[vbriodeau]

Hello everyone,

I am new to this forum. I read the guide to ask questions and the FAQ.

The issue is very simple. I have already been interpreting some UDFs, yet I now try to use User defined scalars (UDS). I wrote the following (useless) UDF and could interpret it, yet, when the calculation launches, it shows the following message:
chip-exec: energy_source: argument 3: incorrect type (32): int expected

The solver still continues.

I defined a scalar in the scalar panel as well as in the materials>source term. In the UDF, I put a 0 as the index of the scalar (I have only one), or I defined int i=0 and later UDSI(c,t,i). I don't understand why such a message appears. I got the error message in both cases

This has similarities to the following topics:
https://www.cfd-online.com/Forums/fluent-udf/67364-udf-parallel-error-chip-exec-function-not-found.html
Error: chip-exec: function "UDF" not found.

I tried once to compile it, and got an error message, yet I did not persevere in this way.

I did not introduce a memory
code for product of rho and velocity
help needed for accessing previous step values
fluentError: received a fatal signal (Segmentation fault).

#include "udf.h" /* must be at the beginning of every UDF you write */
DEFINE_SOURCE(energy_source, c, t, dS, eqn) 
{
real source; /* local variable */
/* Loop over cells in a thread to get information stored in cells. */
  begin_c_loop(c, t)
    {
     /* C_T gets cell temperature. The += will cause all of the cell 
        temperatures to be added together. */
        C_UDSI(c,t,0) +=C_T(c, t)*0,01;
    }
  end_c_loop(c, t)
return source;
}

Does anyone has an idea where it could come from?

Regards

[obscureed]

Hi Vasili,

I cannot explain that error message. (You have not defined a global variable called "t", have you?)

There are some things to look at in your UDF:
(1) I would try to set up the compiler if at all possible, rather than using the interpreter. As one benefit, you might get better error messages.
(2) Currently, source is never given a value (and C does not initialize to zero by default), which will cause trouble when you actually run.
(3) "0,01" is not a valid number -- at least, not in my locale. Your compiler/interpreter might be different, but I doubt it. Try "0.01".
(4) I understand that this is only a test UDF, but it is starting on the wrong track. DEFINE_SOURCE is called for every cell (which is why you are given c and t). So, if you needed to do something for each cell's value of C_UDSI, you would only need to do it once per visit to DEFINE_SOURCE.

To test the compiler, I would start with a really minimal UDF:

#include "udf.h"
DEFINE_SOURCE(zero_source,c,t,dS,eqn)
{
  dS[eqn] = 0.0;
  return 0.0;
}

Good luck!
Ed

[pakk]

I don't know if it causes the error, but what I do see is that you are using the loop wrong.
You should use DEFINE_SOURCE to tell Fluent what is the source in cell c on thread t. But what you do is tell Fluent what is the source on all cells of thread t by looping over each cell c in that thread... Effectively, you are trying to use variable c twice.

Your UDF should be much simpler:

#include "udf.h" /* must be at the beginning of every UDF you write */
DEFINE_SOURCE(energy_source, c, t, dS, eqn) 
{
   real source=0; /* local variable */
   /* C_T gets cell temperature. The += will cause all of the cell 
        temperatures to be added together. */
   C_UDSI(c,t,0) +=C_T(c, t)*0,01;
   return source;
}

By the way: I don't know if you are allowed to write to a UDS like this. I don't know your ultimate goal, but if you just want to store something, a user defined memory (UDM) is better suited.

[vbriodeau]

Hello pakk and obscureed,

thanks for your replies. I tried to compile and it worked (last time I tried I had a wrong header file). Yet, as i now try to interprete, it also works; I don't know what happened before.

The function you both submitted and mine also worked:
- 0,01 does not seem to be a problem, I already tried so.
- Not putting source to 0 does not seem to be a problem.

Your remarks about the use of the function seem to make sense, although I did not completely understood the explanation.
I built this tricky function because I want to sum all the temperatures over all the cells. This is quite useless, but it is just to train. At the end, this should be the sum of the absoprtion of all the cells to get a mass flow
I started with a source function because it seemed easier. That is alo why I defined a source term, which should be the sum of all the temperatures (yet I forgot the line C_UDSI=source).

A memory might indeed be used, but the code which inspires me does not use it, so I think I don't (C_UDSI_M1 would suffice to call the previous value).

hence the remark

You should use DEFINE_SOURCE to tell Fluent what is the source in cell c on thread t. But what you do is tell Fluent what is the source on all cells of thread t by looping over each cell c in that thread... Effectively, you are trying to use variable c twice.

Did I though right? namely, I would like to understand the logic; I already read the manuel about this.

[AlexanderZ]

As Pakk mentioned
for macros DEFINE_SOURCE you do not need to organize loop over cell/threads because this macros already has cell/threads as arguments

DEFINE_SOURCE(energy_source, c, t, dS, eqn)

on the other hand, fluent has other macros, where you may want to use cell/threads loops (DEFINE_ADJUST,DEFINE_PROFILE and so on)

so you will apply

 begin_c_loop(c, t)
    {
     /* code here */
    }
  end_c_loop(c, t)

for cell loop

best regards

[vbriodeau]

Thank you both for your support,

I understood now. i took this idea from a DEFINE_ADJUST macro, so it does not make sense.

I put the topic as solved.

Best regards and thank you very much