AA77 FDR Data, Explained

[W.D.Clinger]

If this helps you, I used GIS mapping software to create a profile of the terrain under AA77's flight path for the approximate length of time of your data up to the impact point. I plotted it over your second graph to show you the comparison:

[qimg]http://www.internationalskeptics.com/forums/imagehosting/thum_159394ae31a8c73832.jpg[/qimg]

That's very helpful. Thank you.

The calculated peaks in the ground elevation at 1:37:50 and 1:37:44 line up pretty well with green peaks (in the known terrain). The calculated valleys at 1:37:32 and 1:37:13 line up well enough with green valleys, considering the ground speed wasn't constant, the x-axis for my graph is time, and the x-axis for your graph is distance.

That puts the last elevations recorded in Warren's CSV file roughly 3 seconds from impact at the Pentagon. Not knowing when the instruments were sampled within each 1-second interval, the uncertainty is plus/minus 1 second. So the radar elevation of 4 feet would seem to have been obtained roughly 2 to 4 seconds before impact.

That far from impact, it looks like the tallest obstacles along the approach path are the Navy annex, the trees along Columbia Pike, the Citgo station, the tree(s) next to the cloverleaf, and the light poles. From comparing the vertical accelerations with the four seconds of 1.6g needed to hit the Pentagon in level flight after clearing the Navy annex, it looks to me as though the plane is roughly 2 seconds past the Navy annex, making the Citgo, cloverleaf trees, and some of the light poles the most likely things to have been cleared by 4 feet plus the height of the landing gear.

Will

 

[beachnut]

... 1000 foot altimeter lag for dives and spins

An instrument that lags, is plus or minus 75 feet for takeoff.
The terrorist pilot is making erratic dives (not smooth) approaching the Pentagon, there can be a 75 foot lag for his dive (as seen on studies if p4t were able to do real research).

Based on studies with large aircraft altimeter errors approach 150 for airspeeds as high as 77 was doing! Good news, when large jets operate near their approach speeds errors are back to 25 to 50 feet. But relax, these errors are all designed in the flight procedures.

Good grief, -+75 feet, lag of 50 to 75 feet, high speed errors 100 to 150 feet reading high. Uncertainly, 100 to 300 too high. Pressure altitude, this is why when you fly the next plane is suppose to be 2000 feet below or above you flying high and fast.

100 to 300 feet uncertainly in the pressure altimeter for the final seconds is not out of the question. Balsamo understanding 9/11, probability zero.

Balsamo is defending pressure altitude +100 to +300 too high, vs RADALT plus or minus one foot. Can’t wait for him to explain away the pitch angle and the DNA. What a moron.

When you sell lies on DVD, acting as dumb as your p4t members posting at the p4t forum is necessary. Balsamo's best science efforts are failed 11.2G messed up moronic physics/math.

 

[tfk]

WD,

As an aid to visualization, I have created two crude graphs. The first shows both uncorrected pressure altitude and radar altitude, as a function of time, for the final 60 seconds of Warren's FDR data:

http://www.ccs.neu.edu/home/will/Music/Jokes/Balsamo/graph1.jpg

The radar altitude apparently pegs at 2795/4000/4000/4000 when the radar altitude is above 2600 feet or so, but the radar altitude tracks the pressure altitude reasonably well for the last 40 seconds.

I wonder if those two peaks (above 2600') represent roll angle of the plane, i.e., the last roll out before the straight in run. I would assume that the radar antenna is emitting in fixed direction (downward relative to the body of the plane), and that any roll would give you a (1/Cos(Roll Angle)) multiplier (assuming flat terrain).

That might also explain all the positive excursions, since a roll in either direction will be perceived as an increase in height.

Tom

Added: I think that they have to reflect roll angles. That is the only attitude that can change as rapidly, or as much, as they are.

Clearly altitude cannot change that fast. Yaw would have no effect & you aren't gonna change pitch by more than a few degrees.

 

[R.Mackey]

This is where despite Bobby taking you to task for supposing a condition existed by which the air pressure reading was affected by the operating conditions, he is quite willing to suppose that these same operating conditions may have set up severe control issues, specifically 'flutter' of control surfaces that were close to the natrual harmonic frequency of those controls.

Seems he wants it to mean that IF air pressure affected the pressure reading then it must also cause the aircraft to be uncontrolable. Quite amazing and amusing really. Cappy is grasping at straws. The decode radalt indicates the last altitue of just 4 feet agl so he is going off on MacKey's explanation of how the PA could be affected by the operating conditions of this flight.

