Jacked In Home -> A total amateur analyzes the Columbia data

A total amateur analyzes the Columbia data

Sunday, February 09, 2003 -- by Jack William Bell


Warning, I am not a rocket scientist. I am not even an aeronautics engineer. Hell, about the only thing I am qualified for is making those guys coffee, and I am only a gifted amateur at that as I am not a barista either... Please take everything in this document with approximately a metric ton of sodium chloride.

Updates and comments on this document can be found on my Slashdot journal in this entry.


When I first saw the NASA technical briefing slides showing shuttle left-wing sensor readings during the critical eight minutes prior to loss of signal I immediately started to grasp the failure mode. The problem was that the static slides still did not give me a real feel for the order of events.

Well, in the spirit of 'scratching my own itch', I put together an animation of those eight minutes from the NASA slides. Because eight minutes is quite a long time, I time-compressed at around 25 to 1, resulting in a twenty-second clip. Keep the actual time in mind when viewing it:

As you can see, the order of events is pretty clear. The animation begins at 13:52:00 GMT. Approximately 13:52:20 GMT some temperature sensors in the wheel-well start to show an anomalous rise in temperature. Quickly afterwards other sensors, far away in the back of the wing (but with their wiring harness placed between the wheel-well and the leading edge of the wing), suddenly fail. Then, after 13:54:13 GMT, more temperature sensors in the wheel-well show a rise in temperature followed nearly four minutes later by the wheel sensors failing. Twenty seconds later all telemetry is lost.

Other sensors located in the back of the wing, which had a wiring harness behind the wheel well, never did fail prior to loss of signal. Combine that factor with this image taken by high-resolution Air Force cameras:

It seems likely there was a loss of integrity in the leading edge heat shielding right in front of the wheel well. Note that these shields are not silicon tiles! Instead they are made of carbon and are thick blocks with a rounded edge that are bolted to the wing itself (which is flat on the front). These shield blocks are fastened so that they are allowed to expand with high temperature and they have 'T' pieces of carbon shielding that fit between them.

The failure mode itself might very well turn out to be quite minor at first, perhaps losing one of the in-between bits, or the bolts might not have allowed proper expansion. Perhaps some impact may have cracked and/or completely destroyed an entire shielding block. No matter, because the temperatures we are talking about are enough to cut through the aluminum structure of the wing like a hot knife through butter. This 'blow-torch' would not only have allowed hot air into the wing, but would have scattered molten metal droplets about as well. This scenario perfectly explains the sensor data to my, admittedly, uneducated eyes.

It will be interesting to see how close to the mark I am when NASA releases the official report. I will update this page at that time.


For those curious about the animation: I used the NASA slides for image sources and created the animation using Corel PhotoPaint 9. (Remind me next time I do something like this to use Flash or some other software with decent animation facilities. It was flat painful in Corel, plus Corel seems to have problems exporting to animated GIF format.) I set the animation speed to five frames a second and did one frame every five seconds. I also added some frames at the end to provide a second of waiting before it loops. It took me about fifteen to eighteen hours of work all told. If anyone needs the file as an MOV or even AVI, please contact me.

 Jacked In Home -> A total amateur analyzes the Columbia data