The simple answer to this question is that the new mixer provides more intense mixing than the old one.  More specifically, the dispersion made by the new impeller contains smaller droplets that do not separate as quickly as larger ones created by the old mixer.  The new mixer is effectively creating an emulsion.

Reviewing information about glass-lined reactors that use the retreat-curve impellers (RCI) and a single top mounted baffle seem to show that the new mixer sounds like a good choice, but the process results are not what was expected.  The difference lies in a combination of factors.  The retreat-curve impeller on the old mixer was probably either 33” in diameter running at 84 rpm or 41” in diameter running at 68 rpm.  The new hydrofoil impeller is 29.5” in diameter running at 100 rpm.  All of those impeller options are running at about the same tip speed, which typically results in dispersions with similar drop sizes.  However, the off-bottom clearance of the retreat-curve impellers would have been only 7” to 9”.  The retreat-curve impeller is a radial-flow impeller, which sweeps the bottom of the tank.  The new hydrofoil impeller is an axial-flow impeller and was placed 33” off the bottom, which should also sweep across the bottom of the tank.

The problem with impeller type and location is that when the retreat-curve impellers are located near the bottom, they are also well below the bottom of the baffle and the flow is almost like solid-body rotation.  Because the liquid is rotating at almost the same velocity as the impeller, the relative velocity between the blade tips and the surrounding fluid do not cause the velocity gradients necessary for good dispersion.  By raising the hydrofoil impeller off the bottom, it is closer to the end of the baffle and the rotational velocity is less.  The result is higher velocity gradients in the trailing vortices behind the blades, where the drop dispersion takes place.

The simplest solution to the dispersion problem is to reduce the rotational speed of the mixer, which is probably most easily accomplished with a variable speed drive on the motor.  The ability to adjust the mixer speed would provide an ability to optimize the dispersion and separation processes.  The alternatives are placing the hydrofoil impeller closer to the bottom of the tank, well below the baffle or going back to a retreat-curve impeller, also lower in the tank.

Hopefully, this answers your question and provides some background to the problem.