Perhaps because the forces are lower during firing?
keeping it really simple,
change in momentum = impulse = F*t
when it strikes the ground, the time to deccelerate is much shorter (therefore the force is much greater) than when it is fired.
When it is fired, it accelerates over the length of the barrel from the expanding gasses. (Still, a very short amount of time, but more time than when it strikes the object)
Just for a rough idea, let's ignore air resistance and assume that it strikes the object with the same velocity as it has when it leaves the barrel. It goes from that velocity to 0 over a very short distance, say a few centimeters as the object deforms and the shell deforms somewhat during the impact. At the other end, they've got a nice long barrel, so it's accelerating along a distance of a couple meters, maybe 50 times longer than the distance it travels during impact, so the force at impact will be roughly 50 times greater.
Of course, I know nothing about these shells, so I may be wayyy wrong, but it's an idea
I haven't got a clue how they develop those bombs that penetrate 2 layers of concrete to hit deep bunkers.
Here's another example to make it very easy to envision:
Needed: 1 waterballoon, 1 waterballoon slingshot, 1 brick wall.
Fire the balloon... they usually don't break when being fired. Will break when it hits the wall.
2nd experiment: make a very simple rocket by putting fins on a 2-liter bottle, filling it up 1/2 way with water, and pressurizing it to about 100psi. (you would need one of those launchers that are easy to make or can be found at several science supply places). Place egg on top. Fire rocket. The egg will pretty much explode during the first small fraction of a second, almost as if it had been thrown downward and hit the bottle squarely. It can't handle that much acceleration. (IIRC, roughly 100 g's)