[cfdstar]

Hi,

I wanted to define a shear boundary condition as a function of temperature gradient. Following is the UDF I wrote for this purpose. The UDF only returns the the constant value of 0.0 since the temperature gradient is not accessible. However, from the contour plots, I can see that temperature gradients are available after the first iteration.

If I remove the "if" statement, I receive this error:
Error: received a fatal signal (Segmentation fault).
Error Object: #f

I should mention that I keep temporary solver memory from being freed. Any help to solve this problem is highly appreciated.

#include "udf.h"
#include "mem.h"

DEFINE_PROFILE(WSS, t, i)
{
face_t f;
cell_t c;
real TG;
begin_c_loop(c,t)
{
if (NNULLP(THREAD_STORAGE(t, SV_T_G)))
{
printf("Gradient of Temperature is available \n ");
TG = NV_MAG(C_T_G(c,t));
C_PROFILE(c,t,i)=-0.0001*TG;
} else
{
C_PROFILE(c,t,i)=0.;
}
}
end_c_loop(c,t)
}

[gearboy]

Originally Posted by cfdstar

Hi,

I wanted to define a shear boundary condition as a function of temperature gradient. Following is the UDF I wrote for this purpose. The UDF only returns the the constant value of 0.0 since the temperature gradient is not accessible. However, from the contour plots, I can see that temperature gradients are available after the first iteration.

If I remove the "if" statement, I receive this error:
Error: received a fatal signal (Segmentation fault).
Error Object: #f

I should mention that I keep temporary solver memory from being freed. Any help to solve this problem is highly appreciated.

#include "udf.h"
#include "mem.h"

DEFINE_PROFILE(WSS, t, i)
{
face_t f;
cell_t c;
real TG;
begin_c_loop(c,t)
{
if (NNULLP(THREAD_STORAGE(t, SV_T_G)))
{
printf("Gradient of Temperature is available \n ");
TG = NV_MAG(C_T_G(c,t));
C_PROFILE(c,t,i)=-0.0001*TG;
} else
{
C_PROFILE(c,t,i)=0.;
}
}
end_c_loop(c,t)
}

Alloc_Storage_Vars(domain, SV_T_RG, SV_T_G, SV_NULL);
T_derivatives(domain);
.........
your code....
.........

Free_Storage_Vars(domain, SV_T_RG, SV_NULL);
Free_Storage_Vars(domain, SV_T_G, SV_NULL);

[mahdi-united]

Originally Posted by gearboy

Alloc_Storage_Vars(domain, SV_T_RG, SV_T_G, SV_NULL);
T_derivatives(domain);
.........
your code....
.........

Free_Storage_Vars(domain, SV_T_RG, SV_NULL);
Free_Storage_Vars(domain, SV_T_G, SV_NULL);

hi please help me
what are the arguments for this macros for accessing species mass fraction?

[lbj007]

Originally Posted by cfdstar

Hi,

I wanted to define a shear boundary condition as a function of temperature gradient. Following is the UDF I wrote for this purpose. The UDF only returns the the constant value of 0.0 since the temperature gradient is not accessible. However, from the contour plots, I can see that temperature gradients are available after the first iteration.

If I remove the "if" statement, I receive this error:
Error: received a fatal signal (Segmentation fault).
Error Object: #f

I should mention that I keep temporary solver memory from being freed. Any help to solve this problem is highly appreciated.

#include "udf.h"
#include "mem.h"

DEFINE_PROFILE(WSS, t, i)
{
face_t f;
cell_t c;
real TG;
begin_c_loop(c,t)
{
if (NNULLP(THREAD_STORAGE(t, SV_T_G)))
{
printf("Gradient of Temperature is available \n ");
TG = NV_MAG(C_T_G(c,t));
C_PROFILE(c,t,i)=-0.0001*TG;
} else
{
C_PROFILE(c,t,i)=0.;
}
}
end_c_loop(c,t)
}

Hi,have you solve this problem,I'm using C_UDSI_G and can't find a way to solve it.

[AlexanderZ]

from manual

In order to retain the gradient data (when you want to set up user-defined scalar transport equations, for example), you can prevent the solver from freeing up memory by issuing the text command solve/set/expert and then answering yes to the question, “Keep temporary solver memory from being freed?” Note that when you do this, all of the gradient data is retained, but the calculation requires more memory to run.

[lbj007]

Originally Posted by AlexanderZ

from manual

I've already tried this,not working.please let me introduce my situation:
now I have used C_UDSI_G to calculate current density successfully in a cas, it can be seen in post-processing, then I write cas and dat. Now I read cas, and when I read dat it shows:



the process of fluent has already down. I've know the reason for it:when I read dat, it can't access the value of C_UDSI_G immediately because I used define_profile macros and it would first be called:



now if I change the value to constant like this:



and also need to calcalate for one iteration,until then it won't get wrong when I read dat.

so I wonder if there's a way to read dat normally without these procedure...

