CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Laminar separation bubble (

Axilleas Tsompanos August 3, 2004 07:49

Laminar separation bubble
Hello everyone,

Can Fluent capture the laminar separation bubble? I have tried but haven't made it.

Thanks a lot.

Jonas Larsson August 3, 2004 11:17

Re: Laminar separation bubble
If you mean the type of laminar separation bubble that undergoes transition and then re-attaches (common on for example turbine blades) the answer is probably no. Fluent does not include any turbulence models that are capable of modling this type of transition and re-attachement occuring after a laminar separation. When you turn on your turbulence model in Fluent you will most likely get turbulent bounday layers everywhere. Hence, you will not see any laminar separation bubbles. What you could do is to define a small laminar region with a UDF. But this requires that you know where transition/re-attachement occures approximately in order to explicitly turn on your turbulence model at that location.

This doesn't mean that Fluent is a bad CFD code - the vast majority of commercial codes are unable to predict this kind of tricky transitional flows.

Axilleas Tsompanos August 4, 2004 04:16

Re: Laminar separation bubble
Dear Mr. Larsson

Thanks a lot for your reply. It was very helpful indeed. Do you know of any turbulence models that can capture this transition?


Alex Chirokov August 4, 2004 14:29

Re: Laminar separation bubble
Turbulent models that are based on modified laminar N-S equations, work in time averaged sence, and can not capture transition. Only DNS can capture transition phenomena, or some particle based methods like SPH.

Dan Brzozowski August 6, 2004 00:35

Re: Laminar separation bubble
Turbulence models provide are closure for the Reynolds Averaged Navier Stokes (RANS) Equations. These equations are averaged *only* over the timescale of turbulent fluctuations, *not* all time. Using the unsteady solver, you can capture a great deal of unsteady phenomena --- that is, as long as the phenomena occur on a timescale much larger than that of the turbulent fluctuations.

Try using the "Enhanced Wall Functions", which should do a good job modelling laminar boundary layers. Also make sure you have a nice fine mesh near the wall.

Good Luck


Alex Chirokov August 6, 2004 12:24

Re: Laminar separation bubble
You are absolutely right it is well known closure problem. The bottom line is that you have to use something to get a closed system. Thus behavior of your system is usually greatly depend on what closing assumptions you used (read: depend on your turbulence model).

There is no question that you can capture unsteady phenomena with turbulent models but this unsteadiness is not related to transitions at all! Because you already assumed that the flow is turbulent when use introduced closing assumptions! You can use very elaborate closing assumptions that work in low Re # as well as in hi Re #, but you can't simulate the transition.

Again there is not way in fluent (or any other commercial CFD) to predict transitions from laminar to turbulent, and to simulate these effects.

Enhanced Wall Functions will help to simulate turbulent boundary layer (you will need very fine mesh for this, not transitions.

All times are GMT -4. The time now is 04:45.