[ashtonJ]

Dear All,
I am stuck with the following error:
_________________________________
*** INSUFFICIENT MEMORY ALLOCATED *** |
| |
| ACTION REQUIRED : Increase the real stack memory size. |
| |
| Details : |
| Requested space : 201528 words |
| Current allocated space : 191808656 words |
| Current used space : 191757021 words |
| Current free space : 51635 words |
| Number of free areas : 1 |
+--------------------------------------------------------------------+


Details of error:-
----------------
Error detected by routine MAKDAT
CDANAM = IELG251 CDTYPE = REAL ISIZE = 201528
CRESLT = FULL
_________________________________

I've gone through all previous threads on this error and tried to increase the memory up to 8x, but still get the same error.
This cfx set-up used to work with no error; however, after activating transient statistics and defining few new variables, the above error showed up.
Here is a summary of newly defined variable:
New Variables:
ABSPGX: abs(Pressure.Gradient X)
ABSPGY: abs(Pressure.Gradient Y)
ABSPGZ: abs(Pressure.Gradient Y)
PG: sqrt(((Pressure.Gradient X)^2)+((Pressure.Gradient Y)^2)+((Pressure.Gradient Z)^2))

In "Transient Statistics" tab, I defined few cases to find the time-average of above variables over whole flow cycle as well as between specific times.

Would be too pleased if any body can help me.

Thanks
AshtonJ

[JuPa]

You've probably got some obsurdly long file names for your .def file. In execution control, under solver increase the solver memory allocation factor to 2.

Failing that just reduce your file names.

[ghorrocks]

Why do you think file names has something to do with it? It is the real stack which has run out of space and that has no link to the file name to my knowledge.

I have a hunch there may be a problem (ie a bug) in CFX with using Pressure.Gradient X/Y/Z and transient statistics. Can you try doing transient statistics on just Pressure? Try a work around by defining a variable like PGX = Pressure.Gradient X, then define a transient statistic of abs(PGX).

By the way: you have an error in your ABSPGZ function - you are using the Y gradient, not Z.

[ashtonJ]

Thanks Glenn and RicochetJ.
I figured it out, the problem was because of choosing the full list of transient statistics. I changed it to Arithmetic Average and it works now.

[JuPa]

Originally Posted by ghorrocks

Why do you think file names has something to do with it?

I had an error a year or two ago where my file names were absurdly long. I received a similar error to the above. I fixed it by increasing the mem allocation factor.

[ghorrocks]

Weird. But it should have been the string stack which ran out of room, isn't it?

[JuPa]

Originally Posted by ghorrocks

Weird. But it should have been the string stack which ran out of room, isn't it?

I haven't the foggiest. I just increased the memory allocation factor (at the time it made sense - two years later I don't know why). The mesh independent results I achieved were validated by experimental data so I was happy and didn't question anything!

Error out of sight is error out of mind!

[ashtonJ]

I've found out what causes "INSUFFICIENT MEMORY ALLOCATED" error and wanted to share it here:
This error might occur
1- if you define a lot of additional variables
2- if you want to save all transient statistics

I've had this problem for the last month and tried different ways to get rid of that error such as
1- Increasing the memory allocation factor up to 10x for “Partitioner”, “Solver” and “Interpolator”
2- Reducing the mesh from 1300,000 elements to 500, 000 elements

None of them worked

However, I fixed that problem by increasing the memory allocation factor within the CFX and also by ticking "Use Large Problem Partitioner" option and surprisingly the error is gone!

[MaBe]

Originally Posted by ashtonJ

However, I fixed that problem by increasing the memory allocation factor within the CFX and also by ticking "Use Large Problem Partitioner" option and surprisingly the error is gone!

Thxanks a lot for this advice.

For everyone: Try to changed the allocation factor in CFX-Pre instead of the CFX-Solver Manager!

This worked fine for me.

[kano22]

Originally Posted by MaBe

For everyone: Try to changed the allocation factor in CFX-Pre instead of the CFX-Solver Manager!

I can't find the Option in CFX-Pre where the allocation factor can be changed. And I can't do it in the Solver Manager, because I have to start the compution with a command in linux. So I need a .def-file with complete settings.
Can anyone tell me where to find the allocation factor settings in CFX-Pre?

Thanks

[kano22]

Originally Posted by ashtonJ

However, I fixed that problem by increasing the memory allocation factor within the CFX and also by ticking "Use Large Problem Partitioner" option and surprisingly the error is gone!

...and, what is meant by "within CFX"? The "Increasing the memory allocation factor up to 10x for “Partitioner”, “Solver” and “Interpolator” - None of them worked " refers as far as I understand to the Solver Manager, which is part of CFX. Or am I wrong?

Thanks

[Opaque]

cfx5solve -help

will provide you with the different options. For the ones that worked (assuming solver step failure) try

cfx5solve -large -s 2 -def ....

What is not clear from the original post is which component wrote the error? That is, what is the interpolator step, the partitioner step or the solver step. Each one of them have separate settings in the command line as well as in the user interface (CFX-Pre, or CFX-SolverManager)

[kano22]

Originally Posted by Opaque

Each one of them have separate settings in the command line as well as in the user interface (CFX-Pre, or CFX-SolverManager)

...so, where are those settings in the user interface of CFX-Pre? I'm not sure whether it's me not finding them, or it's my CFX Version which doesn't have these settings in the CFX-Pre user interface.

Thanks

[Lance]

Originally Posted by kano22

...so, where are those settings in the user interface of CFX-Pre? I'm not sure whether it's me not finding them, or it's my CFX Version which doesn't have these settings in the CFX-Pre user interface.

Thanks

Right click Simulation control/insert/execution control/solver tab.


But, you can also change it in the command line (see Opaque's suggestion on finding the syntax)

[kano22]

Thanks a lot, the hint helped very much.