Honeybee Energy Optimization with Galapagos Error

Hi Mostapha, Chris and group members,

I made a model with variable shape and window rate. I would like to use Galapagos to optimize the shape and window rate for lowest energy consumption.

Here is three questions:

  1. As the optimization goes, more and more errors of zones appeared, and the line of Galapagos seemed confusing.

  2. When I checked the result again with the push of bottom “Reinstate”, the previous total energy calculation turned “undefined”.

  3. The error is all about “All of your HBSrfs must make a closed volume”.

I attached the gh file. I’ll appreciate it if you give me a hand.

Thanks in advance.

Stage02_Optimization_3.gh (861 KB)

Hi Bi Xiaojian,

I am running your model right now. There seems to be no error. It is running fine. I have one question though. In Galapagos, you have kept the threshold to maximize. Are you sure you want to do that?

Devang

Hi Devang,

Thanks for your immediate reply.

I remembered to set the threshold to minimize, maybe my bad.

The error will appear after about 30 generations. Don’t know if it’s only my problem.

Hi Bi Xiaojian,

How much time approximately did it take for you to reach the that many generations?

Hi Devang,

Sorry for my late reply.

It will take approximately one hour to reach that generation.

So do you encounter the same problems?

Xiaojian

Hi Mostapha, Chris, Devang and other group members,

I checked some issues about Galapagos and Honeybee. Mostapha gave a suggestion to add a delay component before the calculation, so is it right to set the data delay component before the Galapagos fittness and after the Honeybee calculation?

I’m also wondering whether my model setting went wrong because of zone errors about not a closed brep.

Xiaojian

Hi Xiaojian,

The model runs fine at my end. The fitness numbers is reducing as it should. I checked the output of your HBZones. They output closed breps so that’s fine too.

Devang

Hi Devang,

Great thanks for your work and immediate reply, and so thankful for your patience to run my gh file completely. The result sounds excellent and exciting.

So, you mean my gh file runs well with no problem that I mentioned above. Hmm, interesting. Maybe I should change another device to run again.

Last but not the least, I’ll appreciate it if it’s possible to post a capture of result for me. If not, everything is fine, still give my best thanks to you.

Xiaojian

Hi Xiaojian,

I did not record the result last time. I am re-running the model right now. It’s been running since last 2.5 hours and has reached generation 7. I am also eager to find out how this goes. I will share results.

Devang

Hi Devang,

Great thanks for your immediate reply. Wow, that sounds cool, thank you very much for your patient work. Hope to see the result soon.

I re-run the gh file and met the same problem above. I stopped the calculation and checked the error step with parameters setting which caused to crash of zones. Interestingly, the model acted well after I double-clicked the boolean toggle to create zone again. That really confused me.

So hope to hear from you Devang with good news, and hope some guys help me to find out the problems.

Xiaojian

I see the same error now. It has got something to do with the Glazing ratios. That is where the error originates. I am looking more into it.

Hi Bi and Devang,

I haven’t checked the model but generally speaking using intersection and split in optimization studies can make Grasshopper to crash. Intersection can change based on the tolerance values. If you can set up a higher tolerance that works with your model that should help avoiding the crash. If your model is simple enough then I suggest to develop your own code to create the geometry and use Honeybee to create the zones and run the analysis.

Also in some cases it has been an issue with the memory. I have recently made major modification to Honeybee to address the memory issue. If you update to the version available on GitHub you can test it with the new version.

Mostapha

Hi Xiaojian,

My results and observations are following:

  1. Please find attached image named observations. air wall is defined in solve adjacency component and this puts an air wall adjacent to an interior wall. I disconnected this air wall and the model runs fine I ran it for 19 generations for over the period of 7 hours.

  2. Earlier, when I ran the model with air wall connected to altConstruction_, I observed that in the middle of this optimization process, one or many zones will disappear altogether from rhino scene and that is when the results became undefined. In the image you attached, if you see in the top left corner, zone Commercial 1F is missing. I witnessed a similar behavior. This disappearance of zones may explain the erratic highs and lows in fitness numbers that you are talking about.

  3. I also made a few changes in how you apply glazing ratios. Please find file with changes.

  4. My results are attached as well. When I used air wall as altConstruction, the model ran for 7 generations and then started displaying errors as I mentioned in my earlier responses. After removing air wall and making a change in how glazing ratio is applied, the model ran successfully for 19 generations.

Hope this helps,

Devang


Dear Mostapha,

Sure. I did use the latest version.

Thanks,

Devang

Revised definition

Stage02_Optimization_3_01.gh (670 KB)

Hi Prof. Mostapha,

Thanks very much for your timely reply and great advice.

There are two more questions about your advice above:

  1. How to set up a higher tolerance? And what’s the tolerance function about?

  2. So excited to hear the news of major modification of Honeybee, just want to make sure the exact version. As I see, you mentioned the Version available on GitHub, so is it the latest one not the Honeybee 0.0.60? If so, I should download the latest file on GitHub and get the version on my pc (equipped with 0.0.60 right now) replaced with it?

I’ll deeply appreciate it if you take the time to give me a hand. Thank you.

Xiaojian

Hi Devang,

Great thanks for your patient and detailed work! And deeply sorry for my late reply.

As you said, the errors happened in the setting of air wall and glazing ratio. I cheked your uploaded gh file, excellent work! Thank you so much.

I also think whether the volume setting method could work better than the surface setting method, which could simplify the model with reduced possibilities of errors, what do you think?

What’s your final result of 50 generations? What I saw is the errors decreased, at the same time, the calculation could not stop at the 50th generation unless I pressed the stop bottom. Hmm, interesting. How about yours? Maybe there could exist no best value setting of lowest energy consumption? Or should we set a threshold at first?

Xiaojian