Topology Optimization for Thermal Metamaterials

[MinMorsh]

Hello,

I am working on reproducing this paper from the scratch (https://doi.org/10.1007/s00158-023-03717-3) where the authors used a Isogeometric level-set topology optimization technique to optimize thermal metamaterials. I was wondering whether Swan already has these capabilities or if it can be used to develop further to this.

Also, I looked at the 'Tutorials' folder and tried to run a couple of cases (e.g. TopOptTestTutorial.m), but I am getting this error:

Unrecognized function or variable 'mesh2tri'

Do I need to manually add the path of 'mesh2tri.m' file or am I missing something here?

Thanks,
Minhaz

[gervilquin]

Hello Minhaz,

We are glad that you are interested in our code.

I have sent your question to my colleagues working on topology optimization to have a precise answer. As soon as I get a response I'll let you know.

Regarding the tutorial error, mesh2tri was included in the Old and src folder. I have solved this problem and now it should work.
Please, update from master and let me know if you have any other problem.

Sincerely,
Gerard Villalta

[gervilquin]

Hello again Minhaz,

According to my colleagues, the isogeometric level set is not implemented, and we are not planning to include it in the short term. Still, you should find some thermal and metamaterial capabilities in Swan.

In case we can help you with something else, we will be glad to answer.

Best regards,
Gerard Villalta

[MinMorsh]

Hello Gerard:

Thanks for the update. I could see some metamaterials design problems in the SwanLab website. Are they available in this repository? If yes, can you please guide me where I can find them?

Best,
Minhaz

[gervilquin]

Hello @MinMorsh,

One approach is using RVEs with optimally homogenized properties. This is done in the tutorial TopOptTestTutorialMicro.m, where the parameters alpha and beta control the weight given to Chomog. In fact, the cost function is defined as
alpha · (Chomog⁻¹) · beta. For example, if alpha = [1 0 0] and beta = [1 0 0], you are minimizing 1/Exx (which means you are maximizing Exx).

The other approach is for achieving a negative Poisson’s ratio to create metamaterials with hinges. Also base your work on the tutorial TopOptTestTutorialMicro.m, but you need to replace MicroAlphaBetaFunctional with MicroFractionFunctional. For the negative Poisson’s ratio, you should use alpha = [1 0 0] and beta = [0 -1 0].

If you have other doubts, do not hesitate to send another message.

Best regards,
Gerard Villalta