AA77 FDR Data, Explained

[beachnut]

I believe PfT's position is that no data was lost since the last data in the NTSB .CSV file has a time of 9:37:44 and one of the NTSB reports gives an impact time of 9:37:45. That leaves them in the position of refuting any claims and explanations of lost data. If I am wrong, anyone reading this can feel free to reply or PM me through either this or the Pft forums or send an email to me at [email protected].

...

p4t has no theory; balsamo says so; "offer no theory". The party line for the truthNAZI. Is that their theory they don't have the impact time from the NTSB which was faked and made up by the MIB? Irony...

The impact time from the report on the FDR is based on what?

You are correct Balsamo implies there is no missing data because of an arbitrary impact time in a report. 911Files points out there is RADAR data showing 77 in the air at the arbitrary impact time.

There are many reasons the data stopped or is missing from the secure chip; p4t have delusions based on nothing. Many examples of missing data for many reasons; p4t have to sell paranoid based lies on DVD to make money and they could care less if they mislead those who lack knowledge and buy lies.

See if the p4t delusion experts explain the impact time they cherry pick.
But the p4t say "We do not offer theory ", so they have no clue what time 77 impacted and they can't use evidence to come up with a number since they are math, 11.2G failed physics challenged.

They will use your information to support their implied lies and use you as one of their experts to support the cult chant of inside job and other paranoid ideas to spread and market their wares.

And of course you will be banned for presenting facts; before the evil NWO mods ban you thank you for answering questions; welcome again

 

[apathoid]

Wouldn't need to be one bad enough to shut down the engine, just enough to cause a significant spike on the power bus.

I agree we would need a lot more evidence before this hypothesis graduates to "theory," but it would explain AA 77 and AA 587's behavior, maybe others.

How many seconds were missing on AA587? I'm googling and all I can find on 587 is woo, imagine that.

If I'm not mistaken, the #1 engine seperated from the wing just before the crash, that would most likely explain the loss of data in this case, though I must admit I've no experience with Airbus.

 

[wstutt]

<snip>

I was mainly thinking of the Huffman schema and the frame markers with respect to filtering. If you've got data from before applying the Huffman symbols, then you may indeed be quite close to the original data.

If I had the data from before applying the Huffman coding, then that would be the decompressed data that I am seeking and I would be able to decode it. The frame markers are only added to the data, they do not change or replace any of the data.

Note that the Huffman coding produces no loss of data. The NTSB report says "Prior to recording, the serial data stream is compressed using a modified Hoffman encoding scheme. The data can be decompressed and restored to the original ARINC 575/717 format without any data loss."

<snip>

... Just for full disclosure, I don't think this is in any way going to change our thoughts on what happened at the Pentagon ...

I didn't expect it would. My intention is only to add my findings to the debate.

I say this becase there are similar episodes, like AA 587, so the fact that the last few seconds are missing may not be a random event.

I quickly looked through the NTSB report for AA 587 again. Unfortunately, I didn't find anywhere where they addressed the issue of why the FDR recording stopped before impact. I would be interested to know if the NTSB did say why.

This obviously doesn't matter for purposes of reconstructing September 11^th, but it could matter if, for instance, there was a mid-air collision event, and the last few seconds were essential in determining how it had happened. If there is a system vulnerability, we should take steps to address it.

I would have thought the NTSB would also be concerned with any system vulnerability that would result in the loss of the last few seconds of data. Didn't beachnut say that had been fixed since September 11^th?

Warren.

 

[beachnut]

How many seconds were missing on AA587? I'm googling and all I can find on 587 is woo, imagine that.

If I'm not mistaken, the #1 engine seperated from the wing just before the crash, that would most likely explain the loss of data in this case, though I must admit I've no experience with Airbus.

(The FDR stopped recording 13.6 seconds before impact.) Flt 587
http://www.ntsb.gov/publictn/2004/AAR0404.pdf
​

​

Page 5 footnote 18​

 

[R.Mackey]

Too bad the evidence is clearly to the contrary. At 9:37:45, the plane was still in the air according to DCA TRACON radar data.

I quite agree, and the "impact time" estimated by the NTSB is pretty clearly referenced to the end of the FDR, and is therefore off by a few seconds. Why the FDR doesn't have the last few seconds is still of some technical interest, or at least it is to me...

How many seconds were missing on AA587? I'm googling and all I can find on 587 is woo, imagine that.

If I'm not mistaken, the #1 engine seperated from the wing just before the crash, that would most likely explain the loss of data in this case, though I must admit I've no experience with Airbus.

