[Perra]

I am reconstucting a old UDF that is used for a partial slip boundary condition from 3D to a 2D axisymetric swirl model. The UDF works fine with no modifications when hooking only the X and Y components but when adding the swirl component the solution diverges directly.

My thoughts are that when using ND_ SET for the nol vector I dont get the third component, same for the ND_DOT and NV_V_VS oprations. So when I call for vel_normal[2]; it should not exists I guess?

Am I corect in this? And also I wounder if the F_W(f,t) comand works in 2D axisymetric swirl flow to get the swirl component of the flow or if there is another way?

part of the UDF:
begin_f_loop(f, t) /* loops over faces in a face thread */
{
c0=F_C0(f,t);
tc0=THREAD_T0(t);
AN=pow((1.0+epN*pow(-epN,q))/2.0,1.0/(q+1.0));
AT=pow((1.0+pow(epT,q+1.0))/2.0,1.0/(q+1.0));
BN=pow((epN*(1.0+pow(-epN,q)))/(1.0+epN),1.0/(q+1.0));
BT=pow((epN+pow(epT,q+1.0))/(1.0+epN),1.0/(q+1.0));

F_CENTROID(center_face,f,t);
C_CENTROID(center_cell,c0,tc0);

NV_VV(distance,=,center_face,-,center_cell);
distance_cell = 2.0*NV_MAG(distance);
aN=AN-BN;
aT=AT-BT;
bN=AN*distance_cell;
bT=AT*distance_cell;
cN=0.0;
cT=0.0;
ND_SET(nol[0],nol[1],nol[2],0.0,0.0,0.0);
F_AREA(normal,f,t); /* obtaining normal direction of the wall */
U_vel=F_U(f,t); /* obtaining u direction of velocity on the wall */
V_vel=F_V(f,t); /* obtaining v direction of velocity on the wall */
W_vel=F_W(f,t); /* obtaining w direction of velocity on the wall */

mag_normal=ND_DOT(U_vel, V_vel, W_vel, normal[0], normal[1], normal[2]);
NV_V_VS(vel_normal, =, nol, +, normal, *, mag_normal);
velz_normal=vel_normal[2];
velz_tangen=W_vel - velz_normal;

mu=C_MU_L(c0,tc0);
vfrac=C_VOF(c0,tc0);
tao_z=vfrac*mu*(aN*velz_normal/bN + aT*velz_tangen/bT);

F_PROFILE(f, t, i) = tao_z;
}
end_f_loop(f, t)
}

Greatful for coments

[dmoroian]

Hello Perra,
One necessary modification is for the area: in an axisymmetric computation, you should multiply the result with 2*PI. See chapter 3.2.1 in Fluent UDF Manual.

F_AREA(normal,f,t);
normal[0] *= 2*M_PI;
normal[1] *= 2*M_PI;
normal[2] *= 2*M_PI;

Another comment: are you imposing the same value on all 3 velocity components, or is just my impression?

Dragos

[wpdeng]

Originally Posted by dmoroian

Hello Perra,
One necessary modification is for the area: in an axisymmetric computation, you should multiply the result with 2*PI. See chapter 3.2.1 in Fluent UDF Manual.

F_AREA(normal,f,t);
normal[0] *= 2*M_PI;
normal[1] *= 2*M_PI;
normal[2] *= 2*M_PI;

Another comment: are you imposing the same value on all 3 velocity components, or is just my impression?

Dragos

Hi,

I simplified 3D geometry to 2D axisymmetry swirl. And I want to load a heat source about space distribution, but it's wrong when I use Cartesian coordinate system.

What should I do to deal with it?