Lycoming wrote:
Then again, I've seen topology optimization results for a tub that had criss-crossing uni tapes all over it, something like a spaceframe on the inside skin of the tub, but you don't need 3D printing for that. The advantage there would be primarily be, I'm guessing, if your torsional stiffness requirement greatly exceeds your side impact requirement.
The problem with the tub is that it on the outside is constrained by aerodynamics, the inside by a human, so you don't have much freedom to begin with.
Then you have the requirement that it has to be impenetrable for for instance pieces of suspension. That means that the walls of the tub have to be closed and thick.
That means the topology optimization can only play around with placing some stringers here and there, which don't do too much except mitigating shear type deformations.
As I said that topology optimization would probably draw the same box an engineer would.
Topology optimization, at least in the results I've seen, tend to produce truss-like structures. Note that I say this without a real understanding of how the algorithm actually works; it's just an observation based on a handful of results.
One of our modellers described it to me once as doing an FEM calculation and then throwing away the non-stressed material. Learned since that it is slightly more complicated but not by that much.
What I do see is that the human input on the model is still quite large both in placing the constraints and in post processing.
The fact that the final designs end up being quite familiar structures like webbing, trusses, diamond structures etc is more a result of the human input than that they are really the most effective.
