Neha B. Raikar1, Surita R. Bhatia2, Michael F. Malone3, Cristhian P. Almeida-Rivera4, Peter Bongers5, and Michael A. Henson1. (1) Department of Chemical Engineering, University of Massachusetts, 686 North Pleasant Street, Amherst, MA 01003, (2) Chemical Engineering, University of Massachusetts, Amherst, 686 North Pleasant Street, Amherst, MA 01003-9303, (3) College of Engineering, University of Massachusetts, Dean of Engineering, 125 Marston Hall, Amherst, MA 01003-9303, (4) Process Science, Unilever R&D Vlaardingen, Olivier van Noortlaan 120, POBox 114, Vlaardingen, 3130AC, Netherlands, (5) Process Science Deparment, Unilever R&D Vlaardingen, Olivier van Noortlaan 120, POBox 114, Vlaardingen, 3130AC, Netherlands
Previously we have used an inverse method to develop a population balance equation (PBE) model for drop breakage in a simulated oil-in-water emulsion (Raikar et al., 2007). A potential advantage of inverse methods is that the drop breakage functions are constructed non-parametrically without the need for mechanistic understanding of the breakage mechanisms. From an emulsion design perspective, the major shortcoming of inverse methods and many other PBE modeling techniques is that the drop breakage functions do not depend explicitly on formulation and processing variables. Such drop breakage kernels have limited predictive capabilities as the PBE model must be refit to each new set of experimental data in which formulation and/or processing variables were changed.
In this contribution, we demonstrate that drop distribution data from a high-pressure homogenizer can be used to identify relevant drop breakage mechanisms. Two breakage rate functions that depend explicitly on the emulsion properties (disperse phase volume fraction, density, and viscosity, interfacial tension, critical capillary number) and the homogenization pressure were considered. The first function was derived by other researchers under the assumption that drop breakage results from collisions with turbulent eddies, while we derived the second function assuming that breakage is due to turbulent shear. We show that neither breakage rate function nor the combination of the two functions provide satisfactory predictions when binary breakage is assumed and the daughter drop distribution function is a truncated normal distribution. Our experiments reveal that substantial breakage of the pre-emulsion occurs during the first homogenization pass even under zero pressure operation, suggesting a pressure independent breakage mechanism. The use of pressure dependent and pressure independent breakage rate functions along with judicious choice of the other breakage functions are shown to provide superior predictions than can be achieved with a single breakage rate function.