AF447
Join Date: Jun 2009
Location: paradise,bc
Age: 82
Posts: 22
Likes: 0
Received 0 Likes
on
0 Posts
GPS / Attitude Tracking
If I recall correctly, this has been used for a long time as part of fighter jet stabilization technology, and I wonder why it hasn't been brought into commercial aviation. Gee, it first came over my horizon in the early 1990s... Hardly rocket science by today's standards.... ?? / CP
Join Date: Apr 2009
Location: Petaluma
Posts: 330
Likes: 0
Received 0 Likes
on
0 Posts
BOAC
To my memory, the ship had deviated left, the buzz about 'lightning', company 'pirep' for 'fortes turbulence' and Msr. Gourgeon's claim that 447's aircrew had been 'unlucky' in reading the Radar were red herrings.
Been awhile, but is GPS to Mil standard nowadays? Used to be purposely degraded to prevent folks like google earth to have the capability of Uncle's satellites.
GPS in hyper accurate mode x4 would very nearly provide sufficient data to turn a wide body into a 'drone'. Oh-Oh, incoming.
To my memory, the ship had deviated left, the buzz about 'lightning', company 'pirep' for 'fortes turbulence' and Msr. Gourgeon's claim that 447's aircrew had been 'unlucky' in reading the Radar were red herrings.
Been awhile, but is GPS to Mil standard nowadays? Used to be purposely degraded to prevent folks like google earth to have the capability of Uncle's satellites.
GPS in hyper accurate mode x4 would very nearly provide sufficient data to turn a wide body into a 'drone'. Oh-Oh, incoming.
Join Date: Jun 2009
Location: paradise,bc
Age: 82
Posts: 22
Likes: 0
Received 0 Likes
on
0 Posts
GPS Mil vs. Consumer tech.
Mil sets work on a different frequency and use a highly encrypted PN (Pseudo Noise) code, which, if not known by the receiver, prevents use of the inherently higher accuracy mil signals. The consumer grade PN code is, I think, 1023 bits long, while the mil one is "months" long, and is changed at shorter intervals than that, so there are never any repeated binary streams that would give a clue to decrypting it.
On the Civilian (consumer) side, there was something called "selective availability (SA) which caused the satellites to "lie" about what time it was, on a purely random basis, thus increasing the amount of error introduced at a receiver. One of the recent Presidents had that removed from service, primarily to allow aviation to have the best possible (non-mil) GPS accuracy for nav purposes.
SA was turned off, by the way, during the Desert Storm operation, to allow Civilian sets to be used by tanks, as there was a shortage of mil GPS receivers for deployment. And there are other stories and anecdotes, of course.
The phase-tracking used in attitude detection is a different animal. The wavelength at GPS frequencies is roughly (very) 25cm, so the theory predicts that one receiver moving vertically (e.g.) relative to a second one, toward or away from the satellite(s), would experience a detectable phase difference in the order of centimeters. Needs superfast processor and GPS chips, but it works....
On the Civilian (consumer) side, there was something called "selective availability (SA) which caused the satellites to "lie" about what time it was, on a purely random basis, thus increasing the amount of error introduced at a receiver. One of the recent Presidents had that removed from service, primarily to allow aviation to have the best possible (non-mil) GPS accuracy for nav purposes.
SA was turned off, by the way, during the Desert Storm operation, to allow Civilian sets to be used by tanks, as there was a shortage of mil GPS receivers for deployment. And there are other stories and anecdotes, of course.
The phase-tracking used in attitude detection is a different animal. The wavelength at GPS frequencies is roughly (very) 25cm, so the theory predicts that one receiver moving vertically (e.g.) relative to a second one, toward or away from the satellite(s), would experience a detectable phase difference in the order of centimeters. Needs superfast processor and GPS chips, but it works....
Join Date: Aug 2009
Location: Australia
Posts: 92
Likes: 0
Received 0 Likes
on
0 Posts
Hi everyone,
Just a thought, I don't know if this has been raised previously as I don't have the time to read the last 200 pages of posts. Given the speculation on a departure from controlled flight due to either stalling or mach tuck, and the idea that this was due to the discrepancy between indicated airspeed and actual airspeed. Couldn't one consider upgrading the flight computer system to challenge the Air Data based on the aircrafts angle of attack and power settings.
For example, ice blocks the pitots slowly, CAS remains the same but the A/P slowly decreases AoA and increases power. Surely if a certain number of parameters are known ie. performance based on CAS AoA and power settings C of G, Trim etc. The aircraft would be able to tell if performance is not meeting what is expected and alerts the pilots.
