Habsheim

[Owain Glyndwr]

Hi Hazelnuts

You may be right in that, but I'd be willing to bet that the acceleration information is passed through a low pass filter before use in that way, if only to eliminate nuisance signals due to noise. This is going to slug the action I think.

[Chris Scott]

AoA data

Re your interest in the AI report of 1995, I only have a PDF that someone kindly sent me. If you PM me with an e-mail address, I will send it to you.

I see that you and AZR have been using the last (data) column of Tome 1 for "INC" (AoA) data.

There are other "INCIDENCE" (AoA) data in Tome 6. On the line for each second are recorded two values, each to one place of decimals; unlike the single column you are using. It is unclear to me at the moment whether the 2 values are from different sources (e.g., FAC1 and FAC2), or if they represent 2 samples per second from a single source. I favour the former explanation, because next to them are twin values of what appear to be sideslip, which is another parameter handled by the FACs.

Using UTC 12:45:39** as the "t" reference, here are the decoded AoA values, from Tomes 1 and 6 respectively, for the last 20 seconds of flight:

t -19 +08 U08.3 U07.2
t -18 +08 U07.6 U08.2
t -17 +09 U09.3 U09.3
t -16 +09 U09.0 U08.9
t -15 +08 U08.4 U07.9
t -14 +09 U08.9 U10.6
t -13 +11 U11.3 U10.8
t -12 +12 U11.6 U11.5
t -11 +11 U10.9 U10.1
t -10 +10 U09.7 U11.4
t -09 +12 U12.1 U13.1
t -08 +14 U13.5 U13.9
t -07 +14 U13.9 U13.1
t -06 +13 U13.1 U12.7
t -05 +14 U14.4 U14.4
t -04 +15 U15.3 U14.4
t -03 +14 U14.3 U13.8
t -02 +14 U13.8 U15.6
t -01 +15 U15.1 U14.9
t -00 +15 U15.3 D73.8
t +01 -00 D00.2 D00.2

** (i.e., elapsed time 335.0 seconds in the second column)

Can anyone inform us of the precise sources of the above AoAs, and if the Tome 6 pairs of data are sequential, or from two different sources simultaneously?

HN39,
I agree that the sequential times in seconds in column 2 are elapsed times from an arbitrary point, which may be the selection of TOGA thrust at the start of the T/O run. (The report states that the a/c was airborne at "12:41", and that the R/H turn was commenced during the second which followed.)
Perhaps the BEA could supply us with pages 1 - 6 of the DFDR print-outs...

Quote from Owain Glyndwr:
I must admit that the debate has been to a much higher standard than I feared it might be.

Praise indeed...

[HazelNuts39]

Hi Chris,

thanks for the good work. I support your thought that the two columns in Tome 6 probably are from two sources. The values in the first column rounded to the nearest degree correspond to the value in Tome 1.

It is now also apparent what the 1234 in the last column represent. Apparently 4 subframes form one full frame. That permits the number of parameters to be increased, by using a single location in a subframe for multiple alternating parameters, e.g. A and B each sampled every other second, or A, B, C and D once every 4 seconds.

[AlphaZuluRomeo]

Originally Posted by rudderrudderrat

Hi Owain Glyndwr,

Originally Posted by Owain Glyndwr

increasing AoA is not going to give an increase in flight path angle unless accompanied by enough thrust.

I don't quite understand that statement.

Hi,

As I'm not sure many of the readers of this thread can read French, allow me to quote and (try to) translate some relevant sentences from the BEA report:

(end of § 1.16.1.4.2) :

Les autres exploitations effectuées montrent par ailleurs que si la commande de profondeur avait été amenée plus rapidement ou plus tôt en arrière lors du vol de l'accident, une incidence supérieure à 15° aurait pu être obtenue avant l'impact sur les arbres. Elles montrent également que sans une remise de gaz plus précoce, une telle manœuvre, qui aurait conduit à augmenter la trainée instantanée de l'avion, n'aurait pas permis d'éviter l'impact avec les arbres.

Other [data-analysis] operations performed also show that if the sidestick had been brought faster or earlier back during the accident flight, an AoA greater than 15° would have been obtained before the impact on the trees. They also show that without an earlier go around [understand: advancing the throttle levers earlier], such a maneuver, which led to increase the instantaneous drag of the aircraft would not have enabled to avoid impact with the trees.


Also noteworthy:

§ 1.16.1.3 (wrongly labelled 1.16.1.13 in the BEA report/PDF) "Expertises effectuées"

Le fonctionnement normal des commandes de vol n'a jamais été mis en doute par l'équipage de l'A320 accidenté. La commission d'enquête a cependant jugé nécessaire de vérifier ce point de façon approfondie.

The proper functioning of the flight controls has never been questioned by the crew of the accidented A320. The Board of Inquiry, however, found it necessary to check this thoroughly.


