Confirmation of a simulation in a diffuser. RANS flow, sticks to the wall.
 

Hello everyone,

I made this with OpenFOAM's pisoFoam. It's a 2D flow in a 2theta=90° diffuser.

I saw this video https://www.youtube.com/watch?list=P...&v=nuQyKGuXJOs and in time 17:44, there's a 90° diffuser in which the fluid behaves as expected. Meaning central flow and two appropriate vortexes on each side. However, after a while, in my simulation the flow completely sticks to one wall, with one big vortex (backflow) on the other side. Much like it happened with a <90° diffuser in the film at 16:52.

Here's some screenshots (I thought the forum would give only thumbnails, so sorry for huge pictures):

This is the velocity field at the end (t=29s)
http://i.imgur.com/2iwJqql.png

This is the same velocity field ranged down to 0-6 m/s, also at (t=29s)
http://i.imgur.com/HsJWZL3.png

This is the gif I've made of the velocity field and velocity vectors (online video to gif site ate the quality, but the orientation of the vectors is still recognizable). The flow appears stationary, but then it sticks to the wall.
http://i.imgur.com/hP140fc.gif

---------------------------------------------------------------------------------

Please tell me your thoughts on this matter.

The initial case is here:
http://www.filedropper.com/diffuser-theta45

The inlet BC was changed from fixedGradient (0.1 0 0) to zeroGradient at t=13.8.