[diamondx]

Hello everyone,
I started learning UDF today, i created my first code. It's an easy source term that i apply on a SPECIFIC domain of fluid. the source is a x-momentum, here is my code:

#include "udf.h"

DEFINE_SOURCE(test,c,t,dS,eqn)
{
real xc[ND_ND];
real source;

C_CENTROID(xc,c,t);

if ((xc[0] == -0.135608)&&(xc[1] == 0.179728))
{source = 1000;
dS[eqn] = 0;}
else {source = 0;dS[eqn] = 0;}
C_UDMI(c,t,0)=source;
return source;
}

The UDF is interpreted/compiled without any error. this udf applies a source in the location(x,y). I should see a change in the velocity vector, right ? nothing happens, why ? I'm trying to simulate plsma actuators and in all the article that i read, this same udf code is used...

[flotus1]

Because of numerical inaccuracies, you always have to provide a small threshold for statements like "if (position != ..."
The position of the cells in your domain will never be -0.135608, but rather -0.135608000001

Make it "if(position < x_0+threshold && position > x_0-threshold)" instead

[diamondx]

thanks alexander for you reply. went back and change the code into this one:

#include "udf.h"


DEFINE_SOURCE(test,c,t,dS,eqn)
{
real xc[ND_ND];
real source;

C_CENTROID(xc,c,t);

if ((xc[0] <= -0.1557860000)&&(xc[0] >= -0.196412000000)&&(xc[1] <= 0.189188000000)&&(xc[1] >= 0.168735000000))

{source = 1000000;
dS[eqn] = 0;}
else {source = 0;dS[eqn] = 0;}
return source;
}

Still the problem persist... if you are wondering where did i get those coordinates, i went to draw the mesh, and i used mouse probe to to determine the location where i want the source to be applied. i copy paste it in the code...
Thanks again for your help

[flotus1]

Are you 100% sure that there are cell centroids within the ranges you specified?
Remember that sources are applied on a "per-cell" basis.

For debugging purposes, try applying the source to a bigger area with less restrictions, say

if (xc[0] < 0) source= 1000;
return source;

If this still has no effect... well we will have to think again.

[diamondx]

thanks a lot for your support, recently compilation works but i soon as i hit on run calculation it shows me this error i dont know why:

999999 (..\src\mpsystem.c@1149): mpt_read: failed: errno = 10054

999999: mpt_read: error: read failed trying to read 8 bytes: No such file or directory
MPI Application rank 0 exited before MPI_Finalize() with status -1073741819
The Parallel FLUENT process could not be started.

[flotus1]

Are you running in parallel?

Normal UDFs do not work in parallel, but as far as I understood, they can be changed accordingly. Refer to the UDF manual.

For now, try running on 1 core only.

[diamondx]

changed into serial, still error problem.
Error:
FLUENT received fatal signal (ACCESS_VIOLATION)
1. Note exact events leading to error.
2. Save case/data under new name.
3. Exit program and restart to continue.
4. Report error to your distributor.
Error Object: #f

Yeah this is definitely the coordinates.

#include "udf.h"
DEFINE_SOURCE(test,c,t,dS,eqn)
{
real xc[ND_ND];
real source,y;

C_CENTROID(xc,c,t);
y=xc[1];

if (y <= 0.1864891)

{source = 1000000;
dS[eqn] = 0;}
else {source = 0;dS[eqn] = 0;}
C_UDMI(c,t,0)=source;
return source;

if you were me, how will you get the coordinates, will you use the mouse probe ?
the procedure: i load the case, define then interpret the udf. i go to cell condition turn on source term , aplly x-momentum click on ok. then initialize and finally run calculation...
i appreciate your help...

[flotus1]

Here is a source UDF I currently use (stripped it from everything but the coordinate thing):
Copy it and try to run it.

DEFINE_SOURCE(sourceterm,c,t,dS,eqn)
{
double x[ND_ND];
double x0,y0;
double y_m=0.0;
double I_0=1000;

C_CENTROID(x,c,t);
if(x[1] <= y_m)
{
source = I_0;
}
}
return source;

When you say "apply x momentum" you mean you add 1 momentum source like in my attached picture?
The rest of your procedure seems ok.

[diamondx]

finally it worked. My udf was good. it worked after i changed the scenario. i was applying the source in a convergent divergent nozzle. velocity was near transonic, may be that's why nothing was happening. after i changed the geometry into a duct where air was flowing with 1 m/s. i could see the source in the velocity contour plot. thanks a lot for you support alex, you deserve a reputation up...

[kaushik_rai]

how can we differentiate between a 32 & 64 bit udf in fluent???