[iota]

Hello every body and good morning. Hope you all are doing well in this horrible pandemic situation.

I am recently writing a udf to calculate at first and then store in UDM (later) the area weighted average specie mass fraction gradient at the fixed boundary.I can do it very easily manually in the surface integral menu of the post processing.However i want to use it in udf for some other operation.This at first seem quite easy task as a lot of information is present in manual. I come up with the following UDF:

DEFINE_EXECUTE_AT_END(execute_at_end)
{
Domain *domain;
Thread *t;
cell_t c;
real area[ND_ND];
real total_area = 0.0;
real total_area_ave_grad = 0.0;
real G_mean;
int ID = 4; /* outlet ID displayed in Fluent boundary conditions panel */
domain = Get_Domain(1);
t = Lookup_Thread(domain, ID);
begin_c_loop(c, t)
{
F_AREA(area, c, t);
total_area += NV_MAG(area);
total_area_ave_grad += NV_MAG(area)*sqrt(C_YI_G(c,t,0)[0]*C_YI_G(c,t,0)[0] + C_YI_G(c,t,0)[1]*C_YI_G(c,t,0)[1]); ;
end_c_loop(c, t)
}
G_mean = total_area_ave_grad/total_area;/* Weighted average of the mass fraction gradeint */
printf("Area averaged gradient: %g\n", G_mean);
fflush(stdout);
}

The code is complied and hooked without any problem but after running the following error appear as attached.

What i have already done so far: (Not useful at all)
1. Run on serial node.
2. Include mem.h header file.
3. Use solve/set/expert and gradient save option.
4. Run for few iterations then hook the udf.
5. Tested this code for temperature. Which mean problem is with C_YI_G(c,t,0).

Please help me by suggesting necessary solution to this problem as it is frustrating me now. Thank you so much for your kindness.

Kind regards.

[pakk]

The manual says about that variable: "C_YI_G can be used only in the density-based solver. To use this in the pressure-based solver, you will need to set the rpvar 'species/save-gradients? to #t".

[iota]

Originally Posted by pakk

The manual says about that variable: "C_YI_G can be used only in the density-based solver. To use this in the pressure-based solver, you will need to set the rpvar 'species/save-gradients? to #t".

Thank you so much for your reply and nice suggestion. Yes, you are very correct and i am already following this step before hooking this UDF before running the simulation. That is the reason, i can get the mass fraction gradient in the derivative section of the surface integrals that i can calculate manually in post processing. However, this udf cannot still access those and give me this error. By the way i have also tried to use the Reconstructed gradient term as well.

Therefore, i do not know what is the real problem with memory access or may be i am missing something key one.

Thank you so much once again.

Kind regards,

[AlexanderZ]

F_AREA(area, c, t);

F_AREA can be used to return the real face area vector (or ‘face area normal’) of a given face f in aface thread t. See DEFINE_UDS_FLUX for an example UDF that utilizes F_AREA.


arguments are: F_AREA(A,f,t)
you are trying to put cell inside as an argument

so you need loop over faces not cells.
How you code should change i'm not sure. probably you should use gradient from adjusted to face cell using t0 c0 macros

btw it is possible to transfer information with data which you are able to get using TUI commands to UDF, ofc it has some limitations.
how are you going to use this integral further?

[iota]

Originally Posted by AlexanderZ

F_AREA(area, c, t);

F_AREA can be used to return the real face area vector (or ‘face area normal’) of a given face f in aface thread t. See DEFINE_UDS_FLUX for an example UDF that utilizes F_AREA.


arguments are: F_AREA(A,f,t)
you are trying to put cell inside as an argument

so you need loop over faces not cells.
How you code should change i'm not sure. probably you should use gradient from adjusted to face cell using t0 c0 macros

btw it is possible to transfer information with data which you are able to get using TUI commands to UDF, ofc it has some limitations.
how are you going to use this integral further?

Thank you so much Alexander for your kind suggestions. Infact your first hint about the error really works in a fraction of seconds and i modify my code accordingly and it really works. Thank you so much for such a nice complement.
Later on, I do realize the last part of your message which i did not consider previously. Actually i was planning to store this integral in a UDM and call it in other udf function. After your comment i revisit the udf manual and found that UDM exist for face, cell and node and not for any numerical output. I did not find much at present how to import data from TUI to another UDF. This put me in another problem.
I will be more thankful if you provide me another quick hint on this part. I will be very thankful for you in this regard.
I really appreciate your previous comment. It really helps. Thank you so much.

Kind regards,