We have created a fork of the CFD workbench of Qingfeng Xia (https://github.com/qingfengxia/Cfd.git). The aim of this wokbench is to assist OpenFOAM beginners to get up and running rather than to provide advanced functionality (such as fluid-structure interaction). After discussion of the different philosophies we felt that a fork was maybe warranted. Our version with installation instructions can be obtained from https://github.com/jaheyns/CfdOF (Edited: The name has been changed from Cfd to CfdOF to allow the fork to coexist with the original).
Current functionality:
Incompressible, laminar flow (simpleFoam).
Basic material data base.
Flow initialisation with a potential solver.
Tetrahedral meshing using GMSH including multiregion meshing (using FEM workbench functionality).
That is in my opinion the way to go … “get it up and running”, keeping it simple to use, demonstrating that is worth to invest more work.
Hope I find the time to test your work soon.
BR
HoWil
BTW.: Only to get some insight…Is there a special reason why there is a extra workbench for CFD rather then including it into FEM? I think Bernd invested much work in splitting FEM wb stuff to simplify the integration of new solvers
Would an indoor / outdoor building airflow simulation and results visualization (vector) like shown in the following two videos be already possible (ignore the node based workflow of course)?
do you have in mind, there might be people who would like to install both, yours and the one of xia. You might use a different name and thus a different folder name.
We really hope the community would find this useful.
@HoWil. The CFD WB makes heavily use of the FEM functionality, but Bernd suggested a while back to Qingfeng that it might make sense to start a new WB. From our perspective, we want to keep the link as close as possible and ensure there are not duplicating code, but seeing that we aim this fork at the beginner we want to keep it clean and simple. We do foresee some differences between the two in the future that might make sense to have to seperate capabilties. For example, for practical CFD work we need to look at some other meshing strategies. We love GMSH, but we often need to generate large meshes (> 1 M cells) for flow analysis and we think it would make sense to move to cut-cell Cartesian meshes (snappyHexMesh and/or cfMesh). Qingfeng are, however, are interested in FSI and has said that he would like to see an integrated capability. Luckily FreeCAD allow us to do both, the beauty of open source!
@Saso. I, unfortunately, struggled to follow the videos, but in principle the inviscid analysis should be able to provide you with a reasonable first-order estimate of the flow. In the case of blunt bodies such as buildings, unlike airfoils, the drag is dominated by the pressure forces and is less sensitive to viscous drag effects. You would likely get fairly good results if you are interested in the loading, but I am not sure how accurate you would be able to resolve the wake and the flow downwind. We are planning to finish the viscous extension mid 2017, if you are willing to wait.
@Bernd. Thanks for the suggestion. We thought it might be a good idea to use a different name, but we were not sure what to use! Do you have good ideas?
+1 and the name change would facilitate this.
Some additional work would be needed on windows though. The windows build linked above handles the dependencies
I see some stuff in bin/Scripts and bin/Libe/site-packages.
Are the Gnuplot, Paraview, OpenFoam executables required to be on the users system.
I will answer some of these questions myself, but I look at the packaging before testing the functionality.
Great job and thanks for this
Gnuplot, Paraview and OpenFOAM is included in the blueCFD binary of OpenFOAM (please see the readme). We did not want to include/package it because it is really large (+700 MB). We thought it might be best if the user download it directly and install it?!
Thank you for the comments and interest. We will think of a new name and take action on this shortly.
There was a glitch with the Windows build - we left out the README file! This has been updated and hopefully will resolve uncertainty about the dependencies. The ‘.;’ directory was just a glitch and has been removed.
Definitely, this is what I expected but I was confused by the elbow tutorial.
WINDOWS:
● Install the binary. All necessary software
components are included.
Thanks!
Next question, I noticed that some PyFoam stuff is duplicated in FreeCAD/bin/Scripts and FreeCAD\bin\Lib\site-packages\PyFoam-0.6.6-py2.7.egg\EGG-INFO\scripts?
The reason I ask is that I am trying to move the PyFoam stuff to %appdata%/FreeCAD so it is available for all my FreeCAD versions.
During sunday i had some time for attempting a first test drive and i wanted to share some first thoughts.
Background:
I am a mechanical engineer working at a company (200 people) producing valves (shut-on/shut-off/regulation) of different sizes.
I am working in R&D-Department and using now for almost 15 years
SolidWorks 3D for doing designs
SolidWorks Simulation for doing FEM(FEA) studies
SolidWorks FlowSimulation for doing CFD-simulations (99% cv-value simulation)
COMSOL Multiphysics (we had prior one license of ANSOFT MAXWELL 2D) for simulating electro-magnetics.
I didn’t want to touch my live Windows 7 64-bit Installation so i booted my Linux Mint 17.3 Installation (i am quite new to Linux Mint …)
As basis i used your CFD-Tutorial 1 - Elbow.
Observations:
when linking to documents on Github, we FORCE to Show/open the RAW-Version of the document, This way the download will
succeed. I had some struggle to download a working PDF
Installation on LINUX
I was missing the Information that a 0.17-devlopment Version (maybe a Minimum git number ?) is needed.
If you only write that FreeCAD is needed people will install from the repository or the 0.16 stable release …
I was able to install all libraries/dependencies BUT OpenFoam 4.1
I downloaded the deb-package and upon starting Installation, a red error message said “unable to install”.
Since i am quite new to Linux Mint i didn’t see any possiblity to find out what exactly was wrong …
Perhaps if you would try try install on a clean machine maybe you are able to tell more about that ?
Documentation
In SolidWorks/SolidWorks Simulation i am used to model the solid body and model/define the lids.
Per algorithm the program will evaluate the (internal) fluid body excluding cavities without flow condition
and will highlight the fluid region with some sort of “bounding box”.
It took me some time to realize that for simulation in OpenFOAM you need to model the fluid body and apply there
all needed boundaries/goals.
It would be nice to have this Information as introductory sentence in the PDF.
Perhaps writing in the documentation that the “slip”-boundary Need to be applied to the OUTER wall
would make the info clearer (OK in FreeCAD it would be a struggle to select the inner walls,
but more precise Information wouldn’t hurt)
General first observation
Hard to tell for me how much programming work on Linux it is, but upon starting the CFD-Workbench,
maybe checking if all libs and dependecies and FreeCAD-Version-Number is/are OK would be a good idea ?
Sorry that this text got longer than expected, thanks for reading.
We have now renamed this fork of the workbench to ‘CFDFoam’. The Windows build is updated and the git repository moves to https://github.com/jaheyns/CfdFoam.
Neither is actually needed for our usage of PyFoam, AFAIK. These files got there by using the supplied installation script to install PyFoam into the python installation bundled in FreeCAD (but perhaps there is a better method).
We’ll have a look at a better way when time permits.
Good point. We’ve added this information to the Readme.
I’m afraid I have no knowledge of Linux Mint. It would certainly be possible (if nothing else, building from source), but I would recommend installing on one of the officially supported distributions listed at https://openfoam.org/download/4-1-linux/ to avoid a lengthy process. Failing that, the Windows installation is likely to be the safest option to avoid dependency headaches.
Thanks, we’ll try to consider this perspective next time we go over the documentation.
Agreed, and we actually have this in the pipeline.
I tried the new CfdFOAM Workbench today but got a problem writing the case. I defined the path to blueCFD according the README file, but get always the following output error:
Do you have good ideas?