In the post Garbage in garbage out, I talk about how the NTSB made numerous mistakes in their attempts to analyze and reproduce the FDR readout from TWA 841. Misreading the FDR readout is one thing, but what do you call it when the NTSB tries to draw comparisons between FDR readouts when the evidence clearly shows that no similarities exist? Such is the case with the acceleration trace comparison between TWA 841 and the Boeing flight test on October 2, 1980. But first a little background.
A little over a year into the investigation of TWA 841, the NTSB had settled on a theory that the upset and dive was caused by the crew making a positive command input that extended the trailing edge flaps and leading edge slats 2, 3, 6, and 7, followed by a positive command input to retract the flaps and slats but with the number 7 slat not retracting due to a misalignment.
They had no proof that this occurred. The crew adamantly denied any such positive command input. And there was nothing on the FDR readout that corroborated their theory. So after some prodding by ALPA, the NTSB agreed to do a flight test in a Boeing 727-100 in order to compare the FDR readouts between TWA 841 and the flight test aircraft before, during, and after extension of the flaps and slats 2, 3, 6 and 7.
The graphic below shows the actual acceleration trace readouts for TWA 841 and the test aircraft with the autopilot on and off. (You can view a larger image by clicking on the image. (Click the back arrow to return.)
You don’t have to be an expert in FDR analysis to see that neither test condition matches the readout from TWA 841. In fact, there are no similarities. The Investigator in charge, Dean Kampschror, had a huge problem. The flight crew, including the flight attendants, did not corroborate their theory. The passenger statements they had did not corroborate the statements from the passengers aboard the test aircraft who reported an abrupt and startling buffet when the flaps and leading edge slats were extended. And now he had physical proof by way of FDR readout that what ever happened to TWA 841 could not have been caused by the extension of flaps and slats. Kampschror was not alone in his skepticism. The NTSB employee assigned the task of making the comparison between TWA 841 and the flight test wrote the following note after first seeing the FDR results. (E209 refers to the flight test).
Present E209 data would not be convincing that all four slats extended at once. Any additional information Boeing can submit would be greatly appreciated. (Deposition of James William Kerrigan January 10, 1983. P 65 – 66.)
Kampschror also contacted Boeing and asked for an explanation. So the Boeing engineers went over the data and after four months of scratching their heads they had their answer. They can explain everything. The dissimilarities between the readouts was due to a difference in autopilots and a test switch that was left in the test position. Now they hadn’t actually seen the test switch in the wrong position, but based on their calculations and simulator tests that is the only explanation they could come up with. Had the test switch not been in the test position, according to Boeing, the acceleration traces would have been much closer.
But Kampschror wasn’t satisfied. He needed more evidence if he was going to hang the crew. That evidence came from NTSB employee Robert Von Husen. Von Husen decided that the problem was that the instruments they had for plotting the FDR readouts wasn’t accurate enough. So he devised a new technique. Here is how the NTSB described this new technique.
This technique involved making cellulose impressions of the foil traces and photographing the impressions under high (200 power) magnification. The photographs were then combined together and time and amplitude scales were calculated and drawn over the composite traces. (Aircraft Accident Report. “Trans World Airlines, Inc., Boeing 727-31, N840TW Near Saginaw, Michigan April 4, 1979.” NTSB-AAR-81-8. July 13, 1981. P 22.)
According to Von Husen, his new technique for reading the acceleration trace readout now showed an exact comparison between TWA 841 and the flight test aircraft with each showing a frequency of six cycles per second. Believe it or not, that is the evidence the NTSB staff used to convince the Board to vote on the probable cause. This same evidence would be used again in the civil litigation pinning the blame on the crew. The following quote is from the closing argument given by the Boeing attorney.
They [the flight data recorder tracings] tell us what actually happened on the airplane, and they tell us in an irrefutable way. We have that information. It can’t be deposed and changed, and the story can’t be manipulated. It’s hard scientific data. (Ratchye H, Boyd Closing Arguments June 8, 1983. P 2563.)
Here is what ALPA investigator Jim McIntyre had to say about the validity of the NTSB’s evidence.
The primary purpose of the flight test was to compare the FDR trace with that of the accident aircraft. The latest analysis of the autopilot by Boeing appears to throw into question the validity of the test flight results. Because the FDR traces were different, Boeing now wants to contend that this difference was due to (1) the failure of a flight test engineer to notice that a switch was in the wrong position and (2) a difference in autopilots — a factor not considered significant prior to the flight test. In fact, the report simply proves that the flight test did not meet the stated objective, i.e., “to obtain data which could be used for comparison with the accident Flight Data Recorder (FDR) and to assist in establishing the configuration changes that may have occurred on the accident airplane prior to loss of control.
The original readouts of the acceleration traces from both the accident FDR and the flight test FDR indicated different acceleration frequencies, thus indicating a possible difference in configuration between the two. Because this contradicted the staff’s preconceived ideas regarding the extension of all leading edge devices, a new method of readout was developed. The new readout method, according to the staff, now shows the same frequencies for the accident FDR and the flight test FDR.
Our own analysis will show that the results of the new readout merely render any conclusions regarding the acceleration trace frequency comparison as worthless.
Take another look at the readouts. Would you be willing to use that evidence to go against the sworn testimony of the flight crew?