I continually have to revise my estimate of Cap'n Bob's incompetence -- which is why I usually ignore him. The vapor cone effect is real and quite obvious when it happens, and it proves that compressibility can lead to reduction in static pressure. Obviously not for every point on the airframe, but nobody said that in the first place. If he doesn't like it, he can take it up with Dr. Prandtl. No amount of braying on his part will erase the fact that he doesn't know the difference between compressibility and compression.

This "flutter" argument is even stupider, if that's possible. Boeing aircraft typically cruise above their critical Mach number. They're designed to do it. Up at altitude, they cruise around 0.8 Mach or so all day long. This is done using supercritical wing design, boundary layer control, large and carefully thought control surfaces, etc. So the idea that AA 77 suddenly became uncontrollable when it hit 0.72 Mach on the deck is... bewildering.

I really did not intend this to turn (again) into a Let's Analyze Cap'n Bob's Stupidity thread. I'd rather turn it back to Warren's analysis.

 

[W.D.Clinger]

I wonder if those two peaks (above 2600') represent roll angle of the plane, i.e., the last roll out before the straight in run.

No, but that was a hypothesis worth considering. Throughout Warren Stutt's CSV file, whenever the altitude is above the apparent operational limit of the radar altimeter, the RADIO HEIGHT CAPT (FEET) is recorded as 2795 and the other three are recorded as 4000:

RADIO HEIGHT LRRAL (FEET)
RADIO HEIGHT LRRAR (FEET)
RADIO HEIGHT LRRAC (FEET)

The FDR recorded only one of those four measurements per second, cycling between them. For the purposes of my graphs, I lumped them all together as the radar altitude. The two peaks at the beginning of my first graph are just an artifact: two of the 2795/4000/4000/4000 cycles. After that, the altitude is low enough for the radar altimeter to provide real data.

Will

 

[Hokulele]

Can we have another thread for Let's Analyze Cap'n Bobby's Stupidity? I am actually learning a lot from this and would hate to see it end prematurely.

 

[bje]

That's very helpful. Thank you.

The calculated peaks in the ground elevation at 1:37:50 and 1:37:44 line up pretty well with green peaks (in the known terrain). The calculated valleys at 1:37:32 and 1:37:13 line up well enough with green valleys, considering the ground speed wasn't constant, the x-axis for my graph is time, and the x-axis for your graph is distance.

That puts the last elevations recorded in Warren's CSV file roughly 3 seconds from impact at the Pentagon. Not knowing when the instruments were sampled within each 1-second interval, the uncertainty is plus/minus 1 second. So the radar elevation of 4 feet would seem to have been obtained roughly 2 to 4 seconds before impact.

That far from impact, it looks like the tallest obstacles along the approach path are the Navy annex, the trees along Columbia Pike, the Citgo station, the tree(s) next to the cloverleaf, and the light poles. From comparing the vertical accelerations with the four seconds of 1.6g needed to hit the Pentagon in level flight after clearing the Navy annex, it looks to me as though the plane is roughly 2 seconds past the Navy annex, making the Citgo, cloverleaf trees, and some of the light poles the most likely things to have been cleared by 4 feet plus the height of the landing gear.

Will

Here is a version that uses the DCA radar file for both ground path and position times.

The yellow numbers refer to the Segment Index in the Excel file and the associated DCA times and the terrain height above MSL at those times.

The only change I made was the repositioning the impact point from the DCA location to the actual location.

 

[beachnut]

Can we have another thread for Let's Analyze Cap'n Bobby's Stupidity? I am actually learning a lot from this and would hate to see it end prematurely.

11.2Gs of stupidity; strange how someone is able to make that big of a moronic mistake and have people praising his paranoid delusions.

For altimeter systems most of the known errors have to be automatically corrected for the airspeeds and flight conditions the aircraft is certified for.

All aircraft for which a type certificate is issued subsequent to the effective date of this regulation shall have the static pressure system error determined throughout the altitude, speed ranges, and aircraft configurations for which the aircraft is to be certificated.

The 757 was not certified to fly 483 KIAS at 200 feet MSL.

Even with corrections, the range of error can be up to 100 to 200 feet.

Means shall be provided to apply corrections automatically to the indication of each required altimeter if the total error (instrument plus static pressure) exceeds 100 feet at altitudes below 29,000 feet or 200 feet at or above 29,000 feet, except that, for aircraft type certificated prior to the effective date of this regulation, it shall be acceptable to use calibration cards in lieu of the automatic device.

It will be hard to know the error for a plane flying outside the certification airspeeds, but with only the allowable errors we have pressure altitude during operations of 100 feet or greater.