§ 1.16.2 "Fonctionnement des moteurs"

Contrairement au cas des commandes de vol, le fonctionnement des moteurs a été mis en cause par l'équipage immédiatement après l'accident : il a en effet déclaré qu'après la remise de gaz, la poussée des moteurs ne s'était pas réalisée.
Dès les premières exploitations des enregistreurs (C.V.R. et D.F.D.R.), il a été toutefois établi que les moteurs avaient répondu à la commande de remise de gaz effectuée entre 5 et 5,5 secondes avant l'impact sur les arbres

In contrast to the flight controls case, engine performance has been questioned by the crew immediately after the accident; the crew indeed declared that after advancing throttle levers, engine thrust had not been delivered.
However, it was established from the early readings of the recorders (CVR and DFDR), that the engines had responded to the request of go around thrust which was performed between 5 and 5.5 seconds before the impact on the trees

Hope this helps.

[Chris Scott]

Thanks AZR,

I think some of us are enjoying deliberately delaying looking at the BEA's conclusions !

The crew's perception that the two CFM 56-5-A1s were slow to spool up probably indicates more about the effects on their metabolic-rate of the urgency of the situation they suddenly recognised than it does about the performance of the engines.

Why would the two engines, controlled independently by their respective FADECs, accelerate slower than normal; but in perfect unison?

[Owain Glyndwr]

Yes, thanks AZR, that was very helpful.

I would translate it slightly more positively, but essentially those BEA remarks are consistent with what I was trying to say.

Other calculations show that if the sidestick had been brought back faster or earlier during the accident flight, an AoA greater than 15° would have been obtained before the impact on the trees. They also show that without an earlier thrust increase such a manoeuvre, which would have led to an increase in the instantaneous drag of the aircraft would not have permitted (the aircraft) to avoid impact with the trees.

[HazelNuts39]

Originally Posted by Conf iture @#11

The plane refused to deliver alpha max, it kept 2.5 deg short of it.

[Owain Glyndwr]

The quality of the traces in the published BEA report is so poor that I imagine many people would like to see something better. This is an extract from that AI report cited by Chris Scott

[Owain Glyndwr]

and this is a somewhat lower quality record of the Airbus flight test replication of Habsheim.


The crosses are, I think, Habsheim actuals. The AOA trace shows a similar restricted AOA gain to Habsheim over the first few seconds, but goes on to achieve Alphamax a few seconds later.

[Chris Scott]

HN39's Graphs

Good evening HN39,

That's a super graph! I was going to ask you which sources you chose for the Alpha1 and Alpha2 curves, and I see that you have now averaged the two values into one curve.

However, although it doesn't affect the overall picture radically, I remain unhappy with the apparent one-second discrepancy between the time bases you and I have been using. (See also my posts of Dec18/0021z and Dec18/2008z.)

In the text accompanying your first graph, you quoted the BEA as follows:
"...the report puts t=0 at 12:45:39 and states that the airplane entered alpha protection mode at 12:45:34 (t-5) at 122 kIAS and alpha=13 degrees."

Why would Alpha-Prot have engaged at an AoA of +13, i.e., 1.5 degrees early? As previously mentioned, the only reference I've found so far is as follows (1.16.1.2):
"A t - 4 s, commutation sur la loi de pilotage en incidence, la valeur de 14.5 deg ayant ete atteinte, cette loi etant ensuite conservee."
[At t - 4 s, switching into the angle-of-attack law, the value of 14.5 deg having been attained, this law being maintained thereafter.]

It also defines "t" as "GMT" 12:45:39; as you have stated. My interpretation of the DFDR print-out is that 12:45:39 is paired with an elapsed time in seconds of 335.0 - not the "334" which you state.

Referring to the first of the many lines annotated "1245", which must represent 12:45:00, the elapsed time is 296.0. So 12:45:30 would be 326.0, and 12:45:39 would be 335.0. That is the last line on each DFDR "Tome", page 8. (BTW, the elapsed times continue at the top of Page 9, but - curiously - the associated GMT times are erroneous.)

So, using your time base, where "t" (12:45:39) is 334.0, t -5 is 329.0, at which the AoA on Tome 1 is indeed +13 (and the Tome 6 values U13.1 and U12.7).

If I am right, however, the time-vs-AoA figures are as follows:
t -06 12:45:33 (329.0) +13 U13.1 U12.7 (Landing mode of Normal Law)
t -05 12:45:34 (330.0) +14 U14.4 U14.4 (Landing mode of Normal Law)
t -04 12:45:35 (331.0) +15 U15.3 U14.4 (Alpha Prot)
t -03 12:45:36 (332.0) +14 U14.3 U13.8 (Alpha Prot)
t -02 12:45:37 (333.0) +14 U13.8 U15.6 (Alpha Prot)
t -01 12:45:38 (334.0) +15 U15.1 U14.9 (Alpha Prot)
t -00 12:45:39 (335.0) +15 U15.3 D73.8 (Alpha Prot)
t +01 12:45:40 (336.0) -00 D00.2 D00.2

I suggest that the numbers annotated on your time axis need to be moved one space to the right. Also, the first two values of "AlphaCmd" need to be deleted, because Alpha Prot did not start until t -4.

