well I think I have answered my question.
I created a simple assembly, tested two configuration, one with the weight untouched and one with the weight overwritten and factored by 10, and the results are exactly the same.
I guess one way to adapt my problem is to tune the drop height so I get constant potential energy between the real product and the simplified drop test assembly?
Correct, all SW simulation ignores mass properties overrides.
I frequently start all my simulation work with a static 1g study. I can use this study to apply materials, setup all my contacts and make sure the model will mesh. I can verify the materials are setup correctly by examining the reaction force and verifying it is the same as that reported by mass properties. These setup items can then just be dragged/copied over to the study of interest.
If simplification for FEA is minor, like removing the components from a PCB, I'll just create a custom FEA material for that part with a hand-calculated density set to recreate the correct weight.
For the drop test maybe create a super-simple representation of the missing geometry and apply a custom FEA material for just this body so the weight is correct. Give it a very coarse mesh. This way the weight and weight distribution are about right, but the calculation time is still vastly simplified.
the custom material route was on my mind, but I tried to increase the drop height instead, to values maintaining the potential energy of the drop (so from 1.22m to... 28m!). Results look sensible so far.
I shall try the weight method as well, see if the results compare.