AF447 final crew conversation - Thread No. 2
 

Thanks for the link, John.

HN39;

I won't forget our early exchange on stall AoA and later also with Owain Glyndwr the effects of Mach on AoA and buffet thresholds. The discussion had a profound effect on my understanding of my airplane and of high altitude-high speed flight and AF447 specifically. I would recommend to anyone wanting to know more from those that really do know their stuff, to find this and other exchanges with these two posters.

Re, "...would it help prevent future accidents if sophisticated Flight Guidance and Aircraft Monitoring systems, having rejected three sources of airspeed information as unreliable, and 'knowing' conditions such as configuration, weight and altitude, provide more specific information as to the appropriate response?"

I need to think about it! ;-) Along with weight, altitude & configuration it would use AoA and perhaps also have a running historical window, say the last five or ten minutes, of all relevant indications upon which to build a "safe envelope"? Just thinking out loud.

Initially I think, Yes, it would help, but in the Birgenair and Aeroperu cases more than here. Reason I say this is, most of the Airbus UAS events lost airspeed indication for less than two minutes and from what can be gleaned from the Interim Report 2, those crews affected didn't seem to do anything but wait.

Because pilots are creatures of visual, auditory and tactile habit, a switching to a "novel" presentation may present as much surprise as the event itself and take a moment to get used to...which means it also has to be trained in the sim, which available time these days is heavily-challenged already.

But the Birgenair/Aeroperu cases are completely different and I think such a system/presentation would have saved the aircraft.

Organfreak;

I didn't watch the BBC series, preferring the BEA work plus some discussions on the side. I prefer original sources to anything that will perhaps innocently but most assuredly have a point of view and a theory and must keep an audience entertained to be economically viable. I avoid the "Discovery Channel" kind of stuff for this reason. The needs of television are not those of good investigations because such work is enormously detailed and for most, tedious and certainly would not entertain. "The story" is certainly interesting and I can see where those not familiar with the event would enjoy learning about it though and getting enough from such programming to ask further questions.

Not sure if the BBC was showing Vasquez's work, but it showed up early here and was an enormous help in understanding the weather aircraft transiting the ITCZ deal with on a regular basis.

Hi,

Well .. here's yet another has appeared in a few days .. and in which journalists claim to show how the accident really happened :ok:
The final report of the BEA about what happened in advance ... :)
Google Traduction

Pièces à conviction revient en mars sur le crash du Vol AF 447 Rio / Paris - Le zapping du PAF

I must admit I was impressed by Owain Glyndwr's well thought out dissertation on THS v Elevator, AoA and Mach buffet, and some time ago took the liberty of putting a couple of his posts together as a webpage.

THS v Elevator and AoA

Thank you..........

Tv, Polacko and Otelli storytelling
 

That for are we Ppruners !

It might be esoteric, but what I got from Owain was two behaviours. In the long term, the a/c has an all flying tailplane. In the short term, the a/c has a standard PITCH control, one that degrades to "neutral" over time. The aircraft is always seeking cruise, (or attitude: establish), and this conflicts with the need to maneuver, at times.

Which could lead to confusion, unless the configuration (instant) can be sussed visually (screen), or by feel? "What's it seeking?"

Why would this a/c seek to "establish" and hold an attitude that required constant attention from the handling Pilot? An attitude that was intended to be short term, perhaps? PF had the ROLL under control because he knew what was going on, albeit badly. Right, wrong, or sideways, he did NOT get what the PITCH required. This speaks to the a/c's programmimg in ALII. For whatever cause, the handling pilot was unable to 'get'. Is the a/c always correct?

Just a question.

Lyman:
In ALT2???

:suspect: :sad: := :( :hmm:

Lyman

Quote:
[what I got from Owain was two behaviours]
unquote

Wherever you got that, it wasn't from me!

Even in Alt2 the C* law is a manoeuvre command system for goodness sake. The control laws don't 'seek' anything. They hold what's there until told by the pilot to change it. There is no conflict except in your mind.

Owain.

If from the initial PITCH up command, the THS had been trimming the command, the a/c would have STALLED well before the top of climb, where there was no energy left, and the STALL was unfelt, unobserved (per your comment).