Just a thought, I am no expert at all but it seems with so much automation and cross checking by the automation systems it could be wise to have this.
J
Just a thought, I don't know if this has been raised previously as I don't have the time to read the last 200 pages of posts. Given the speculation on a departure from controlled flight due to either stalling or mach tuck, and the idea that this was due to the discrepancy between indicated airspeed and actual airspeed. Couldn't one consider upgrading the flight computer system to challenge the Air Data based on the aircrafts angle of attack and power settings.
For example, ice blocks the pitots slowly, CAS remains the same but the A/P slowly decreases AoA and increases power. Surely if a certain number of parameters are known ie. performance based on CAS AoA and power settings C of G, Trim etc. The aircraft would be able to tell if performance is not meeting what is expected and alerts the pilots.
Just a thought, I am no expert at all but it seems with so much automation and cross checking by the automation systems it could be wise to have this.
J
Join Date: Jul 2009
Location: SUSSEX UK
Age: 76
Posts: 57
Likes: 0
Received 0 Likes
on
0 Posts
Air France crash victims' families to meet judge
Join Date: Nov 2006
Location: SoCalif
Posts: 896
Likes: 0
Received 0 Likes
on
0 Posts
TCAS Fail Unrelated to Other Faults
Back to your post 4435 on 21 Sep, Belgique:
The ADMs, Air Data Modules, used in the A330 can be pretty dumb sensors, equivalent to the blind encoders that came out 25 years ago: just an absolute pressure transducer and a digital output. I'm not intimate with the A330, but here's a logical way to build a pitot/static system with modern devices:
Each ADM for static pressure puts out a 429 word labeled, ALT, which is fed to its companion ADR with minimal monitoring. The ADR passes the ALT on to the using devices, such as pilot displays, AP and transponder. The using devices receive ALT from two or three ADR for redundancy and/or voting.
No airspeed input is needed or used in the altitude calculation.
For airspeed measurement, we use an identical ADM, but connected to a pitot probe, and its output word coded for impact pressure. The ADR merely subtracts ALT from impact pressure to calculate IAS, which is then forwarded on. There is no reason for the ADR to report ALT Fail if it can't compute IAS. Each output from an ADIRU stands on its own.
So, I come back to my original premise: blocked pitot tubes do not cause TCAS Fail to be reported. There must be another cause of the TCAS Fail that was reported by the ACARS.
Beyond that, if there were ALT Fail out of the ADR it would have caused the transponder to revert to Mode A,
thereby causing the TCAS to report TCAS OFF to the ACARS, and not TCAS Fail. There is no direct connection between the ADR and the TCAS processor.
TCAS Fail is reported in case of:
TCAS processor computation fail
TCAS directional antenna fail
Loss of valid data from transponder
The first ACARS reports mentioned a TCAS Antenna Fail, which should have been coded in the report.
Again, I agree with the BEA report: the TCAS Fail is unexplained. It may be a symptom of an event that was otherwise unreported.
GB
The ADMs, Air Data Modules, used in the A330 can be pretty dumb sensors, equivalent to the blind encoders that came out 25 years ago: just an absolute pressure transducer and a digital output. I'm not intimate with the A330, but here's a logical way to build a pitot/static system with modern devices:
Each ADM for static pressure puts out a 429 word labeled, ALT, which is fed to its companion ADR with minimal monitoring. The ADR passes the ALT on to the using devices, such as pilot displays, AP and transponder. The using devices receive ALT from two or three ADR for redundancy and/or voting.
No airspeed input is needed or used in the altitude calculation.
For airspeed measurement, we use an identical ADM, but connected to a pitot probe, and its output word coded for impact pressure. The ADR merely subtracts ALT from impact pressure to calculate IAS, which is then forwarded on. There is no reason for the ADR to report ALT Fail if it can't compute IAS. Each output from an ADIRU stands on its own.
So, I come back to my original premise: blocked pitot tubes do not cause TCAS Fail to be reported. There must be another cause of the TCAS Fail that was reported by the ACARS.
Beyond that, if there were ALT Fail out of the ADR it would have caused the transponder to revert to Mode A,
thereby causing the TCAS to report TCAS OFF to the ACARS, and not TCAS Fail. There is no direct connection between the ADR and the TCAS processor.
TCAS Fail is reported in case of:
TCAS processor computation fail
TCAS directional antenna fail
Loss of valid data from transponder
The first ACARS reports mentioned a TCAS Antenna Fail, which should have been coded in the report.
Again, I agree with the BEA report: the TCAS Fail is unexplained. It may be a symptom of an event that was otherwise unreported.