[AlexanderZ]

unfortunately, no idea, how to deal with it.

try to remove reading from .dat step.
simulate everything without exit from fluent

[lbj007]

Originally Posted by AlexanderZ

unfortunately, no idea, how to deal with it.

try to remove reading from .dat step.
simulate everything without exit from fluent

OK，thanks.BTW, will this situation happen when using other macros of gradient,such as C_T_G? or it just goes wrong when using UDS gradient.

[Shoonya]

Originally Posted by lbj007

Hi,have you solve this problem,I'm using C_UDSI_G and can't find a way to solve it.

DEFINE_ADJUST(adjust_sources,d)
{
    Thread *t; 
    cell_t c;    
   
    double  del_psi_delx, del_psi_dely;

    thread_loop_c(t,d)    
    {
    if (NULL != THREAD_STORAGE(t, SV_UDS_I) && 
        NULL != T_STORAGE_R_NV(t,SV_UDSI_G) ) 
        
        begin_c_loop(c,t)         
        {            
            del_psi_delx = C_UDSI_G(c,t,0)[0];
            del_psi_dely = C_UDSI_G(c,t,0)[1]; 

       }
}

See the line SV_UDS_I, SV_UDSI_G. This way you would be able use the UDSI and will not receive the segmentation fault or fetal error.

Best

[lbj007]

Originally Posted by Shoonya

DEFINE_ADJUST(adjust_sources,d)
{
    Thread *t; 
    cell_t c;    
   
    double  del_psi_delx, del_psi_dely;

    thread_loop_c(t,d)    
    {
    if (NULL != THREAD_STORAGE(t, SV_UDS_I) && 
        NULL != T_STORAGE_R_NV(t,SV_UDSI_G) ) 
        
        begin_c_loop(c,t)         
        {            
            del_psi_delx = C_UDSI_G(c,t,0)[0];
            del_psi_dely = C_UDSI_G(c,t,0)[1]; 

       }
}

See the line SV_UDS_I, SV_UDSI_G. This way you would be able use the UDSI and will not receive the segmentation fault or fetal error.

Best

Have you proven its effectiveness？cause I tried using "if (NNULLP(T_STORAGE_R_NV(t0, SV_UDSI_G(0))))" and it didn't work out, I think I don't need to use if (NULL != THREAD_STORAGE(t, SV_UDS_I)

[Shoonya]

Originally Posted by lbj007

Have you proven its effectiveness？cause I tried using "if (NNULLP(T_STORAGE_R_NV(t0, SV_UDSI_G(0))))" and it didn't work out.

Hi,
Yes, I have checked its effectiveness and in fact I am working on a problem where 3 UDSIs are present and I am using gradients of 2 UDSIs. Without that "if(NULL!=....)" statement my UDF was not working because no memory was allocated to them and when I call them in UDF (i.e., for example C_UDSI_G(c,t,0)[0]), it threw the segmentation fault. When I used that statement my UDF worked fine. Also I checked those gradients by storing them in UDMI's, I and found reasonable results.

Originally Posted by lbj007

I think I don't need to use if (NULL != THREAD_STORAGE(t, SV_UDS_I)

Ok, what I meant was to use the following (I am directly copying from the code I am using):

if (NULL != THREAD_STORAGE(t, SV_UDS_I(0)) && 
    NULL != T_STORAGE_R_NV(t,SV_UDSI_G(0)) && 
    NULL != THREAD_STORAGE(t, SV_UDS_I(1)) && 
    NULL != T_STORAGE_R_NV(t,SV_UDSI_G(1)) &&
    NULL != THREAD_STORAGE(t, SV_UDS_I(2)) && 
    NULL != T_STORAGE_R_NV(t,SV_UDSI_G(2)))

rest your wish, I just thought to respond to this thread, although I was not linked to this thread. And I only write on those problems in cfdonline forums in which I too have faced problem at certain stage.

[lbj007]

Originally Posted by Shoonya

Hi,
Yes, I have checked its effectiveness and in fact I am working on a problem where 3 UDSIs are present and I am using gradients of 2 UDSIs. Without that "if(NULL!=....)" statement my UDF was not working because no memory was allocated to them and when I call them in UDF (i.e., for example C_UDSI_G(c,t,0)[0]), it threw the segmentation fault. When I used that statement my UDF worked fine. Also I checked those gradients by storing them in UDMI's, I and found reasonable results.






Ok, what I meant was to use the following (I am directly copying from the code I am using):

if (NULL != THREAD_STORAGE(t, SV_UDS_I(0)) && 
    NULL != T_STORAGE_R_NV(t,SV_UDSI_G(0)) && 
    NULL != THREAD_STORAGE(t, SV_UDS_I(1)) && 
    NULL != T_STORAGE_R_NV(t,SV_UDSI_G(1)) &&
    NULL != THREAD_STORAGE(t, SV_UDS_I(2)) && 
    NULL != T_STORAGE_R_NV(t,SV_UDSI_G(2)))

rest your wish, I just thought to respond to this thread, although I was not linked to this thread. And I only write on those problems in cfdonline forums in which I too have faced problem at certain stage.

