[jhz]

Hello,

I am using a DEFINE_ADJUST macro in a transient flow simulation. I am also using the documented thread/cell loops to access all cells, but I've found that for the first iteration of a timestep ONLY, there appear to be no cells inside the loop at all. I've included a very general, application-independent code snippet that can reproduce this phenomenon below.

When I print to console the counter number after the loop has ended, for the first iteration of the timestep it will always be 0. For every subsequent iteration in that same timestep until the solver converges, it will be a non-zero value appropriate to how many cells are actually inside the domain.

I am wondering whether there is an explanation for why this happens (possibly inherent to the way Fluent does its calculations)? If this is well-known, can someone point me to a specific part of the manual explaining this? Alternatively, if this is something that "just happens" then is there any way to circumvent this in implementing code that would calculate values in DEFINE_ADJUST which are passed to DEFINE_SOURCE? If for some reason I cannot loop over cells for the first iteration then that makes it seemingly impossible to calculate cell-dependent source terms.

Any help is appreciated, thanks!

#include "udf.h"

DEFINE_ADJUST(test_adjust,d)
{
    Thread *t;
    cell_t c;

    int counter = 0;

    thread_loop_c(t,d)
    {
        begin_c_loop(c,t)
            counter = counter + 1;
        end_c_loop(c,t)
    }

    Message("---------------------------------\n");
    Message("counter ended at %i\n", counter);
}

[AlexanderZ]

DEFINE_ADJUST is executed before each iteration
and your code is not an exception