GB
Join Date: Nov 2006
Location: SoCalif
Posts: 896
Likes: 0
Received 0 Likes
on
0 Posts
The TCAS antennas are roughly overhead and beneath the forward fuselage just aft of where the nose transitions to constant diameter. A diagram of antenna locations on the A330 was posted in this thread, back about 9 June.
GB
GB
Join Date: Jun 2008
Location: Cambridge UK
Posts: 192
Likes: 0
Received 0 Likes
on
0 Posts
Accuracy of differential GPS
I feel that the accurate statements previously made about GPS absolute accuracy give a misleading impression of the utility of present GPS systems for measuring civilian aircraft attitude.
There are lots of causes of GPS inaccuracy; such as deliberate encoding, atmospheric effects, and engineering imperfections. However these effects are similar for two GPS readings taken at positions close in space and time.
This effect is used in the various differential GPS systems in operation, in which base stations broadcast correction factors. For measuring aircraft attitude its even easier -- the error is surely [almost entirely] canceled when you take the difference between two GPS readings; e.g. the difference in altitude of the two wingtips.
Wikipedia suggests that differential GPS -- with all the bells and whistles -- can provide accuracies of the order of 10 cm. I cannot see why local differencing should not be at least as accurate.
I'm not advocating such a system, just commenting on its engineering feasibility.
Regards Peter
Unsettling idea of the week.
Some years ago I read in the popular scientific press of an idea for increasing runway throughput. Basically you bolt two wide-body jets together. They share a wing and of necessity fly in close formation. You only have pilots in one body; and rely on a data link between the cockpits and GPS sensors all over the place to ensure that the autopilot in the second body "doesn't overstress anything".
There are lots of causes of GPS inaccuracy; such as deliberate encoding, atmospheric effects, and engineering imperfections. However these effects are similar for two GPS readings taken at positions close in space and time.
This effect is used in the various differential GPS systems in operation, in which base stations broadcast correction factors. For measuring aircraft attitude its even easier -- the error is surely [almost entirely] canceled when you take the difference between two GPS readings; e.g. the difference in altitude of the two wingtips.
Wikipedia suggests that differential GPS -- with all the bells and whistles -- can provide accuracies of the order of 10 cm. I cannot see why local differencing should not be at least as accurate.
I'm not advocating such a system, just commenting on its engineering feasibility.
Regards Peter
Unsettling idea of the week.
Some years ago I read in the popular scientific press of an idea for increasing runway throughput. Basically you bolt two wide-body jets together. They share a wing and of necessity fly in close formation. You only have pilots in one body; and rely on a data link between the cockpits and GPS sensors all over the place to ensure that the autopilot in the second body "doesn't overstress anything".
Join Date: Jun 2009
Location: paradise,bc
Age: 82
Posts: 22
Likes: 0
Received 0 Likes
on
0 Posts
Differential GPS - NOT
Depends upon a fixed station having a known and precise location, somewhere nearby, and thus able to quantify the magnitude and direction of the "lie" or inaccuracy at any given moment. This parameter is then used by the GPS receiver that does NOT know its precise location, so that it can apply the error measured by the reference station to its own computed location, to provide additional accuracy. Operative word is "nearby", else the unknown GPS could be experiencing a totally different error, to say nothing of getting the differential error reading from the ref. station.
Join Date: Jun 2008
Location: Cambridge UK
Posts: 192
Likes: 0
Received 0 Likes
on
0 Posts
Differential GPS - NOT
cribbagepeg
Sorry for any lack of clarity. Differential GPS is well-known strategy for correcting the effects of local-ish GPS errors when calculating absolute positions. I was pointing out that when you calculate relative positions the local GPS errors cancel out anyway, so no such correction is required.
As I stated:
There are lots of causes of GPS inaccuracy; such as deliberate encoding, atmospheric effects, and engineering imperfections. However these effects are similar for two GPS readings taken at positions close in space and time. ... This effect is used in the various differential GPS systems in operation, in which base stations broadcast correction factors.
As you state, this base station needs to be at a known location, so that the [local] GPS error can be measured, and transmitted to interested parties. These parties can then apply this as a correction to their readings to get first-order correction for their GPS errors [if they are close enough].
However if you want to know the difference in altitude between your wingtips you are interested in relative not absolute positions. You may know that there is a local GPS error, but you don't actually need to worry about its value. As the error is the same for all local GPS measurements, its value disappears when you take differences.
So I repeat:
For measuring aircraft attitude its even easier -- the error is surely [almost entirely] canceled when you take the difference between two GPS readings; e.g. the difference in altitude of the two wingtips.
