Since version 2021 Solidworks Simulation will sometimes take an inertial load and apparently amplify it by a wild factor. It looks like everything in the assembly works as expected, the loading is just grossly over-applied.
[I documented this, and other problems, and the support team told me to simplify things and 'use more bonded contact', picking out details of the setup and never acknowledging the gross errors in the solution. I've stuck with version 2020 since then, after I personally got a dozen SPRs pushed through until 2020 was a sweet spot for performance and stability. But 2020 has a hard limit on problem size and this assembly just has too much going on to keep the mesh count low.]
These two studies are consecutive runs, one 'duplicated' from the other. The only load is a gravity vector. At first the vector was defined on a line in a 3D sketch. Many years ago there was a problem with using non-solid geometry to define load vectors, so I tried again with some solid pyramids to give the gravity directions. I can delete and re-make the gravity load and get a different result every time. One time in ten or so it will look 'reasonable', but there's no way to be sure that solve is correct.
Has anyone else seen this and found a way to deal with it? I can apply forces instead of gravity, but each subcomponent adds significantly to the total loads so I'd have to find a force for each one and a place to apply it near each CG.
My long-term solution is that Ansys is being set up on my machine right now. But in the mean time I need to get this study completed.