Had it not remained at -3.4, then, the a/c would have fallen out, buffeted, and the two man crew would be left with inescapable evidence that STALL recovery was a necessity. Point of fact, the a/c would have recovered on her own had she been allowed to STALL. Instead, the a/c drained energy until at full power, when it mushed into an attitude that was to remain until water contact.

After the loss of energy, the THS then started to trim in NU, and remained at 13.2 (-) until water contact. It stands to reason A) you are not wrong, and B), the lack of SA, CUES, and CRM prevented the crew from sussing the STALL.

No judgment, no accusation. It remains that had the a/c STALLED earlier, the cues would have been present and the outcome may have been different. No?

Transient in Feedback Systems
 

A33Zab

Will post on the subject applied to:

1) Sub heated Pitot's
2) RH not recorded

I am elaborating and will post ASAP.

AF447
 

Lyman,

I'll give it one go then I'm back into hibernation.

[quote] If from the initial PITCH up command, the THS had been trimming the command, the a/c would have STALLED well before the top of climb, where there was no energy left, and the STALL was unfelt, unobserved (per your comment). [unquote]

You keep writing as if the THS has a life of its own. It hasn't. It is slaved to the elevator through an integral term (see the block diagram PJ2 posted a while back). It WAS trimming in the initial phase of the second stall approach - it moved from around 2.5 deg to around 3.4.

It matters not one jot whether the pitching moment comes from elevator, THS or some combination thereof. The aircraft responds to the total pitching moment commanded by the pilot and there is absolutely no evidence to the contrary. That being so, the THS movement would have no effect on the point at which the aircraft stalled - it was merely sharing out the commanded pitching moment to minimise elevator movement.The time at which stall occurred was driven purely by pilot input as a commanded 'g' history.

I did NOT say the stall was unfelt or unobserved. I said that it was difficult to distinguish between turbulence buffet and stall buffet and that there were no pitching moment breaks. As a nonpilot writing in a professional pilot's forum I try to avoid sticking my neck out where it can get chopped off, so I was careful to say that the other clues indicating stall (high attitude, inability to arrest rate of descent) were best left to pilots to evaluate.

I did not add of course that there was this little matter of a voice calling STALL STALL every five seconds or so.

All in all I cannot accept that the stalled condition was unrecognisable.

[quote] Had it not remained at -3.4, then, the a/c would have fallen out, buffeted, and the two man crew would be left with inescapable evidence that STALL recovery was a necessity. Point of fact, the a/c would have recovered on her own had she been allowed to STALL. Instead, the a/c drained energy until at full power, when it mushed into an attitude that was to remain until water contact.[unquote]

Since stall AoA is independent of THS setting, your argument won't hold. The aircraft WAS allowed to stall and it did buffet. Just allowing the aircraft to stall won't get you out of it unless you remove the pitching moment that is holding you in there. This didn't happen.

[quote] It stands to reason A) you are not wrong, [unquote]

True ;)

[quote] and B), the lack of SA, CUES, and CRM prevented the crew from sussing the STALL. [unquote]

Also truehttp://images.ibsrv.net/ibsrv/res/sr...y_dog_eyes.gif

[quote]No judgment, no accusation. It remains that had the a/c STALLED earlier, the cues would have been present and the outcome may have been different. No?[unquote]

No

THS
 

Hi,

Owain Glyndwr,

In your opinion what kind of improvement(s)* could and should be made to the aircraft Systems in order to effectively "help" the crew in similar situations in respect to THS operation.

Kudos for your analysis.

Rgds

(*) Algorithms, indications, etc.

"DON'T PULL" flashing on the PFD in time with the stall warning?

Transient in Feedback Systems (I)
 

Hi,

Feedback loops are designed (and potentially capable) to improve a System.

For example, the AS probes in auto mode receives just the required electric power.

Feedback loops are designed with "time constants" and eventually with techniques like hysteresis. The engineering takes into account essential aspects like stability, etc.

I posted some comments assertively saying the AS probes the industry is still using, are "subheated".

I am based in several facts and i can discuss each one when necessary.