So for the measurement of absolute positions differential-GPS is far more accurate than normal GPS (as it removes the effect of local GPS errors). While for the measurement of relative positions the two systems provide the same high accuracy (as local GPS errors have no impact). Making the accurate measurement of aircraft attitude, based on civilian GPS technology, a very real possibility.
Again:
I'm not advocating such a system, just commenting on its engineering feasibility. ... [But] I feel that the accurate statements previously made about GPS absolute accuracy give a misleading impression of the utility of present GPS systems for measuring civilian aircraft attitude.
Regards, Peter
If you prefer an algebraic argument.
Let the altitude of the left and right wingtips - as measured by their GPS receivers - be l & r.
Let the altitude component of the GPS correction required locally be c (perhaps as transmitted by a friendly neighbourhood differential base station).
The corrected altitudes of the wingtips are (l+c) and (r+c).
The corrected altitude of left wingtip relative to the corrected altitude of the right wingtip is then ( (l+c) - (r+c) ) = (l-r).
Which is, of course, the same as the simple difference in GPS readings (l-r); as any differential GPS correction simply cancels out in the calculation.
Sorry for any lack of clarity. Differential GPS is well-known strategy for correcting the effects of local-ish GPS errors when calculating absolute positions. I was pointing out that when you calculate relative positions the local GPS errors cancel out anyway, so no such correction is required.
As I stated:
There are lots of causes of GPS inaccuracy; such as deliberate encoding, atmospheric effects, and engineering imperfections. However these effects are similar for two GPS readings taken at positions close in space and time. ... This effect is used in the various differential GPS systems in operation, in which base stations broadcast correction factors.
As you state, this base station needs to be at a known location, so that the [local] GPS error can be measured, and transmitted to interested parties. These parties can then apply this as a correction to their readings to get first-order correction for their GPS errors [if they are close enough].
However if you want to know the difference in altitude between your wingtips you are interested in relative not absolute positions. You may know that there is a local GPS error, but you don't actually need to worry about its value. As the error is the same for all local GPS measurements, its value disappears when you take differences.
So I repeat:
For measuring aircraft attitude its even easier -- the error is surely [almost entirely] canceled when you take the difference between two GPS readings; e.g. the difference in altitude of the two wingtips.
So for the measurement of absolute positions differential-GPS is far more accurate than normal GPS (as it removes the effect of local GPS errors). While for the measurement of relative positions the two systems provide the same high accuracy (as local GPS errors have no impact). Making the accurate measurement of aircraft attitude, based on civilian GPS technology, a very real possibility.
Again:
I'm not advocating such a system, just commenting on its engineering feasibility. ... [But] I feel that the accurate statements previously made about GPS absolute accuracy give a misleading impression of the utility of present GPS systems for measuring civilian aircraft attitude.
Regards, Peter
If you prefer an algebraic argument.
Let the altitude of the left and right wingtips - as measured by their GPS receivers - be l & r.
Let the altitude component of the GPS correction required locally be c (perhaps as transmitted by a friendly neighbourhood differential base station).
The corrected altitudes of the wingtips are (l+c) and (r+c).
The corrected altitude of left wingtip relative to the corrected altitude of the right wingtip is then ( (l+c) - (r+c) ) = (l-r).
Which is, of course, the same as the simple difference in GPS readings (l-r); as any differential GPS correction simply cancels out in the calculation.
Join Date: Apr 2009
Location: Petaluma
Posts: 330
Likes: 0
Received 0 Likes
on
0 Posts
From another thread, EASA are considering an AD on Trent 800 for Airbus 330. It was only a matter of time. The FOHE inlet/interface is being altered for 777 equipped a/c. A Northwest 777 had rollbacks over the Rocky Mountains subsequent to a suspicion of icerestricted FOHE's. This following BA038's short arrival at LHR.
Odd that on this large thread the possibility of engine rollbacks hasn't gotten much interest. I believe NWA incident occurred at 35,000 feet.
Ice is Ice.
For clarity, the EASA is considering extending the AD to include Trent 700 and 500.
Odd that on this large thread the possibility of engine rollbacks hasn't gotten much interest. I believe NWA incident occurred at 35,000 feet.
Ice is Ice.
For clarity, the EASA is considering extending the AD to include Trent 700 and 500.
Join Date: Dec 2004
Location: YORKSHIRE
Posts: 13
Likes: 0
Received 0 Likes
on
0 Posts
Anybody Remember a Canberra climbing over a TS?
