Chairmen: L. Tadrist, Weiyang Fei


Yoshinori Kawase and Kazuhiro Shimizu
Biochemical Engineering Research Center
Department of Applied Chemistry, Toyo University
Kawagoe, Saitama 350, Japan

The shear stress in liquid-liquid dispersion was examined when the continuous phase is a non- Newtonian fluid. Although in the chemical and biochemical industries two-liquid phase dispersion is a process of considerable importance and non-Newtonian fluids are frequently encountered, information available on drop breakup in non-Newtonian liquids is greatly limited. Drop sizes and drop size distributions were measured in agitated non-Newtonian fluids using a 0.09 m diameter mechanically stirred tank. Agitation was provided by a six flat-blade Rushton- type turbine impellers. Palm oil was the dispersed phase and aqueous solutions of carboxymethyl cellulose and xanthan gum representing non-Newtonian flow behaviors were the continuous phase. The influence of non-Newtonian flow behaviors of the continuous phase on drop breakup and shear stress acting on the drop was discussed. Experiments were carried out at several impeller speeds. It was experimentally found that the non-Newtonian characteristics cause the increase in the maximum drop size particularly at low impeller speeds and lead to widening of the drop size distribution. The proportionality between the Sauter mean drop diameter and the maximum drop diameter was found to be independent of non-Newtonian flow characteristics.

The boundary layer model and the viscous force model for liquid-liquid dispersion in stirred tank reactors were developed for pseudoplastic non-Newtonian fluid systems. The boundary layer model is based on the mechanism in which drop break-up occurs in the laminar boundary layer on impeller surface. The viscous force model for very small drops assumes that the drop breakup occurs in the viscous shear region. The applicability of the models·including the Reynolds stress model was examined. The predictions of the models and the experimental results were compared. It was found that the experimental data correspond to the boundary layer model. The characteristic shear stress responsible for liquid-liquid dispersion in stirred tank reactors may be able to be estimated using the boundary layer model.


Leonid N. Braginsky and Yuri V. Kokotov
VISIMIX Ltd.., Park Center, Har Hotzvim, P.O. Box 45125, Jerusalem, 91450 Israel

The stable-state distribution of drop sizes, as well as mean diameter of droplets formed under effect of turbulent mixing, are a result of the equilibrium between two simultaneous processes - break-up and coalescence of drops.

There are two different approaches to the problem of coalescence. One of them, applied usually to relatively stable emulsions, consists in analyzing the properties of the boundary between the two phases based on DFLO theory. Another approach, typical for chemical engineering, is based on statistical modeling of collisions; the probability of junction of droplets is assumed to depend on mean parameters of turbulence, and properties of a continuous-phase liquid.

Experimental research on emulsifying in mixing vessels has shown that coalescence rate for droplets of a given size depends actually on both factors, and that their relative influence varies with the intensity of turbulence. On the other hand, the increase in viscosity results in the increase in the mean drop size, contrary to the predictions of theoretical models mentioned above.

This data can be explained if we consider an individual act of coalescence as occurring due to the effect of an instant random turbulent fluctuation of pressure that takes place in the following conditions: (1) the fluctuation occurs in the vicinity of two or more droplets during their contact, and (2) the amplitude of the fluctuation is high enough to overcome the repulsive pressure of double layers. A mathematical model developed by us on the basis of these assumptions and the description of break-up kinetics proves to be in good correspondence with experimental results of drop size measurements. Its predictions have been verified in a wide range of turbulence intensities in stable and non- stable-state conditions.

The results described above show that the size of droplets formed in various mixing vessels depends not on the surface tension only, but also on the value of the boundary repulsive pressure. The higher is the intensity of turbulence (or shear stress values) in emulsifying equipment, the higher is the repulsive pressure needed to prevent coalescence.

The mathematical models developed on the basis of these results proved to be applicable to describing the dependence of drop sizes on vessel and agitator geometry, properties of phases and duration of mixing in a wide range of experimental conditions. These mathematical models are used as a basis for Liquid-Liquid section of VisiMix - a new software for technical calculation of mixing processes.


