About Eddy

Eddy is a plugin for outdoor airflow and pedestrian wind comfort simulation tightly integrated into Rhino and Grasshopper

Eddy is developed by Patrick Kastner and Timur Dogan at Cornell University.

This category is for Eddy related questions.

3 Likes

https://www.eddy3d.com/

1 Like

Hi,

thanks a lot for the promising plugin.

Right now the same problem keeps occurring eg. in the simple wind analysis:

After I ran the mesh and simulation (the gnuplot graph doesn’t pop up), the VisProbes component still gives me the error “1. The mesh folder does not exist. Please create a mesh first.”

Even though the folder exists and is unblocked in the properties.

I’m using the GNUPLOT Version 5.2 patchlevel 8

Hi,

Can you confirm that BlueCFD has been installed as described here?
If yes, can you also confirm that you see command line windows pup up after starting the meshing and simulation procedure?

If you can answer both questions with “yes”, the workflow should be as shown in this video.

1 Like

Thanks for the quick answer - seemingly the directory of BlueCFD mattered.

We are absolutely loving having a play with the capabilities of Eddy. Is there a way to view the mesh and thus visually check its quality in GH/Rhino before launching the simulation?

Thanks, James

Hi, I’ve a problem with the VisProb. Someone who knows a solution for this?
“1. Parsing of the probes failed. This data does not exist yet. Please run the probing component.”



Kindest regards
Alexander

Hi Alexander,

Thanks for reporting this. I will try to provide better error messages in the future.
Can you make sure that a proper mesh was produced and that the simulation converged?
Check this (3:30) on how this can be done in Paraview.

Best,

Patrick

I’ve used Eddy and I love it! Congratulations on the work you guys have done!
It would be awesome if indoor simulations are supported though.

2 Likes

Hi Felipe, thanks for the feedback! Yes we have discussed adding indoor simulations and will add that at some point in the near future. Please feel free to add some more specific comments of features that you would like to see!

1 Like

Hi guys!

Awesome work. Seems really easy to use and well put together. I am however experiencing issues with the annual outdoor comfort script, specifically the MRT component is giving me an error. I guess it’s not writing the file as it should? The only thing I have changed from your template is using my own epw file.

1 Like

Hi Patrick,

I started testing Eddy. It looks great. I have a few questions though.

Currently the meshing refinement for buildings is combined for all buildings. It is usually the case that we have a particular set of buildings of concern and the area around them but we have a lot more context buildings surrounding them, which may need to be included but do not need the same level of accuracy. Is there a way to provide two sets of geometry inputs - buildings and context buildings, both of which have separate controls for mesh settings?
The same applies to adding an area of refinement… If we are only looking at pedestrian level comfort on street, it might make sense to have a more refined meshing for the lower part of the domain than the upper one, is there a way to include that?

Is there a way to visualize in Rhino/GH the mesh after refinement, before running the simulation?

Thanks a lot!

Sarah

@jensbuchjohansen hi, you need to install diva. This components looks for energyplus inside diva.

Hello community, first of all i like to thank the eddy3d team for such an awesome plugin.
Just have a query related to natural ventilation studies. Is it possible to do a indoor natural ventilation to visualize the airflow. I have performed a simulation using a small model with two openings set among a building context.

I request please have a look at it and help me out if it is possible to do this way. I am not sure if it is valid with Kepsillion model.

Thank you @timkado
Eddy3d.gh (21.8 KB) EDDY3D1.3dm (288.0 KB)

Hi,

I installed the plugin last week but can’t seem to can’t seem to get the simulation to run correctly. The cmd window opens after run mesh and run sim is activitated but run simulation doesn’t run iterations. Folders are created but not much appears to be in them.

Our IT security means we can’t run the simulation directly from the C drive but instead from a sub-folder. could this be an issue?

Any ideas would be welcome,

Hi,

I am just starting off with Eddy and I am still running the template exaples to make sure my set-up has been done correctly. It all seams to be ok except that the simulation is not reconstructing all time steps. If I look into the processor folders I see all time steps but in the 0-folder I only see the 0 and 100 steps. Am I doing something wrong here?

Thanks,
David

Can anyone help me out here please

Hello,

I have started using Eddy3D recently for calculating pressure forces on building envelopes. It has mostly been great and easy to follow. I managed to run a simple simulation of my own and get some reasonable results using the k-epsilon turbulence model, but for some reason when I try running the exact same model with the k-omega SST turbulence model it fails to converge.

