Flight director ALT HOLD mode

[Hardy Heinlin]

Good morning,

quiz question:


Situation:

- QNH 1013 is set
- ALT HOLD mode is engaged
- A/P engaged
- Aircraft maintaining 5000 ft


2 minutes later, ATC says QNH 1015. You set 1015 and your indicated altitude moves up a bit.

What will the A/P do?

A) The A/P will climb
B) The A/P won't climb, it'll stay below the new altitude


Cheers,

|-|ardy


Not to confuse with ALT SEL. ALT HOLD doesn't refer to the value selected in the MCP altitude window.

[Dennis B]

I'd guess that we will stay where we are. Alt HOLD should refer to vertical speed only, isn't it?

[kOOk]

Once we did the test in cruise, on board a 737 classic. ALT HLD mode engaged on the FMA, we turned out our altimeters setting, up and down, ready to regain control of the aircraft. Nothing happened.

Apparently, the system takes a pressure reference at the time he reaches the desired level selected on the MCP or at the time the button ALT HLD is pushed, and then keeps it as his definitive reference to maintain.

I'll investigate on the shematics, 'cause it seems too simple to me

[Hardy Heinlin]

My current theory is this:

When pressing ALT HOLD, a barometric altitude reference will be set that is rounded to the next 100 feet increment. One can set the bug exactly on this reference thereafter. E.g. the reference won't be 5032 or 4985, but 5000.

Aircraft reference is the barometric altitude shown on the PFD (the exact ADC source depends on the current source selection).

F/D action on an increasing altitude difference:
- If the IRS V/S and the baro V/S agree, the F/D commands a prompt correction.
- If the IRS V/S is neutral and only the baro V/S is non-zero, the F/D commands a very slow correction.

Altitude flying under IFR is always a matter of bug flying and references rounded to 100 feet; the ALT HOLD mode supports this principle, I think. If ALT HOLD is annunciated and 5000 feet is maintained, it would be confusing if this mode would ignore long term baro changes and later on stay at 4930 feet or whatever. Instead, it will slowly tend back to 5000. Definition of "slow": Possibly as slow as a weather induced or flight route induced baro change.

Theory.


Cheers,

|-|ardy

[kOOk]

Today I'll do the check again.

If you read in the press this afternoon that a 737 fell into the mediterranean sea, you know why

[John Golin]

It's a chance we're willing to take...

[effte]

If I, as an aero engineer, had designed the ALT HOLD function, I'd have it latch onto the barometric pressure at the time of activation and then feed the difference  between barometric pressure at the time of activation and the current barometric pressure into the FLCS. This would make the aircraft follow the isobaric surface it was on when ALT HOLD was activated.

Browsing around on the net, I find sources such as these:

http://www.piac.com.pk/TrainingSection/PDFFiles/737/737-Operation%20Manual/Ch-4-Automatic%20Flight.pdf

which support this, stating that ALT HOLD "commands pitch to hold uncorrected barometric altitude at which switch was pressed".

Sorry Hardy, but I do not see any point in the autopilot attempting to be smart and holding any other altitude, or rounding anything off to present the pilot with a nice, even figure. In fact, automation trying to second guess the pilot's intentions is something I'd very much frown upon.

If I tell the aircraft "stay right here" and it decides that I probably mean for it to stay a couple of feet higher, as the number presented then has more trailing zeroes - I wouldn't be a happy camper!

In other words, if the above source is correct you can fiddle with the Kohlsmann all you want - the altimeter reading changes but the aircraft keeps doing what it was doing.

Edit: Philisophy, assumption and googling done. Now we only lack Wikipedia reference why we wait for kook to give the answer.

Cheers,
  Fred

[kOOk]

So, back from the test flight

Both in ALT HLD and in VNAV PTH, the aircraft doesn't move a foot if you change the altimeter setting (CPT or FO, we've checked both).

That is, the ship keeps a perfect level flight, even with an altimeter baro setting turned to show 500 feet above the initially selected altitude (let's say 35500ft for a FL350 level flight), and/or the other one 500 feet below...

On a 737.

