You’re not showing the important screen shot, which is the “collision view” in the static mesh editor.
There are two modes of collision: “simplified” and “detailed.” The “simplified” mode assumes the object is “smallish,” e g, somewhere between soda can to car in size. You can give it a box, or some number of convex hulls, as its shape. The benefit of this collision mode is that it’s generally efficient/fast, and it can generate good physical response (including realistic mass distribution) when the object moves in the world.
The other mode is “detailed,” which uses per-triangle collisions. This can be a lot more expensive if you have a lot of triangles and they each are small, but it does make the collision match up to what’s visible on the screen. This is ideal for things like ray-casting to figure out where to place a decal.
In the collision setup for a static mesh object, you can check a box that says something like “treat detailed as simple,” which means that the physics engine will use the exact triangles, even for physical simulation. As long as your object isn’t simulated on its own (e g, it’s static ground/walls, not spinning around in the world) then this can still be efficient enough for what you need to do.
So, long story short: try “treat complex as simple” for collision and see how it goes. Keep an eye out for performance problems.