[oliveira1820]

I am using the k-epsilon turbulence model in my simualtion for modelling the water behaviour inside a recipient. For the boundary lenght , density fluid value, dynamic viscosity and typical fluid velocity values of my simulation, the estimated wall distance for obtaining a y+ value of 30 is 0,002 m. The problem is that my recipient is only 0,004 m, long, therefore, the first mesh element will occupy half the recipient.

Is this right, or what can I do to oberride this and have a mesh which trully correpsonds to the k-e model?

[kyaha]

For k-e model, 30＜Y+, is ok

[Julian121]

Originally Posted by kyaha

For k-e model, 30＜Y+, is ok

k-e is a high Reynolds turbulence model which means that it does not have good prediction in a boundary layer. So, y+ should be larger than 11.

[ghorrocks]

Are you saying your pipe is only 4mm long? What is the diameter of the pipe? Are you modelling anything else upstream or downstream?

[ghorrocks]

Julian - your comment is a bit mixed up. k-e does not model the laminar sublayer well which is y+<11. This is why a standard wall function boundary condition should not be used under y+<11. But k-e models the loglayer pretty well and this is part of the boundary layer, so this part of the boundary layer is well captured most of the time.

So your comment should be k-e does not have good prediction in the laminar sublayer, so y+ should be >11.

[oliveira1820]

The "pipe" is conical and is only 4 cm long and has initial 2cm and final1 cm of radius. The cone is filled with water which is heated and water doesnt enter or exits, only flows inside the cone.

[ghorrocks]

What is driving the flow? Is it driven by convection from the heating?

[oliveira1820]

Yes, it os driven by convection from heating

[ghorrocks]

Have you calculated the Rayleigh Number of the flow? Is it laminar or turbulent? Steady or transient?