CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM (https://www.cfd-online.com/Forums/openfoam/)
-   -   PressureJump Errors simpleFoam for Ducted Fan (https://www.cfd-online.com/Forums/openfoam/220203-pressurejump-errors-simplefoam-ducted-fan.html)

vLhomme August 28, 2019 14:19

PressureJump Errors simpleFoam for Ducted Fan
 
Hello foamers !

All pictures and files can be found at :
https://www.dropbox.com/sh/qcgk5ajlp...q1t7sNF-a?dl=0

I'm trying to simulate the behaviour of a ducted fan(picture 1) by representing my propeller with a pressure Jump on a baffle in a 3D model of my duct. But I faced some complications (especially singularities in the pressure field on the rotor disk as shown in picture 2, I have similar singularities in the U field) and would be delighted if you would share experience or intuition on this case.
https://www.dropbox.com/s/hlx8lfbm5r...0Mesh.JPG?dl=0
picture 1 at (https://www.dropbox.com/s/hlx8lfbm5r...0Mesh.JPG?dl=0)
https://www.dropbox.com/s/g4jp3qe84l...rrors.JPG?dl=0
picture 2 at (https://www.dropbox.com/s/g4jp3qe84l...rrors.JPG?dl=0)

Introduction
I'm running on OF18.12, a steady state, incompressible, turbulent (using K-omega model) with simple Foam on a cartesianMesh generated by cfMesh (I also tried to apply pMesh and to use snappyHexMesh, more details below).

I'm running it on various amount of processors (usually between 70 and 96) using scotch decomposition.

Meshing
What I tried to do both with CfMesh and Snappy Hex Mesh was to mesh all my domain and duct and in that mesh create a baffle on which I impose a pressureJump. (I originally did it with cfMesh and we thought maybe it was the origin of the problem but I found the same thing with snappy).

The two cases with mesh parameters and output (especially checkMesh but I cannot provide the mesh nor the geometry) are available in the dropbox folder.

with CfMesh
I use this mesh dict https://www.dropbox.com/s/vwl5wkdabcbz15s/meshDict?dl=0 and after my mesh was created I used the createBaffles utility using a .stl file from Salome to represent my rotor disk and the following createBafflesDict https://www.dropbox.com/s/mtpa3z7ces...fflesDict?dl=0.
On the patches created with this utility I applied a pressure jump (or in the case in the dropbox a groovyBCJump to get a radially non uniform pressure jump, but I ran other case with cfMesh with fixed Jump and I got the same problems).

with SnappyHexMesh
I adapted the use case from https://www.erikrotteveel.com/tag/ducted-fan/: during the volume mesh I created a faceZone from a .stl file in the snappyHexMeshDict https://www.dropbox.com/s/es2ntrz42w...xMeshDict?dl=0 that I defined as a baffle, and from this face zone created patches to apply my cyclic pressure jump.

Simulation
For the two cases solver parameters and output are available in the dropbox folder.

Observation
Pressure Jump Errors
After running with simpleFoam I realised that on some points of the baffle the pressure jump was not properly implemented : to observe this I got pressure values on FanIn and FanOut from paraview, compare them and marked in picture 3 points for which the deltaP between the two patches was not equal to the pressure jump I imposed.
picture 3 at (https://www.dropbox.com/s/cxxqz6cma2...ition.JPG?dl=0)
Note: Pictures 2 and 3 are taken from the snappyHexMesh case but same irregularities were found with cfMesh (c.f. picture 4 and 5)

Faces Orientation
In another post I saw that wrong faces orientation could trigger these issues but it does not seem to be the case here (checked with paraview filter 'generate surface normals')

Questions
  • Is my method to represent the fan with a pressure Jump valid ?
  • Is my method of creating a baffle from only one surface should work ?
  • If yes two the two question above, have you any clue why I get these singularities ? And why the fixed jump condition isn't applied only for some isolated cells ?
  • Subsidiary : in SnappyHexMesh the baffle created cannot reach the duct's walls (cf Picture 6 https://www.dropbox.com/s/9pm1jzohdo...20SHM.JPG?dl=0) since the boundary layers are added after the faceset is defined, is there a way to counter this ?


All times are GMT -4. The time now is 22:46.