Learning Forum

[h.a.maldonadodeleon]

Hi all,

I'm trying to set a UDF to set the bubble diameter for an Eulerian simulation while simulating the turbulence under the Dispersed model (as the density ratio is way beyond 1). While a similar UDF works perfectly for the Mixture model (for turbulence), it does not work with the modifications I've deemed necessary (i.e., pointing to the liquid - primary phase thread using the super_thread macro). Has anyone managed to define a similar UDF successfully? Any input is appreciated as the current formulation results in SIGSEV error. 

#include "udf.h"
#include "math.h"

DEFINE_PROPERTY(calderbank_db, c, t)

{
  /* Accessing the liquid domain */
  Thread liq = THREAD_SUPER_THREAD(0); /* Pointing to liquid phase */

  /* Define the variables */
  real db; /* Bubble diamter placeholder */

  /* Get local conditions from the dispersed (gas) phase */
  real eps = C_D(c, liq); /* Obtain turbulence dissipation from liquid phase */
  real eG = C_VOF(c, t); /* Volume fraction of the gas phase */

  /* Liquid properties */
  double density_liq = 998.2; /* Density of liquid phase (kg/m³) */
  /* Ensure epsilon is positive to avoid division errors */
  if (eps <= 0.0)
  {
    eps = 0.0000000001; /* Small value to prevent singularity */
  }
  if (eG > 0.3)
  {
    db = 0.0000001; /* almost zero db */
  }
  /* Calderbank's model (1958) for pure liquid */
  /* db = C1 + C2 * ((surf_tension / rhoL)^0.6 / eps^0.4) * eG^0.5 */
  db = 0.0009 + 4.15 * pow(0.072 / density_liq, 0.6) * 1 / (pow(eps, 0.4)) * pow(eG, 0.5);
  return db;
}

[Rob]

The primary phase doesn't have a diameter - that's the second phase(s). 

[h.a.maldonadodeleon]

Hi Rob, 

Thanks for your input. However, the model requires as input the turbulent energy dissipation rate (EDR) of the continuous phase (liquid), which is why I need to access that variable from the secondary phase. From what I got when reading the manual, once I hook the UDF to the diameter of the secondary phase (gas), the thread is pointed to the gas phase, and with the super_thread macro I should be able to reach the liquid one. Is this the case? 

[Rob]

The turbulence is typically set at the mixture level, unless you've been altering the options in the Viscous panel?  

[h.a.maldonadodeleon]

 

Hi Rob, 

As you mentioned, I’m making some changes to the viscous settings for the Turbulence Multiphase model (i.e., dispersed) due to the characteristics of my system. For now, the solution I found was the following – which seems to work for now: 

 

1. Defining the mixture and liquid threads
2. Get/Define the respective pointers to those threads

#include “udf.h”

#include “math.h”

Thread *mix; /* Define the mixture thread */

Thread *liq; /* Define the liquid thread */

DEFINE_PROPERTY(calderbank_db, c, t)

{

  /* Accessing the liquid domain */

  mix = THREAD_SUPER_THREAD(t); /* Pointing to mixture thread */

  liq = THREAD_SUB_THREAD(mix, 0); /* Pointing to liquid phase */

  /* Define the variables */

  real db; /* Bubble diamter placeholder */

  /* Get local conditions from the dispersed (gas) phase */

  real eps = C_D(c, liq); /* Obtain turbulence dissipation from liquid phase */

  real eG = C_VOF(c, t); /* Volume fraction of the gas phase */

  /* Liquid properties */

  double density_liq = 998.2; /* Density of liquid phase (kg/m³) */

 

  /* Ensure epsilon is positive to avoid division errors */

  if (eps <= 0.0)

  {

    eps = 0.0000000001; /* Small value to prevent singularity */

  }

 

  if (eG > 0.3)

  {

    db = 0.0000001; /* almost zero db */

  }

  /* Calderbank’s model (1958) for pure liquid */

  /* db = C1 + C2 * ((surf_tension / rhoL)^0.6 / eps^0.4) * eG^0.5 */

  db = 0.0009 + 4.15 * pow(0.072 / density_liq, 0.6) * 1 / (pow(eps, 0.4)) * pow(eG, 0.5);

  return db;

}