Parallel EnergyPlus runs with Colibri or optimization components


I wonder if for parametric EnergyPlus runs (for example with Colibri) or multiple EnergyPlus runs based on optimization algorithms (with components that I am not aware of for the time being), we can use the full capacity of local processor in parallel simulations.


@farhang.tahmasebi the answer is true. You can try this example.

1 Like

@minggangyin I meant simultaneous runs using multiple processor cores. As I ran that example file, EnergyPlus models were running one by one.

If you want to use the plugins as it is then the answer is no, and it’s because how Colibri, Galapagos and all the other iterative plugins in Grasshopper work.

Whenever you change a value in Grasshopper it starts a Solution and once the solution is Expired and Grasshopper starts a new Solution there will be no access to the data from the Expired Solution. This will be a challenge for a plugin like Colibri since they need to collect the results for each Solution before starting a new Solution.

On the other hand if you want you can create your own custom workflow to achieve what you want by separating two steps:

  1. Generate all the design options without running them.
  2. Collect the path to all the generated files and run them in parallel. For EnergyPlus and OpenStudio Honeybee has components that runs idf/osm files in parallel.
  3. Run collibri again and this time only load the result files and create the output csv file.

2 posts were split to a new topic: How to create a new name for each study based on input parameters?

Hi @mostapha,
thank you for this suggestion,
I’ve been having a problem when I follow these steps with Electric lighting energy, ElectricEquipement energy, Natural Ventilation energy and total peoples gain which becomes 0 in this method,( this values are not 0 when I do the parametric study in the simpler way by generating OSMs, running them and exporting the results in each iteration)

Hi @mostapha How can I write all the design options? Manually?

Automated Iterative Modeling | Ladybug Tools Academy check this out

@farhang.tahmasebi this is probably a dull comment, but could you look at writing idf files using the colibri structure, then execute them using something like JEPlus or JESS, the read the results and feed back into colibri to format the data into something like design explorer?

Yeah, you can do this workflow.As for my experience, I will change the non-geometry parameter using Python for loop, it is more quickly than using colibri.After generate a lot osm or idf file, you can use run osm/idf on parallel CPU.