as well as climb above the altitude constraint. In these procedures, VNAV will incorrectly delete the speed constraint prior to reaching the waypoint if the altitude constraint has been satisfied. When this happens, VNAV will command speed to accelerate to ECON speed (or SEL speed) prior to reaching the constrained waypoint. This anomaly exists on all Boeing 747 / 757 / 767 / 777 airplanes equipped with the Honeywell FMC.
Honeywell is aware of this anomaly and has planned changes for the 747-8.

I don't know what your thoughts are on replicating real aircraft bugs but this one certainly has some operational significance and good training value in it.

Cheers

Owl

[Hardy Heinlin]

Hi Owl,

thanks for the hint. Will the constraint be deleted when the aircraft is, say, 50 feet above the target altitude? Is it mentioned in any manual? I didn't know this feature. Maybe there's also a time delay.

I wouldn't call it a "bug" if the deletion is a thought-out feature (a bug is an error). This feature is probably just not well-thought-out :-)

As the NG FMC allows speed-only constraints, the NG will probably delete the altitude only. Do you know more about it?


Regards,

|-|ardy

[Owl]

Will the constraint be deleted when the aircraft is, say, 50 feet above the target altitude? Is it mentioned in any manual?

It's strangely absent from any manual I have, the only record of it I can find is the safety bulletin that it causes. We encounter this so often on the line that it becomes second nature to just open the speed window straight away.

Again due to the absence of documentation I couldn't tell you much about it's exact logic, other that the constraints are usually cleared within maybe 2-3 seconds or so of passing the level. Whether there's a fixed time delay or the function is just being throttled for performance reasons who knows?

If anyone else could chime in and add anything that would be great (75/6/7 also welcome!). If not I can try and record it happening next time I'm on the jumpseat.

As for the NGFMC I don't have any experience with it, however the bulletin does apply to all 747's so I would be inclined to say it hasn't been fixed there either.

I wouldn't call it a "bug" if the deletion is a thought-out feature (a bug is an error). This feature is probably just not well-thought-out :-)

Unintended behaviour maybe if we're being polite

Some other gems that are still current on the 747:
Uncommanded Turns When LNAV is in Use
Uncommanded VHF Frequency Changes
FMC Performance Predictions Anomaly
Flight Deck Display Unit Blanking Anomaly

Owl

EDIT: Found a perfect example in this video at the linked timestamp, you can see the 8000A disappear from the ND waypoint DATA block a few seconds after passing through the level.

https://youtu.be/2qXrrLA-UGw?t=2648

[Hardy Heinlin]

EDIT: Found a perfect example in this video at the linked timestamp, you can see the 8000A disappear from the ND waypoint DATA block a few seconds after passing through the level.

https://youtu.be/2qXrrLA-UGw?t=2648

Nice. I guess the trigger is "100 feet above" to be on the safe side in case of turbulence or baro disagreements.


|-|ardy

[Owl]

I wonder if it would also apply in reverse in the case of an at-or-below restriction during the DES phase?

[Hardy Heinlin]

If I were to set a speed-only constraint in DES (in the legacy FMC) with a "B" altitude, I would set that "B" altitude above the aircraft -- just as I would, vice versa in CLB, set an "A" altitude below the aircraft. So if I were to set a "B" in DES above the aircraft, I would immediately see the autodeletion and wouldn't do this again. Therefore I guess everyone would set speed-only constraints always with an "A", both in CLB and DES, and always far below the aircraft to avoid a level-off in DES. And therefore I guess nobody has ever seen such an autodeletion in DES. Hope my comment wasn't too complicated :-)

[Owl]

If I were to set a speed-only constraint in DES (in the legacy FMC) with a "B" altitude, I would set that "B" altitude above the aircraft -- just as I would, vice versa in CLB, set an "A" altitude below the aircraft. So if I were to set a "B" in DES above the aircraft, I would immediately see the autodeletion and wouldn't do this again. Therefore I guess everyone would set speed-only constraints always with an "A", both in CLB and DES, and always far below the aircraft to avoid a level-off in DES. And therefore I guess nobody has ever seen such an autodeletion in DES. Hope my comment wasn't too complicated :-)

Yeah it makes sense that it wouldn't be seen, not to mention the fact that B constraints are generally much rarer than their A counterparts in the wild. Also outclimbing a SID is considered a beneficial thing to do, whereas descending below profile obviously not.

Thinking further about this:

I wonder how many legs the autodeletion function iterates through. It could only be the active leg or it could be all CLB legs. I'm not sure what it would do if you had for example 2 or more legs with the same A constraint. I would be inclined to think it iterates through all of them but I haven't paid that close attention to it until now. Hmmm 

[Hardy Heinlin]

Actually, I thought it was a recommended trick to set speed-only constraints with an A-altitude that should be as low as possible, e.g. at airport elevation. I think it's also in Bulfer's FMC Guide. That's why I always thought auch an autodeletion would never occur. The trick won't work. Maybe it was just for the DES.

Wouldn't that be a horror if you were to descend below the A in manual flight inadvertently and the autodeletion would support your erroneous path because you just lost the true target altitude in the FMC which you should re-climb to?

[Hardy Heinlin]

Perhaps the logic is a bit more complex. It could be that the autodeletion in CLB isn't armed when the constraint is already below the aircraft before takeoff, or before that waypoint becomes active. Something like that. Similar to the autosequencing on final which can only occur when the aircraft was in front of that waypoint on leg course.

I'll keep my current model as is until the logic is 100% clarified. Otherwise there'll be too many possibly unwanted side effects.


|-|ardy

[Owl]

I'll keep my current model as is until the logic is 100% clarified.

I'll see what I can come up with. Were there some particular edge cases you had in mind to test?

[Hardy Heinlin]

For example:

In flight, in CLB, at the next waypoint after the active, set a 000A constraint (sea level). When will the FMC delete it?

In DES, at the next waypoint after the active, set a "B" constraint 200+ feet above the current aircraft altitude. When will the FMC delete it?


Thanks!

|-|ardy

[Hardy Heinlin]

Re pilot entered "A" constraint below aircraft: I just found the source of this trick in Bulfer's FMC Guide on page 97, bottom right. He mentions this in context of the descent only. So this supports the idea that this trick may not work in climb (due to autodeletion). Now this is motivating me to risk the modification in the next PSX update :-)

Actually, in climb, you could also insert a speed constraint with an "A" using an altitude way above the predicted altitude, so that it won't be autodeleted before passing that waypoint. VNAV SPD will climb with THR REF anyway. The only disadvantage is that the "UNABLE NEXT ALT" message might appear. But that won't influence the planned path anyway.

I don't know if the NG FMC will autodelete the speed as well. I will modify it so that the NG FMC keeps the speed. I can remodify this later when pilot reports confirm the exact behaviour.
Edit: Altitude-only deletion looks bizarre; I changed my mind: I think the speed should always terminate at the altitude-point (not way-point) if the original entry was a SPD/ALT data pair. So I'll apply my entire modification to the NG FMC as well (not just to the legacy FMC).


Regards,

|-|ardy

[Hardy Heinlin]

Autodeletion of "at or above" climb constraint is now implemented in PSX update 10.158:

https://aerowinx.com/board/index.php/topic,4191.0.html


|-|ardy