Hi you all! I tried to solve something in fluent and I received this error:

"Error: Out of memory in Resize_Vector_AMG_Chunk

Error Object: ()"

I continue iterating and I received another one:

" iter continuity x-velocity y-velocity z-velocity Cd-aripa Cd-placa_p time/iter

reversed flow in 255 faces on outflow 5.

reversed flow in 3783 faces on outflow 6.

reversed flow in 25847 faces on outflow 7.

Error: 1000000: Failed to allocate 31.87MB memory (..\..\src\amg.c:2779)

Error Object: ()"

Does anyone know what may be the causes for this error? I' m working on a machine that have: Intel(R) Xeon(TM) CPU 3.2 GHz, 4.00 GB of RAM. That 31,87MB it is variable: 8Mb, 24Mb.. etc but not more that 32 Mb and in the conditions that I never use more than 60 % of machine resource.

how big is your mesh? (number of cells)

I have approx. 2.3 millions cells.

try to coarse your mesh and retry. Or if you have the possibility, switch on parallel architecture

thanks, parallel architecture is not possible, but I ' ll try to modify the mesh.

[Ludo]

Originally Posted by costel
;154197

thanks, parallel architecture is not possible, but I ' ll try to modify the mesh.

Hello

Have you the solution for this error because I have the same problem

Thanks

[IvanCFD]

Hi all,

I've just come across with the same problem.

I thought my mesh wasn't very large - 1316055 cells.

The thing is I've run a steady simulation on this mesh and it's gone great.

Any solution?

Thanks a lot beforehand,

Ivan.

[Chris D]

If you change the solver settings, turbulence model, etc, then you change the amount of memory required for a given number of gridpoints. For example, if you run a case with the density-based explicit solver and the S-A (1 equation) turbulence model, and then run the same grid with the implicit solver and k-omega (2 equations), you'll need more memory for the second case, even though the grid is the same.

[IvanCFD]

Hi Chris,

thanks a lot for your explanation.

Essentially I want to run a laminar unsteady model. I've got the following settings for the solver:

- Discretization:

Pressure - Second Order
Momentum - Second Order Upwind

- Pressure-Velocity Coupling:

SIMPLE

I'd like to preserve this settings since my steady simulation with them is pretty close to experimental data.

Then, for the time marching I've got Second Order Implicit.

[Chris D]

It seems like you might not have enough memory to run that case, though. Maybe you could try running first-order accurate for your time marching scheme, but set a smaller timestep? That might reduce the amount of memory you need. How much memory does your computer have, and can you increase the size of your swap? One more thing: close all the other programs on your computer except for FLUENT. That should free up some memory, too. (If that's actually the problem.)

[IvanCFD]

I see what you mean...

Currently, with the Second Order Implicit I wanted to run with a time step of 1e-4 seconds. I thought that was not too big...

I'll try what you say.

Regarding my computer, it's a 32 bits Windows XP machine with an Intel Xeon and 4GB of RAM.

Would you please let me know how to increase the size of my swap? I don't even know what that is... Sorry.

Cheers,

Ivan.

[IvanCFD]

Hi Chris, I have already tried to switch it to 1st Order and changing the time step size from 1e-4 sec to 1e-5. Unfortunately, I've got an error again, this time different:

Error: 1000000: Failed to allocate 59.9MB memory (..\..\src\amgif.c:1068)

So sad...

[Chris D]

Explanation of swap.

Swap and windows.

Also, check out 6.6.2 Memory Usage in the FLUENT user's guide. That way you can make sure it's a memory problem. Did the memory that it failed to allocate decrease?