[souria]

Hey everyone,

I'm tracking particles (surface injection) in steady flow using DPM. I used this UDF to impose to particles the velocity of the flow when they will be injected(just for the initialisation). Herein the lines :

#include "udf.h"
#include "dpm.h"

DEFINE_DPM_INJECTION_INIT(velocity_magnitude, I)
{

Particle *p;
cell_t c;
Thread *t;

loop(p,I->p_init) /*Standard Fluent looping Macro to get particle streams in an inejction using unsteady particle tracking*/
{
c = P_CELL(p); /*Get the cell that the particle is currently in*/
t = P_CELL_THREAD(p); /*Get the thread that the particle is currently in*/
P_VEL(p)[0] = C_U(c, t); /*Longitudinal velocity*/
P_VEL(p)[1] = C_V(c, t); /*Spanwise velocity*/
P_VEL(p)[2] = C_W(c, t); /*vertical velocity*/
}


}

But, when I check partciles velocity for the first step it gave me ( u = 0, v = 0 and w= 0). So, it didn't take in account the UDF !!!
Please tell what it's wrong with my UDF !!!

Thanks for your help

[Sun]

might be silly, but did you remember to hook up the UDF?
and what if you use the very same code in a

DEFINE_ADJUST

macro to check if you can get the velocity for time steps other than the first one.

[souria]

Originally Posted by Sun

might be silly, but did you remember to hook up the UDF?
and what if you use the very same code in a

DEFINE_ADJUST

macro to check if you can get the velocity for time steps other than the first one.

Hey Sun, Yes I load and built the UDF and put it in initialization case (image1 : inisialization)
But, in point properties the velocities still equal to 0 (second image), do you think that it comes from this ??

Thanks for your reply

Souria

[souria]

I fund out that I have no access to the User-defined Functions Hooks !!!

Any help about this ???

[CeesH]

Hi Souria,

Do you only see the velocity 0 in the setup screen, or do you actually `measure' it to be 0 when iterating in particle tracks?

[souria]

Originally Posted by CeesH

Hi Souria,

Do you only see the velocity 0 in the setup screen, or do you actually `measure' it to be 0 when iterating in particle tracks?

Yes, Ceesh, after the dpm iterations I get "0" for the first step !!! So, it didn't intialize the velocity of particle to these of the flow !!

any idea CeesH ??

[CeesH]

So far I don't see any error (I assume the flow is already steady at the point you inject the particles)

Are you sure the define_adjust function is properly compiled? It doesn't give any warnings or so? I've had some problems with mass transfer functions in which the function compiled properly, but a warning for udf_names.h was given and the function did not appear in the proper hook menu - nothing like that here?

[souria]

Originally Posted by CeesH

So far I don't see any error (I assume the flow is already steady at the point you inject the particles)

Are you sure the define_adjust function is properly compiled? It doesn't give any warnings or so? I've had some problems with mass transfer functions in which the function compiled properly, but a warning for udf_names.h was given and the function did not appear in the proper hook menu - nothing like that here?

It gives no error or warning when I complied and built the UDF !
Yes, Im in steady flow but the DPM is unsteady.
To be clear, for hooking the UDF, in my case, I just did it in the injection box (not in User defined Function Hooks), is it correct ?

Thanks for your answers CeesH

[CeesH]

Yes, the initialization is via the injection box. The define_adjust should be via function hooks. but it is weird the define_adjust doesn't pop up in the menu when you do compile it...

[souria]

Originally Posted by CeesH

Yes, the initialization is via the injection box. The define_adjust should be via function hooks. but it is weird the define_adjust doesn't pop up in the menu when you do compile it...

"the define-adjust" ? what's it's role, here?

[CeesH]

As sun mentioned, maybe you can use define_adjust with a similar code (a loop over all particles to adjust their velocity in the 1st timestep) to achieve the same result as the particle initialization UDF. You do have to rewrite the code a little bit, it's a little less straightforward but the result should be the same: at the end of the 1st timestep the velocity of all particles is adjusted to the cell velocity. I guess it is most easy if you inject all particles in a single timestep. is this the case? If so, the pseudocode would look like:

if NStep = 1 (or flowtime is 1dt){
{- loop over all particles
- set P_VEL(p)[i] = C_i(c,t)}}

[souria]

Originally Posted by CeesH

As sun mentioned, maybe you can use define_adjust with a similar code (a loop over all particles to adjust their velocity in the 1st timestep) to achieve the same result as the particle initialization UDF. You do have to rewrite the code a little bit, it's a little less straightforward but the result should be the same: at the end of the 1st timestep the velocity of all particles is adjusted to the cell velocity. I guess it is most easy if you inject all particles in a single timestep. is this the case? If so, the pseudocode would look like:

if NStep = 1 (or flowtime is 1dt){
{- loop over all particles
- set P_VEL(p)[i] = C_i(c,t)}}

I tried to do this but it gave this message : 'time_step : is not a member of particle_struct' !

