[skumar112]

Hello

I am trying to employ the inert heating law as a UDF In fluent however when I run it in parallel it diverges. This does not happen when run in series, in the fluent manual it states that no additional adjustments need to be made for DPM UDF's run in parallel. Any suggestions what may be going wrong here?

DEFINE_DPM_LAW(Heating,tp,ci)
{
/* get the cell and Thread that the particle is currently in */
cphase_state_t *c = &(tp->cphase[0]);
Thread *t = TP_CELL_THREAD(tp);
mp = TP_MASS(tp);
Dp = DPM_DIAM_FROM_VOL(mp / TP_RHO(tp));
Ap = DPM_AREA(Dp);
Tp = TP_T(tp);
dt = TP_DT(tp);
real Pr = c->sHeat * c->mu / c->tCond;
real Nu = 2.0 + 0.6 * sqrt(tp->Re) * pow(Pr, 1./3.);
real h = Nu * c->tCond / Dp;
TP_T(tp) = Tp + dt * h * Ap * (c->temp - Tp) / (mp * tp->Cp);
}

[pakk]

Originally Posted by skumar112

Hello

I am trying to employ the inert heating law as a UDF In fluent however when I run it in parallel it diverges. This does not happen when run in series, in the fluent manual it states that no additional adjustments need to be made for DPM UDF's run in parallel. Any suggestions what may be going wrong here?

No answer to your real question, but you should clean your code! Declare your variables (I am surprised you were able to compile this...), and don't introduce variables that you never use (such as t).

DEFINE_DPM_LAW(Heating,tp,ci)
{
/* get the cell and Thread that the particle is currently in */
   cphase_state_t *c = &(tp->cphase[0]);
/* the line above is not the normal way to find the cell! 
   instead use cell_t c = TP_CELL(tp);
*/
   real mp = TP_MASS(tp);                               
   real Dp = DPM_DIAM_FROM_VOL(mp / TP_RHO(tp));      
   real Ap = DPM_AREA(Dp);                                   
   real Tp = TP_T(tp);                                
   real dt = TP_DT(tp);
   real Pr = c->sHeat * c->mu / c->tCond;
/* all "c->" contructions are under-the-hood constructions, that might work now but fail later. Don't do this, just use the normal Fluent way:
   c->mu should be C_MU(c,t) for example.
  To do this, you should re-add the thread t that I removed. */

   real Nu = 2.0 + 0.6 * sqrt(tp->Re) * pow(Pr, 1./3.);   
   real h = Nu * c->tCond / Dp;                   
   TP_T(tp) = Tp + dt * h * Ap * (c->temp - Tp) / (mp * tp->Cp);
 /* tp->Cp is actually DPM_SPECIFIC_HEAT(tp,t)? */
}

[skumar112]

I had declared the variables globally outside of the function I have managed to fix this thank you for your help

[ahouse8]

Hi, I am having the same problem with a relatively simple UDF of mine. When I run in serial, it converges with no problem. In parallel, it diverges at random timesteps, but will always diverge.

What steps did you take to solve your problem?