[Cooper24]

I am new to writing UDFs and new to coding itself in general. I have written a UDF for adding some source terms in my model. I use 10 cores on my PC for FLUENT in parallel. My UDF compiles and the simulation is also running.
Today, I learnt about something called parallelization of UDF. As I understood, the UDF should be parallelized for using multiple cores by adding some special commands in the UDF. I do not have any special commands in my UDF related to parallelization or number of cores to be used. Is it that my simulation is still running in serial even after selecting 10 cores on Fluent startup? As I said that my UDF compiles without error and the simulation is also running. Do I need to make changes in my UDF?

[pakk]

It depends on the UDF.

If you are doing something that involves the complete domain (integrating, finding a maximum, something like that), you need to change something. But in that case you would also get clearly wrong results if you used the serial code in parallel.

If your UDF is writing to a file, you also need some changes.

But if you just set viscosity as function of local pressure and temperature, you don't really need to change anything.

By the way: typically 99.9% of the time is spent on the "normal" Fluent code to calculate pressures, velocities and temperatures for the next iteration. This code is parallelized by the Ansys programmers, who did it quite well. The <0.1% of time spent in your UDF code is not significant. It might be suboptimal for speed, but it does not matter. If your simulation takes two days to finish, you don't care about the 30 seconds you could gain by improving your UDF. So as long as you don't make your UDF incredibly inefficient, you're fine.

[Cooper24]

Thanks for your reply. As you mentioned, I am not integrating any quantity or finding maxima-minima. I am just using the UDF to apply source terms and temperature dependent properties. So I hope I don't need to change anything.

[gauravkmr]

I agreed with pakk, and as far as your question, you didn't need to parallelize your code.

[Cooper24]

Thanks a lot for clarifying.

[AlexanderZ]

manual says

Some other types of operations that require parallelization of serial source code include the
following:
• Reading and Writing Files
• Global Reductions
• Global Sums
• Global Minimums and Maximums
• Global Logicals
• Certain Loops over Cells and Faces
• Displaying Messages on a Console
• Printing to a Host or Node Process

[Cooper24]

Thanks for your post. I am using loop over cells to add source terms but not sure whether I need parallelization.