I believe about the early sixties over Sudan. Engines suffered N over Root T Surge - (or so the Board of Enquiry speculated -or perhaps the crew were talking to somebody). Double flameout due to low IAS. Aircraft broke up as it plunged into TS.
I don't suppose modern bypass engines suffer N over Root T Surge.
Any relevance to the Air France accident?
The thought of this occurring to me frightened me for most of my 50 year flying career. The nearest I let myself get to any TS was looking at them from 10 miles away (at night). No Weather Radar in the ancient old aircraft I have flown.
Totally irrelevant I suppose but I keep getting messages from PPRUNE saying 'you haven't posted for a long time'
I don't suppose modern bypass engines suffer N over Root T Surge.
Any relevance to the Air France accident?
The thought of this occurring to me frightened me for most of my 50 year flying career. The nearest I let myself get to any TS was looking at them from 10 miles away (at night). No Weather Radar in the ancient old aircraft I have flown.
Totally irrelevant I suppose but I keep getting messages from PPRUNE saying 'you haven't posted for a long time'
Join Date: Jun 2009
Location: NNW of Antipodes
Age: 81
Posts: 1,330
Received 0 Likes
on
0 Posts
Where were they?
I'm throwing a new graphic into the mix, but before going any further lets have a look at the AOC reported positions. On departure Rio de Janerio the 10 minute AOC reports were noted at 09, 19 etc.., but prior to passing NATAL it is assumed that AF Operations reset the report timing to 10.5, 20.5 etc., probably so that they would get regular sequencing of reports from multiple aircraft, rather than a disorganised dump every 10 minutes. To complicate the issue the BEA produced a route chart showing the report times as clearly occurring on the minute. Based on the timings produced by the BEA, many moons ago I deduced that AF447 was overhead INTOL at 01:32:32 whereas it can now be assumed the time was 01:33:02. This is important because there is currently no rational for the apparent deviation from the track originally noted at 02:10:00 to have occurred prior to the upset that happened at 02:10:06 when the ACARS commenced phoning home.
02:10:10 - AUTO FLT AP OFF
02:10:16 - AUTO FLT REAC W/S DET FAULT
02:10:23 - CTL ALTN LAW
02:10:29 - FLAG ON CAPT PFD SPD LIMIT
02:10:34 - 02:10:30z AOC Position Report [injected into ACARS sequence as priority]
02:10:41 - FLAG ON F/O PFD SPD LIMIT
02:10:47 - AUTO FLT A/THR OFF
02:10:54 - NAV TCAS FAULT
02:11:00 - FLAG ON CAPT PFD FD
Other than the AOC position, it can be assumed that remaining items all originated at 02:10:06 (4 secs to receipt at AF Maintenance Base). Note the AUTO FLT AP OFF, AUTO FLT A/THR OFF and NAV TCAS FAULT are possibly due to discrepancies in Pitot ADM and Baro ADM static pressure data [Note:: The rejection by the EFCS of an invalid IAS due to more than 30KTS reduction in 1 second, doesn't happen until just over a minute later]. On the otherhand, the initial ACARS fault messages could have been caused by control parameters being exceeded, e.g. extreme turbulence, updrafts / thermal elevator, or extreme wind shear involving Loss of Control in what would be a Coffin Corner situation. A Mach Critical event seems to have been avoided.
A larger scale graphic is available at -
http://i846.photobucket.com/albums/a...7-bank-lge.png
A turn radius for a 20 degree bank (grey circle), and the 02:10:30z AOC position would have resulted in a deviation from the track initiated at 02:09:30z. Indications are that the GS from the 02:00:30z position to the 02:10:30z AOC position had remained constant at 463KTS (475KTAS). However, what was happening to the OAT?
Consider each of the following scenarios -
(1) That the Cb cell updrafts in the ITCZ were conveying an unusual amount of water vapour/latent heat which was carried out into the anvil stratus cloud radiating from each cell. The a/c encountered this cloud around 0208z and the incremental ice-up of the pitot tubes, some other surfaces and potentially the engine cores took place over the next two/three of minutes. Likewise the ATHR augmented decreasing IAS/CAS with increasing power. Whether the left turn that is already evident in the 02:10:30z AOC position was crew initiated or took place as a result of a LOC event is unknown - my supposition is that its related to the 02:10:06z events.
A turn radius for a 35 degree bank (white circle), and the 02:10:30z AOC position would have resulted in a deviation from the track initiated at 02:10:00z.
