Dear @mostapha and @chris,
Is there any suggestions regarding modeling a high-rise building with almost 800 zones? What are the possible approaches to simplify the algorithm? I am planning to model each floor separately at the moment and then merge all floors together in one file after I solved any possible fatal errors at each level, but it seems its gonna be a very heavy algorithm to simulate eventually. I was thinking also to make clusters of some definitions, do you think it might help in running the HB definitions faster?
Why do you need so many zones? Will not many of them be identical or almost identical?
The architectrual drawings dictate such a massive modeling… Each floor has 4 residential blocks and each block has almost 15 to 20 zones… Besides there are shared spaces in each level and we have two floors including lobbies and offices… Plus three underground floors…
@AMIRTABADKANI Please upload some screenshot of your model.According to your description, you should simplified your model and run it using hight performance computer.
Thanks @minggangyin for your reply,
Here is one of the floor’s algorithm without assigning windows, HVAC or Natural ventilation definitions for now … Consider I will have 15 floors …
I decided to model surface by surface as I ll be able to edit each surface material and settings easier than if I wanted to start modeling as masses to hbzones (because in this case I had to explode all the zones each time I wanted to assign a new settings and again merge them all, besides controlling interior and exterior walls would be much harder) …
Do you think clustering approach can help? or any other recommendations are appreciated.
9thFloor-KrajResident-970907.gh (1.4 MB)
I would use one or more of the following methods
- Divide the model into smaller geometrical problems if you could ignore hest tranfer between the models.
- Use miltipliers for identical floors unstead of modelling all of them.
- Use a simplier model than Energyplus.
- Use larger zones.
- Cloud computing with paralization similar to the first method
- Divide the model in time with some overlap to take heat storage into account, maby a month with two weeks startup.