I discovered a bug in the release 1.0.0 that prevented symmetry boundary condition from being applied in certain cases. Please upgrade to version 1.0.1

Free CFD 1.0.0 is finally ready for download. It is major version update. Most of the code is rewritten so it might have a few bugs to be found. Here is a list of new capabilities:

• Multiple grids, multiple coupled problems (For example, coupled fluid  flow and solid heat conduction problems. Documentation on this is currently being added.)
• Variable viscosity and specific heat options
• Venkatakrishnan limiter
• List of flow output variables can be specified
• Integrated force and moment output
• VTK output option brought up to date
• Mass flow inlet
• Stagnation (total pressure and temperature) inlet
• Heat flux specification at walls
• Hope I am not missing anything else

Downside is, the dual time stepping and preconditioner options are currently not there. I want to add those together with second order time integration. Download it, enjoy it and let me know what you think.

It has been a long time since I announced the completion of my PhD and promised to resume development of Free CFD. The fact is, my PhD work dragged on with thesis revisions after the defense. At last, the whole thing is over. Now, all the time I can find out of my day job is dedicated to Free CFD.

Needless to say,  many new ideas, recommendations and plans have piled up in the mean time. But I am off to a start already. I am going through a painful process of  largely rewriting the code to allow for a level of flexibility that will yield itself to quick development of a whole set of new features.

As a glimpse of what is to come, imagine a solver where you can read multiple grids (not multiple-block grids) and specify exactly what equation set to solve on each and how do they interact all at run time. I am talking about overset, conjugate simulations, cartesian background grids…

The first step was to modularize the grid class and isolate it from flow variables. That is done. The current development version can now read in multiple grids, partition and distribute each. I am now building the current basic functionality over this framework.

grids {
grid_1 (file=sq_cavity.cgns; dimension=2;);
grid_2 (file=backstep.cgns; dimension=2;);
transform_1 (gridId=1; function=translate; begin=[0.,0.,0.]; end=[1.,0.,0.];);
transform_2 (gridId=1; function=scale; anchor=[-0.5,-0.5,-0.5]; factor=2.;);
transform_3 (gridId=2; function=stretch; anchor=[-0.5,-0.5,-0.5]; begin=[0.,-0.5,0.5]; end=[0.5,-0.5,0.5];);
transform_4 (gridId=2; function=rotate; anchor=[-0.5,-0.5,-0.5]; axis=[0,0,1]; angle=45;);
}

There are a number of things already developed but never released. These include dual time stepping, dynamic load balancing, new limiter and so on. I am keeping busy putting these all together toward a new major release.

There are many ways to support development.

• Spread the word. Link to Free CFD from your websites, mention it to your colleagues.
• Are you an expert in C++? You can help kosher up the code.
• Do you know the GNU make system? You can help setup a smooth installation process.
• Are you a CFD researcher? You can help test Free CFD, put in new models or simply provide your feedback.
• Don’t feel like participating in development but would like to see the project go on? You can definitely help by clicking on that orange donate button on the top right.

Free CFD development has slowed down for the last few months. The reason is, I have been finishing up my PhD degree. I just defended this Monday and passed. After some revisions, probably a few weeks, I can finally get back and pick up where I left off. So, this would be a good time to ask for your feedback. Please let me know your general experience with Free CFD and what features you’d like to see added so that I can plan/prioritize the tasks. You can comment here, e-mail me or add a wish list post to the forum.

This one is a bug-fix release addressing:

• When grid dimension was not specified in the input file, it didn’t fall back to its default value of 3, resulting in a smaller stencil (of node averaging) than normally would be used.
• Another bug fix affecting the accuracy of 3D node averaging.
• Entropy fix in Roe solver was not correct.

There are no changes to the input file interface. If you have run cases without specifying the dimension in the input file, you are likely to see noticable improvement with the new version.