code_saturne User's Forum

Hello users,

I came across following errors when I was trying to run a serial simulation via Code_Saturne 4.0.2:

Computing geometric quantities (0.755 s)

ALMAX = 0.77674E+02 (Characteristic length )
ALMAX is the cubic root of the domain volume.

ALMAX is the length used to initialize the turbulence.
SIGSEGV signal (forbidden memory area access) intercepted!

Call stack:
1: 0x7fa923489788 <intprf_+0x23> (libsaturne.so.0)
2: 0x4038a7 <cs_user_initialization_+0x47a> (cs_solver)
3: 0x7fa9231c978e <inivar_+0x82e> (libsaturne.so.0)
4: 0x7fa9230e9ce5 <caltri_+0x2235> (libsaturne.so.0)
5: 0x7fa9230cc15e <cs_run+0x3fe> (libsaturne.so.0)
6: 0x7fa9230cbc52 <main+0x112> (libsaturne.so.0)
7: 0x7fa92273bec5 <__libc_start_main+0xf5> (libc.so.6)
8: 0x403369 <> (cs_solver)
End of stack

I am not sure where is the problem? Is there something wrong with the initialization files?

The source files are included in the attachment.

Best regards,

Yu.

Hello,

I have limited network connectivity this week, so ly answer wil be short...

You probably are missing initialization of parameters to iniprf in cs_user_initialization, so check those.

Regards,

Yvan

Hi Yvan,

I think what you really mean is 'intprf'. I did include it in the cs_user_initialization. the following is the code:

! --- 2. Turbulence Components
do iel = 1, ncel

zent = xyzcen(3,iel)

call intprf &
!==========
(nbmetd, nbmetm, &
zdmet, tmmet, umet , zent , ttcabs, xuent )

call intprf &
!==========
(nbmetd, nbmetm, &
zdmet, tmmet, vmet , zent , ttcabs, xvent )

call intprf &
!==========
(nbmetd, nbmetm, &
zdmet, tmmet, ekmet, zent , ttcabs, xkent )

call intprf &
!==========
(nbmetd, nbmetm, &
zdmet, tmmet, epmet, zent , ttcabs, xeent )
! --- Velocity components
cvar_vel(1,iel) = xuent
cvar_vel(2,iel) = xvent
cvar_vel(3,iel) = 0.d0

! ITYTUR est un indicateur qui vaut ITURB/10
if (itytur.eq.2) then

cvar_k(iel) = xkent
cvar_ep(iel) = xeent

elseif (itytur.eq.3) then

cvar_r11(iel) = d2s3*xkent
cvar_r22(iel) = d2s3*xkent
cvar_r33(iel) = d2s3*xkent
cvar_r12(iel) = 0.d0
cvar_r13(iel) = 0.d0
cvar_r23(iel) = 0.d0
cvar_ep(iel) = xeent

elseif (iturb.eq.50) then

cvar_k(iel) = xkent
cvar_ep(iel) = xeent
cvar_phi(iel) = d2s3
cvar_fb(iel) = 0.d0

elseif (iturb.eq.60) then

cvar_k(iel) = xkent
cvar_omg(iel) = xeent/cmu/xkent

elseif (iturb.eq.70) then

cvar_nusa(iel) = cmu*xkent**2/xeent

endif

if (iscalt.ge.0) then
! On suppose que le scalaire est la temperature potentielle :
call intprf &
!==========
(nbmett, nbmetm, &
ztmet, tmmet, tpmet, zent , ttcabs, tpent )

call field_get_val_s(ivarfl(isca(iscalt)), cvar_scalt)
cvar_scalt(iel) = tpent

endif
enddo

I copied this from the ''cs_user_initialization-atmospheric.f90''. Is there any problem?

I also attached the 'cs_user_initialization.f90' can you please have a look.

Kindest regards,

Yu

Yvan Fournier wrote:Hello,

I have limited network connectivity this week, so ly answer wil be short...

You probably are missing initialization of parameters to iniprf in cs_user_initialization, so check those.

Regards,

Yvan

Hello,

Are you running using the atmospheric model ? From the files you posted, you don't seem to be (but the listing and setup.log would help here, that's why they are in the forum usage recommendations).

If you are not using the atmospheric module with a meteo profile, some definititions (nbmetd, nbmetm, zdmet, tmmet, umet) may be missing, which explains your crash.

Try using simple (non-specific physic) boundary conditions, or do not forget to activate the meteo options (in general, setup your case progressively, activating main options first, adding complex profiles second).

Regards,

Yvan

Hi Yvan,

Thanks for all of your help. The initialization problem solved.

But the simulation stopped at the first step with the following error:

/home/martian/APP-INST/code_saturne-4.0.2/src/fvm/fvm_selector_postfix.c:1281: Fatal error.

Error parsing expression:
Group 28
^
Expected operator instead of operand.

I do not get the meaning of this statement. Can you please have a look?

The cd_user_boundary conditions.f90 and the listing.log had been uploaded.

Thanks very much.

Kindest regards,

Yu.

Yvan Fournier wrote:Hello,

Are you running using the atmospheric model ? From the files you posted, you don't seem to be (but the listing and setup.log would help here, that's why they are in the forum usage recommendations).

If you are not using the atmospheric module with a meteo profile, some definititions (nbmetd, nbmetm, zdmet, tmmet, umet) may be missing, which explains your crash.

Try using simple (non-specific physic) boundary conditions, or do not forget to activate the meteo options (in general, setup your case progressively, activating main options first, adding complex profiles second).

Regards,

Yvan

Hello,

Is your group named "Group 28", or simply 28.

If you have whitespace in your group names, you need to protect that whitespace inside the name.

This can be done by addding a '\' (but the C compiler "eats" one, so you need 2 in C, I'm not sure in Fortran, or mixing quotes:

"'Group 28'"

should work. The user documentation has more details on this.

Regards,

Yvan

Hi Yvan,

I thinke there is something wrong with my salome 7.6.0. I used it to generate the mesh and export as .cgns. However, the name I gave to the boundaries are missing. The 'Group 28' is given the software. However, it is not the case when I used the salome-meca 2014.1 to generate the mesh.

Regards,

Yu.

Yvan Fournier wrote:Hello,

Is your group named "Group 28", or simply 28.

If you have whitespace in your group names, you need to protect that whitespace inside the name.

This can be done by addding a '\' (but the C compiler "eats" one, so you need 2 in C, I'm not sure in Fortran, or mixing quotes:

"'Group 28'"

should work. The user documentation has more details on this.

Regards,

Yvan

Hello,

Yes, SALOME's favored format is MED. Importing CGNS from SALOME is OK, but for export and use by Code_Saturne, MED is recommended.

Regards,

Yvan