[srv537]

i am using ansys fluent in super cluster virgo but when i am using UDF for parabolic profile at inlet and create .cas and .dat file in my system and run in virgo then it shows some error.
i dont know what should i do please help me
thanks in advance

[AHF]

Originally Posted by srv537

i am using ansys fluent in super cluster virgo but when i am using UDF for parabolic profile at inlet and create .cas and .dat file in my system and run in virgo then it shows some error.
i dont know what should i do please help me
thanks in advance

well , whats that errors ?

[srv537]

Originally Posted by AHF

well , whats that errors ?

Reading "/IITM_GPFS_FS1/userfs1/ms/am15s014/s1/jet.dat"...
Parallel variables...
Error: chip-exec: function "inlet_x_velocity" not found.
Error: chip-exec: function "inlet_x_velocity" not found.
Error: chip-exec: function "inlet_x_velocity" not found.
Error: chip-exec: function "inlet_x_velocity" not found.
Error: chip-exec: function "inlet_x_velocity" not found.
Error: chip-exec: function "inlet_x_velocity" not found.
Done.

Warning: An error or interrupt occurred while reading the journal file.
Some commands may not have been completed.

> Halting due to end of file on input.

[sajjan]

Hello,

I am facing a similar issue. Do you remember how did you solve the problem?

[vinerm]

This error implies that Fluent has a UDF hooked for something, in this case a velocity profile at a boundary, but that UDF library is not loaded in Fluent. Once the library is loaded, the error will disappear.

[sajjan]

Originally Posted by vinerm

This error implies that Fluent has a UDF hooked for something, in this case a velocity profile at a boundary, but that UDF library is not loaded in Fluent. Once the library is loaded, the error will disappear.

Hi Vinerm,

Thank you for the reply. I receive similar message but for a different problem. But even if I get the message, my simulation runs. Let me explain you my problem in detail.

I have a UDF (attached here) which contains two macros: DEFINE_INIT and DEFINE_PROFILE. I use INIT to define the initial temperature distribution in the domain while PROFILE to define the temperature in the walls. I interpreted the UDF in "parallel fluent" with 12 cores on windows system and it works fine.
But when I save case and data file and transfer these to cluster and submit the job, it displays the following error. I modified my UDF code before transferring it to cluster and added “#if PARALLEL…#endif “, to make it compatible on the Linux system.

Parallel variables...
Error: chip-exec: function "temperature_profile" not found.
Error: chip-exec: function "temperature_profile" not found.
Error: chip-exec: function "temperature_profile" not found.
Error: chip-exec: function "temperature_profile" not found.
…………………………………

I wonder if my approach is correct and what could be the problem.

[vinerm]

First of all, few changes should be made in the code. if condition to check for the IDs of the zones should be outside the begin_c_loop so that loop is executed on for these three cell zones. Secondly, do not use F_PROFILE within if. Define a new C variable of type real and then calculate value within if. In the end, assign the value to F_PROFILE.

Use compilation instead of Interpreter. Interpretation or compilation needs to be done separately on Linux. So, either compile it before hand and transfer the library to the working directory or the first command in your journal file should be to compile or interpret the UDF before loading case file. That will ensure the availability of the UDF when the case file is loaded.

[sajjan]

Originally Posted by vinerm

First of all, few changes should be made in the code. if condition to check for the IDs of the zones should be outside the begin_c_loop so that loop is executed on for these three cell zones. Secondly, do not use F_PROFILE within if. Define a new C variable of type real and then calculate value within if. In the end, assign the value to F_PROFILE.

Use compilation instead of Interpreter. Interpretation or compilation needs to be done separately on Linux. So, either compile it before hand and transfer the library to the working directory or the first command in your journal file should be to compile or interpret the UDF before loading case file. That will ensure the availability of the UDF when the case file is loaded.

Thank you Vinerm. It worked.
Just in case someone is coming to this thread: Compilation of UDF can be done in your PC (I did in windows 10) . Once you compile and load your UDF, save your case and data file and transfer them to cluster.
For some reason, interpreted UDF does not work in cluster but only works in windows PC. So as Vinerm mentioned above, compilation is recommended over interpretation.
Regards,
Sajjan