I have some questions to the Pilots (daily facing UAS threaths) and to the Technicians here:

1) Is the power being applied to the AS probes enough in order to their proper operation?
2) The delay between a sudden atmospheric temperature drop and the "reheating" of the probes could explain the "failures" leading to UAS?
3) The differences between US and FR probes may be explained by "different" time constants?
4) Is the max power sufficient to provide a safe operation in most environments?

Is there "recording capability" in current Systems in order to analyze the UAS incidents more precisely?

Could be better to enter WX using MAX instead of AUTO? (Operating "open loop")?

RR NDB

good start but to better understand what questions are applicable first search on High altitude "ice crystals" in aviation

I suspect that you may better understand the time differentials between sensing and applying enough heat.

Note that you car windscreen will still ice up and then shed while the defroster is on high, as the sleet density vs your speed changes rapidly.

Time constant
 

lomapaseo,

i suspect US probe is "faster" to adapt to a "new environment".

The max power is the same to both probes.

Anyway i will go deep in this subject as you suggested.

IMHO the System should be STABLE under "brief failures". Like AS probes anomalies.

K.I.S.S.
 

Hi,

Quote:

---

I suspect that you may better understand the time differentials between sensing and applying enough heat.

---

In AUTO mode you will have a delay from "sensing" to "correct" to the required probe temperature.

if you apply maximum available power, ("open loop condition") no delay at all :)

At which cost? Roughly 1 KW being used, i estimate.

And the environment is frequent, i understand.

Are you saying is difficult to decide (System or crew) when apply max power?

Not sure that I understand the question. If the amount of heat is the problem, then it was obviously not enough to prevent blockage of the pitots in the particular conditions encountered by AF447. However, as stated by PJ2's #1420 in thread no.1: The recorded data show the opposite of a temperature drop. At 02:09:30 the temperature was ISA +10.6°C, trending downwards (Interim #3, p.48), and at 02:10:00 it was ISA +15.6°C (p.86).

1st question
 

Hi,

HazelNuts39,

Quote:

---

Not sure that I understand the question.

---

Is the power being applied to the AS probes enough in order to their proper operation?

The power UNDER STEADY STATE CONDITIONS (operating at MAX POWER) is enough?

Operating at MAX power there is no delay, i.e. you are in a "open loop" mode wasting energy and stressing materials when not necessary.

I guess the max available power is enough. (Just guessing).

And i suspect the amount of heat are not being sufficient when encountering ice. This "encountering" was mentioned to be heard on CVR.

I took into account PJ2 post when formulating all 6 questions,

I am talking in the transient, just after noise increased in CVR due ice crystals "collisions" to windshield an nose cone surfaces.

I also took into account the recorded temperature, (higher than expected and imposing a FL375 REC MAX) not relevant to my rationale. I am looking to the transient, the so called t0+, just after encountering ice crystals when the heat MUST BE PRESENT immediately to assure proper AS sensors operation.

Pitot heating was in auto or manual mode?
 

Hi,

F-GZCP was operating with Pitot heating in which mode? Auto or manual mode?

This was mentioned in BEA reports on AF447 case?

FDR records enough data on this issue?

In other 32 UAS incidents Auto mode was selected, as recommended (default)?

Most crews use Auto or manual mode when closing in specific WX conditions?

The timing required for AP/ATHR quit is "compatible" with "heating time constant" of probes when encountering ice crystals when auto mode is selected?

PS

If the temperature BEFORE encountering ice crystal is higher (as in this case) WORSE.

The "required heat" will takes longer to set in the probes.

Quote:

---

The recorded data show the opposite of a temperature drop. At 02:09:30 the temperature was ISA +10.6°C, trending downwards (Interim #3, p.48), and at 02:10:00 it was ISA +15.6°C (p.86).

---

Therefore a fact in that flight.

RR NDB;

Where do you see a drop in temperature? "The background noise changed rapidly around 2 h 09 min 46." (p.73). The ice crystals have the same temperature as the air that carries them. Several UAS reports* notice a sudden rise in temperature, which is to be expected because the rising air in which the ice crystals are formed derives its buoyancy from being warmer than its surroundings. See also Tim Vasquez' temperature profile at FL 350.

