(533a) Batch Emulsification Using An Inline Rotor-Stator in a Recycle Loop of Varying Volume | AIChE

(533a) Batch Emulsification Using An Inline Rotor-Stator in a Recycle Loop of Varying Volume


Martin, P. - Presenter, The University of Manchester
Manning, J. - Presenter, The University of Manchester
Kowalski, A. J. - Presenter, Unilever R&D Port Sunlight

In industry emulsions are produced by recirculating the contents of a stirred tank through an inline mixer located in a recycle loop. The distribution of drop sizes in the stirred tank depends on the number of batch volumes that have been pumped around the loop. When scaling up pilot trials this value is kept constant. One factor that changes between these scales is the size of the recycle loop relative to the size of the tank. The effect of this factor is unknown since existing models neglect the volume of the recycle loop.

This study extends an existing model of Baker (1993) to include the effect of a finite residence time in the recycle loop. Larger loop volumes are shown to lead to narrower distributions within the stirred tank and more rapid reduction of the fraction that has not passed through the mixer. On scaling up to industrial scales the recycle loop normally becomes proportionally smaller. Consequently if the number of batch volumes is held constant the results will not be as good as the trials: the distribution will be wider and less material will have passed through the mixer at least once.

An experimental study was conducted to investigate these predictions. At small recycle loop volumes the results from the literature were accurately reproduced. At larger recycle loop volumes it was possible to detect characteristic features of this extended model. However the shortcomings of the available inline mixer limited the contrast between the existing model and the proposed extension.

A rotor-stator was used as the inline mixer. A new method of representing the dispersive process as a matrix transformation has been developed. This allowed determination of the daughter droplet distributions without a priori assumptions of their form. These have been shown to be broader than the distributions normally assumed in the literature.