(2) Take onboard the first paragraph in (1), then assume that the a/c effectively penetrated the active Cb cell, hit the wall and rode the thermal elevator, then the 02:10:30z AOC position represents where it had been thrown as it was spewed out of the mesoscale system. The BEA know a lot more about the 0210z position than they have let on - enough said.
Throw in a double flame-out, the AP/ATHR disconnect, ADIRU's disagreeing, unbelievable IAS along with sundry chirps and bells, then it doesn't take you long to realise the only way is down and how the hell do you manage it. On top of all that the FBW software, when it gave up, gave you back the a/c with control limitations!
A turn radius for a 45 degree bank (cyan circle), and the 02:10:30z AOC position would have resulted in a deviation from the track initiated at 02:10:06z, i.e. AP/ATHR OFF.
The debris distribution and current data indicate that the a/c continued in a westerly direction and descended in probably heavy turbulence to possibly FL100 over the next 3min 10sec - averaging nearly 8,000 feet/min with a KTAS of about 408 (+/- wind) and GS of 400KTS. If a high altitude 2 x flame-out had occurred, then sufficient turbine rpm was being maintained to power the essential service bus during this descent. Attempts to relight may have been hampered by turbulence and ice load. Reducing the rate of descent near FL100 then resulted in electrical power loss at 02:13:10z and it is presumed the APU was started as evidenced by resumption of the SATCOM link at 02:13:40z.
Remember, the a/c was deep in the ITCZ descending through numerous cumulonimbus cells and associated out-flowing anvil cloud with the propensity for ongoing icing highly likely.
Putting together a new graphic to represent more clearly what may have occurred, I looked at the 3 separate positions I had previously calculated for the possible impact position and decided they represented the proverbial "cocked hat". Bisecting the internal angles through the center of each opposite side gave me a position of 3°03.4'N 31°04.5'W, and this position when plotted gave the track (magenta line) from the Last Known Position as plotted tangential to the 35 degree bank curve.
As previously suggested, the Cabin Vertical Speed advisory gives an indication of when the a/c was passing through about 8,000 feet - 5 secs after the external pressure exceeded the cabin pressure and cabin vertical speed exceeded 1,800 ft/min.
In summary:-
(1) WX radar faulty or crew didn't detect Cb cell(s). Probably former.
(2) LOC at FL350+/- was as a result of extreme conditions encountered when penetrating an active mesoscale system.
(3) Icing that took place was probably 'rapid' as opposed to 'gradual'.
(4) Intake blanking in the updrafts resulted in a 2 x flame-out.
(5) Recovery was achieved, but relights were unsuccessful due to engine core ice load.
(6) The low level LOC could have been compounded by an aft C of G problem.
(7) Tail yawing to port? A 'flat spin' or just surface wind 025°T x 21/35KTS.
Finally, I'm just the messenger!
mm43
02:10:10 - AUTO FLT AP OFF
02:10:16 - AUTO FLT REAC W/S DET FAULT
02:10:23 - CTL ALTN LAW
02:10:29 - FLAG ON CAPT PFD SPD LIMIT
02:10:34 - 02:10:30z AOC Position Report [injected into ACARS sequence as priority]
02:10:41 - FLAG ON F/O PFD SPD LIMIT
02:10:47 - AUTO FLT A/THR OFF
02:10:54 - NAV TCAS FAULT
02:11:00 - FLAG ON CAPT PFD FD
Other than the AOC position, it can be assumed that remaining items all originated at 02:10:06 (4 secs to receipt at AF Maintenance Base). Note the AUTO FLT AP OFF, AUTO FLT A/THR OFF and NAV TCAS FAULT are possibly due to discrepancies in Pitot ADM and Baro ADM static pressure data [Note:: The rejection by the EFCS of an invalid IAS due to more than 30KTS reduction in 1 second, doesn't happen until just over a minute later]. On the otherhand, the initial ACARS fault messages could have been caused by control parameters being exceeded, e.g. extreme turbulence, updrafts / thermal elevator, or extreme wind shear involving Loss of Control in what would be a Coffin Corner situation. A Mach Critical event seems to have been avoided.
A larger scale graphic is available at -
http://i846.photobucket.com/albums/a...7-bank-lge.png
A turn radius for a 20 degree bank (grey circle), and the 02:10:30z AOC position would have resulted in a deviation from the track initiated at 02:09:30z. Indications are that the GS from the 02:00:30z position to the 02:10:30z AOC position had remained constant at 463KTS (475KTAS). However, what was happening to the OAT?
