Optimization with Galapagos

Hi Jon,

I finished the modelling part of the project. What I wanted to do is to optimize the structure with Galapagos with the data that Geometry Gym obtained from GSA.

I set different variables and one fitness function that needs to be minimized (the weight).

When I start the optimization solver, the program switches to 'not responding' as well as Grasshopper and Rhino. When I try to use less variables, or even only one variable (genome), this still happens.

Is there a way to prevent this from happening? I know this is a complicated model, but eventually Galapagos needs to gather the data from each variable, right? I also tried to hide the model in Rhino, but this doesn't help much..

I hope you can help!

Best regards,

Stefan

Attachments:

6 - Tube bridge - zelfde straal gehele brug.gh, 118 KB

Replies to This Discussion

Permalink Reply by Jon Mirtschin on May 28, 2015 at 3:25pm

Hi Stefan,

Looks good. I didn't study your definition too closely but I'm assuming the objective also includes a stiffness check (else to minimize weight you simply put as few elements in as possible at the smallest size).

I started Galapagos to take a quick look. I noticed that some models don't close in GSA, and that they then seem to have thousands of members with no load or boundary condition applied. Maybe you can help with trouble shooting, If you can capture a position of the sliders where then making a manual change to the slider causes the same issue, then I can trace the code and see why.

I would suggest as a starting point reducing the range of some of the sliders and test with a smaller range of options. For example allowing the number of vertical segments to increase to 100 for a 9m span isn't a good option to test and in combination with a large number of vertical segments is a huge model that takes a lot of time to transfer to GSA. Letting Galapagos try and solve something like the image below will take a lot of time (And might make it seem like Grasshopper has frozen). But it's also the fact if models are being left open in GSA that there is likely a bug that I might have to fix.

I will try and help more, but if you can test with some simpler runs this can help.

Cheers,

Jon

Permalink Reply by Stefan Slangen on May 29, 2015 at 2:29am

Hi Jon,

I changed the range of the number sliders and I let the optimization run for a bridge with a 50 meter span. The optimization ran the whole night and when I checked the progress a few moments ago Galapagos was finished. The only problem was that most of the results were 0 (kg). I also got some error messages from GSA. I think that is the bug you mentioned. When the sliders have some specific values, I get an error that GSA needs to shut down, but when I manually close GSA, the optimization continues. So I have to watch the progress, but that is no problem for me. Because I ran the optimization during the night, I didn't recognize any specific combinations of the number sliders. But I will keep an eye on it the next time and I will let you know.

I was thinking about adding a stiffness check, because that improve the results. I checked online and I found that the specific modulus is used for optimization purposes. So I have to divide the elastic modulus by the total weight of the structure, and this value needs to be maximized. But is this elastic modulus just the 2,1*10^5 N/mm2 of steel, or do I have to calculate this for every element by dividing the stress by the strain? I'm not sure how to apply this stiffness check..I looked for some example files, because you once posted on the geometry gym blogspot a truss optimization project. But in the added version of the file, no stiffness check was performed.