From a paper Sciencesplaining how it happened:

it was observed that if all structural members of the wing are lumped together and smeared into a box beam of equivalent mass, its thickness becomes over 100 mm which is ten times larger than the 9.5 mm thickness of the hollow external column of the Twin Towers.

They modelled it by creating a box beam. A wing is not a box beam. It's not even close to a box beam. It is a long thin honeycomb which has failure modes completely different than a box beam. It's not designed to handle point like impact forces on the leading edge. There are no substantial forces in that direction in the sky, not counting the occasional errant bird which is substantially squishier than steal beams. In the case of birds, it is estimated that 15% of bird strikes on the leading edge of airplane wings cause structural failure in the wing because they aren't built to handle those forces. They are just too rare to compromise the design.

If a wing can barely survive a bird, how is it going to cut through a box beam made of tempered steel?

Their initial assumptions were not inline with reality, and any engineer or physicist understands this. They call it a "first approximation," but it's full of bad assumptions.

After playing with their model for long enough, they got a result that "helped explain what we saw" in a "rational way."

That's all it takes for The Science to become reality, a plausible explanation. You keep adding the assumptions you require, and fiddle with the parameters until you build a model that gives the desired answer. No testing required.