CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Boundary conditions/solver (

chrisausAUS May 9, 2012 22:45

Boundary conditions/solver

I have to simulate a screw extruder.

I have set up the geometry - I have an inlet and an outlet needle.
I have the inlet as one part, the screw fluid as one part and the needle part.

I meshed within ANSYS workbench. Named the inlet and outlet as well as the screw wall.
I set up: Using laminar flow, defining a new fluid (PCL) with values I found in the literature.
For the cell zone condition I put fluid everywhere. For the screw fluid body I also tick frame rotation with rotation around - y axis with a certain speed.
Boundary conditions: Inlet-pressure 7e5 Pa, outlet-pressure 0 Pa gauge pressure, screw_wall is rotating around - y axis with the same speed as the frame rotation.

Solver is Presto and second order upwind momentum.

Monitors is the convergence turned off.

Then I initialize.
And then run until stable around 4000 iteration steps.

PROBLEMS that I have:
1) it is stable, but in the 10^3 range rather than 10^-4 range.
2) I tried standard/ second as well as standard/first, but the problems remains.
3) I didn't rotate the frame. And the same thing happens (stabilized at around 10^3)
4) If I make the whole extruder out of one part, then I get stabilizing in the 10^-4 to 10^-16 range for standard/second. If I use presto - I get still 10^3 and oscillating like crazy.

That brought me to the thought, that something must be wrong with my parts and their connection maybe.

Please give me some suggestions what I can try out next.


P.S.: I just realized that by setting the extruder up out of three parts, the coordinate system is not in the same position, so that the rotation is not around the rotation axis of the screw. Ho do I solve that?

narendra.gadwal May 10, 2012 00:01

Check the mesh quality, scale in Fluent.
Try to run in first order for some iterations.
Create a monitor point in the domain and plot the values of velocity at that location wrt no of iterations.

chrisausAUS May 10, 2012 00:27

Mesh Quality
It tells me that maximum cell skewness is greater than 0.98 - I refined the mesh, but it seems that this mistake does not disappear.

chrisausAUS May 10, 2012 00:30

First order
I tried running it in standard and first order (a friend told me till it stabilizes) - but unfortunately it doesn't stabilze in the minus range.

chrisausAUS May 10, 2012 00:31

Monitor point
Do you mean within FLUENT? Can you please explain a bit more - or send me a link whre this is explained better (I'm new to FLUENT)

chrisausAUS May 10, 2012 00:34

Rotation Origin

I created a new coordinate system within ANSYS Mesh and could read out the values to the global coordinate system.

Then I tried to change the coordinates within the rotation origin in:
1) both, the cell zone condition for the screw part AND the screw wall in the rotation in the boundary conditions
2) just in the cell zone
3) just in the boundary screw wall
4) making the frame stationary

All didn't converge.


t.krishnamohan May 10, 2012 00:58

Complete geometry should be a single part

narendra.gadwal May 10, 2012 01:38

Skewness is greater than 0.98 means your mesh is not of good quality.
Please look at the mesh you created and see that the quality is good.

narendra.gadwal May 10, 2012 01:38

if possible send me a schematic diagram

chrisausAUS May 10, 2012 01:58

Schematic setup
1 Attachment(s)
Here is the schematic set up.

I model the fluid, such that the screw part is hollow (from the screw inside)

I hope it is understandable - otherwise please ask again.


narendra.gadwal May 10, 2012 02:14

The geometry doesn't look that complicated. Do you have enough number of cells in the needle ?
The case also is like a basic laminar flow with known properties of the fluid. I don't think it should not create a convergence issue. Do you have ICEM to generate the mesh ?

chrisausAUS May 10, 2012 18:58

Number of cells in the needle: I did the face sizing option to mesh in the first place and then I used the refinemnt number 2 to refine the needle. (I thought that this would be good enough)

But even now - standard/second, presto/second, standard/first all go up after 18 iterations step to around 10^3.

What is ICEM? If you mean the ANSYS ICEM CFD something - I can't see in my list of the workbench.

narendra.gadwal May 10, 2012 23:14

Can you send me the geometry ?

All times are GMT -4. The time now is 21:24.