CFD Online Discussion Forums

CFD Online Discussion Forums (
-   SU2 (
-   -   an error when Paraview try to input flow.vtk (

shirazbj January 28, 2013 08:24

an error when Paraview try to input flow.vtk
2 Attachment(s)

I am playing with TestCases\rotating\open_rotor.

When input flow.vtk with Paraview, I got an error sasying Error reading assii dat.

I opened the vtk file and saw the lookup_table for scalars mac_number like this:

Is this the problem?



shirazbj January 29, 2013 00:34

1 Attachment(s)
Just tried another testcase, and found its output to the terminal also have the -1.#IND00 value.

no wonder the VTK will have the same value.

economon January 29, 2013 15:23

Thanks for trying the code and letting us know about issues with Paraview. A couple of thoughts:

1. What type of machine/OS are you using?

2. While the open rotor case is provided with the code (it is a little older, maybe out of date), I would recommend trying the rotating NACA 0012 which is simpler. If you try to view the output in Paraview with this case does it work? What about the transonic NACA 0012 case (not rotating)?

It could be something related to the Paraview file writer that is specific to rotating cases, so please let us know if you can isolate it with a few tests.

shirazbj January 29, 2013 21:04

Hi Thomas,

It's only the Mach number has this problem for this open rotor case. But from my second post using another test case, I realised the simulation is generating this data and output to screen.

I googled "1.#QNAN00000000" and found out it's the way "printf" deal with a huge number. So something is sort of divided by zero.

I am using win7-32b. I compiled the source with gcc for win and can now trace the program now. I found in output_structure.cpp when caculate Mach number:

                WriteInOutputFile(geometry, solution_container[FLOW_SOL], ConsVar_file, "Mach_Number", scalar,  0, "GetMach", config);
In WriteInOutputFile sub,

                if (!incompressible) {
                      mach_number = sqrt(solution_container->node[iPoint]->GetVelocity2())/solution_container->node[iPoint]->GetSoundSpeed();

the value of SoundSpeed() in my case is ZERO. The program can save a few times but give up at the end.

Here should be a compressible calculation, but I give a very low mach number in cfg file.

Just wondering is "bool" value of "incompressible" set in CFG file? Which one? The solver is a compressible one, where to let it deal with incompressible case?



economon February 1, 2013 03:17

I can confirm with some of my own tests that there is something odd happening in the Paraview output for rotating frame simulations. We will look into this issue, but in the meantime, do you have access to Tecplot (or another package that can read Tecplot ascii files)? These output files seem to be working just fine for rotating frame problems.

Lastly, yes, SU2 can indeed handle incompressible cases by using an artificial compressibility method. Please see some of the included test cases or other posts where this is discussed: the config options and required inputs are somewhat different for these cases. In general, an incompressible simulation can be run using the following option:

% Incompressible flow using artificial compressibility (NO, YES)

shirazbj February 1, 2013 03:45

If I set FULLMG=YES with MGLEVEL= 2, i'll get zero velocity and zero sound speed. No matert if I output Tecplot or Paraview data.

All times are GMT -4. The time now is 08:33.