Takumi Kinugasa*, Kunio Watanabe*, Tsuneo Sonobe*, and Hiroshi Takeuchi**
* Department of Industrial Chemistry, Niihama National College of Technology, Niihama, Japan
** Department of Chemical Engineering, Nagoya University, Nagoya, Japan

Liquid-liquid dispersions are frequently encountered in a number of industrial operations such as solvent extraction, direct contact heat transfer and heterogeneous chemical reactions. These dispersions are produced by mechanically agitating the two phases in a vessel in which the dispersion is to reside. In a stirred vessel the application of external energy subdivides one of the phases into droplets, the dispersed phase, while the other liquid forms the continuous phase. A point is eventually reached at which the addition of more dispersed phase causes inversion to take place: the continuous phase suddenly becomes dispersed and the original dispersed phase does continuous. Phase inversion has been studied in terms of the volume fraction of dispersed phase at inversion, viz., inversion holdup. In the present study, the effects of stirring speed and liquid properties have been investigated in a variety of types of oil-in-water (W/O) and water-in-oil (O/W) dispersions under stirred conditions.

A stirred tank consisting a glass vessel of 0.17 m I.D., fitted with four baffles, was used. Agitation was performed by a six-blade disk-turbine impeller. The vessel was immersed in a water bath controlled at 298 K. Two liquids, water and oil, were placed in the vessel and agitation was started by the impeller. At an appropriate stirrer speed, a suitable amount of the liquid to be dispersed was added into the dispersion after removing the same volume of the dispersion, to keep the total dispersion volume constant. Consequently, the volume fraction of the dispersed phase was periodically increased before phase inversion. In the present study, inversion point was determined by the visual change of the dispersion with the sudden alteration in the hydraulic characteristics.

Figure 1 shows the phase inversion behavior for kerosene-water system as a plot of the volume fraction of the oil phase, , at inversion against the stirring speed, n. This indicates that two curves divide the dispersion into three zones: oil continuous, water continuous, and ambivalent regions 1,2,3. In the oil and water continuous regions, the respective liquids can only exist as the continuous phase, whereas in the ambivalent region the dispersion type will become either W/O or O/W which depends on the methods by which the dispersion was formed. In Fig. 1, we also found that no effect of the dispersion volume on the inversion holdup.

Figure 1. Effect of total volume of dispersion on phase inversion for kerosene-water system.

Figure 2 shows the effect of viscosity of the oil phase containing liquid paraffin on the phase inversion point. On adding liquid paraffin into kerosene and cyclohexane, the points at W/O-to-O/W inversion became closer to each other; however, the inversion point for the carbon tetrachloride system deviates from those for the two systems. We cannot but conclude that it is impossible to explain the phase inversion behavior in terms of only viscosity effect. This seems to be influenced by other factors which have been considered in the present study.

Figure 2.Effect of viscosity of oil phase containing liquid paraffin on phase inversion for kerasene, cyclohexane, and carbon tetrachloride systems.

Figure 3 shows the inversion behavior for the systems of n-hexane and isobutyl alcohol. Both the density and the viscosity of n-heptane are smaller than those of kerosene. Selker and Sleicher 2found that as the viscosity of a phase increases, its tendency to be dispersed increases. In the case of the kerosene-water system, the holdup at O/W-to-W/O inversion was higher than that at the W/O-to-O/W; this result agrees with the previous reports. At inversion in the n-hexane-water system, however, a region where the water phase to be more viscous than n-hexane becomes dispersed is very narrow.

Figure 3.Behavior of phase inversion for n-hexane-water and isobuthyl alcohol-water systems.

For the isobutyl alcohol system, both inversion holdup data were similar to the kerosene system. Although the interfacial tension between isobutyl alcohol and water is significantly smaller than in the kerosene system, no effect of interfacial tension was found. Clarke and Sawistowski 3 stated that the drop size rather than the interfacial tension is of the primary factor in consideration of inversion. In this study, since the drop size was not measured, its effect could not be clear; however, we have found that for all the systems investigated, there is no correlation of the inversion holdup data with the Weber number, which is known to be well correlated with the drop diameter.

In conclusion, a lot of the findings were no better than fragmentary physical explanation for the phase inversion characteristics, and we were unable to find any clear trend.

