[Sabomb]

I get this velocity contour (attached) when I simulate with inviscid assumption. As you can see there are clear viscous effects visible. I also checked the total pressure at inlet & outlet. There is a definite dissipation of energy taking place.

What's happening here? Are viscous terms creeping in while discretization happens?

[LuckyTran]

All I see are velocity gradients, and this does not mean viscous effects. You can have velocity gradients in inviscid flows.

But yes you do lose energy due to discretization, so called numerical dissipation because you always truncate higher order terms.

[Sabomb]

Hi tran, the loss in total pressure is of the order of tens of kilo Pascals. Isn't that a little too steep for a discretization error.

Also, I dont face this problem if the channel was a simple straight path. But the moment a bend is introduced, viscous layers begin to appear.

[Sabomb]

Here is the case file, if you can take a peek it would be great!

[LuckyTran]

Stop making up names and calling them viscous layers, you have not proven there are viscous effects.

The flow must turn around the bend, but it cannot do an instantaneous 90 degree turn. Hence it separates from the surface. However, because of lack of viscosity the flow cannot mix. Hence any gradients that appears, stays.

I will not comment on the total pressure drop because it is too ambiguous and often times ppl mistakenly assume 1d flow theory and try to apply it to 3d. If you start with a uniform velocity profile and leave with a different velocity profile, "total pressure" can change depending on how you define "total pressure at the outlet". It boils down to what type of averaging do you use at the outlet?

What I'm trying to say is, don't completely freak out because the answer isn't what you expect.

[Sabomb]

Thanks for the insight on the flow bending and not being able to reattach due to absence of viscosity. Except that in this case it is not a 90 degree bend. This is an axi-symmetric case. The bottom line is the axis of the flow.

As for the total pressure I don't think there is any ambiguity involved. The facet average of inlet and outlet shows a massive difference!

[LuckyTran]

Flows can also separate in less than 90 degree bends. You're just arguing pedantics now.

There are tons of ambiguities. Facet averaging of total pressures is probably one of the most stupid ways to calculate an average total pressure because nothing is conserved, neither momentum nor energy, nor work rate. See for example: doi:10.1115/1.2098807

Again, based on just the velocity plot, I am not convinced there is really anything wrong with the result.

[ishan2033]

Hi Luckytran,
I am modelling A stilling Basin in Fluent but Velocity at outlet is coming out to be greater than the inlet and also velocity contour shows velocity at top portion of outlet but Streamlines are not present there. Please help.

Thank you
https://drive.google.com/open?id=0By...Dd6dmE1LWxHT0k