Learning Forum

[s.sethuraman]

Hi, 

I have a question regarding writing and implementing UDFs for Fluent in Python. I understand that it is only available in the Beta Feature in 2025R1. But I have a hard time implementing them. 

The Link (33.1. Fluent GPU Solver Beta Features) explains how to implement them and since they are in beta version, you can only use TUI command. I used this command (

(rpsetvar 'gpuapp/python-udf-files '("myudf.py"))

) with my UDF. I think the scheme command hooks my UDF, but the UDF is never implemented during iteration. With the above command, I don't encounter any errors either. So, I think there is no error in UDF.py. But Fluent silently ignores the UDF?

In addition, I observed that only a handful of functions are currently available in Python UDF (Table 33.5: Supported Hookup Functions). I cannot seem to find any relevant function which is equivalent to DEFINE_NET_REACTION_RATE (name, c, t, particle, pressure, temp, yi, rr, jac) in C UDFs. 

If anyone has experience in handling the above two issues, please let me know. 

Any suggestion is highly helpful. :)

Thank you!

[Rob]

As it's a beta feature there's a good chance it's not fully functional. Additionally, the gpu solver hasn't (yet) got all of the additional models available: so the chemistry UDFs may simply not be available (yet). 

What is your UDF trying to do? 

[s.sethuraman]

Hi, Rob. 

The main purpose of the UDF is to define net reaction rate for every species in the model. As premilinary test, I was trying a Python UDF to perform logging of velocity components, but it did not work as I expected. The model case is fairly simple - Laminar Combustion.

I guess I could perform the same in C. But I wanted to implement it in Python since my Custom Chemistry Calculation program is in Python. 

Thanks in advance!

[Rob]

I suspect that's not available in the gpu solver at present. The fact you needed a rpsetvar to switch something on generally means it's not quite ready. 

[Rob]

And for once the person I needed was in the same time zone and available. It's not quite there yet: keep watching the release notes as it's a rapidly evolving solver.