I have tried improving my mesh accuracy and increasing maximum number of iterations, but with no luck. Any suggestions out there? Thanks.

Dear users,

Thank you for all the questions, please find the detailed answers below!


@smokhtar

Currently the meshing refinement for buildings is combined for all buildings. It is usually the case that we have a particular set of buildings of concern and the area around them but we have a lot more context buildings surrounding them, which may need to be included but do not need the same level of accuracy. Is there a way to provide two sets of geometry inputs - buildings and context buildings, both of which have separate controls for mesh settings?

Not as of now but I will add that feature in the future.

The same applies to adding an area of refinement… If we are only looking at pedestrian level comfort on street, it might make sense to have a more refined meshing for the lower part of the domain than the upper one, is there a way to include that?

AccGround should take care of that.

Is there a way to visualize in Rhino/GH the mesh after refinement, before running the simulation?

Unfortunately not after the refinement and you probably wouldn’t want that in GH. Unfortunately, GHs display pipeline is not very performant for these things which is why I would rely on Paraview to do any kind of mesh inspections.
However, we have a button to start Paraview for you! Check 3:30 min here on how to do that: https://vimeo.com/375755963#t=3m30s


@Asisintel

Just have a query related to natural ventilation studies. Is it possible to do a indoor natural ventilation to visualize the airflow. I have performed a simulation using a small model with two openings set among a building context.

This is possible. Suppose you have a geometry similar to this one below [1]. That way, you’ll achieve cross-ventilation through your geometry.

I am not sure if it is valid with Kepsillion model.

You can use any turbulence model with this approach.


@jensbuchjohansen

Awesome work. Seems really easy to use and well put together. I am however experiencing issues with the annual outdoor comfort script, specifically the MRT component is giving me an error. I guess it’s not writing the file as it should? The only thing I have changed from your template is using my own epw file.

We’ll release a new implementation for this very soon, please stay tuned.


@james.lacy-smith

I unfortunately can’t speculate on the IT setup from a distance but you shouldn’t need Admin credentials to run any simulations after installing the plugin.


@davidand103

I am just starting off with Eddy and I am still running the template exaples to make sure my set-up has been done correctly. It all seams to be ok except that the simulation is not reconstructing all time steps. If I look into the processor folders I see all time steps but in the 0-folder I only see the 0 and 100 steps. Am I doing something wrong here?

It could be that your Iter and your WriteInt was set to 100. That way, you would simulation until 100 iterations have completed but only write out the 100th iteration. If you would like to write out more steps in between, please set the WriteInt to a small integer, i.e. 5.


@LloydE

I have started using Eddy3D recently for calculating pressure forces on building envelopes. It has mostly been great and easy to follow. I managed to run a simple simulation of my own and get some reasonable results using the k-epsilon turbulence model, but for some reason when I try running the exact same model with the k-omega SST turbulence model it fails to converge. I have tried improving my mesh accuracy and increasing maximum number of iterations, but with no luck. Any suggestions out there? Thanks.

That is surprising. You are right, the turbulence models shouldn’t differ in their convergence behavior so much. I will look into this and get back to you.

Best,

Patrick


[1] Kastner, Patrick, and Timur Dogan. “A cylindrical meshing methodology for annual urban computational fluid dynamics simulations.” Journal of Building Performance Simulation 13.1 (2020): 59-68.

1 Like

This is a great app and long overdue.

I have had a few errors using the annual template:

  1. “The average offset between simulated wind directions and wind directions in the weather file is 5.64°.” Not sure what this means.

  2. “Some probed velocities with very large values (likely because they weren’t inside the simulation domain) have been replaced with 0s.”

  3. Occasionally one or two directions will go a bit crazy. The mesh will look fine and the other directions seem to match a Simple Analysis, but for two directions the vectors are all over the place and very high.

  4. The Wind Vectors seem to Cull correctly but the Pedestrian Comfort does not. There is noise under the buildings.

Requests:

  1. I love the simplicity of the tool but can there be an indicator in between meshing and running the analysis just to let someone know what step they are in the process?

  2. Can you do a short video looking at inputs for the boundary layer and domain components?