The last 5 seconds of RADAR Altimeter readings. p4t experts failed to decode the last five seconds after they were given all the data Warren had; which they also never knew they had the last 4 or 5 seconds. 273, 233, 183, 89, 57, 4 ; 273 is what p4t decoded years ago, now we see the rest of the values. The best part about this, I never needed the raw data full decode of the FDR to know Balsamo is a paranoid fringe nut case conspiracy theorist selling lies on DVD.
Hard to find dumber poppycock than Balsamo's DVDs, save his drones blindly praising his delusions at p4tf.

57 feet AGL will not make it over the Pentagon. 4 feet AGL will not make it over the Pentagon.

Hope Warren adds the pitch angle to his decode. The average in the last few seconds was -5 degrees. Is there a VVI in the data?

 

[WildCat]

I see Cap'n Bob's money-o-meter is stuck at $0.00, far short of the $1,500 he's asking to keep his "research" going...

 

[apathoid]

Some systems though are completely unaffected by the outside air in any way. One of those of course is the radar altimeter. It will function the same at all speeds and for all clear air, pressure conditions one might experience on Earth.

Seems that it has a max range on 3000 feet agl but by the time one is at 3000' agl one would not really need a precise altitude measurement (just need to stay at assigned level +/- 50 feet IIRC for IFR flight, and all VFR flights I have been on stayed below 10K feet)

I wonder if those two peaks (above 2600') represent roll angle of the plane, i.e., the last roll out before the straight in run. I would assume that the radar antenna is emitting in fixed direction (downward relative to the body of the plane), and that any roll would give you a (1/Cos(Roll Angle)) multiplier (assuming flat terrain).

I believe that the Radalt receivers compensate for roll up to a limit and then they "flag" inop beyond that. Radalt has an upper range of 2500', above that - the cockpit display either flags or disappears(depending on carrier preferences).

 

[W.D.Clinger]

Here is a version that uses the DCA radar file for both ground path and position times.

The yellow numbers refer to the Segment Index in the Excel file and the associated DCA times and the terrain height above MSL at those times.

The only change I made was the repositioning the impact point from the DCA location to the actual location.

[qimg]http://www.internationalskeptics.com/forums/imagehosting/thum_159394ae350a71bb62.jpg[/qimg]
[qimg]http://www.internationalskeptics.com/forums/imagehosting/thum_159394ae350c3c5be5.jpg[/qimg]

Thank you again. Comparing peaks and valleys of the calculated ground profile with the profile you supplied, it appears to me that the times associated with the FDR altitude data are within a second or so of the DCA times. It also looks as though the altitudes recorded in the last FDR subframe (at 1:37:52) are closer to 4 seconds away from impact than to 2 seconds. The DCA times support that as well. Backing away from my previous speculation, that makes the Navy annex and the trees along S Columbia Pike more likely candidates for the last recorded radar altitude of 4 feet.

That's just my speculation based on eyeballing the two graphs. I don't know how to take it further, so I'll return to lurking. I really appreciate your help, and I'll close by thanking Warren once again for uncovering the last four seconds of FDR data.

Will

 

[bje]

Thank you again.

You're quite welcome, thanks.

 

[apathoid]

The vapor cone effect is real and quite obvious when it happens, and it proves that compressibility can lead to reduction in static pressure. Obviously not for every point on the airframe, but nobody said that in the first place. If he doesn't like it, he can take it up with Dr. Prandtl. No amount of braying on his part will erase the fact that he doesn't know the difference between compressibility and compression.

Ryan, couldn't the engineers program the Air Data Computer with the known effects of compressibility at speeds in the 757's certification range? I know that the ADC factors in AoA vane angle above MACH 0.55(I know this because recently on a 767 static accuracy test, I was reading between 120-250' low at my test points because I had the AoA protracters swapped). I also know that you can't simply swap an ADC between a 757 and a 767, even though the Airspeed Indicators and Altimeters can be swapped, because the ADC has plumbing lengths, and others airplane specific data programmed in.

But even if the ADC's are programmed for this effect....since AA77 was above VMO and out of its certification range, perhaps the error could still be quite large.

ETA: I was wondering if the AoA vane angle would also throw off the reading, if the ADC factored it in above M0.55 as it does the 767.

 

[R.Mackey]

Ryan, couldn't the engineers program the Air Data Computer with the known effects of compressibility at speeds in the 757's certification range?

Probably. If the airflow is consistent rather than buffeting in that area, it's possible, and they do, but...

But even if the ADC's are programmed for this effect....since AA77 was above VMO and out of its certification range, perhaps the error could still be quite large.

Exactly. Like I said above, I am willing to bet that no 757 has ever flown this fast, this low, and survived. It's totally out of calibration. There's no reason to assume the behavior is linear, so you can take a guess at it from lower speed testing, but it's only a guess.

