Thank you, Dan, for the file and the trust.
I understand this is only part of the whole set up, and my reply can only use this fraction of it. My short answer is, it is not so much hidden inside an object that I would call a container, it is more a question, what will be visible when the camera is set (but there might be always a change)...
My suggestion would be:
The file comes in Groups, place all of them in separate Layers. This gives you an better management of the file. The file hold 223 object, and 223744 polygons. This works even fine on my five year old laptop. I wouldn’t worry to much about too many polygons right now. The top group has many sub-groups, these sub-groups can be established in the Layer Browser as well
After the Layers are established, you can “Solo” them, which results in an much nicer “review” of the whole scene. Perhaps you leave that step out and go directly to:
The next step is to find out what parts will a) never move, b) move, c) need any kind of deformer to move [hydraulic tubes eventually, if flexible] and d) move dependently, but not deform.
After you got all the group a) in one new Layer, you go ahead and create groups for each element that moves independently b) or c) and then place d) accordingly to b) or c)
There are hierarchies, some parts move others and at the same time they are moved by other itself. This needs to be clear, before you start. I use normally a paper for that, after the first rough organization. This helps normally to know where to start, and to not get lost in the amount of single objects.
If you select all (Object Manager) you can set their color now to Attribute Manager>Basic>Use Color>Layer. This helps the next steps a lot. You can switch this color of in the same way.
All of that work has familiarized you already a lot with the parts. This is, from my point of view, a needed step, as in watching footage before one can edit them.
The parts that needs animation might need a new Null to group them nicely [each time], if done in a clean way, the animation should be simple. (Ask me if there is a problem, or something needs to be cleared)
The amount of polygons is not the main concern here, my concern would be a clean mesh. There are a lot of disconnected polygons, etc, which will not render nicely. I have attached a tiny image [the first object in the list], cropped enough to give nothing away, where I used only the Mesh>Commands>Optimization function. Which helped a lot.
Again if you go through the polygons first, you might feel overwhelmed, but grouping them and color them will give you a better idea what really could be deleted. The question remains—if that makes really sense. What I would do first would be a exploration of how much the “Mesh>Commands>Un-Triangulation might help to improve. The first object had 11k polygons after that not even 9K, with a single function. A second one, a Hexagon-Socket-Screw, went from 210 to 110 polygons. (Object Manager>Objects>Object Information), but each use might show some other problems. Note that the smallest polygon in this case must be larger than the “tolerance value”, otherwise you might mess up the mesh. You need to check the Phong Tag as well, I set it to the values in the image. It is an function that needs control. Just using it for the sake of getting a little bit lighter model might end up in many hours work, to replace parts. Hence, I’m not so certain to invest a lot of time here in reduction. Does it really slows your computer down? Keep in mind that all quad polygons will be triangulated during render time. Since we have a mechanical animation here, certainly for the most parts, I see also based on that only a limited advancement here.
Just as a personal side note: The model that you got is compared to many CAD files I saw in my life—very nicely done.
Let me know if there is more to discuss. The mesh is a typical CAD mesh, long triangles and such, not the most desired way for visualization. But that might be a longer theme, and you let me know what kind of quality is expected after all.
One more concern to reduce it, in this case—not as a general rule—what would happen if the “client” decides after all is done, to have an explosion animation to see all parts involved, and then move all parts back, or animate this “explode” view for explanations. If that happens, and clients are that way, you might start over again.
I checked out the options of an Polygon Reduction, but many polygons are not connected, or reduce differently, so the points don’t match—many parts showed gaps after that—even with settings as low as 20%.
My best wishes for the project.