Since you have a relatively small and simple weldment here I would try running it as a solid body. This will save you the trouble of trying to get the correct contact sets working between the solid bodies and the beams and you will for sure be able to use the no penetration contact. This will require more resources though and the use of a no penetration contact set will dramatically increase the solving time. Since you have a large body (x-ray enclosure) in your model you will have to create a mesh with a very coarse global size so that you do not create a lot of pointless and resource draining nodes and elements. This will probably cause the other bodies to fail but you can use mesh controls to create finer meshes on these parts.
Just wondering, when you ran it as a weldment did you use the remote load/mass or just equivalent forces? This may give you another method to compare results to.
I ended up running it as a weldment with loads applied to the beams and constraints at the bottom joints. I did some hand calculations to verify the results. I tried running it as a solid, but it took too long.
Mark, does the box just sit on the frame in the real world? Is it bolted at a few points? Welded in? I ask because if it is actually welded in, it will probably stiffen the structure dramatically.
Also, may I ask why you aren't exploiting symmetry? It appears the model has a high degree of symmetry -- you could probably get away with modeling 1/4 section. This would be particularly valuable if you are going to proceed with a model using solid elements.
The lead box just sits on the frame and can be lifted off with a crane. So I divided the weight in half and applied it to the two supporting beams. I also applied gravity because the frame weighs some 650 pounds. Notice the bottom crossmembers are split at the longitudinal members. These are forklift tubes. The designer put in 16 gussets to beef it up at the junctions. Those are the points of highest stress. This analysis was run as beams with no gussets.
Here's the stress plot.