 

[beachnut]

FDR has impact recorded ?
LATERAL ACCELERATION (G's) -0.564 word 227/256
LONGITUDINAL ACCEL (G's) -1.083 word 225/256
The longitudinal accel on take off = .363 max.

155/800 second critical word 243

 

[Reheat]

I believe that the Radalt receivers compensate for roll up to a limit and then they "flag" inop beyond that. Radalt has an upper range of 2500', above that - the cockpit display either flags or disappears(depending on carrier preferences).

The ones that I'm familiar do display an off flag past x degrees of bank. Sorry, I don't remember the numbers.

 

[Reheat]

FDR has impact recorded ?
LATERAL ACCELERATION (G's) -0.564 word 227/256
LONGITUDINAL ACCEL (G's) -1.083 word 225/256
The longitudinal accel on take off = .363 max.

155/800 second critical word 243

Can't say for sure yet, but he sure as hades wasn't pulling up to overfly the Pentagon while pulling those negative G's!

Is the longitudinal accelerometer in error? It should be more on take off or is the index zero for that measurement? Even so, it seems to me it should be more.

ETA: Of course that index would be zero as that is the normal unaccelerated state! Boing, (as he conks himself on the head!)

 

[beachnut]

Can't say for sure yet, but he sure as hades wasn't pulling up to overfly the Pentagon while pulling those negative G's!

The lateral accer in level flight is typically 0.001 and longitudinal accer 0.039 for this flight prior to the murderers flying.

The impact, or last reading for vertical accer, 1.945 /max vertical accer for flight 2.264, min 0.306G.

The lateral acceleration being negative is which direction?



Can we get this parameter decoded?
Uid: RAD_H_F
Abbrev: RAD_H_F
Name: RADIO HEIGHT F/O
Units: FEET
Minimum Value: -256
Maximum Value: 255.5
Digits Displayed: 2
Signed Value: Yes
Parameter Type: Linear
Format is y = m*x + b: m = 0.5, b = 0
Sampling Freq.(hz): 1
Number of bits: 10
Locations/value: 1
Frame(s) Subframe(s) Word Start Bit End Bit
ALL 1234 204 3 12
Number of Tests: 0

pa word 29/256
PA f/s f/min
307
239 68 4080
173 66 3960 - time 09:37:44
106 67 4020
35 71 4260
-40 75 4500
-99 59 3540
Based on the pressure altitude alone during the last recored second, early in the subframe, we have a 3540 foot descent rate.

 

[apathoid]

The lateral accer in level flight is typically 0.001 and longitudinal accer 0.039 for this flight prior to the murderers flying.

The impact, or last reading for vertical accer, 1.945 /max vertical accer for flight 2.264, min 0.306G.

The lateral acceleration being negative is which direction?



Can we get this parameter decoded?
Uid: RAD_H_F
Abbrev: RAD_H_F
Name: RADIO HEIGHT F/O
Units: FEET
Minimum Value: -256
Maximum Value: 255.5
Digits Displayed: 2
Signed Value: Yes
Parameter Type: Linear
Format is y = m*x + b: m = 0.5, b = 0
Sampling Freq.(hz): 1
Number of bits: 10
Locations/value: 1
Frame(s) Subframe(s) Word Start Bit End Bit
ALL 1234 204 3 12
Number of Tests: 0

pa word 29/256
PA f/s f/min
307
239 68 4080
173 66 3960 - time 09:37:44
106 67 4020
35 71 4260
-40 75 4500
-99 59 3540
Based on the pressure altitude alone during the last recored second, early in the subframe, we have a 3540 foot descent rate.

The impact of the light pole and subsequent engine "burp" could have caused some accelerometer noise.

 

[tfk]

No, but that was a hypothesis worth considering. Throughout Warren Stutt's CSV file, whenever the altitude is above the apparent operational limit of the radar altimeter, the RADIO HEIGHT CAPT (FEET) is recorded as 2795 and the other three are recorded as 4000:

RADIO HEIGHT LRRAL (FEET)
RADIO HEIGHT LRRAR (FEET)
RADIO HEIGHT LRRAC (FEET)

The FDR recorded only one of those four measurements per second, cycling between them. For the purposes of my graphs, I lumped them all together as the radar altitude. The two peaks at the beginning of my first graph are just an artifact: two of the 2795/4000/4000/4000 cycles. After that, the altitude is low enough for the radar altimeter to provide real data.

Will

I believe that the Radalt receivers compensate for roll up to a limit and then they "flag" inop beyond that. Radalt has an upper range of 2500', above that - the cockpit display either flags or disappears(depending on carrier preferences).

The ones that I'm familiar do display an off flag past x degrees of bank. Sorry, I don't remember the numbers.

Always good to close the loop to be able to distinguish "an idea" from "a good idea".



Thanks, guys.

Tom