I just came across this article that might help explain things somewhat.

I've been suggesting for a few days now (a href=http://www.habitablezone.com/space/messages/30084.html>here and here) that the insulation might've been harder than modeled because it was so cold. The article suggests that ice may have formed behind the foam when rain seeped into the insulation and froze. In this scenario, the ice would've weakened the bond between the insulation and the tank and might've made the piece a lot heavier than the estimated 2.5 pounds.

The ice forms because the cyrogenic propellants in the External Tank (ET) are incredibly cold. The LOX is almost -300 degrees Fahrenheit and liquid hydrogen is well below -400 degrees F. The cold temps cause condensation from the water vapor in the air (Florida is very humid), which then freezes on the insulation. Putting some sort of heater circuit on the foam might reduce the ice formation, but it also adds weight and complexity to the tank. It might even increase the possibility of an accident due to an electrical short. Perhaps there's another way to accomplish the same thing by spraying a coating on the foam, perhaps some form of antifreeze to reduce freezing or a slick coating to cause water to run off faster.

I came across one complicating factor with the foam insulation on the ET recently - it seems the ET shrinks about 6 inches after being filled due to contraction from the cold temperatures. Metal and foam have different expansion coefficients, so the foam might not contact as much as the metal. That could leave voids and weaken the bonding between the insulation and the metal.