(*)From Interim #2 p.52:

@RR_NDB
 

You will locally heat-treat the neighboring skin with insufficient cooling.

I don't have exact data for the probe, but AOA operating temp is 120 ºC,
suppose this will be the same for the probe.
---

Don't have sleepless nights about the probes anymore.
The affected thales probes are obsolute and replaced by other probes per AD 2009-0195.

Quote:

---

After 07 January 2010, do not install a Thales Avionics P/N

---

System is automatic, on ground 115V to the Low side if 1 engine is in run state.
115V is applied to the Normal side in Flight.
Manual switch (from AUTO to ON) bypasses the Engine Run & Flight logic.
Temp controlled by the probe itself, the PHC contains the switching logic and monitoring circuits.

HN39 questions and A33Zab comments
 

Hi,

HN39,

Will process pertinent information and review my books on the subject before. Thanks for questioning. Will answer ASAP.

A33Zab,

Any other implication if we cross the specific WX area in MANUAL mode?

:)

Question for you:

Time constant of the "closed loop" configuration when AUTO mode is selected for Pitot heating seems relevant to be take into account in the 30+ UAS cases?

What is your feeling on this particular point.

I remembered of this possibility when preparing second post on "Transient": Transient in Feedback Systems (II)

Regarding FD data just after UAS (in the transitory)

Related to "What PF saw at his (RH) side", not recorded.

THS changes
 

RR NDB

Quote: [In your opinion what kind of improvement(s)* could and should be made to the aircraft Systems in order to effectively "help" the crew in similar situations in respect to THS operation.]

Nothing other than the obvious I'm afraid.

In normal law any further THS nose up movement is inhibited when alpha protect switches in. I can't for the life of me see why this logic was not carried over to Alternate and stall warning respectively.

Apply this simple change and I suspect that most of the other controversial points discussed in earlier threads decline into insignificance.

Pitot anti icing
 

Don't you think that the problem might be both more subtle and more complex than just the amount of heat applied?

Why did it seem to be almost limited to one particular design of probe? (well by about 10:1 anyway)

What was peculiar about that probe? It must I think be critically dependent on the shape and finish of the probe internal geometry which is a closed book to all except the probe manufacturers.

It wasn't blockage of the drainhole by ice because the sensed pressure dropped to ambient static; so the blockage must have been upstream of that. Did the crystals melt and refreeze? Are the internal diameters critical? Or the radius at the corner of the 'L'? Or the actual location of the drainhole? Would the distribution of the heating along the length of the 'horizontal' bit of tube be important? When they went to the AA probe Thales also made some changes to the surface finish to improve corrosion resistance. Was this important?

I don't know - does anybody else?

Pitot anti icing and THS changes
 

Hi,

Owain Glyndwr

Don't you think that the problem might be both more subtle and more complex than just the amount of heat applied?

Certainly is for both FR and US probes

Both probes showed limitations. My feeling is:

FR, probably many factors, like you put (geometry, etc)

US, less critical and probably requiring a little bit more heat when facing conditions near limit.

Look, i am trying to simplify (not always possible) because the AS probe principle of operation is VERY SIMPLE. What you need:

-A "certain geometry" (Henri Pitot invention)

-Heating above a certain threshold when facing conditions near limit.

-Other characteristics derived from the environment (moisture, heat, need to drain, etc.)

The short duration of the "failures" suggest that:

The heating may be near the required amount or not being applied fast enough due thermal inertia due geometry, geometry of the "heater" or a combination of aspects related to thermal characteristics.

I am anxious to analyze deeply the 30+ UAS cases to look for:

AUTO or ON mode selected and recorded (consistent) data during the transitory.

I understand the valid information is not recorded during the incidents like AS SIMPLY not measured in AF447.

My feeling is, we are near the required heat. Statistically based and considering AS probes (simple devices) did not fail. Just a brief erratic output.

Good questions. I don't know.

Thank you for direct answer!

Rgds,

@RR_NDB:
 

No, switch to ON bypasses the auto logic but is the same circuit as in flight and at least 1 eng in RUN state.

