[shirazbj]

which post process software to view the result?

Thanks

[arjun]

Originally Posted by shirazbj

which post process software to view the result?

Thanks

inavier writes output in tecplot , fieldview and ensight format. This means if you use visIt you can read tecplot or if you use paraview you can read ensight gold. If i remember correctly inavier exported in binary and ascii format of ensight.

Anyway, there was a big reason i was not pushing inavier. The reason was that inaver used additive corrective multigrid. Which is not very suitable for diffusion dominated problems.

This means that pressure correction equation suffers and inavier is not good for very large meshes.

Here is current update.

Based on pyamg liberary, i have re-written smoothed agregation multigrid which works very good even for very large problems.
I have replaced this new multigrid solver in old inavier code and it is working very well.

So the old code with new multigrid is not in public domain because my multigrid uses 3 functions from pyamg library and i am not sure if i can put it online without proper license.

Further update.

Another problem with old inavier code was that it was not organised well.

So i have now re-written everything with new multigrid , i now call it iNavier Version 2.

At the moment of writing this message, this code in single processor is working very well. I am in process of writing a parallel version of this code based on shared memory. I can say 70% of it is already converted in 2 weeks time it will be working well with shared memory.


The biggest feature of new inavier is that it can work as an external liberary in any other C code and could be used as solving general partial differential equations.

Other than adding source terms of these equations solver takes care of everything. That means allocating memory, setting up equations, applying boundary conditions and exporting results in output files.

Its very very easy to use now and very very flexible and easy to use.

I am still not sure what form it will be available to people though. I need to talk with my company if they want to push it and own it. Or it goes out as open source code.

Plus I am now spending time thinking about gpu version of this multigrid. If this is done, you might see GPU based version of ultra fast LES code.

I have one more working immersed boundary based DNS code, that my company uses. Unfortunately since my company uses i can not share it outside.

[shirazbj]

Hi Arjun,

Thanks for your quick reply.

I am using paraview.

I just found this code and hope I can learn CFD from it. My first goal is a 3d code to solve a step flow to see how theory is turned into a code.

Regards,

Cean

[ztdep]

I strongly suggest you to use OpenFVM, it is a powerful c code. fvm on the unstructured grid system, with VOF multiphase model.

[arjun]

Originally Posted by shirazbj

Hi Arjun,

Thanks for your quick reply.

I am using paraview.

I just found this code and hope I can learn CFD from it. My first goal is a 3d code to solve a step flow to see how theory is turned into a code.

Regards,

Cean

i would suggest not to use the current version that is available. I wanted to remove it but sourceforge does not allow me.

This code works well with small cases and structured meshes. But anything larger than 1 million and it can have problems. (though i have run some calculations around 3-4 million cells on it).

If you intend to use larger meshes and parallization, then openFOAM i think is good thing to spend time with.


New version of iNavier shall be available in June , IF IT IS DECIDED TO PUT IT ONLINE.

The reason why it will be available in june is, this time i intend to create documentation properly.

After that IT MAY NOT be updated because of two reasons:

1. It will be used as external liberary that solves partial diff equations on unstructured meshes. (plus can solve navier stokes + turbulence). So i do not think i can modify it further.

2. OpenFOAM is already there in unstructured grids and segregated section. What is needed at the moment is - highly parallel LES,DNS code that can support immersed boundary and thus can allow moving boundaries.
This type of solver with cartesian meshes , i have already written and working. But i am working on advanced version which i am half way. (that is between unstructured and cartesian type, that is hybrid).


(half way means - I can mark moving solids very very fast in such meshes (of sizes say 500 million or more) , and pressure equation solver is working well now).

So i think immersed boundary and GPU based methods are two thing other do not offer and i shall spend my time on.

[shirazbj]

Which result file I should open with Paraview 3.6.2.

I tried the Box2d_rotation example. I thought it was solvered. What is the file extension name for Ensight?

Thanks

[arjun]

Originally Posted by shirazbj

Which result file I should open with Paraview 3.6.2.

I tried the Box2d_rotation example. I thought it was solvered. What is the file extension name for Ensight?

Thanks

there must be a file with extn case, with other files with extension .geo etc etc. If i remember correctly you should read .case file.

try searching on ensight gold format.


i would still suggest not to use old inavier, because soon there will be newer version and everything is changed in new one. Nothing matches with old.

I will be busy preparing test and tutorials and little docs in next few days.