  1. Watanabe, K. and Takeuchi, H., Kagaku Kogaku Ronbunshu, Vol.7, pp 538-540 (1981)
  2. Selker, A.H. and Sleicher, Jr., C.A., Can.J.Chem.Eng., Vol.43, pp 298-301 (1965)
  3. Clarke, S.I. and Sawistowski, H., Trans.Istn.Chem.Eng., Vol.56, pp 50-55 (1978)


Takaaki Mochizuki*, Yasuhiko H. Mori**
*Dept. of Technol. Education, Tokyo Gakugei University, Tokyo, Japan
**Dept. of Mech. Engng., Keio University, Yokohama, Japan

ABSTRACT. If a liquid drop carrying some net charge is suspended in another weakly conducting liquid, which fills the space confined by a pair of parallel-plate electrodes, the drop may be driven by a Coulomb force toward either of the electrodes, brought into contact with it, thereby exchanging the charge on the drop for a different one which is opposite in sign, and then driven back toward the other electrode. Once it is bounced back from either electrode in this way, the drop will fall into a continuous shuttle motion across the electrode spacing. If the electrodes are set along the axis of a contactor (a heat exchanger or an extractor), which may be vertical or tilted from the vertical by an acute angle, each drop released in the electrode spacing will have a translational velocity higher than its gravitational terminal velocity, depending on the Coulomb force superposed on the gravity. While a significant increase in the magnitude of the translational velocity is obtained in this way, its axial component can readily be lowered below the terminal velocity by increasing the tilt angle of the contactor axis. Thus, the device composed of tilted parallel-plate electrodes ensures a simultaneous increase both in the coefficient of heat or mass transfer to or from each drop and in the residence time of the drop in the contactor with a given axial length. The number density of drops in the contactor may be raised by arranging positive and negative electrodes alternately to form a lateral array of several (or even more) narrow-spaced drop passages.

This paper aims to estimate the performance of the contactor particularly in terms of the volumetric heat/mass transfer coefficient and the electric power consumption. The calculation scheme constituted for this purpose is outlined below. First, we specify the contactor-geometrical parameters (the electrode spacing, the axial length, the nozzle diameter and the lateral nozzle spacing), the operational conditions (the electric field strength and the flow rate of the continuous- phase liquid), and the dispersed-phase-side transfer efficiency to be achieved. The flow rate of the dispersed-phase liquid is tentatively specified such that we can estimate the diameter of drops to be released from each nozzle and the axial interval of those drops. The scheme of numerical calculation of the motion of single drops passing through an electrode spacing1 is then utilized to estimate the heat/mass transfer to or from the drops and the minimum axial length required to attain the specified transfer efficiency. Subsequently we utilize the drop-interaction calculation scheme2 to estimate the minima of the lateral spacing between two neighboring nozzles and of the axial drop interval, respectively, required to prevent the drops from mutual collision and coalescence during their axial passage through the electrode spacing. Since the mutual coalescence of drops may break an otherwise stable operation of the contactor, the lateral nozzle spacing and the axial drop interval specified before should be larger than the corresponding minima thus estimated. If they are not so, the lateral nozzle spacing and the flow rate of the dispersed-phase liquid must be increased and decreased, respectively, before performing the above calculations again. Through a repetition of these calculations, we can attain the possible maxima of the dispersed-phase holdup and the volumetric heat transfer coefficient, respectively. The electric power consumption is also evaluated by taking account of the charge conveyance by each drop through its repetitive passage across the electrode spacing. The results of the calculations indicate that leaning-tower-type contactors equipped with multi-plate-electrodes have a high potential for realizing high transfer efficiencies and their own compactness at the same time.


  1. Mochizuki, T., Mori, Y.H. and Kaji, N., Bouncing Motions of Liquid Drops between Tilted Parallel-Plate Electrodes, AIChE J., Vol.36, No.7, pp.1039-1045, 1990.
  2. T. Mochizuki, Mori, Y.H. and Kaji, N., Drop Interactions in Electric Fields Across Tilted Parallel-Plate Electrodes, JSME Int. J., Series B, Vol.36, No.4, pp.628-635, 1993.


