Comfort Mapping Recipe Longwave MRT Running Out of Memory

Thanks for the updat @chriswmackey . So if I assign a grid in Hb and convert to osm, will that grid not transfered ?

For the recipie issue I still didnt get it ,my model is valid and runs successfully locally! Is there a way I could run it on web !

@chriswmackey I again tried the recipe and its failing again.

Correct. There is no way to store a sensor grid in the OSM.

For the other issue, this appears to be different than my original explanation as the run is failing. @antoinedao or @mostapha , do you have any idea why this node is being killed? Is it running our of memory or something?

FYI, @asisnath . If that node is running out of memory, you should be able to get it to succeed by running a shorter run period. You said that you had no issues running it locally, right?

@chriswmackey Thanks for the update. I will give a try for a particular month.

Looking at the the files going into that node, I’m pretty sure we are running out of memory. The SQLite file coming out of E+ is 4 GB:

… and we’re probably loading half of this data into the node to be able to perform the longwave MRT calculation.

Running the simulation for a month will cut this file size down to 1/12, which should succeed. Another option is just running the simulation for the particular rooms that you care about instead of a 600-room model.

But I think the only way we’ll get the original large simulation to succeed is by being able to allocate larger resources to this particular node since I think I have gotten this task in the recipe to be as optimized as possible. This is just the pinch point of the recipe.

1 Like

@chriswmackey . I gave sensor points for 2 rooms only rather than 600 zones. I may be wrong but though I am using 2 rooms for sensor grid it still performs energy simulation for 600 rooms. I will run for a month in an hour and post the update.

So is there a reason why you need all of the other rooms of the model? Can’t you just put those two rooms that you care about into their own model and run the recipe with that? The recipe is forced to run an energy simulation for the whole Honeybee model that you input so there’s a lot to be gained by just building a model for the part of the building that you care about.


What I was thiking @chriswmackey is that as this was a multizone model and each zone will be effected by adjacent zones thermal condition, it need to have energy simulation of total 600 rooms. If I only put individual rooms than it would be like simulating a shoebox model. I may be wrong but for this reason I did this way!

If all the spaces are conditioned and they have a similar usage pattern (e.g. similar occupancy schedule, loads) and facing a similar direction (similar solar radiation) then you don’t need to model all of them. There will be minimum heat transfer between them.

Even if you want to model the adjacent rooms you can model the closest ones and model the rest as shade objects. It is unlikely that you need to model 600 rooms for comfort studies in two rooms.

1 Like

The error message is not great but running out of resources seems to be the reason.

@asisnath, we have a development task to allow users adjust the resources for large studies to avoid issues like this. I will keep you posted.


The tough thing with this model is that the whole building is passive and so there is likely some heat flow between neighboring rooms that you would be missing by using an Adiabatic BC. Still, I’m sure that you wouldn’t be too far from the mark if you only modeled those two rooms and used Adiabatic Faces for the interior conditions. And, if you are really worried about accounting for heat flow into the neighboring spaces, you can just model those two rooms and their immediate adjacent rooms (using Adiabatic Faces for the indoor faces of the immediate adjacent rooms). At that point, the error is far enough away from the rooms you are studying that it should not matter.

Whatever the case, there’s no need to simulate 600 rooms when you only care about the results in two of them.

1 Like

Thanks @chriswmackey @mostapha for valuable insight on the workflows. Really appreciate!

1 Like