[swapbitra]

Hi all,

I am doing a simple simulation of a Taylor Couette Flow. I want to vary the rpm of the inner cylinder wall as a function of the flow time. I have used a simple UDF

# include "udf.h"
DEFINE_PROFILE(unsteady_angularvelocity, thread, position)
{
face_t f;
real t = CURRENT_TIME;
begin_f_loop(f, thread)
{
F_PROFILE(f, thread, position) = 600 + 60 * t;
}
end_f_loop(f, thread)

}
The udf is interpreted but the results are erroneous and also I want to print the current rpm in the console window. Please anybody help me to properly formulate the udf and achieve desired results.....

Regards

[pakk]

Originally Posted by swapbitra

The udf is interpreted but the results are erroneous

You probably mean that the UDF gives results, but these results are not what you expected. If that is the case, then you should tell us what you expected, and what you received. I can not change the UDF to your expectations, if I don't know what your expectations are.

and also I want to print the current rpm in the console window.

Add

Message("time %f: angular velocity %f\n",t,600 + 60 * t);

at the end of your code.

[swapbitra]

Dear Pakk

At a flow time of 0.1 sec, as per udf
angular velocity should be 600 + 60 * 0.1 = 606 rpm, corresponding to this rpm

Wall velocity should be 2 * pi * 606/60 *radius of inner cylinder(in my case 0.0595 m) = 3.775 m s^-1

But the fluent with udf gives 3.61 e+01 m s^-1. Please see the picture attached

and also
Code : Message("time %f: angular velocity %f\n",t,600 + 60 * t); gives an error message of "Invalid : -1"

Without udf at individual 606 rpm, the results are correct.

Regards

[obscureed]

Hi Swapbitra,

Fluent requires UDF to supply profile values in its default units -- in this case, [rad/s]. It does not matter whether you have changed the GUI panel to use [rpm]. This exactly explains the discrepancy.

I cannot see anything wrong with Pakk's Message code, unless there is some quirk of interpreted UDFs. (I would recommend using compiled UDFs unless you are forced to use interpreted.) For reasons of speed and maintainability, you should evaluate and type the formula only once, for example:

# include "udf.h"
DEFINE_PROFILE(unsteady_angularvelocity, thread, position)
{
  face_t f;
  real rate; /* in [rad/s] */
  rate = (600. + 60.*CURRENT_TIME) * M_PI / 30.;
  /* "*2*M_PI" to convert revs to rad, then "/60" to convert /min to /s */
  begin_f_loop(f, thread)
  {
    F_PROFILE(f, thread, position) = rate;
  }
  end_f_loop(f, thread)

  Message0("time %g s: angular velocity %g rad/s\n",CURRENT_TIME,rate);
}