[msd]

Hello All,

I am trying to write a UDF for the pressure outlet boundary condition and the underlying condition is,
P = R*Q.
P = Mean pressure at the outlet (i.e. it remains constant at the whole face)
R = constant taken from literature
Q = flow rate
At the inlet, the pressure boundary condition is specified.

The code written goes as follows,
#include "udf.h"

#define R 12e9 /*Value taken from literature*/

#include "mem.h"
#include "metric.h"

DEFINE_PROFILE (outlet_pressure, thread, position)
{
real A[ND_ND];
face_t f;
Thread *t;
real Ar, V, Qfinal;
real Q = 0;

begin_f_loop(f, thread)
{
F_AREA(A, f, thread);
Ar = NV_MAG(A); /* Used for the computation of the magnitude of the area vector */
V = F_U(f, thread);
Q = Q + (Ar*V);
}
end_f_loop(f, thread)

Qfinal = Q;

begin_f_loop(f, thread)
{
F_PROFILE(f, thread, position) = R*Qfinal;
}
end_f_loop(f, thread)
}


Can someone please tell if there is some logical error in the code because the relation P = R*Q is not getting satisfied.

Thanks in advance!

[AlexanderZ]

right before line: Qfinal = Q;
you need to add global summation macro to make code works in parallel

Q = PRF_GRSUM1(Q);

also I'm not sure if you need to use full outlet area of each finite are (as you did) here:

Q = Q + (Ar*V);

check it

[Yasser]

  q += a[0] * v

q += a[0] * v;

[msd]

@Yasser, while inculcating this change, should I not calculate the magnitude of the area vector?

[Yasser]

Hello,

If you are considering F_U, you should be multiplying this time the area in the x-direction only!

[msd]

Thanks, Yasser for your input, I will check it.

[msd]

Thank you Alexander for your input, I will check if it works.

[msd]

Originally Posted by AlexanderZ

right before line: Qfinal = Q;
you need to add global summation macro to make code works in parallel

Q = PRF_GRSUM1(Q);

also I'm not sure if you need to use full outlet area of each finite are (as you did) here:

Q = Q + (Ar*V);

check it

Dear Alexander,

I tried incorporating the change suggested by you and the modified code is as follows:
#include "udf.h"

#define R 12e9 /*Value taken from literature*/

#include "mem.h"
#include "metric.h"
#include "prf.h"

DEFINE_PROFILE (outlet_pressure, thread, position)
{
real A[ND_ND];
face_t f;
Thread *t;
real Ar, V, Qfinal;
real Q, P;

begin_f_loop(f, thread)
{
F_AREA(A, f, thread);
Ar = NV_MAG(A); /* Used for the computation of the magnitude of the area vector */
V = F_U(f, thread);
Q = Ar*V;
}
end_f_loop(f, thread)

Q = PRF_GRSUM1(Q);
Qfinal = Q;

begin_f_loop(f, thread)
{
F_PROFILE(f, thread, position) = R*Qfinal;
}
end_f_loop(f, thread)
}

But I have a couple of questions, as in will the final pressure obtained be the same all over the face or some more modification is required? Also, I don't know why my relation is still not getting satisfied.

Can you please help?

Thanks!

[msd]

Dear Alexander,

I am facing one more problem, the volumetric flow rate calculated by the UDF (as I printed and checked the value), and the surface integral is coming different and it is differing by a value of order 10^(-3). But the relation is satisfied by the value of the volumetric flow rate obtained from the UDF and not from surface integral.

Can you please tell what can be the possible reason behind it?

Thanks in advance!

[pakk]

Q = Ar*V;

This is wrong, you only keep the value of the last cell.

Q += Ar*V;

[msd]

Originally Posted by pakk

Q = Ar*V;

This is wrong, you only keep the value of the last cell.

Q += Ar*V;

Thank you, I will try incorporating the changes.