The figure explains the function of switch and logic:

http://i474.photobucket.com/albums/r.../ProbeHeat.jpg

Thank you for the schematic. Looks like the probe temperature of the probe is controlled by the actual heater coil itself which is probably made like a an old style wire wound ballast resistor. The temperature coefficient of resistance would then regulate the current & heat. The PHC is just switching the units between On and Off, ground and air mode, and also monitoring the current to be sure each heater is actually working and has not burned open.

I have an old Rosemount pitot probe on my desk, and although these things look fairly uniform between models, there is all sorts of room for variation as to how thick the metal is in various areas (controlling the thermal mass and time constant to conduct heat to the other side), input wattage, heating coil location, what metal is used, and where the drain holes are relative to the elbow and heat sources.

I would think they would avoid putting the heating coils where moisture can actually impinge since you would then be generating steam and pressurizing the probe.

The 3 channel design of the Airbus does provide sensible redundancy for everything downstream of the probes, but no real redundancy for a common mode sensing problem. A better design would involve alternative sensors such as a laser airspeed sensor (but I note that several websites touting such equipment a couple of years ago are now out of service).

Another concept would be to place the probe behind a vane that would bend the airflow into the pitot inlet, but which would centrifugally separate ice particles so that they would miss the inlet for reasonable yaw angles under cruise conditions.

Amen to that.
It is if someone in conceptual design had decided that the AOA systems were not to be trusted.

Fully in agreement. :ok:

[Certainly is for both FR and US probes]

I hate to confess ignorance, but FR probes? US probes? Either side of the Atlantic or something else?

[Both probes showed limitations.]

Agreed, but one design (from a maker with a previously good record) starts to show problems after being in service for what, seven years? I am not aware of any changes to route structure and if you exclude global warming :) the atmosphere hasn't changed much, so what did? I don't think one can rule out some sort of time dependent change in the physical state of the probes.

[Look, i am trying to simplify (not always possible) because the AS probe principle of operation is VERY SIMPLE.]

Again agreed, but as always the devil is in the detail

[The short duration of the "failures" suggest that:

The heating may be near the required amount or not being applied fast enough due thermal inertia due geometry, geometry of the "heater" or a combination of aspects related to thermal characteristics.]

Again agreed, and Machinbird's listing of the possible apparently minor variations in probe design is exactly what I had in mind when I suggested the answer might be too subtle for outsider analysis (but we still try!)

Like RR NDB, I feel that the heat being applied is probably near what is required, since the UAS events have been transitory, but it may not have been applied in the right area. I have heard of problems on other aircraft where pitot icing was linked to the absence of an insulating washer between probe and fuselage. Extrapolating that thought, the fuselage (and the unheated parts of the probes) constitute a thermal mass that would be at much lower temperature than the heated part of the probe. There will be a temperature gradient along the probe and maybe the details of that matter.

I must say that this is one of the more sensible discussions on this threadhttp://images.ibsrv.net/ibsrv/res/sr...lies/wink2.gif

Machinbird
[It is if someone in conceptual design had decided that the AOA systems were not to be trusted. ]

Which is not consistent with the exhortations to "Believe the stall warning". If AoA is good enough for triggering stall warning it is good enough to freeze the THS.

PS - could somebody please remind me how to put quotes in those little blue boxes?

Hi Owain Glyndwr,

Good to read you back again. The technical solution to the pitot icing problem has probably been solved already.

I'm more bothered about a technically qualified crew who didn't know what attitude they needed to maintain FL 350 after they lost the FD guidance.

Thanks RRR

Very probably - wouldn't it be nice to know how ;) But at least they now have definitive icing requirements that cover the problem altitude/temperature bands

Quite!

SPECIFICATIONS
 

Thank you !!! I like to see finally inside of inside !

Specifications :ok: :ok:are often written only when the project is finished ! It is a good thing, that means that we can cross out, delete, change what is wrong in the "philosophy" of the aircraft : We don't fly "philosophy":}. We don't fly ideology. :}

My key takeaway from A33Zab's schematic is that there is no direct feedback from any of the probes controlling the power level. Temperature control is reactive rather than proactive. If there were temperature feedback sensors in the probe, it would be possible to more finely control the probe power input. That is where the next big area for improvement in Pitot tube design/performance lies.
All the PHC does is supply 115 V to the Flight or Ground circuits of the Probe heater . There is no High or Low setting as some may understand the concept.