[CeesH]

do you get such an error for:

#include "udf.h"
#include "dpm.h"

DEFINE_ADJUST(whatever it needs here)
{
Injection *I;
Injection *dpm_injections = Get_dpm_injections();
Particle *p;
int T = N_TIME;

if(T == 1)
{
loop(I,dpm_injections)
{
loop(p,I->p)
{
c = P_CELL(p); /*Get the cell that the particle is currently in*/
t = P_CELL_THREAD(p); /*Get the thread that the particle is currently in*/
P_VEL(p)[0] = C_U(c,t)
P_VEL(p)[1] = C_V(c,t)
P_VEL(p)[2] = C_W(c,t)
}}}}

(note, I did not run or debug this)

[pakk]

I would be interested in what happens with the following code:

#include "udf.h"
#include "dpm.h"

DEFINE_DPM_INJECTION_INIT(velocity_magnitude, I)
{

Particle *p;
cell_t c;
Thread *t;

loop(p,I->p_init)           /*Standard Fluent looping Macro to get particle streams in an inejction using unsteady particle tracking*/
{
c = P_CELL(p);              /*Get the cell that the particle is currently in*/
t = P_CELL_THREAD(p);       /*Get the thread that the particle is currently in*/
P_VEL(p)[0] =7;    /*Longitudinal velocity*/
P_VEL(p)[1] = 8;    /*Spanwise velocity*/
P_VEL(p)[2] = 9;    /*vertical velocity*/
}
}

Do you see velocities (0,0,0) or (7,8,9)?

[souria]

Originally Posted by pakk

I would be interested in what happens with the following code:

#include "udf.h"
#include "dpm.h"

DEFINE_DPM_INJECTION_INIT(velocity_magnitude, I)
{

Particle *p;
cell_t c;
Thread *t;

loop(p,I->p_init)           /*Standard Fluent looping Macro to get particle streams in an inejction using unsteady particle tracking*/
{
c = P_CELL(p);              /*Get the cell that the particle is currently in*/
t = P_CELL_THREAD(p);       /*Get the thread that the particle is currently in*/
P_VEL(p)[0] =7;    /*Longitudinal velocity*/
P_VEL(p)[1] = 8;    /*Spanwise velocity*/
P_VEL(p)[2] = 9;    /*vertical velocity*/
}
}

Do you see velocities (0,0,0) or (7,8,9)?

Pakk, Weird weird !! I got (0,0,0)
PS. In the Box Injection > Point Properties : the variables value of each velocity is "0" (I never change this box), did this affect the function of the UDF ? Or genrally, should I put other values, but which values, if I want to set the particles velocity to that of the flow at the injection ???

Thanks for your help

[souria]

Originally Posted by CeesH

do you get such an error for:

#include "udf.h"
#include "dpm.h"

DEFINE_ADJUST(whatever it needs here)
{
Injection *I;
Injection *dpm_injections = Get_dpm_injections();
Particle *p;
int T = N_TIME;

if(T == 1)
{
loop(I,dpm_injections)
{
loop(p,I->p)
{
c = P_CELL(p); /*Get the cell that the particle is currently in*/
t = P_CELL_THREAD(p); /*Get the thread that the particle is currently in*/
P_VEL(p)[0] = C_U(c,t)
P_VEL(p)[1] = C_V(c,t)
P_VEL(p)[2] = C_W(c,t)
}}}}

(note, I did not run or debug this)

Results : I could pop up the "adjust fuction" (in User defined Function Hooks) but still get (0,0,0) for the fisrt step !!

[pakk]

Originally Posted by souria

Pakk, Weird weird !! I got (0,0,0)

What if you replace

loop(p,I->p_init)

by

loop(p,I->p)

?


About your DEFINE_ADJUST code: I thought T_INIT would be 0 in the first iteration step, not 1.

[CeesH]

another one about the define_adjust; are you actually injecting in the first transient timestep?

what you could try is setting a small timestep, then adjusting the velocities in the first timestep after initializing, rather than the timestep of initializing. Yes, this is a very crude test, but maybe it helps. If it works, you can reset the timestep to the normal value and continue calculating. The one `fake' first step shouldn't make too much a difference.

But of course, a working DPM initialization would be preferable.

[souria]

Originally Posted by pakk

What if you replace

loop(p,I->p_init)

by

loop(p,I->p)

?


About your DEFINE_ADJUST code: I thought T_INIT would be 0 in the first iteration step, not 1.

I's great PAKKKkkkkk, it works, Ive just changed loop(p,I->p_init to loop(p,I->p) !!

Thanks a lot

[souria]

Originally Posted by CeesH

another one about the define_adjust; are you actually injecting in the first transient timestep?

what you could try is setting a small timestep, then adjusting the velocities in the first timestep after initializing, rather than the timestep of initializing. Yes, this is a very crude test, but maybe it helps. If it works, you can reset the timestep to the normal value and continue calculating. The one `fake' first step shouldn't make too much a difference.

But of course, a working DPM initialization would be preferable.

Thanks a lot CeesH for all your replies (I learned so much thing). Right Now, it works and gave me the result I'm expected