[I-mech]

Let me introduce what i want to analyse with Fluent by Workbench.
I'm trying to analyse the 3D flow between 2 surface.
The upper surface is a moving wall (translating on x axis), the lower one is a stationary wall and represents a pivoted body.
My scope is to find the body rotational equilibrium by changing the angular position of the body.
The image attached represent the problem scheme with boudary conditions applied.

So, what i desire to do is create a script that automatically does what follows:
1. assign a starting value to the angle
2. update geometry and mesh
3. update setup e solve
4. check results and change the angle if center of pressure doesn't coincide with the pivot (i accept a tolerance for numerical stability)
5. repeat 1-4 until the convergence criterium over attitude angle is reached.

If i manually change the angle value, refresh geometry, reset the setup and setup from zero the problem, initialise and solve, i have no problem.
When i try to do it, running a script, i have two different error message during the update of the system:
a) Warning: ST_Malloc: out of memory.malloc_storage: unable to malloc pressure redistribution.
Or in alternative:
b) Unable to allocate ... Mbyte of memory
To create the script I record journal in WB during the manual setup. I try to do it also by resetting and setting from zero but script doesn't work.

I don't think it's a problem of computer capabilities because manually it works really fine.
I hope someone can give me a tip to understand where is the problem so that i could find the solution.

Thanks to all.

Ps. I hope i have described the problem in the right way (I don't have the gift of being sintetic ), ask for more details if necessary.

[ghost82]

Hi Marco,
it seems your error is related to ram; maybe you can post your script to look for something which can cause this error.

[I-mech]

Originally Posted by ghost82

Hi Marco,
it seems your error is related to ram; maybe you can post your script to look for something which can cause this error.

Hi Daniele,
I really appreciate your reply.
I have attached a my recorded journal.
This script just want to change a parameter input (the angle) and update the solution).
It doesn't have an iterative algorithm because it has problems in update a value change, so i need to understand the problem before implementing iterations.

Running this script sometimes a third error compares, saying "Out of memory in Resize_Vector_AMG_Chunk"

I think the scope is too simple for Ansys to be impossible, but i don't understand the problem.
I think it's something related to update and initialization, because it works if I do what follows:
1. update geometry and mesh
2. reset the setup and solution
3. edit the setup (model, material, BCs, monitor, method and so on)
4. edit solution (initialization, checking, solve)

Although if i manually do a general update of the system, it fails.

[ghost82]

How many cells do you have after the script remesh and what are your system specifications (OS and ram)?
Are you using double precision solver?turbulence (what model)?first or second order?

My feeling is that the remesh creates too many cells to be handled for your system specifications.

[I-mech]

Originally Posted by ghost82

How many cells do you have after the script remesh and what are your system specifications (OS and ram)?
Are you using double precision solver?turbulence (what model)?first or second order?

My feeling is that the remesh creates too many cells to be handled for your system specifications.

I am using Ansys 14.5 on a Windows 8.1 x32 4Gb Ram (i can move to a x64 computer if necessary)
Number of cell created is 750000 (hexaedral), and this number is constant because in geometry i decide the number of division of the edge so i think that a geometry change means a volume change of the elements.
I use Double precision solver with laminar flow and Second order for Moment and Pressure too.
The domain has a lenght (x axis) of 50 mm and a width (z axis) of 40 mm.
The thikness is variabile as you have seen but it's in the microns order.

What i don't understand is why the same change and update works if done manually and fails if done by run a script.

Thanks for your interess.

Well, it really seems to be a Ram problem. A coarser mesh not present the error. I am a bit confuse because with a finer mesh the problem appears only sometimes.
We can say the problem it's solved but i think it's something i will explore more deeply in future.