Consider each of the following scenarios -
(1) That the Cb cell updrafts in the ITCZ were conveying an unusual amount of water vapour/latent heat which was carried out into the anvil stratus cloud radiating from each cell. The a/c encountered this cloud around 0208z and the incremental ice-up of the pitot tubes, some other surfaces and potentially the engine cores took place over the next two/three of minutes. Likewise the ATHR augmented decreasing IAS/CAS with increasing power. Whether the left turn that is already evident in the 02:10:30z AOC position was crew initiated or took place as a result of a LOC event is unknown - my supposition is that its related to the 02:10:06z events.
A turn radius for a 35 degree bank (white circle), and the 02:10:30z AOC position would have resulted in a deviation from the track initiated at 02:10:00z.
(2) Take onboard the first paragraph in (1), then assume that the a/c effectively penetrated the active Cb cell, hit the wall and rode the thermal elevator, then the 02:10:30z AOC position represents where it had been thrown as it was spewed out of the mesoscale system. The BEA know a lot more about the 0210z position than they have let on - enough said.
Throw in a double flame-out, the AP/ATHR disconnect, ADIRU's disagreeing, unbelievable IAS along with sundry chirps and bells, then it doesn't take you long to realise the only way is down and how the hell do you manage it. On top of all that the FBW software, when it gave up, gave you back the a/c with control limitations!
A turn radius for a 45 degree bank (cyan circle), and the 02:10:30z AOC position would have resulted in a deviation from the track initiated at 02:10:06z, i.e. AP/ATHR OFF.
The debris distribution and current data indicate that the a/c continued in a westerly direction and descended in probably heavy turbulence to possibly FL100 over the next 3min 10sec - averaging nearly 8,000 feet/min with a KTAS of about 408 (+/- wind) and GS of 400KTS. If a high altitude 2 x flame-out had occurred, then sufficient turbine rpm was being maintained to power the essential service bus during this descent. Attempts to relight may have been hampered by turbulence and ice load. Reducing the rate of descent near FL100 then resulted in electrical power loss at 02:13:10z and it is presumed the APU was started as evidenced by resumption of the SATCOM link at 02:13:40z.
Remember, the a/c was deep in the ITCZ descending through numerous cumulonimbus cells and associated out-flowing anvil cloud with the propensity for ongoing icing highly likely.
Putting together a new graphic to represent more clearly what may have occurred, I looked at the 3 separate positions I had previously calculated for the possible impact position and decided they represented the proverbial "cocked hat". Bisecting the internal angles through the center of each opposite side gave me a position of 3°03.4'N 31°04.5'W, and this position when plotted gave the track (magenta line) from the Last Known Position as plotted tangential to the 35 degree bank curve.
As previously suggested, the Cabin Vertical Speed advisory gives an indication of when the a/c was passing through about 8,000 feet - 5 secs after the external pressure exceeded the cabin pressure and cabin vertical speed exceeded 1,800 ft/min.
In summary:-
(1) WX radar faulty or crew didn't detect Cb cell(s). Probably former.
(2) LOC at FL350+/- was as a result of extreme conditions encountered when penetrating an active mesoscale system.
(3) Icing that took place was probably 'rapid' as opposed to 'gradual'.
(4) Intake blanking in the updrafts resulted in a 2 x flame-out.
(5) Recovery was achieved, but relights were unsuccessful due to engine core ice load.
(6) The low level LOC could have been compounded by an aft C of G problem.
(7) Tail yawing to port? A 'flat spin' or just surface wind 025°T x 21/35KTS.
Finally, I'm just the messenger!
mm43
Last edited by mm43; 27th Sep 2009 at 20:15. Reason: spelling!
Join Date: Jul 2009
Location: Tranquility Base
Age: 68
Posts: 53
Received 0 Likes
on
0 Posts
Greybeard.... Your query is certainly worth further study. There is likely to be an intitiating event followed by further contributing factors that caused this accident. I used to research airborne lighting suseptability for a major avionics company and the TCAS antennas appear to be in areas that are prime for lightning attach points. Couple that with the photos showing rudder static dissipator tufts missing and it points to a possible event that needs to be eliminated or studied further. The root cause of the initiating event (beyond flying into a highly convective area) is not known, so all possible causes have to be fully explored and analyzed.
Join Date: Mar 2001
Location: us
Posts: 694
Likes: 0
Received 0 Likes
on
0 Posts
mm43, thanks again for your effort and analysis.
I am hoping that the next BEA report contains the transcript of all relevant communications between ATLANTICO and the three flights immediately preceding and following AF447 on the track that night (IB6024, AF459, and LH507). LH507 preceded AF447 by 20 minutes and deviated west because of the weather appearing on its radar. Presumably communication about that deviation between LH507 and ATLANTICO should have been overheard by AF447, as AF447 should have been on ATLANTICO's frequency from 0133 until 0220 hours.
