I'll try to articulate my question more clearly. An SUV is maybe not the best example but I'll try.
If you take the volume of each of the thousands of components that make up the SUV, one could simply sum these volumes and say that that's as compressed as the SUV will ever get, assuming we're not trying compress it to something that resembles a portion of a neutron star. This is a trivial exercise if given the data and one doesn't actually have to do the measurements.
But I'm thinking a bit more pragmatically. If one were to actually pack all the parts, say, for transport, and one wanted to fit all these parts into the smallest container possible, there are bound to be larger components, for example, the roof, that will stick out and force one to use an otherwise larger volume container. A way around this is to make the roof in smaller pieces that can be assembled when the shipment gets to its destination. But now the roof needs to have components that join it together increasing the total volume. Even if that means weld seams. On the other hand if it were possible to 'extrude' and 'inflate' a car in one piece (impossible, I realize), you would have far less joining components for all the pieces reducing material volume but the overall volume would be larger because a car has some empty space within its extremities and it can't be packed.
A (normal) car is not really a good example because it is made up of so many pieces and most of them are relatively not too big compared with others. What I'm trying to get at is this: is there a general method for optimizing a device for packing which takes into consideration its final form and function.
People actually buy all sorts of things based partly on the attribute that they 'pack flat'. But as far as I can determine that's about as quantitative as this property gets. I don't even know the name of this attribute except to call it 'compressibility'. I'm confident I'm not inventing a new science here. There must be lots of people who have thought about, named, and even found mathematical relationships for this.
Note: I'm note including any self-assembly attributes here.