Thanks for the new videos about topology optimization. Not only 3D but also for 1d which is often understimated. The videos are well done and extremely clear.
Probablly the best rib pattern I have seen on the internet. I am also running topology optimization on a bracket using Normal modes as a load case. Unfortunately I am not getting such a nice rib pattern. All I am getting are big holes in the bracket. Do you have an example where you can achieve similar rib patterns using Normal mode as a load case ?
I don't know if you'll read this or know what I'm talking about, but it seems like the penalty factor for SIMP is too low. This is because your final result has a full range of densities from 0 to 1, but with a higher penalty factor, density would be pushed toward either 0 or 1, making it a more realistic solid-void design.
I'm working with a topology optimisation algorithm called BESO for my research, I'll run the same case and see what result I get (though I can't apply all the same constraints you have with my implementation)
Please keep me updated about your research. I want to use BESO for a highly nonlinear optimization problem. I want to optimize the inner core structure of a sandwich sheet that undergoes a U-Bending. It shall be done by topology optimization, but so far I face to many difficulties with SIMP and RAMP such as errors an non-satisfactory results. Would you recommend using BESO instead? Regards Lukas
That's not how this type of topology optimisation works. The result is actually a continuous density field, where efficient material ends up with high densities and unnecessary material ends up with low densities, even if a max stress constraint is applied. By using something called "SIMP", the algorithm pushes material toward 100% dense (solid) and 0% dense (void), but there are still intermediate densities left over. The user has to pick a "cut-off" density, above which we consider material to be solid and below which we consider it void. Most FEA programs implement this with a view cut along iso-surfaces of density, so here the cut-off density is just called the "current value". There are ways to avoid this problem (e.g. increasing the SIMP exponent as the optimisation progresses, using other topology optimisation algorithms like BESO), but most cause the optimisation to take even longer, or they have their own issues. Hope this makes sense, its my PhD topic lol
![Topology Optimization using Hypermesh [Optistruct Tutorial]](http://i.ytimg.com/vi/-Uw7YWTy5so/mqdefault.jpg)
Thanks for the new videos about topology optimization. Not only 3D but also for 1d which is often understimated.
The videos are well done and extremely clear.
Probablly the best rib pattern I have seen on the internet.
I am also running topology optimization on a bracket using Normal modes as a load case.
Unfortunately I am not getting such a nice rib pattern.
All I am getting are big holes in the bracket.
Do you have an example where you can achieve similar rib patterns using Normal mode as a load case ?
I don't know if you'll read this or know what I'm talking about, but it seems like the penalty factor for SIMP is too low. This is because your final result has a full range of densities from 0 to 1, but with a higher penalty factor, density would be pushed toward either 0 or 1, making it a more realistic solid-void design.
Awesome... how to i output the optimised design as a STEP file?
I'm working with a topology optimisation algorithm called BESO for my research, I'll run the same case and see what result I get (though I can't apply all the same constraints you have with my implementation)
Please keep me updated about your research. I want to use BESO for a highly nonlinear optimization problem. I want to optimize the inner core structure of a sandwich sheet that undergoes a U-Bending. It shall be done by topology optimization, but so far I face to many difficulties with SIMP and RAMP such as errors an non-satisfactory results. Would you recommend using BESO instead?
Regards
Lukas
how to block geometry in hypermesh?
What I have never understood is this "current value". Why does the program not choose the optimial value to withstand the loads?
That's not how this type of topology optimisation works. The result is actually a continuous density field, where efficient material ends up with high densities and unnecessary material ends up with low densities, even if a max stress constraint is applied. By using something called "SIMP", the algorithm pushes material toward 100% dense (solid) and 0% dense (void), but there are still intermediate densities left over. The user has to pick a "cut-off" density, above which we consider material to be solid and below which we consider it void. Most FEA programs implement this with a view cut along iso-surfaces of density, so here the cut-off density is just called the "current value".
There are ways to avoid this problem (e.g. increasing the SIMP exponent as the optimisation progresses, using other topology optimisation algorithms like BESO), but most cause the optimisation to take even longer, or they have their own issues.
Hope this makes sense, its my PhD topic lol