SPECIFICATIONS
 

It proves Airbus is K.I.S.S.!

The switch needs only to be touched to ON in de-icing like PJ2 stated and if dual or all PITOT HEAT FAIL message is shown.

Uhhh...isn't that what is inherent to feedback controls? You need a certain deviation to correct the situation.
(we are awaiting RR_NDB Transient in Feedback Systems (II))

Proactive......than you call it predictive! or not?
if any possible UAS is predictive, wouldn't it be better to DONT GO THERE?

I don't have the pitot internals by hand but the low heat (ground) setting is to protect the A/C skin when less cooling is available.

OK, Looks like I need to explain the situation better.

One form of current stabilization (old fashioned now but I have used it in equipment) is the old fashioned wire wound ballast resistor. As the current goes up, the temperature of the wire windings goes up, the resistance of the specially selected wire increases and limits the current. These things generally operate at a dull red glow. They are likely using this type system internally for pitot heat stabilization/regulation.

No moving parts, no electrical connection other than normal regulated voltage power supply, and you get a heat source. Properly selected, it is a simple means of generating heat for situations that are entirely predictable and slow changing. The aircraft climbs to altitude, the probe cools a bit, the wire cools a bit and the current increases which increases the power input. KISS to the extreme. To increase the power further, you need to increase the voltage.

The problem occurs when the heat demand changes rapidly. In the case of AF447, we had 19 seconds between the sound of ice crystals impinging and the AP disconnect. It took some time for the ice crystals to cool the probe internal surface and then the heating coils enough so that the power would then increase and melt the ice. When overloaded with ice crystals, this type probe goes through cycles of icing, melting, purging the melt water, and icing again.

If the icing had been detected in the first couple of seconds by a thermocouple, the power could have been cranked up then to meet the new demand before the blockage occurred, but this means that probe heating would be solely under the control of an external control system using feedback to sense and defeat icing. The dynamic control range would be much higher.

This is my take on what is likely the means of pitot tube heat control based on the evidence of the schematic you posted A33Zab. The externally controlled feedback system is probably what is needed to fix the problem.

If the ice particles are like snow flakes, they may not conduct heat very well between the heated wall of the tube and the space within.

P.S.
How much heat is required to melt 1.6 grammes of ice at -40°C? (Tube opening diameter 10 mm, 20 s at TAS 250 m/s, TWC 4 gm^-3 (NPA 2011-03), 100% catchment efficiency)

Isn't the challenge to melt XX grammes per second ?

The meltal temp may be above ice-melting but the 3rd or fourth layer of ice crystals hasn't begun to feel it yet.

My grand daddy use to tell me about sleeping in an unheated drafty room on a cold snowy night with layers of blankets to keep him warm, yet the snow flakes would still glisten on the outside layer

Hi HazelNuts39,

Assume snow requires 4.18 J/(g•K); & Latent heat of fusion of ice is 333.55 J/g

Then 1.66gm ice at -40 would require 4.18*1.66*40 = 277 J to be raised to 0C as ice.
Plus 1.66*333.55 J to melt at 0C = 554 J.
Total = 831 J.

If that amount of snow could accumulate in 20secs, then in 1 hour you would need 149,580 J or extra power source of about .04 KWh.

It doesn't sound much - but I can't spot the error.

I'm with lomapaseo. The problem is rate, not temp. Tightly packed granular microcrystalline water ICE will not conduct heat well. Just as a solid crystal of quartz is a nice heat sink, sand is not. Besides, the crystals only melt to form ICE, they have no exit. So the problem is to expel the 'pack', not prolong the obstruction by trying to 'melt' it? Heating the Stainless steel is a waste of energy.

Some way to expel the slug of compressed particles? compressed air, forward?

A snow cave is great insulation against a bitter exterior, as loma has said. Heat is slow to escape, why would it be quick to enter? Some new approach is needed, it is taking too long to fly without Pitot tubes?