Hi everyone,
I would like to ask you something about the optimization process. I’m trying to optimize a facade morphology taking into account the daylighting values in lux and 4 geometric variables (each parameters with max 4 steps). The idea is to implement the process adding the energy values. Right now, I can’t even start the solver that Rhino crashes. Is it something possible? What can I do to speed up the process?
Thanks.
Angelo
The first step is to figure out why Rhino is crashing. Does it also crash if you run a single model? How would you get that fixed?
Yes. Several people have done similar studies.
For total of 4 * 4 = 16 options you can just run them all. Do you really need to run optimization?
Dear @mostapha,
Thanks for your answer.
With “Is it something possible” I was referring to crashes that occurred to Rhino. I know that optimization is a well-known topic in this field of investigation but I don’t know which technical characteristics of the machine it requires (CPU, RAM, Free space on disk). Are there any specific requirements? In my case, Rhino crashes with the optimization but not with normal gh codes.
About the parameters, they aren’t 16 because there is also interrelation among the different 4 parameters of the 4 variables…5 parameters = 4(4) = 4 x 4 x 4 x 4 = 256 simulations.
Is it probably depending on the fact that with energy and daylighting analysis the optimization has to run energyplus and radiance in the background each time? What can be done to speed up the process but most important to make the optimization able to run?
Thanks.
Angelo
What are you using for optimization studies? Galapagos or other available plugins?
Does it always crash on the same option? There us no specific requirements that I know. Did you try to check how the resources are used during the study?
You are right! This is still something that I would let run for all the options. If you set up the optimization study with a population of 50 you will reach 250 runs in 2 + 3 generations which in that case your solution most likely is not yet converged. If you use Colibri and run all the options you’ll have the results for all the possibilities at that time.
PS: Your profile picture is missing!
Dear @mostapha,
I was using Galapagos for the first trial but right now I’m trying to use Opossum that seems to be a little bit faster. At least, now the optimization runs but remain a slow process. In the next step, I will try to use Colibri to expand the numbers of performative parameters.
Thank you for your availability.
Angelo