I am hoping that the next BEA report contains the transcript of all relevant communications between ATLANTICO and the three flights immediately preceding and following AF447 on the track that night (IB6024, AF459, and LH507). LH507 preceded AF447 by 20 minutes and deviated west because of the weather appearing on its radar. Presumably communication about that deviation between LH507 and ATLANTICO should have been overheard by AF447, as AF447 should have been on ATLANTICO's frequency from 0133 until 0220 hours.
Re high altitude icing, engine issues and so on, I see that there is a "Winter Operations Conference" being hosted by the Air Canada Pilots Association at the Royal York Hotel in Toronto on these and other topics. The dates are October 7 & 8 with registration on the evening of the 6th. The home website is here and the agenda is here.
mm43;
First, I acknowledge the "messenger" status and the "perhaps" status of your observations. So it is with respect that I offer some small points.
With regard to the radar serviceability, I don't think there is likely justification for saying, "probably former" (in the quoted passage), as there are no ACARS messages in the mix received indicating a radar failure. From experience, there are definite ECAM and Maintenance messages associated with such a failure. There may no issues at all regarding why the encounter, or there may be "sole causes" as a result of the encounter either as a result of lack of knowledge are pure unavoidability. In short there is nothing we can say about the crew. We simply do not know what went on in the cockpit. I believe we can say in the absence of ACARS messages however, that the radar was working.
As I posted in July, while we cannot state that the engines were flamed out, while we don't have any hard data from the manufacturer, the likelihood of the engines continuing to run with a high angle of incidence is low; hydraulic power from the RAT would be similarly reduced due to the relatively low forward speed as described in the BEA Report. The RAT requires at least 140kts over the blades to generate some hydraulic pressure. Whether ice or high incidence would cause re-start difficulties cannot be stated, nor can we say the APU was running although that would certainly be something the crew may have considered/attempted at some point.
I have earlier posted reasons why I think an aft CG would not be a strong factor in any LOC but could be a contributing factor in abnormal attitudes or post-LOC. In other words, I doubt very much whether an aft CG would in and of itself, even in heavy turbulence or reduced airspeed, (though still well above Vls), would cause a loss of control.
The reasons why the autoflight system "gives the pilot the airplane" has been discussed and isn't unusual in transport aircraft. No autopilot is designed or certified to handle an aircraft beyond the accepted limits indicating loss of control. All other transports do the same thing and I think including the statement is incongruous with your excellent analysis. Whether it is causal or not is another matter, mainly for the designers. Suffice it to say that an autopilot capable of recovering an aircraft from unusual attitudes/LOC while managing power and 'g' loads is a long way off; -the best responder is still the human one.
These aren't meant as criticisms of your continuing fine work but observations which I believe must attend any such mapping speculations.
best,
PJ2
mm43;
(1) WX radar faulty or crew didn't detect Cb cell(s). Probably former.
With regard to the radar serviceability, I don't think there is likely justification for saying, "probably former" (in the quoted passage), as there are no ACARS messages in the mix received indicating a radar failure. From experience, there are definite ECAM and Maintenance messages associated with such a failure. There may no issues at all regarding why the encounter, or there may be "sole causes" as a result of the encounter either as a result of lack of knowledge are pure unavoidability. In short there is nothing we can say about the crew. We simply do not know what went on in the cockpit. I believe we can say in the absence of ACARS messages however, that the radar was working.
As I posted in July, while we cannot state that the engines were flamed out, while we don't have any hard data from the manufacturer, the likelihood of the engines continuing to run with a high angle of incidence is low; hydraulic power from the RAT would be similarly reduced due to the relatively low forward speed as described in the BEA Report. The RAT requires at least 140kts over the blades to generate some hydraulic pressure. Whether ice or high incidence would cause re-start difficulties cannot be stated, nor can we say the APU was running although that would certainly be something the crew may have considered/attempted at some point.
I have earlier posted reasons why I think an aft CG would not be a strong factor in any LOC but could be a contributing factor in abnormal attitudes or post-LOC. In other words, I doubt very much whether an aft CG would in and of itself, even in heavy turbulence or reduced airspeed, (though still well above Vls), would cause a loss of control.
On top of all that the FBW software, when it gave up, gave you back the a/c with control limitations!
These aren't meant as criticisms of your continuing fine work but observations which I believe must attend any such mapping speculations.
best,
PJ2
Last edited by PJ2; 28th Sep 2009 at 16:03.