Learning Forum

[eric1234598765]

Hi,

I'm using Fluent 2020R2 to obtain laminar viscosity and velocity derivative via "C_MU_L(c,t)" and "C_DUDY(c,t)".

The trouble UDF is tested in VOF model with single processor using SST-K-OMEGA, which is going to compute the viscous drag of the "interface" which Fluent can't handle.

Here's UDF:

==================================================================

#include "udf.h"

DEFINE_ON_DEMAND(Viscous_Drag)

{

    Domain *d = Get_Domain(1); /*Get the domain ID*/

    face_t f;

    cell_t c;

    /*Get the thread id of that surface*/

    Thread *t  = Lookup_Thread(d,6); /* Interface_Fluid_porous1 */;

    Thread *t1 = Lookup_Thread(d,7); /* Interface_Fluid_porous2 */;

    real force;

    real avgf;

    int ii;

    if (!Data_Valid_P())

    {

    return;

    }

    force=0.0;

    avgf=0.0;

    ii=1;

    /* Loop over all cell*/

    begin_c_loop(c,t)

    {

        force  = force + C_MU_L(c,t);

    }

    end_c_loop(c,t)

    Message("Viscous drag is = %8.4f ", force );

}

=================================================================

After debugging, I found that the Fluent would pop out "Received signal SIGSEGV." error and show the zero value when I using either "C_MU_L(c,t)" or "C_DUDY(c,t)" in the begin_c_loop, and with no problem when using other UDF cell macro such as C_T(c,t), C_K_L(c,t), and so on. However, laminar (molecular) viscosity was checked by xy-plot which is not zero.

Obviously, the problem is the misusing of the C_MU_L(c,t) and C_DUDY(c,t), therefore, I looking forward to any help.

Thanks!

[eric1234598765]

Update:
Same error as stated when the VOF model is off.

[Rob]

Check if C_MU_L(c,t) needs a phase domain for VOF. I'm guessing, UDFs aren't one of my strong points.

[eric1234598765]

Hi I've solved the problem by pointing the face thread to the face centroid, which is
Thread *t = Lookup_Thread(d,6); /* Interface_Fluid_porous1 */
Thread *t0 = t->t0; /* Access t0 (Pointer to the adjacent cells thread.) from t (face thread) */

And using adjacent Cell Index in the f_loop , which is:
c0 = F_C0(f,t);
Now, the pointers I set are : t, t0, c0. Using c0 and t0 instead of c and t solve the problem in the loop. So the UDF is like:
==================================================
DEFINE_ON_DEMAND(Viscous_Drag1)
{
Domain *d = Get_Domain(1); /*Get the domain ID*/
face_t f; /*Index that identifies a wall face.*/
cell_t c0; /*Cell index that identifies the cell next to the wall.*/

/*Get the thread id of that surface*/
Thread *t = Lookup_Thread(d,6); /* Interface_Fluid_porous1 */
Thread *t0 = t->t0; /* Access t0 (Pointer to the adjacent cells thread.) from t (face thread) */
real viscosity;

if (!Data_Valid_P())
{
return;
}
real NV_VEC (A); /*Area normal vector*/
/* Loop over all surface faces*/
begin_f_loop(f,t)
{
F_AREA(A,f,t); /*Get the area vector*/
c0 = F_C0(f,t);
viscosity += C_MU_EFF(c0,t0);
}
end_f_loop(f,t)
}
=========================================
Best