[LuckyTran]

So why do you have a free-slip wall? That's not a rectangular channel. A rectangular channel has four no-slip walls. Symmetry boundary conditions are not physical in LES because of the kinematic blocking that occurs.

Are you simulating a laminar or a turbulent boundary layer? If laminar then the laminar solver is better. If turbulent then you need to completely specify all the turbulence at your inlet (with both spatial and temporal fluctuations). You can use the spectral synthesizer or a synthetic eddy method to do this. If you do not, then you will have a laminar inflow and you will end up solving a laminar boundary layer even if your Reynolds number is high enough for it to become turbulent. You must give some perturbations at the inlet to have a turbulent flow. Ideally you would set these perturbations from experimental data or from doing LES of everything upstream of your channel. Otherwise, garbage in => garbage out.

[roi247]

It's an open channel which has the shape as a rectangular. Maybe I should call it rectangular open channel flow. That's why I use free-slip wall boundary condition.(It's a flume in the experiment)

I'm simulating a turbulent boundary layer along the channel. My Reynolds number is high enough to achieve turbulent flow. I thought it is unnecessary to add perturbations myself.

So I will try to use spectral synthesizer to generate the perturbations. The parameter I'll leave it as default to see what happens.

"Ideally you would set these perturbations from experimental data or from doing LES of everything upstream of your channel." It seems that I should set the parameters based on my experience or find more information from the paper.

Thank you so much!




(Lol (garbage in => garbage out) Maybe I don't understand myself, that's why fluent doesn't understand me. )

And you can delete the other same topic of mine if you have the authorization.

[roi247]

Originally Posted by LuckyTran

So why do you have a free-slip wall? That's not a rectangular channel. A rectangular channel has four no-slip walls. Symmetry boundary conditions are not physical in LES because of the kinematic blocking that occurs.

Are you simulating a laminar or a turbulent boundary layer? If laminar then the laminar solver is better. If turbulent then you need to completely specify all the turbulence at your inlet (with both spatial and temporal fluctuations). You can use the spectral synthesizer or a synthetic eddy method to do this. If you do not, then you will have a laminar inflow and you will end up solving a laminar boundary layer even if your Reynolds number is high enough for it to become turbulent. You must give some perturbations at the inlet to have a turbulent flow. Ideally you would set these perturbations from experimental data or from doing LES of everything upstream of your channel. Otherwise, garbage in => garbage out.

Hello Lucky Tran,

I took you advice. I add perturbations 10% turbulent intensity at the inlet. However, what I got is almost the same as that of no perturbation. Total time steps gives a simulation time 4.5 seconds, which is 3 times the water flowing through time. time step size is 0.003 s, which is smaller than the value of cell length dividing the velocity. As can be seen from the attachment, the velocity time series at a point stop changing after around 3 seconds

My inlet is UDF, outlet is pressure outlet, Will that wall shear have a chance in the end to be horizontal?

The wall shear keeps going down. In the experiment, wall shear is 0.6 is the result. The authors measured this value when they thought the flow is fully developed at the location 11.97 m to the inlet.