Dear All,

I am getting the message:-

"reversed flow in 102 faces on pressure-outlet 11"

What I need to do, in order avoid seeing this message?

Expecting your experienced replies.

Lopez

You can kill the message, as you command "/solve set flow-warnings? no", on NOT GUI but TUI mode.

Jinbok

can the reversed flow influence the result?

No, this reverse flow can not influence the result but it'll influence the convergence. Don't worry keep going.

bye

Alamgir

I would suggest that you first use a solver such as a 1st order momentum and energy, and then initialize. Solve and reach a convergence, and then "upgrade" to the 2nd order solutions. I find that this often removes the "reversed flow" phenomenon...which was I used to get, when I always tried to converge right from the start with 2nd order equations.

Dear All,

Thanx for the responses. Chris, can u please elaborate (step-by-step procedure)how to do what u have mentioned. Waiting for reply.

Lopez

First solve using 1st order momentum and/or energy until a convergence is reached (a).

Then change to a higher-order momentum and/or energy, and continue the iterations from (a) until the final convergence is reached.

Lopez,

This a good idea given by Chris, if u get a result with most economical way and then use that values to get precision results (ie u then have approx results, that results r initials for the next scheme).

Alamgir

[myty]

I got same problem but my solution is not converging, can anyone suggest something to improve. Thanks

[Ganapathy]

The cause of the "Reverse Flow" phenomenon is improper initialization. If possible initialize with the actual expected values of the flow in the direction of the flow. If there are different domains with different flows, initialize and patch these domains.
Reverse flow does cause changes in result. If you have a pipe with reverse flow at the outlet, the solution near the outlet is influenced by the BC at the outlet. For eg : If you set teh outlet temperature to 150°C, and if there is reverse flow, the outlet cells and nearby cells will be at 150°C even though the solution may be at 500°C

[flotus1]

The FAQ-Section for Ansys CFX applies to this problem aswell.

Ansys-FAQ

[Pat07]

Helo frnds, i try to create a circular domain around an airfoil, but it is not converging...the message says reversible flow in faces on pressure outlet can any1 help me plzz..thxx

[Ganapathy]

Can you share the gemoetry?

[farzadmech88]

I have the same problem in the centrifugal compressor,
1- I have reverse flow in both pressure-inlet and pressure-outlet
2- my solution is not converged.

if you know how can I fixed it, please help me.

[myty]

I would suggest you to vary your mesh, it could help you for convergence and get rid of reverse flow problem.

[Ganapathy]

Hi,
Try using mass flow inlet and outflow.
I know that mass flow inlet will be a wrong BC since you want to simulate the mass flow and not give it as a BC, but try with some guess values at mass flow inlet.
As the solution progresses, switch to Pressure inlet

[farzadmech88]

Thank you for your wise answer. When I change "mass flow inlet" into "pressure inlet"(after a relative convergence), is it necessary to change outflow to another BC?
Because I think that "pressure inlet" and "outflow" cannot be in a same geometry as BCs.

[Ganapathy]

Thank you for your wise answer. When I change "mass flow inlet" into "pressure inlet"(after a relative convergence), is it necessary to change outflow to another BC? Because I think that "pressure inlet" and "outflow" cannot be in a same geometry as BCs.

Yes please also change the outflow to a Pressure outlet.

[Agad15]

Ganapathy,

I am trying to simulate flow through a micro channel of 250 x 120 micron dimensions with unsteady vof model. I have given outlet BC as " pressure outlet" . Reverse flow is occurring after some time. How to avoid the problem?

[Ganapathy]

Has the solution converged ?

Usually this will disappear once the solution converges.

Is the pressure outlet boundary real ? is it really a pressure outlet in the actual physics too?