Large Scale UTCIMap use reduce run time

Hello all.

I’m starting to do some test runs on a area using the UTCIMap tool and was wondering if there are any good ways to help speed up the analysis process or reduce the load on the component to help it produce results slightly quicker.
The area I’m looking at is around 500,000m² and I’m currently using a grid size of 4.5m for the analysis (which is realistically the highest I’d like to go).

Previous tests had it running for an hour with not a huge amount of progress, and that was with no building geometry or natural shading.

Any suggestions would be great, and I’ll keep tinkering as well.

Thanks,
Alex.

Hey @alexsdiamond,

I’m not sure that the UTCIMap is the best approach for that large an area. What are the key outcomes you’re trying to get from the study?

Best,
Charlie

I will sound like a Pollination salesperson at this point since I just referred someone to use Pollination ¯_(ツ)_/¯ - but this is one of those cases that cloud computing can help. I would still probably break down the study into smaller chunks.

1 Like

I’d agree with Mostapha. Cloud computing on Pollination would be the fastest but I’ll add that a high end desktop would likely make a world of difference if you’re currently trying to run it on a laptop. I have run studies of that size on my desktop with an i9 and 24 processors without too much wait time.

FYI, you can also try visualizing the steps of the recipe execution using the HB Visualize Recipe Execution component. There are a lot of different steps to the comfort mapping recipes and you may be able to use that to sense whether the EnergyPlus simulation is that part that’s taking a while or it’s all in the Radiance calculation. If it’s the EnergyPlus part, you may be able to simplify the model a bit to get that part to run faster.

The aim of the assessment is to determine thermal conditions felt by people in the outdoor areas of the site we’re looking at and then using that information to influence design changes that would improve the thermal comfort of people.
We’re wanting to get mean radiant temperature and UTCI results for the whole site using a reasonable grid size, then focusing on certain sections that can be identified as problem areas that can be adressed with different methods.

@mostapha @chris I have asked for the possibility of pollination cloud computing license to be considered, but we do also have access to some virtual machines in house, as well as our modelling laptops being fairly ‘chunky’ (11th gen i7 processor and NVIDIA A3000 RTX GPU 64GB Ram).

I have the steps visulaised in the queenbee.exe popup command box and have noticed that in one of my more recent runs it got ‘stuck’ on the 41st step in the recipe looking at the DirectSky section, and a time before it haulted at 42nd step TotalSky.

I have adjusted the grid size to be slightly larger to see if that can aid in speeding up results, as well as simplifying all geometry to help reduce points of analysis, but as it is a large site, I think it’ll just take its time.