I'm trying to write a UDF of a transient temperature BC to be applied on a wall. The data that I will use for this purpose is a thermo-couple data obtained from the process that I'm interested in. The data is 27 records of time and temperature. So how to write the correct UDF for this case? Can I use the code shown below but using my own values? then how to develop it to include all the 27 records of time and temperature? Can I just repeat the:
else if(flow_time
First of all you need to define a C function which would take time as an argument and return the temperature, something like
real mytemp(real mytime)
{
/* some kind of interpolation that takes your 27 points {ti, Ti}, i=1..27 */
return result;
}
This function should be at least continuous.
This function you will use in your DEFINE_PROFILE in the line
Assuming that the temperature will have the same time-dependent values everywhere on the boundary, your DEFINE_PROFILE simplifies to this:
DEFINE_PROFILE(transient_temp,th,i)
{
face_t f;
real flow_time = CURRENT_TIME;
begin_f_loop(f,th)
{
F_PROFILE(f,th,i) = mytemp(flow_time);
end_f_loop(f,th);
}
}
Now the most interesting question is how to implement the interpolation and which interpolation scheme to choose. Depending on your problem an the character of the temperature time-curve you choose your interpolation scheme. You can always start with the piecewise linear interpolation. You can also use approximation here, it all depends on your temperature-time data points.
There are two ways to implement this. Either you can program it yourself in C (google "pieceweise interpolation" for formulae) or use a library such as GNU Scientific Library (https://www.gnu.org/software/gsl/doc/html/interp.html?highlight=interpolation) or https://people.sc.fsu.edu/~jburkardt/c_src/pwl_interp_1d/pwl_interp_1d.html
I am sure there are other libraries available. This is a very often occuring task.
It depends on what is easier for you: to program your own code, or to get the UDF to call a library function and to compile with the external library.
I had good experiences calling GSL library subroutines from UDF. It is quite easy to do under Linux.
I hope this helps. Good luck!
@ Anton Ishmurzin
Hi Anton, thank you for answering my question. I'm an absolute beginner in coding with C++, so I'm looking for the easiest way to write my UDF. The UFD will be interpreted in ANSYS Fluent software. I did some search and I found a schematic diagram of the UDF for wall temperature for a case similar to mine, please see attached file. Now my question is do I still need to use the library that you mentioned in your answer? is there a simple way to code the function shown in the schematic diagram?
Thank you
The absolute minimum of programming would be to try to approximate your temperature-time curve using Excel or similar software and get an analytic formula that would give you the values "close" to your 27 points. Then you take the formula and program it into your UDF, it will be only one line of code. See the example in the attached screenshot. First you need to plot an XY-Plot with your data, then do "insert trend line", there will be an option (checkbox) to show the approximation formula.
@ Anton Ishmurzin YOU ARE MY HERO. This is really good information and good start to code my UDF. I have found so many examples on coding the transient temperature using formulas, so this should not be so hard especially after your explanation on how to obtain the formula out of my data.
I really can't find enough words to thank you for your effort on helping me with my question. Regards
- Badges
- Science topic
- Similar topics
- Correction