[HazelNuts39]

Hi Chris,

In reply to your post#107 I already admitted in post #109 that I was in error for the time of entry into alpha-protection mode, which is not t-5 but is t-4. I'm on the move right now but will remove the AlphaCmd point at t-5 as soon as I have retrieved my laptop from my luggage.

At 12:45:34 airspeed is122 kIAS and Alpha is 13 degrees is recorded at 329.0s
Five seconds later at 12:45:39 t=0 so that must be at 329.0s + 5 = 334.0s

[AlphaZuluRomeo]

Hi Owain Glyndwr,
Thx for your comment (the same goes to you, Chris Scott)

Your translation is better than mine, I agree

[Chris Scott]

Hi HN39,

Sorry to catch you "on the hoof"! When you have time to have a closer look, our one-second discrepancy in timings is identified in your answer.

Quote from your response (my emphasis):
"At 12:45:34 airspeed is122 kIAS and Alpha is 13 degrees is recorded at 329.0s"

My interpretation is that 329.0s = 12:45:33 (see explanation offered in my previous post). But I have little experience in DFDR readings.

Quote:
"Five seconds later at 12:45:39 t=0 so that must be at 329.0s + 5 = 334.0s"

According to my interpretation, that should read "six seconds later". As you can see in my previous post, I have the AoA datum of +13 at t -6.

PS
I now understand what you meant about all the parameters in a one-second time frame not necessarily being recorded at the same instant, but possibly sequentially during the whole second. Do you or anyone else have any news on that?

[awblain]

Re: rudderrat - exchange speed for height

You're right. They could have exchanged another 5kt (2.5m/s) of airspeed, from 110 to 105 (~55m/s). for height h (in m) ~ v.Delta-v/g ~ 55(2.5)/10 ~ 14m.

The key question is… did they have that 5kt of airspeed to exchange (FBW doesn't change the physics), while avoiding a stall.

There was also an interesting point about the motion of the aircraft when the elevator changes position - there is indeed an immediate tail-down pitch acceleration, about the center of mass, plus a linear downward acceleration, due to an increase in negative lift from the horizontal stabilizer... the change to the attitude then gives an upward acceleration from the larger, increasing wing lift, traded for forward speed. These changes in response to an instant movement of the elevator would extend over ~ a second.

[HazelNuts39]

Hi Chris,

My timeline is based on the information given in the report for 12:45:34 and 12:45:39. That would put frame no. 296 at 12:45:01.

While I agree that one would expect the first 1245 frame at 12:45:00, that doesn't fit with the two points mentioned in the report. Maybe BEA's GMT's are one second off, but who cares? What matters are the seconds to some reference time in the recording. The report on page 10 also puts the impact with the trees between 12:45:39 and 12:45:40. The data clearly show a discontinuity between frames 334 and 335, so that fits also. Another fit is 114 kIAS at 'ts'.

[jcjeant]

Hi,

If not yet read .. you can read this Airbus report
"The A320 Habcheim Accident
An Airbus Industrie response to allegations made in television programme and other medias"
Rapport Airbus.pdf - Petit Fichier

[Chris Scott]

Possible benefit of "Zoom Climb", using any available airspeed

Owain Glyndwr's translation of a BEA conclusion:
"Other calculations show that if the elevator had been brought back faster or earlier during the accident flight, an AoA greater than 15° would have been obtained before the impact on the trees. They also show that without an earlier thrust increase such a manoeuvre, which would have led to an increase in the instantaneous drag of the aircraft would not have permitted (the aircraft) to avoid impact with the trees."

I notice that OG has so far not specifically disputed rudderrudderrat's calculation that a trade of kinetic energy by zoom climb from an airspeed of 112 kt to 107 kt (assuming a steady wind) would provide an altitude gain of 52 ft. (I assume that the 112 kt was based on the IAS recorded at DFDR time-frame of 333.0 secs, which is only 1 or 2 seconds before the defined point of impact.)

awblain has also taken up the cudgels (my emphasis):
"You're right. They could have exchanged another 5kt (2.5m/s) of airspeed, from 110 to 105 (~55m/s). for height h (in m) ~ v.Delta-v/g ~ 55(2.5)/10 ~ 14m.
"The key question is… did they have that 5kt of airspeed to exchange (FBW doesn't change the physics), while avoiding a stall. "

The short answer is in the negative. The GW was about 59 tonnes, with "Flaps" (config) 3 and L/G down. For an A320-100 (no winglets), the VS1G is a CAS of 114 kt. Given that VS1G is said to be defined on the A320 as the steady airspeed at the alpha-max in this confiuration of 17.5 deg (not the CL-MAX), provided the Nz (normal acceleration) is 1G, that begs the question of why the AoA coincident with 112 kt on the Habsheim fly-past was only +14.

