reverse flow
 

can any one explain me about the reverse flow of the fluid in a channel flow???

Yeah same question

In the theory of fluid dynamics there is no phenomenon that has been identified as reverse flow.
It is mainly an industrial and economical term (Google for "reverse flow" and you will find many applications where the flow reverses direction, reverse flow heat exchanger, reverse flow combustors, reverse flow check valves, .....etc
If you check this same CFD forum, you will find posts about outlet boundary condition where some velocity vectors are not leaving the domain of flow but entering, the reasons are explained in these posts.
But in the theory of fluid dynamics we have secondary flows and vortices,so check this movie (http://web.mit.edu/hml/ncfmf.html ) scroll down the list till you find it, and read the film notes, enjoy your time.

how can I jude whether it will produce the reverse flow?

that's because of Control Volume..

In real, The control volume is the earth..
but in simulation, specific volume should be selected..
So, reverse flow can be occured the end of the control volume, if there's exist recirculation near the end.

If you want to use the outflow b.c. you have to ensure that it fits to what real happens. That means:
- there should not recirculation at this point. You may check it if you double the channel length and analyse the place where you intend to set the outflow
- the zero gradient condition needs to be fulfilled. Violating this is not as problematic as recirculation, but still physical not correct. You may check this the same way.

In your case the flow is gravity driven and it may be taht you never reach a point where you may set an outflow b.c.
I see two ways around it
- use the outflow but make the channel much longer than needed. Ignore results calculated for the region near the outflow
- use another b.c. Fixed velocity comes in mind. This would mean that you simulate a physical problem with a kind of sieve or porous media at the outflow. It is not your problem, but it is physical correct. You still may make the channel longer in the simulation.

Or for short: Outflow b.c is often problematic :rolleyes: