[.Thomas.]

Dear All,

I have a problem when I impose a UDF flow profile in a rectangular channel.
I don't understand if it is a normal physical phenomenon or only my mistake but when the flow field changes from an high value to a low one (for example after 1 minute the flow rate moves from 500µl/min to 150µl/min) I can notice that there are a lot of faces in the outlet with reversed flow.
The number of reversed flow faces is decreasing every time step (∆t), for example in the first time step there were 600 faces with reversed flow and after 10 time step (in this case ∆t=0.001s) there are 285 faces with revered flow.
Let me clarify the problem, it is a transient and pressure-based problem, as boundary condition I imposed a mass flow rate in the inlet and a zero pressure in the outlet (because my outlet is opened to the atmosphere), as solution method I am using PISO, Least Squares Cell Based, Second Order, Second Order Upwind, First Order Implicit (in order).

Thank you so much for your kind answer.

Thomas

[CeesH]

Qualitatively, that doesn't sound too weird right? If you have a sudden drop in the inlet velocity, inertial effects will kick in - the fluid wants to keep it's velocity; but obviously that is not possible because the fluid density is constant. The inertial effects result in vortices (probably, you will see local pressure minima if you plot the pressure field), causing backflow. As you say, it's only a transient issue - the flowfield will recover and the reversed flow will go away.

[.Thomas.]

Thank you do much CeesH for your explanation, I understood what you said and it make sense.
I was trying to understand the phenomenon and I found out that using less time steps I don`t have this backflow. When I have the 500µl/min flow rate, I discovered that using only 25∆t (size ∆t=0.001s) I can follow the flow profile development. Than I choose 2 big time steps of t=10s, so the simulation achieve the time at which the flow field change to the low peak, without computational efforts wastes. To follow the fluid develment here I choose again a small ∆t of 0.001s. In the first ∆t (after the 2 big ∆t) I have some faces with reversed flow only in the first iterations, but before converging it recovers without showing backflow effects.
When I posted the problem I was using the same settings except the first part, instead of 25∆t of 0.001s I was using 100∆t with the same size.
Now I am really surprised that with this change in number of ∆t I have a big change in the solution. Is it a numerical problem?
Thank you so much again,

Thomas

[CeesH]

I wasn't expecting that. Did the flowfield reach the same steady state after the 2 big dt, compared to the 100 small dt? It could be a difference in starting conditions.