The FDR on AA 587 conked out approximately 14 seconds before impact, 10 seconds before engines were torn free, and 6 seconds before major structural distress indicated by white streaks of smoke or ejected fuel. These times are found by comparing the FDR to the CVR, which recorded all the way to impact without interruption. At the time the FDR record ends, both engines are still running well, as reported on the FDR itself and verified by the sound on the CVR.

I doubt there's a plane crash in history that some idiot hasn't claimed is "suspicious."

If I had the data from before applying the Huffman coding, then that would be the decompressed data that I am seeking and I would be able to decode it. The frame markers are only added to the data, they do not change or replace any of the data.

I should have been more clear, my apologies. I meant before the Huffman decode, i.e. I thought you were strictly looking at decoded parameters rather than raw bits. Obviously you can't access data before being encoded and put onto the FDR in the first place.

Note that the Huffman coding produces no loss of data. The NTSB report says "Prior to recording, the serial data stream is compressed using a modified Hoffman encoding scheme. The data can be decompressed and restored to the original ARINC 575/717 format without any data loss."

Right. Huffman is variable compression rate but lossless. Because of this, you are almost certainly correct that the flash isn't written a block at a time, but probably written only a word at a time.

I didn't expect it would. My intention is only to add my findings to the debate.

Understand, and it's appreciated. Whatever the data says, we should try to figure out.

I quickly looked through the NTSB report for AA 587 again. Unfortunately, I didn't find anywhere where they addressed the issue of why the FDR recording stopped before impact. I would be interested to know if the NTSB did say why.

As far as I know, they don't speculate why. My own guess is the engines stall shortly after pitch-roll sets in, and this upsets the power bus. But there could be some other mechanism, and potentially a common one with AA 77.

I would have thought the NTSB would also be concerned with any system vulnerability that would result in the loss of the last few seconds of data. Didn't beachnut say that had been fixed since September 11^th?

I sure hope so. Could be that more modern recorders aren't as vulnerable even without learning from AA 77 etc. But the technology sometimes gets better, sometimes worse. The FDR in this case performed pretty well, certainly better than the old days of metal tape! But AA 587 is a case where the (analogue) CVR worked better than the FDR, and it's not the only one, either. It's a pity the CVR of AA 77 was destroyed in the fire.

Again, thanks for your comments.

 

[apathoid]

The FDR on AA 587 conked out approximately 14 seconds before impact, 10 seconds before engines were torn free, and 6 seconds before major structural distress indicated by white streaks of smoke or ejected fuel. These times are found by comparing the FDR to the CVR, which recorded all the way to impact without interruption. At the time the FDR record ends, both engines are still running well, as reported on the FDR itself and verified by the sound on the CVR.

...
{good stuff}
...

As far as I know, they don't speculate why. My own guess is the engines stall shortly after pitch-roll sets in, and this upsets the power bus. But there could be some other mechanism, and potentially a common one with AA 77.

Thanks Ryan. That would be my guess as well, definite possibility of a compressor stall with all that yaw....

I wouldn't be a bit surpised to see the FAA address data loss in the future. The capabilty is already there to all but eliminate the possibility. But sadly it will take a major crash, where the cause is very much in doubt, coupled with similar critical data loss events for them to issue an AD or Service Bulletin. I personally have reported several service difficulties that needed addressing, but I've yet to even hear back. Unfortunately, its reactive culture, not a proactive one.

/soapbox

I sure hope so. Could be that more modern recorders aren't as vulnerable even without learning from AA 77 etc. But the technology sometimes gets better, sometimes worse. The FDR in this case performed pretty well, certainly better than the old days of metal tape! But AA 587 is a case where the (analogue) CVR worked better than the FDR, and it's not the only one, either. It's a pity the CVR of AA 77 was destroyed in the fire.

There have been no major changes that I'm aware of, don't know what Beachnut found. AA77s FDR was a Loral Fairchild FA2100. I pulled one out of an airplane just last week....they are still probably the most common DFDR around today.

 

[jaydeehess]

On edit: In case anyone was wondering what the problem I alluded to earlier ending up being - apparently maintenance miswired a terminal strip by putting a 28VDC wire into a data bus(ARINC) AC return terminal block. This 28 volts wasnt always present, so the FDAU worked most of the time.
Not the first time I've seen something like that and it wont be the last!

