For a liquid fueled rocket you have many opportunities before launch such as firing the engine in a test stand, etc. For the whole rocket you can do a static fire, or even a full duration fire to test the components. Mostly when you are moving you are worried about the mechanical loads on the rocket instead of having to worry if parts were put together correctly. Of course, even this level of testing takes a lot of money, capital, and of course labor.
For a solid fueled rocket, you're right that pretty much you have one shot, and you can't turn it off if something goes wrong.
> My gut feeling about 3D printing is that it just makes it worse. How do you inspect a welding line as long as a rocket's dry mass?
In general, the way you do this is to move it lengthwise through an x-ray scanner, so you can inspect the full length by moving the rocket through. I'm not sure if 3d printing makes this easier (because there's not really welds, it's one piece) or harder (you have to be able to inspect the inside of something), but it is certainly tricky.
I guess you can think of it that way, but it probably depends on the technology used by the printer at that point. My thought was that for the 3d printer, at least it's one consistent piece with no seams or edges that are later welded together. The welds can have imperfections/weaknesses and need to be inspected. Materials can also have weaknesses and imperfections, either on a batch level or individual area level, and need to be tested/inspected. For a printed piece, I would expect the strength to be generally consistent throughout the piece (of course still might be weaknesses based on design shape, thickness), but of course, you'd still want to inspect it.
PS - totally not an expert on 3d printers or materials or welding
I am not an expert on either, but in general 3D printers lay out the material in thin layers (with each layer laid out line by line). Thus you have lots of potential seams everywhere two points laid at different times touch.
I think modern technologies prevent seams forming at most of those points, but the potential (impurities, dirt, etc.) still remains so must be tested.
I could imagine some technology that takes advantage of future layers not being there yet to validate the most recent layer on the same pass, and the one that the current layer will be built upon (as that might have suffered in the time since it has been laid) in a way that is exclusive to this manufacturing method. But as you move to bigger and bigger one-piece parts, you would probably also want your process to support undoing a few layers on failure detection to forego scrapping the whole part.
For a solid fueled rocket, you're right that pretty much you have one shot, and you can't turn it off if something goes wrong.
> My gut feeling about 3D printing is that it just makes it worse. How do you inspect a welding line as long as a rocket's dry mass?
In general, the way you do this is to move it lengthwise through an x-ray scanner, so you can inspect the full length by moving the rocket through. I'm not sure if 3d printing makes this easier (because there's not really welds, it's one piece) or harder (you have to be able to inspect the inside of something), but it is certainly tricky.