While I can't peek at CAESAR's code, I would suspect that when you input things correctly, it assumes a uniformly corroded pipe for purposes of strength, but a brand new pipe for purposes of weight calculation.

Thus, the second method (with the 0 CA) is "lighter" in the eyes of CAESAR, and therefore has several interactions. Support friction doesn't work as before exactly. Piping appears to be stronger.

But that would be my guess.