butterfly
butterfly copied to clipboard
ladybug-tools
Postprocessing
As I mentioned before, there is a need for visualise results efficiently. For instance, at current, vector display is not eonugh, because you cant see the vectors perpendicular to the cutplane, its not valid for commercial cfd packages, so accurate projection is required.
Secondly, contourmaps, isosurfaces should be created, as to isosurfaces I am trying to create a component but until now I am not succeed. Lastly I have too many ideas and its not a good idea to write them at once, because there are so many issues to be handled now. But if there is a plan to write definitions for contourmaps and isosurfaces, it would be amazing.
Hi @mimarilker
I feel this is difficult to do inside BF for a couple of reasons.
First of all, parafoam is already extremely efficient in producing these visualizations. It is an open source tool which comes along the installation of OF. Since one of our goals is to educate I feel we should not completely abandon the parafoam route. I feel we discussed in the past about somehow connecting BF to PF in a better way, especially since PF (to my knowledge) has a python API.
The second reason I see is that the current main visualization tool for BF is the probe function tool. I find it very inefficient to use probes for post-processing a finished case. To do this with probes we would probably need to re-run the simulation for one iteration to get values for the new position of the cutting plane for example. I think it might be more efficient to use OF utilities like sampledSurface to do post processing, which extract data on a surface at the selected time step. Actually I had many ideas as well about which utilities BF can use (we can use all of them but some are more interesting than others).
But you are right the post-processing and visualization part of BF needs improvement. I feel the main reason we haven't done it so far, apart from available time, is that OF completely overhauled their post-processing API during the last updates. BF isn't yet up to date with that, even though the way we have been working seems to be aligned almost perfectly with the new approach OF is taking.
I think it's great that you are working on new ideas and components! I just wanted to note that I feel we should aim that any improvements / additions to the post-processing tools should be aligned to the newer versions of OF. What does everyone think?
Kind regards, Theodore.
I am exactly agree with you on aiming at improvements of postprocessing, its very clear actually. But I didnt find reasonable the idea that there is an existing tool and lets use it. At current, we are depending on OF, its enough actually. Btw, there is a tool which is being developed on Lattice Boltzmann method, it has no external software dependencies. Its at the link https://vimeo.com/150931537
Cutplane Tool : currently works for xy and yz planes (seperate comp.), user assigns a value between 0-100 (means start and end of the domain in percentage) with slider, and the comp. gives feedback as to real location in meters. The output testGeometry should be linked to "generate test points comp "(honeybee).
Now that I read this again, and after the discussion that we had I think the best approach for this sort of problems is to use either the postprocessing commands from OpenFOAM which is already submitted here (#61) or use ParaView's python library. I did a couple of tests inside paraview and it's a very powerful library which covers all the functionalities that we are looking for. There is also a trace option under tools which generates the python code for each click.
My suggestion is that to use postprocessing and probes for the upcoming release and eventually use paraview's library for advanced visualizations.
I am exactly agree on that. Lets make the release first, then BF will be get improved with newer integrated postprocessing capabilities like streamline, contour maps, isosurface etc.
Besides flow trajectories is very good option to visualise results, check the video that I have created for a research study: https://vimeo.com/198602499
@mostaphaRoudsari That would be amazing! Parafoam is one of the best visualization tools and its library is also made for (massive) parallel visualizations. I think connecting to it would be superior to using OF utilities. It would also possibly be smth to look for in the (future) cloud versions.
On the other side, there are a few FOs and utilities that are only in OF. Ofc you can almost do everything in parafoam as well with custom functions but I feel it would be like re-discovering the wheel. Maybe we first see which objects we use from OF and then we tap into parafoam for the rest?
@thinklikeanarchitect have you seen this? http://www.openfoam.com/version-v3.0+/post-processing.php