Learning Forum

[AGAAMAZ]

[AGAAMAZ]

and it appears again in another modelingncan any one help me please in this nn

[Amine Ben Hadj Ali]

Means the Fluent Cortex Process has crashed and might be related to either communication or network issue.nnI recommend updating to a new supported version.n

[YasserSelima]

I am facing the same problem with Academic R2020 R2. nI am using CG_MOTION UDF and I am supposed to move few objects. The CG_MOTION function works to the last line and the last Message is printed. And then, it crashes and I receive this messagen

[Rob]

That could be the UDF, or could be the solver failing due to the network. What happens if you run on one core parallel? n

[YasserSelima]

Hi Rob,nThanks for replying. I found that when I define 14 global variables, it works as intended. However, once I define the 15th variables, I get this error. I changed the variable name many times and still the same issue. I even, used it with same name and removed one of the other 14 variables and the UDF worked fine. Now I am running the simulation with 14 global variables and I defined the 15th as local variable. It is working but I need to compile every time to change this variable the 15th.nI am just wondering is this a limitation in FLUENT, or I am doing something wrong? (I am doing this using an academic version)n

[Rob]

I'm not aware of any limitation, how are you defining the variables?n

[YasserSelima]

I define them as double at the top of the file. And I initialize them through DEFINE_INIT and through Excute_ON_DEMAND when required. I also tried to define them as static double but this made no difference.nI define them at the top because I have to keep track of them. Also I report them through DEFINE_REPORT_DEFINITION.nOne of my trials was to make all of them in one vector, but it did not make a difference.n

[Rob]

Please post the code. I (or any other Ansys staff member) won't be debugging the UDF but there may be something obvious or another contributor may spot something. n

[YasserSelima]

Hey Rob,nIt is 500 lines code. But here is the part of defining the variablesndouble velo_old_C, y_cg_C; ndouble velo_old_N , y_cg_N ; ndouble velo_old_S , y_cg_S ; ndouble velo_old_NW , y_cg_NW ; ndouble velo_old_NE , y_cg_NE ; ndouble velo_old_SW , y_cg_SW ; ndouble velo_old_SE , y_cg_SE ; ndouble my_tube_mass, my_tube_stiff, my_tube_length, my_tube_damping;nI have 7 DEFINE_CG_MOTION functions and each one uses 2 of the variables defined in lines 1 to 7. The last 4 variables, defined in the 8th line, are being calculated once at the initialization and all the functions use them. When running the simulation, I reach the end of one or two of the CG_MOTION functions and the message is printed correctly. Then, I get the error of fl process could not be restarted.nWhen I calculate the last 4 variables externally and define them locally inside the 7 functions, with their calculated values, and comment the 8th global variable definition line, everything works as intended.nThank you for attention!

[Adam Anderson]

If you are using global variables in a file like this, it's best to make them staticstatic double my_tube_mass, my_tube_stiff, my_tube_length, my_tube_damping;nBut that all looks fine otherwise. There must be something else happening with your my_tube_ variables that's causing the issue.nBe careful about parallel UDFs, treat DEFINE_CG_MOTION UDFs as if they are only run on the compute nodes.nAlso putting these lines at the top of each DEFINE_CG_MOTION UDF is advised:nif (!Data_Valid_P())n {n   return;n }nnn

[YasserSelima]

Using static does not change the error. I will try to use if (!Data_Valid_P ()) .. hopefully it works. Thank you very much.n

[YasserSelima]

Surprisingly, the problem is solved and I don't know how. Now I am saving more than 14 variables and the fl process does not stop. This works on the Academic Research License as well as the student license. So, there is no limitation as I though.n