[Sherlock_1812]

Hi !

I would like to apply a varying temperature boundary condition at one of the boundary faces in my geometry by reading the temperature values from a data file (.txt) present in the case folder.

I read the data into a global variable (as an array), but am confused as to how I would assign each of those temperatures to the face centroids while looping over the faces of the boundary. (The current way assigned in the C file attached, obviously throws an error while interpreting- incorrect type conversion)

This is my first time writing a udf, so any help is welcome. pls find the C file attached.

[pakk]

I think you made a good start. Here is a hint for the rest:

DEFINE_PROFILE(dropSurfaceTemp,thread,position)
{
  real coord[ND_ND];	/* this will hold the position vector */
  real x;
  face_t f;

  begin_f_loop(f,thread)
  {
    F_CENTROID(coord,f,thread);
    x=coord[0];
      /*Here you need to find i such that x1[i] is close to x */
    F_PROFILE(f,thread,position)= Temp[i];   /*Notice the [i] added */
  }
  
  end_f_loop(f,thread)
}

How to find that i is not so trivial, otherwise I would have added it.

[Sherlock_1812]

How silly of me!

Thank you!

[Sherlock_1812]

Hi!

I am attempting to assign a temperature variation on a face using a polynomial, with coefficients defined in the UDF.

However I'm getting a strange error where the temperatures vary from 265-275 K on the face when the actual variation according to the polynomial should've been between 289-294 K. I don't seem to find a reason for the mismatch.

Could it be because of the data type used, or the extremely large coefficients? Should I increase precision for them?

Open to suggestions. Attached are the UDF and the plot.

[pakk]

A solution that might work:
change your evaluation scheme to something numerically more stable. What I mean in practice: replace

(p1*x1*x1*x1*x1) + (p2*x1*x1*x1) + (p3*x1*x1) + (p4*x1) + p5;

by the mathematically equivalent but numerically better

p5+(x1*(p4+x1*(p3+(x1*(p2+x1*p1)))));

A better solution: rethink your model. Your coefficients are far too large, it seems like you are overfitting. Why do you fit with a sixth-order polynomial? Because you have six data points? If you would use a linear fit, which characteristics do you lose? Or a piece-wise linear fit?

[Sherlock_1812]

So you think that it could be because of the extremely large polynomial coefficients? I'm afraid I can't use a piecewise linear or a linear fit. I shall try the 6th order polynomial with the expression suggested.

Thank you.

[pakk]

What do you mean, you can't use a piecewise linear fit? Computationally it is no problem, it will be easier to implement, and easier to understand. And the results will be virtually indistinguishable from the sixth order polynomial.

Where does your sixth order polynomial come from? If it is really from a fit, I would strongly suggest to use a piecewise linear fit. What makes you say you have to use the sixth order fit???

[nsaeidi]

Hello Srivaths!

Did you manage to make your code work?
I am trying to apply a UDF to the inlet boundary of my porous zone. My C file is at the end of this text.
After I compile the UDF and Hook the read_data function, this message shows up in Console: “temperature limited to 1.000000e+00 in 141630 cells on zone 4”
Zone 4 is my porous zone. My temperature limit setting is from 1K to 5000K.
I don’t face this problem when I set constant inlet BC or when I don’t have a read_data function in the UDF.
Do you have any idea what could be the problem?

Here is my Code:

/* ************************************************** ********* */
// UDF for time-varying temperature at the inlet
/* ************************************************** ********* */

#include "udf.h"


double Temp[15];
FILE *fp;

DEFINE_INIT(read_data, d)
{
int i;
fp=fopen("Env-Temp.txt","r");
for (i=1;i<15;i++)
{
fscanf(fp,"%ld", &Temp[i]);
}
fclose(fp);
}


DEFINE_PROFILE(Tinlet, thread, position)
{
face_t f;
real t = CURRENT_TIME*pow(1,-1);
int i = t;

begin_f_loop(f, thread)
{
F_PROFILE(f, thread, position) = Temp[i];
}
end_f_loop(f, thread)
}

Here is the text file that I read from:

273
278
283
288
293
298
293
288
283
278
273
268
263
258
253