[krihamm]

Hi,

My solver is failing with return code 1, but without any error message. Exact same project that worked a few days ago doesn't work now. Tried duplicating and running a simulation that worked fine before, but still same problem. Does anyone know what the problem could be?

Thanks!

[ghorrocks]

If it gave you an error code it gave you an error message. The error code is not much use in CFX as there is no table of what those codes mean (in fact I don't know why they bother giving error codes....). You need the error text and the context which it occurred in to work it out.

[krihamm]

This is all the information I'm getting:

[ghorrocks]

Context is important. Please attach the entire output file.

[krihamm]

Originally Posted by ghorrocks

Context is important. Please attach the entire output file.

Output file is attached below:

[ghorrocks]

Now I can see the CCL is read correctly, the partitioner completes successfully and the run fails as it starts the solver.

I suggest you:
* Run this single processor and see if it runs
* If it does run single processor then check your parallel setup
* If it does not run check you have adequate memory and disk space available

[krihamm]

Originally Posted by ghorrocks

Now I can see the CCL is read correctly, the partitioner completes successfully and the run fails as it starts the solver.

I suggest you:
* Run this single processor and see if it runs
* If it does run single processor then check your parallel setup
* If it does not run check you have adequate memory and disk space available

Thank you for taking time to help me!

I tried running on single processor and this time the solver started, but crashed on the first iteration. This time I got a warning message saying insufficient memory allocated. So I guess we found the source of the problem (although strange since the exact same setup worked before and with 58 GB RAM available). After that I tried with "Large problem" checked and still single processor. Now it seems to work, but takes forever to iterate.

Also tried allocating more memory and with parallel setup, but with no luck.

I will try to decrease the number of elements in my mesh to make the problem smaller, but it is a problem that requires a very large fluid domain and also a very fine mesh around areas of special interest so this will likely decrease the accuracy of my simulations.

Once again thank you for your help!

[AtoHM]

In execution control, there should be a solver -> solver memory -> memory allocation factor. Try to increase it like 1.2, 1.4, 1.6 until it works.

[krihamm]

Originally Posted by AtoHM

In execution control, there should be a solver -> solver memory -> memory allocation factor. Try to increase it like 1.2, 1.4, 1.6 until it works.

Tried increasing it to 2, but still same problem. How much can you reasonably increase it?

[ghorrocks]

If the error is now about the integer stack size, this is because at the start of a run CFX estimates the memory it will use and allocates that space. If the estimate is very wrong it fills the space it will crash with a stack size error. Note that this can occur on quite small simulations as it is does not mean you have filled the physical memory - just the estimate CFX made at the start of the run.

So first try to increase the stack to 2 or maybe 3 times the original guess. There is no point going beyond 2 or 3 as if this does not work it means something is fundamentally wrong with your simulation and it is gobbling up all the space you allocate. This can be caused by excessively complex GGI interfaces, CCL which is too long, excessively complex CEL, too many boundary conditions and many other causes (but all to do with something being excessively complex)