On the ground I was called to investigate a problem with a back up ground-air comm transmitter. Pilots complained of a squeal anytime ATC was using the back up tx. Turned out that facilities had just installed new cabling and had inadvertantly wired it so that the 48 Vdc push-to-talk would energize both the main and back up tx's anytime they selected the back up. It also split the balanced audio so that both the main and back up were 6 db quieter than they should have been. (of course I know the tech that did it and I know how fast he can finsih a bottle of Scotch Whiskey)

 

[jaydeehess]

I quite agree, and the "impact time" estimated by the NTSB is pretty clearly referenced to the end of the FDR, and is therefore off by a few seconds. Why the FDR doesn't have the last few seconds is still of some technical interest, or at least it is to me...

Exactly. The NTSB simply took the end of recording and marked it as impact. It made no difference whatsoever to the cause of the aircraft crash investigation since it was already quite apparent that Flt 77 had been flown into the Pentagon deliberatly.

The NTSB might have been more accurate by stating that impact occured shortly after 9:37:45
As pointed out there is physical evidence that at 9:37:45 the aircraft was still on radar and thus still in the air, not in the Pentagon.


It is of some interest to wonder what became of the data between 9:37:45 and actual impact but only as far as fallibility of the DFDR goes. I have asked PfT on many occassions why they have not pressed L3, ICAO and the pilot's unions on this since it would be very much in their interests to know why it did not record parameters from a supposedly fully functional aircraft in the final second before destruction.
The ingestion of lamp post debris may well explain that the aircraft was not in a fully functional state though, in which case it would follow much that same as the other crash linked to above where loss of power resulted in loss of data recording.

No answer...............

 

[BCR]

I think you guys have offered some significant clues as to what might have happened. Back when I was investigating the Paik brothers claim that an antenna was bent at the VSP tower (behind the VDOT), Pat Creed offered this observation.

Striking the first tower [VSP tower] might have produced some of the smoke that was believed to be coming from somewhere on the plane, possible the right engine. I never believed there was enough time for something to get ingested into the engine and for it to produce smoke, if it was part of a light pole, it would have been less than a second. But striking the antenna might have done enough damage to cause it.

As most of you are aware, I have been unable to verify the damage described by the Paik brothers and other witnesses. However, a closer examination of the area of the last FDR return is proving interesting.

This image is the elevation contour between the final DCA returns, with the final RO2 position about midway. Just prior to this, it passes directly over a building that is located at ~200 feet MSL. The plane altitude is given as 238 (538 feet adjusted) and a radar altitude of 273 feet. The buildings in this area rise about 120-150 feet agl, meaning that the plane cleared them by around 125-175 feet.

It seems reasonable that the approaching plane would have caused birds perched on the roofs of these buildings to scatter, making the bird-strike hypothesis quite reasonable. If indeed the right engine was damaged in this area, it would have been adequate time for the smoke trail to have developed and may have led to the bus power failure. In any event, with the loss of the FDR in this area and known damage to the plane before impact (smoke) certainly warrants further study this area.



Reference footage taken in the area in April, 2007 by me.

 

[jaydeehess]

But AA 587 is a case where the (analogue) CVR worked better than the FDR, and it's not the only one, either. It's a pity the CVR of AA 77 was destroyed in the fire.

Being in the Tv industry I have to tell people many times that with digital the picture will be either crystal clear or not there at all.

The good point of analogue recording of anything was that one can still get something usable from a very degraded signal. The bad is that degradation starts to occur as soon as you exit optimum conditions.

The CVR might have recorded some sound from a bird strike or some reaction to a loss of electrical power from those in the cockpit (ie. in Arabic "what was that?")

 

[jaydeehess]

does anyone know how prevalent bird strikes are near Reagan? It is next to a water course after all.


I haven't checked but I can imagine the reaction over at Pft about a bird strike hypothesis.

 

[wstutt]

Here is the NTSB report on the FA2100 FDR from Flight 1400 that made an emergency landing due to fire in one engine, no injuries fortunately. But the FDR lost power twice and stopped recording during flight, loosing large amounts of data, before it regained power and continued to record. The document also lists the conditions necessary to keep the FDR powered. It could be of interest.

<snip URL>

Thanks for that.

I see the report just says the data was lost due to the FDR losing power. Did the NTSB just presume that? From what I read from the report, they don't seem to have looked in to the issue of whether the amount of data lost by the FDR is equal to the amount of time the FDR was without power.

Warren.

 

[wstutt]

Warren, I just noticed that you are posting here now. Welcome and might I add how impressed I am with your work on the FDR. Just an FYI for JREF'rs, Warren and I compared notes on the AAL77 FDR around a year ago and he was kind enough to provide me with some of his early C# code that helped me in putting together my own VB code for working with the raw FDR file.

Thanks John for the welcome and kind words.