OK，thanks for sharing I'll try it after.

[lbj007]

Originally Posted by Shoonya

Hi,
Yes, I have checked its effectiveness and in fact I am working on a problem where 3 UDSIs are present and I am using gradients of 2 UDSIs. Without that "if(NULL!=....)" statement my UDF was not working because no memory was allocated to them and when I call them in UDF (i.e., for example C_UDSI_G(c,t,0)[0]), it threw the segmentation fault. When I used that statement my UDF worked fine. Also I checked those gradients by storing them in UDMI's, I and found reasonable results.






Ok, what I meant was to use the following (I am directly copying from the code I am using):

if (NULL != THREAD_STORAGE(t, SV_UDS_I(0)) && 
    NULL != T_STORAGE_R_NV(t,SV_UDSI_G(0)) && 
    NULL != THREAD_STORAGE(t, SV_UDS_I(1)) && 
    NULL != T_STORAGE_R_NV(t,SV_UDSI_G(1)) &&
    NULL != THREAD_STORAGE(t, SV_UDS_I(2)) && 
    NULL != T_STORAGE_R_NV(t,SV_UDSI_G(2)))

rest your wish, I just thought to respond to this thread, although I was not linked to this thread. And I only write on those problems in cfdonline forums in which I too have faced problem at certain stage.

Now it's weird,I can't even calculate after adding this, segmentation fault. here is my profile code,I only changed this when I calculate:

begin_f_loop(f,f_thread)
	{	
		c0=F_C0(f,f_thread);
	    t0=THREAD_T0(f_thread);
		c1=F_C1(f,f_thread);
		t1=THREAD_T1(f_thread);
		tem3=C_T(c0,t0);
		diff=33400*exp(-10300/tem3); 
		if (NULL != THREAD_STORAGE(t0, SV_UDS_I(0)) && NULL != T_STORAGE_R_NV(t0,SV_UDSI_G(0))) 
			NV_V(J0,=,C_UDSI_G(c0,t0,0));
			ja=diff*NV_MAG(J0);
		concen0=1000*C_R(c1,t1)*C_YI(c1,t1,0)/2;
		concen1=1000*C_R(c1,t1)*C_YI(c1,t1,1)/18;
		
		nesta=(8.31447*tem3/(2*96485))*log((concen1*11.555)/(concen0*0.3573778));  
		acta=8.31447*tem3/2/96485*ja/20000*11.555/concen0*0.3573/concen1;
		uright=1.11-nesta;
		C_UDMI(c0,t0,0)=ja;
		C_UDMI(c0,t0,1)=uright;
		C_UDMI(c0,t0,2)=nesta;
		C_UDMI(c1,t1,3)=ja/2/96485/(1e-4)*0.002;  
		C_UDMI(c0,t0,4)=acta;

[pakk]

You forgot to put brackets after your if statement. Indentation is only for the programmer, not relevant for the compiler.

[lbj007]

Originally Posted by pakk

You forgot to put brackets after your if statement. Indentation is only for the programmer, not relevant for the compiler.

so I just need to add brackets like this,no need to add else judgment statement

begin_f_loop(f,f_thread)
{	
	c0=F_C0(f,f_thread);
    t0=THREAD_T0(f_thread);
	c1=F_C1(f,f_thread);
	t1=THREAD_T1(f_thread);
	tem3=C_T(c0,t0);
	diff=33400*exp(-10300/tem3); 
	if (NULL != THREAD_STORAGE(t0, SV_UDS_I(0)) && NULL != T_STORAGE_R_NV(t0,SV_UDSI_G(0))) 
	{
		NV_V(J0,=,C_UDSI_G(c0,t0,0));
		ja=diff*NV_MAG(J0);
		concen0=1000*C_R(c1,t1)*C_YI(c1,t1,0)/2;
		concen1=1000*C_R(c1,t1)*C_YI(c1,t1,1)/18;
		
		nesta=(8.31447*tem3/(2*96485))*log((concen1*11.555)/(concen0*0.3573778));  
		acta=8.31447*tem3/2/96485*ja/20000*11.555/concen0*0.3573/concen1;
		uright=1.11-nesta;
		C_UDMI(c0,t0,0)=ja;
		C_UDMI(c0,t0,1)=uright;
		C_UDMI(c0,t0,2)=nesta;
		C_UDMI(c1,t1,3)=ja/2/96485/(1e-4)*0.002;  
		C_UDMI(c0,t0,4)=acta;

		F_PROFILE(f,f_thread,position)=C_UDMI(c0,t0,1);
	}
}
end_f_loop(f,f_thread)

[djendara]

you can prevent the solver from freeing up memory by issuing the text command solve/set/expert yes
you will find the result after computation