Thank you for your feedback.
In truth, the fact that this "flooding" occurs is not a huge concern. The main issue with this model is that it is extremely sensitive to any small adjustments I make to it trying to implement your feedback. That was why I was questioning the recirculation zone in the first place.
I've been fiddling around with this model for a while now, testing to see if the mesh size/shape would affect it, whether or not I should use symmetry axes or expand the model to contain neighboring segments, as well as testing different under-relaxation factors. It seems like I rarely am able to replicate my previous results, even when I try to change the settings back to my previous ones. I also tried with your suggestion for mesh settings, but this does not seem to give a good convergence.
Do you have any idea or other suggestion on how to refine my model?