We both reached something of a dead end with the Hoffmann encryption, but got far enough I think to verify the frame structure. Warren, have you found anything since we last communicated to explain the 4-6 seconds of missing data in the published NTSB CSV file?

I haven't got any further with decompressing the .FDR file yet to see if any extra data is there or not.

It's interesting that you used the word encryption. Some papers I have read expressed the opinion that data encoded with Huffman coding is as good as encrypted, however I don't think this was the intention of the FDR manufacturer. I'm sure their goal was to compress the data without loss.

Reading the replies here has led me to review some information I previously downloaded, and this has given me some ideas of some new approaches to attempt to decompress the .FDR file. It's still going to require quite a bit of time and effort though. I'll let you know if and when I'm successful.

Warren.

 

[beachnut]

does anyone know how prevalent bird strikes are near Reagan? It is next to a water course after all.


I haven't checked but I can imagine the reaction over at Pft about a bird strike hypothesis.

http://wildlife.pr.erau.edu/database/select_i.php

769 reports from 1990 at KDCA. 39 in 2001.

 

[jaydeehess]

Thanks for that.

I see the report just says the data was lost due to the FDR losing power. Did the NTSB just presume that? From what I read from the report, they don't seem to have looked in to the issue of whether the amount of data lost by the FDR is equal to the amount of time the FDR was without power.

Warren.

It would be interesting to know since if data was lost for a time period prior to power loss it would support the idea that data in the buffer is significant.
How one would determine this, other than lab testing which would have a lot of variable parameters, is another question.

 

[jaydeehess]

http://wildlife.pr.erau.edu/database/select_i.php

769 reports from 1990 at KDCA. 39 in 2001.

approx average of 3/month under normal circumstances. A loud, fast, low aircraft during a migratory period might increase the probability of a strike.

 

[R.Mackey]

I haven't checked but I can imagine the reaction over at Pft about a bird strike hypothesis.

Just to be clear, I'm not necessarily proposing a bird strike hypothesis -- instead, I'm proposing that something interfered with one or both engines, blipping main power and sending the FDR into a recycle. I imagine every sane person reading here understands this, but just in case...

A bird is one possibility, but there are others -- bits of antenna, bits of trees, or sheer violent maneuvers, for example.

As I said, there is some evidence that at least one engine was in distress, namely the (few) reports and (awful) pictures of a smoke trail. Assuming this is true, it may be correlated with the FDR outage.

By comparison to other engine problems, primarily bird strikes, I only want to see how plausible this scenario is. Doesn't mean I'm betting any farms on it. I doubt we'll ever know for certain.

 

[beachnut]

approx average of 3/month under normal circumstances. A loud, fast, low aircraft during a migratory period might increase the probability of a strike.

Flying low over the city is stupid! I would fear a bird strike if I was not familar with the area and habits of birds. Hani was below 2200 feet for less than 31 seconds. He had the engines at 100.6 and 100.4 percent until the last 15 seconds, they started to roll back on their own; he DID NOT move the throttles back they are at the firewall; full forward.

last 15 secons, RPM percent N1

Engine RPM for the final 15 seconds; right engine started back first.

The EGT was at the limit I believe.
EGT for the last 21 seconds.

EGT last 21 seconds, degree C
The terrorist pilot pushed up the throttle angle (pushed up the throttles and left them at the max angle both throttle for the last 30 second of FDR. The throttle postion did not change for the EGT and RPM above. The airspeed went from 300 or so to 470 knots in thirty seconds.

Poor fan engine down low is going to have problems at max throttle position and max speed.

Too bad p4t have not studied the FDR. I had to get the new EXCEL to get to column ARR and find some of this stuff.

Are the fans happy at 400 plus knots? I was going to ask if anyone saw flames come out the front or hear choo-choo or some bang or pops. But the flame would be force back down the bypass I think at 400 knots.

Looks like the engines were rolling back on their own for some reason; who is an engine person?

 

[BCR]

Good find Beach! Hani may well have simply 'blown' the engine without any help from birds.

 

[R.Mackey]

Not blown, stalled. Holy cow are those high EGTs.

You'd lose N1 just because of crap efficiency. High EGT means either a fire or (more likely) high inlet temperature, which means less air mass through the core and therefore less thrust and less cooling. Couple that with the high speeds and you are begging for compressor stall.

At those temperatures the engine's control laws would probably try to compensate by running rich, blowing excess fuel to cool off the core, and that could also create a smoke trail. Of course, I don't know exactly how that particular aircraft behaves, so it might not -- you can also cool EGT by leaning out the mixture, but that's a bit more dangerous.

But is this enough to cause havoc on the power bus? Could be...