Galapagos multiple variables

Hi, I have a reasonably complicated script generating structural vibration results for a timber glulam and CLT floor plate, testing for all the acceptable combinations of geometry (genome n=5) that give an adequately low value for vibration. Script all runs fine and matches expected results for a few test cases. Phew. Note that in excel form this calculation runs over 6000 cells of look up's and arrays, it's not a trivial calc that can practically be rearranged to make it work back from an acceptable value.

But in my results I have 99 sets of vibration-acceptable outcomes, which I'd now like to examine to see which one uses the least material in its particular combination of beam thicknesses, slab thicknesses and other stuff to find the overall most efficient system. The plan at the moment is to copy the solver Record over into Excel, extract the %'s for the different variables and post-process the info to sort the acceptable outcomes by weight.

It feels like there must be a better way that avoids taking the data out and having a gap in the parametric thinking. Is there a way to ask Galapagos to give me all (or at least, loads of) combinations for which R < 8 and then test those for the minimum weight? Can I automatically take the winning results from fitness test 1 out into an array of data that feeds into a second fitness test in the same grasshopper space?

Replies to This Discussion

Permalink Reply by David Rutten on June 24, 2016 at 11:24am

Hi Rob,

not in Galapagos no, it's almost impossible to harvest any data from the process. You can however add some custom storage, but even that will require a significant amount of custom coding. What would be needed is to have an additional component which pays attention to your vibration value and is aware of all the sliders that are relevant. Then, whenever a solution yields a R < 8, this custom script will either add a state to the document or maybe writes all relevant data out to a text file, or maybe even just aggregates all possible genomes as an output.

If the weight calculation is also part of your file, then the script could 'remember' the lightest genome which results in R < 8, although you'll have to manually reset it before each Galapagos run.

Does that sound like a viable option?

RSS