code_saturne User's Forum

Hi,

I'm currently setting up the ahmed car body case using DDES, and I want to use the vortex method at the inlet.

I have a data file which contains u,v,w,u,v etc from a precursor RANS simulation (32798 lines).

When I go into usvort I therefore set ndat to be 32798, e.g the number of lines in my data file. However code_saturne gives me an error saying that the maximum size of ndat is 10,000, please check vortex.h.

So I copy vortex.h into my SRC directory and change ndatmx to be 40,000.

However it appears to ignore this when I run the simulation.

I'm using a version of saturne which is installed on a cluster which I don't have admin permissions so I cannot easily recompile it.

I was just wondering if firstly an include file can be read from the local SRC directory of a case, and if it isn't, is there any work around to set ndatmx to be 32798 but place ndatmx in one of the subroutines instead of a n include file?

Thanks very much,

Neil

Hello,

If an include file (or a module for Code_Saturne versions > 2.0) is added to user subroutines, only additions may be done without risk; modifications are extremely dangerous, as the rest of the code was compiled with the unmodified version, and may place other variables just after a modfied one assuming its initial size... A perfect recipe for disaster.

So the best solution is to recompile the whole code. Even if was compiled by an admin, nothing prevents you from compiling it in your user directory (nothing in the code requires administrative privileges, such as starting or ending system services).

Ideally, if the build directory in which the code was compiled still exists, you may check the configure command at the top of the config.log file, adapt the paths, and build the code with the same options on your account.
Otherwise, the installation manual will be your friend. The only essential aspects to compiling the code on your own account will be to use the same MPI compiler wrappers (often in the default path, unless multiple versions are installed, in which case setting the correct environment, possibly indicated with "module" lines in the reference install's runcase), and batch system (to be defined for version 2.0 using --with-batch=<type>).

The more info you can obtain or check for yourself on the reference/admin install, the easier this will be.

Regards,

Yvan