A procedure involving high speed tine photography and novel optical probes has been used to study droplet interaction phenomena in liquid-liquid dispersions. Coalescence and breakup events were observed and the rate of coalescence was measured at various positions in a stirred tank for dispersions of methlyisobutyIketone in water. For the conditions studied, drop breakup occurred near the impeller and droplet coalescence predominated at other locations, as expected. However, the extent of this behavior was unexpected. Beyond distances from the impeller region of order of only l/6 the impeller diameter, breakup was virtually nonexistent. Outside the impeller region, extensive coalescence measurements showed (I) collisions between droplets are extremely inefficient for this chemically equilibrated system-at most 10% of collisions result in a coalescence, (2) only binary coalescence occurs even at the highest dispersed phase concentration investigated, (3) coalescence rate shows little preference on drop size, and (4) the coalescence rate is directly proportional to turbulence level; that is, the highest coalescence rates occur closest to the impeller. On the basis of these measurements, drop balance methods and a circulation path model were used to relate the drop size distribution at various locations in the region where coalescence predominates. In this case good agreement was obtained between measured and predicted drop size distributions.
Rational design of liquid-liquid contacting equipment requires not only a knowledge of dynamic equilibrium properties of the dispersion, such as drop size distribution and residence time, but also the dynamic rate characteristics of droplet coalescence and breakup rate. These rate parameters can have a profound effect on the concentration distribution of a third species in the dispersed phase. In turn, the rate of mass transfer between phases of this third component or its reaction within the dispersed phase are greatly affected (Curl[S], Rietema[31],. Furthermore, drop size distribution itself is a direct manifestation of these droplet interaction rates. Valentas and Amundson [42] and Valentas et al. [41] have presented a coherent formulation of the relationships between these interaction rates and equilibrium drop size.
Essentially all previous attempts to measure droplet interaction rates in agitated liquid-liquid dispersions have involved indirect methods [7,[16][17][18]23,24, 261. Shah et a[.[351 recently have reviewed these and other related studies. In these studies, interaction rates have been deduced from models which beforehand assume a mechnism of interaction. Although the type of experiment differed from investigation to investigation, in every case the model relating the interaction rate to the measured variable (typically interphase mass transfer rate or sauter mean drop diameter) assumed (1) homogeneity throughout the vessel, and in most cases that (2) drops break up immediately following coalescence and (3) monodispersed drops.
The applicability of the homogeneity assumption will depend on the conditions of the experiment. However, it will be most valid for vessels with short circulation times or for low coalescing systems (typically characterized by high interfacial tension and low dispersed phase fraction).