My suggestion is that the a/c was in a slight bunt-manoeuvre. The Nz at that stage is around 0.93G, and the pitch has been reducing. Although the IAS had fallen from 116 in the previous second, the ground-speed remained the same (112 kt). That suggests a mini-tailwind shear in that second. The following second shows an IAS of 114 at an AoA of +15 with 1.00G. However, it appears that either the a/c was overperforming slightly, or the ZFW (zero fuel weight) may have been lower than calculated on the loadsheet.

I think it might be more realistic, therefore, to postulate a zoom climb initiated at time frame 329.0 sec; 5 or 6 secs before impact. That is the point at which TOGA thrust had been commanded, at IAS 122. Therefore, the PF was already anxious to expedite his go-around from that point. If rudderrudderrat's energy calculation is good, that should provide at least as much height-gain as he calciulated for his.

As for the zoom climb per-se, I would be very interested to hear the views of Owain Glyndwr, John Farley, and others as to whether it is in any way analogous to the ski-jump concept used operationally, 6 years earlier, to increase the take-off performance of the Hawker/BAC Harrier to permit - in effect - overweight take-offs in STO mode.

I'm wondering if the gain in altitude achieved at the expense of IAS, followed by a semi-ballistic trajectory - maintaining alpha-max at an Nz below 1G - might have provided two advantages, compared with maintaining height/altitude into the treetops.
(1) The initial impact would be avoided (and the engines were rapidly spooling up).
(2) The semi-ballistic segment would delay the stall, providing even more time for the engines to add energy to the total-energy equation. In the case of the Harrier, the engine is already at its TOGA thrust (and, admittedly, with thrust-vectoring). The A320 at Habsheim was light, of course, with very great surplus performance at TOGA, and less than two seconds short of achieving TOGA thrust at the time of impact.

[Owain Glyndwr]

@Chris

I notice that OG has so far not specifically disputed rudderrudderrat's calculation that a trade of kinetic energy by zoom climb from an airspeed of 112 kt to 107 kt (assuming a steady wind) would provide an altitude gain of 52 ft.

Partly because as an energy calculation it is unexceptional, partly because of proccupation with another task in conjunction with a mutual friend and partly because I didn't want to get too involved in a debate about what is actually a complex issue.
The missing parameter is, once again, time. A zoom climb being essentially a pull up manoeuvre one has to consider the 'g' available to execute such a pull up. One might expect the lift curve to be nonlinear up near alphamax, so that is one complication. Ignoring any nonlinearity but allowing for the zero lift AOA, we might expect an available 'g' of about 1.1 at 14.5 deg AOA falling to 1.00 at 17.5 deg.

That is going to mean that the latter part of any zoom climb is going to be prolonged, and in practice indistinguishable from an (unsustainable) steady climb. If a full zoom is going to take a long time then if the intent were to clear the trees by another 50ft or so the zoom would have to have been initiated well back down the runway. In fact of course any zoom is only needed to give just enough clearance to avoid impact so that only the first part of any pull up is relevant and once again timing of such a manoeuvre is critical.

For me, that issue is covered by the BEA report statement I quoted in earlier posts.

Other calculations show that if the sidestick had been brought back faster or earlier during the accident flight, an AoA greater than 15° would have been obtained before the impact on the trees. They also show that without an earlier thrust increase such a manoeuvre, which would have led to an increase in the instantaneous drag of the aircraft would not have permitted (the aircraft) to avoid impact with the trees.

PS I agree with your comments on 'stall' getting in the way.
PPS I'll let John Farley make any comments on ski-jumping.

[DozyWannabe]

Originally Posted by Chris Scott

Given that VS1G is said to be defined on the A320 as the steady airspeed at the alpha-max in this confiuration of 17.5 deg (not the CL-MAX), provided the Nz (normal acceleration) is 1G, that begs the question of why the AoA coincident with 112 kt on the Habsheim fly-past was only +14. [/SIZE]

*Dons the Helmet Of Speculation*

If I were to make an educated guess, I'd say it was because the CAS trend indicated a steady loss of airspeed, and the logic based its pitch limit on the assumption that the trend would continue - which it did right up until t-1.

[C_Star]

Hi,

Another thing regarding the zoom climb - the calculation presented assumes that all of the kinetic energy from the speed loss would be converted into height.

I think there would be some loss due to increased drag due to higer AoA required to pull g's and to maintain lower speed afterwards. Any idea how that could affect the height gained?