Personnaly I’m using the py script developped in the discussion below simpledrainage_te.gh (92.6 KB)
1 Like
Thanks!
I’ll look into it and try to implement it with my workflow.
Have you tried it with Eddy3D?
Since I’m not very familiar with field lines, would I need a 3D grid of points that covers my area of interest in the wind tunnel with appropriate velocity vectors or just starting points at the beginning of the wind tunnel with their vectors?
Yes 3d grid points give better result
I export vrml file from paraview and import it
I see, thanks again.
THe definition isn’t not perfect but maybe it can inspire you 
caseA.gh (50.2 KB)
1 Like
Nice script, of course paraview better but this give nice result
1 Like
Hi @timkado . I am trying to use annual outdoor comfort. Everything runs well but Rhino crashes when I toggle MRT component. I also installed Radiance 5.3 in c drive program files. Is there a way to solve this repetitive error.
1 Like
Anyone can provide some assistance here?
Hi! I have a basic knowledge of Grasshopper and I’ve just started with Eddy. I’m trying to run the “SimpleWindAnalysis” example following the official web site tutorial but I have problems when I run “Visualize Probes”. It should create the wind vectors but there is a mistake and the vectors appear in all directions and oversized. There is also a message with the next information: “can’t be probed within the simulation domain and have been discarded”.
Hi,
It is hard to tell exactly what your issue could be. A few things pop into mind:
-
It seems you are trying to probe inside a mass that is in your simulation? If this is the case, it could explain your issues. You cannot probe right now inside of a mass.
-
How many Iterations did you run? If you ran only a small number of iterations (say 100) than this could also explain the extreme and unrealistic vectors since it likely could not converge the solution enough. If you ran a large number of iterations and it is like this, then check the Residuals plot and see if that specific wind direction completed all the iterations - or if it stopped short. This would indicate it “broke”, for some reason and was not able to complete. Generally I would run no less than 500 iterations for a reasonable simulation outcome.
Honestly though it could be many other things as well. Hope this helps.
Has anyone run this annual wind analysis thing on a site that is 1x1km large? Been running the simulation since yesterday 10pm, its been 12h…
Hi, how big is your site? Am struggling with my 1x1km site…
Hi, @aeaechan96
You should be able to run sites that are much bigger than 1km x 1km. How detailed is your model? What meshing settings did you use? How many iterations? It would be helpful to see a screenshot of your model and grasshopper script setup as it could be many things.
That said, depending on the power of your computer and # of CPUs you can use, it could take a long time to do a big simulation. For reference, I did, on my laptop, a 1km x 1km simulation a couple weeks ago w/ 4CPUs & 16GB RAM for 1,000 iterations and a basic mesh and it took almost 12 hours for 16 wind directions.
Thanks,
Remy (Eddy Team)
Hi Remy,
Would you mind sharing the mesh settings you used for such a large model please? The roughness (z0) would also be very helpful for reference.
I’ve struggled to yield sensible results for a masterplan geometry (1.3km radius and height), with the ground’s boundary layer seemingly being too large (zero velocity at pedestrian level; the ABL looks sensible upwards of that height though). I’ve had the same issue for even very basic small box geometries (e.g. 4m cube); I’ve found it to respond to a desired cell size of 0.25m (3 layers), but this seems impractical for a masterplan. (Meshing my masterplan with even DC = 0.5m gave a mesh of well over 16GB)
Any help is greatly appreciated.
Kind regards,
Julia
Hi Julia,
For my large model I used the CellSize component in Eddy w/ a BS of 10 in the Domain Cyl Component. The 10 feeds into BS on Domain Cyl and Cellsize Component. I used 3 for DC. Acc from the CellSize gets fed into my AccBuilding and Accfeatures and AccGround. This was suitable enough for my initial modelling. If you are doing a final simulation, maybe you want to bump it up a bit to higher resolution.
Again though it highly depends on the computer you are running. Make sure your model is more massing and less detailed. The fewer openings the better (ideally no window openings), no complex meshes. I would say for a masterplan a 0.5m mesh is very fine and maybe too fine. 1.3km height is also very tall, I do not know what your terrain is, but 300m height is what I used.
I do not have any specific z0 numbers to share, but if you look online in papers you can get a good idea of what people use for different materials.
Hope this helps.
~Remy
1 Like
Hi Remy,
Thank you ever so much for this and for being so quick to reply, that’s very helpful.
Did you use layers for this?
For a pedestrian comfort study we’d ideally have 3 layers below 1.5m height. Looking at the snappyHexMeshDict (firstLayerThickness with relativeSizes), am I right in thinking the first layer height is 0.3 x DC, and from there the layers increase in height by a factor of 1.3 (expansionRatio). If so I think I’d need a minimum DC of 1.25, which isn’t unreasonable. I get very strange results for that though (virtually zero flow below 2m height) even after 2000 iterations. (I have simulated the same in ANSYS Fluent for comparison)
I would also be interested to know whether the user can change the settings in Eddy3d, for instance turning relativeSizes off. It lets me edit and save the file, but upon running the simulation it reverts back to the default. Apologies, my knowledge of OpenFOAM is fairly limited.
Thank you very much,
Julia
Hi Julia,
Can you clarify what you mean by Layers? In Eddy there is an option in the MSet component for the Mode, Snapping, Snapping w/ Layers, No Snapping. Snapping is usually where I start, and for a final simulation I might use Snapping w/ Layers. You just need to specify which you want, not the thickness of the layers.
I am not sure about the relativeSizes settings, I will talk to other members of the team about that. Why do you want to change this?
Hi Remy,
Thank you for your response. Yes the layers you mention as part of MSet is what I mean (for modelling a boundary layer grown from the ground).
Looking into the snappyHexMeshDict script, it seems the height of each layer is relative to DC (see firstLayerThickness = 0.3, with relativeSizes = true). This is my understanding from having read section 5.4.8 of:
http://foam.sourceforge.net/docs/Guides-a4/OpenFOAMUserGuide-A4.pdf
If that is correct then the requirement of 3 layers below 1.5m is only possible using a fine mesh (DC < 1.25). If I were able to set relativeSizes = false, then perhaps I could use a more coarse mesh and specify the absolute firstLayerThickness instead.
Kind regards
Julia
Hi Julia,
Gotcha. Unfortunately right now setting the relativeSizes is not an option w/in Eddy. We are working on implementing it though, so keep an eye out in the future. As you mentioned, if you absolutely need to change it then you can do so manually in the snappyHexMeshDict.
Thanks for using Eddy!
~Remy