CFD Online Discussion Forums - Periodically occuring reverse flow after residuals have converged

Greetings CFD users
I have found an interesting problem on a tutorial on electronics cooling by forced convection

https://confluence.cornell.edu/displ...lectronics+Box

As mentioned I have used velocity inlet and pressure outlet.

The graph of residuals for momentum and continuity will become a straight zero slope line.So naturally people will assume convergence and stop the iterations.
However after every 100 or so iterations after that,there will be continous messages of reversed flow at pressure outlet for the next 20 iterations
After this the number of reverse flow iterations steps will increase to the point that all steps show reverse flow.
and finally divergence also occurs.

This is true even if velocity at inlet is increased substansially.
Thus no matter how many times this process is done there will be this same pattern repeated.

In such a case can I trust the results have converged?If not what things do I need to change?
Also being a physics student could someone tell me why this "repeating"reverse flow occurs ? :)
Thank you