 

[Jeroen D]

If I, as an aero engineer, had designed the ALT HOLD function, I'd have it latch onto the barometric pressure at the time of activation and then feed the difference between barometric pressure at the time of activation and the current barometric pressure into the FLCS. This would make the aircraft follow the isobaric surface it was on when ALT HOLD was activated.

Interesting. Couple of thoughts; If you're flying through a front, you could see some pretty rapid changes in barometric pressure and thus vertical movement of the plane following the isobar.

Also, I'm not sure if this would be in line with current rules/regulations/procedures. I'm being told to keep my barometric setting as accurate as possible. So when I do a cross country I keep listening to various weather services, be it a near ATIS, ASOS or other weather services and keep adjusting my altimeter accordingly. Every plane in the vicinity would be doing the same. Using the same sources as I am.  But all of these readings will be somewhat off the true barometric pressure, because of time lag in adjusting them for instance.

With your solution you could be further off, because you're the only one out there that has instant actual reading. Not sure if that would be a real life problem.

Jeroen

[effte]

When you are flying an altitude, you are flying the isobaric surface. The only times you switch isobaric surface is when you change altimeter setting or when you change your altitude. Both conscious acts, where you'd adjust your held altitude if in ALT HOLD.

If you use ALT HOLD when assigned an altitude by ATC, you're likely flying at that altitude with the correct altimeter setting anyway, so you'll simply capture that altitude - freeze your altimeter, so to speak.

If given a new QNH, I wouldn't want the aircraft to start wandering about as I dial it in. All changes in flight path should have to be acknowledged by the pilot, just as you have to acknowledge a change in the selected altitude on the A/P.

As for keeping the exact half-circle altitude in uncontrolled airspace, where there is no altitude separation enforced... well... that's the vertical equivalent of remaining on the exact airway centerline, and I consider SLOP a Good Thing(tm).

Cheers,
  Fred

[Jeroen Hoppenbrouwers]

You can rephrase this problem.

You are a 172 pilot, holding altitude on the last known regional QNH 1015 using the Mk II Eyeball. No A/P in this aircraft.

ATC broadcasts: "All stations, regional QNH now is 1014".

You reset your altimeter. The dial hands move a bit.

Now -- you need to do something to move the hands back to where they came from. THIS IS THE THING. You are holding your altitude properly. You need to initiate a vertical movement consciously to get the dial hand back in place. ALT HOLD does not do this. But mentally pushing your LVL CHG button does.


Jeroen
working on avionics as we speak -- hey, I need a few test flights once in a while   :-)

[Hardy Heinlin]

Sorry Hardy, but I do not see any point in the autopilot attempting to be smart and holding any other altitude, or rounding anything off to present the pilot with a nice, even figure. In fact, automation trying to second guess the pilot's intentions is something I'd very much frown upon.

If I tell the aircraft "stay right here" and it decides that I probably mean for it to stay a couple of feet higher, as the number presented then has more trailing zeroes - I wouldn't be a happy camper! :)

You convinced me promptly, Fred. I agree :-) My theory was false.

And the stuff with the slow and fast correction for IRS or baro probably affects selected altitudes only (as shown on the FMC or MCP).


Cheers,

|-|ardy


Edit: That 737 manual says ALT HOLD mode can be engaged also at the MCP altitude in which case the HLD light extinguishes. I think for this mode there must be a certain tolerance window for the system to decide whether the aircraft is currently at the MCP altitude or not. Perhaps here the 50 feet rounding is used, i.e. one half bug height. (Still talking about the 737 now.)

[Hardy Heinlin]

That is, the ship keeps a perfect level flight, even with an altimeter baro setting turned to show 500 feet above ...

Good to know. Thank you! -- One more question if I may: How long did you wait, how long did you keep the 500 feet offset approximately? For a couple of seconds or rather for a minute?


Cheers,

|-|ardy

[Hardy Heinlin]

Has anyone tried this in the full flight simulator? Engage V/S mode with 0 fpm set. The F/D wants to maintain 0 fpm and also altitude X since there is no climb/descend to do. Now disturb the path with the yoke. The F/D will want to regain 0 fpm. But will it also want to fly back to altitude X, or will it stay at the new altitude?

In other words, the V/S mode with 0 fpm might use the same algorithm as the ALT HOLD mode.


Cheers,

|-|ardy


Edit: I think V/S-0000 must be really the same mode as ALT HOLD. Imagine, in turbulence, without a fixed reference altitude, the aircraft would go up and down and stay wherever the wind is putting it. There would be no stability regarding non-climb and non-descend. You could also turn the F/D off and the result would be just as bad. I compare it with the ATT roll mode which you automatically get along with the V/S mode when turning the F/D on in flight. The ATT mode aims at a fixed roll angle, not at a roll velocity. When turbulence disturbs the roll angle, the F/D will not use the new roll angle but turn back to the initial roll angle. It does not fix a zero roll velocity, but the roll angle. Analog to this, V/S-0000 doesn't fix a zero altitude velocity but the altitude, I think.

[kOOk]

I would say 30 seconds. BUT I've also tried on the ground (by following a AMM test) and the FD stayed level thus...

Concerning the VS test, again I'll check in flight today (gently, poor passengers)

[effte]

Again, I'd expect it to do what it says on the tin and hold vertical speed. If I set vertical speed to 0 fpm, I don't want to see the aircraft lose speed by climbing at anything above 0 fpm - even if the altitude was changed by a downdraft. In ALT HOLD, you regulate towards a fixed constant barometric pressure. In V/S, you regulate towards a constant rate of change in barometric pressure.

Imagine hitting turbulence. You decide that it places undue stress on the airframe to maintain a fixed altitude, so you ask ATC for permission to drift up and down a bit which is granted. Now, what A/F mode do you enter? One option would certainly be V/S, with V/S set to 0000. In that situation, the aircraft doing anything but maintaining the set V/S, such as trying to maintain altitude, would be bad and certainly could come as a surprise to the pilots.

This is speculation/philosophy, of course, but based on sound engineering and human factors principles.

I'd compare it to heading hold vs wings level as found in some autoflight systems. One will bank you back towards a given heading, the other will just try to keep whatever heading there is.

Cheers,
  Fred

[Hardy Heinlin]

I would agree with you in case of significant drowndrafts. But what if we encounter just minor smooth fluctuations around +/- 20 feet? If the V/S 0000 mode accepts every tiny fluctuation, there would be no V/S control at all. It would be nearly the same as letting the aircraft fly freely in its current trim. (I don't think the natural phugoid effect would restabilize at a precision of +/- 20 feet and go back to the initial altitude; instead it would also just keep 0 fpm primarily for small fluctuations.)

Edit: Perhaps "pitch hold" would be a better mode name then. It choses a pitch that keeps 0 fpm in the average of the current conditions; fpm fluctuations are accepted and are secondary, while pitch hold is the primary goal.

Edit 2: In other words, if the fpm correction is fast, it can lose only little altitude. The one depends on the other. It's neither plain pitch hold nor plain fpm hold. If the fpm correction would happen in 0 time, the altitude would be fixed 100%. The parameters are interdependant. And in such a case, if the correction is fast and the altitude deviation is only a couple of feet, I wonder if the mode really doesn't use a small altitude window as a reference as long as it is not extremely exceeded. Errors wouldn't accumulate so easily then.


|-|

[effte]

V/S is intended for maintaining a vertical speed. If you want to maintain an exact altitude, there are other A/F modes. In normal operation, set it to 500 fpm and it'll be fine if it's between 400 and 600 - so that's doing the intended job. If you try to use it as a faux altitude hold, I think you will have to accept random influences causing some Brownian motion, or even climb/descent through built-in or externally caused hysteresis. It can't be pitch hold, as that would mean the V/S would vary with up- and downdrafts.

Cheers,
  Fred

[Hardy Heinlin]

Well, your suggestions make my programming work only easier :-)

Good! :-)


Cheers,

|-|ardy

[John H Watson]

More food for thought:

http://www.pprune.org/tech-log/398441-inertial-altitude-boeing.html