Ege University, Chemical Engineering Department, Bornova, Izmir, Turkey

ABSTRACT. W/O high internal phase ratio emulsions (HIPRE) with three different dispersed phase volumes of 80, 88 and 94% were prepared with 1% sorbitol solution as the aqueous phase, light mineral oil as the oil phase and polyoxyethylene (2) oleylether as the surfactant. Complex modulus of viscoelasticity and shear viscosities of the emulsions were measured as a function of the rate of premixing, and relaxation time over a period of several hours.


High internal phase ratio emulsions (HIPRE) constitute a class of emulsions where the volume of the internal dispersed phase is greater than 74% corresponding to maximum compaction of spherical dispersed phase drops. External phase of HIPRE is still continuous but remains only as a film separating the dispersed phase drops which can now only exist in the form of tessellated polygonal cells. Relations given in the literature correlate the rheological properties with the phase ratio, average drop size, shear rate and the physical properties of interfacial tension and interactive forces between the surfactant molecules of the emulsions. This work addresses the determination of the rheological properties of (W/O) type of HIPRE to evaluate the applicability of these models.


W/O type emulsions were prepared with polyoxyethelene(2)oleylether with the compositions as given in Table 1.

Table 1
Compositions and Drop Shapes of the Emulsions

Dispersed phase (% volume) 80 88 94
Drop shape (Lissant) RDH RDH /(TKDH ) TKDH
% Paraffinic oil (volume) 10.00 5.55 2.94
% POE (volume) 10.00 5.55 2.94

The emulsions were premixed at 200 rpm and 1000 rpm prior to measurement of the rheological properties.

Measurement of Rheological Properties:

Apparent viscosities were measured in a rota-viscometer of Rheology International with ASTM spindles to determine the aging effects and index of thixotropy. Relations between shear stress and shear rate were determined using the DIN spindles, with the cylindrical sensor rotating in a concentric cylindrical cup. Modulus of viscoelasticity, relaxation time and stresses at the elastic limit were determined by the controlled strain method in Instron food testing apparatus.


The variations of the shear stresses with shear rate are given in Fig.2 for 80, 88 and 94% emulsions measured 10 minutes after being stirred at 200 or 1000 rpm. An initial film formation was observed in most of the cases. The scatter in the K and n values raises doubts on the similarity of structure of the reformed emulsion layer at the surface of the spindle with that of the rest of the emulsion in the bulk.

Table 1
Variation of Shear Stress with Shear Strain

Emulsion 80% 88% 94%
Relaxed = 0.1+26.90.14 = 9.2+27.10.35 = 15.7+37.10.15
Premixed at 200 rpm = 22.9+39.30.30 = 9.4+18.30.53 = 68.7+33.10.28
Premixed at 1000 rpm = 42.9+51.60.16 = 47.5+21.00.17 = 82.3+16.30.56

As the shear stresses in cylindrical DIN spindles were excessive, viscosities were also measured with ASTM spindles as a function of pre-stirring rate and time lapse after pre-stirring (aging). At 200 rpm prestirring rate, consistency and viscosity index show little variation if any, with respect to time and phase ratio, indicating the stability and homogeneity of the emulsion with respect to drop size distribution. Large variations are observed after 1000 rpm. stirring, corresponding to 3.80 kJ/m3 of emulsion. As this energy is applied locally by the impeller blade during the pre-stirring period, very small drops are formed around the impeller. Consistency index decreases drastically (by 50%) in 88% emulsions and to a lesser extent (25% decrease) with 80% emulsions within five hours. Whereas the consistency of 94% emulsions remain almost constant with time, viscosity index increases greatly (almost three times ), indicating a decrease in pseudoplasticity, or equivalently, strong thixotropy. Similar increases in n, though to a much lesser extent, are observed with 80% emulsions.

Various parameters associated with the viscoelastic behavior of emulsions are given in Table 2.

Table 2
Elastic Behavior of the Emulsions

%94 %88 %80
Rate of Prestirring 200 rpm. 1000 rpm. 200 rpm. 1000 rpm. 200 rpm. 1000 rpm.
Complex Modulus of Viscoelasticity G* (Pa.s) 3396 4480 2349 6471 - 3598
Shear Modulus of Elasticity E (Pa.s) 4270
s= 383
s = 443
s = 662
s= 1575
897 6697
s = 1042
Relaxation Time (s) 4.3 7.5 6.4 8.8 - -


According to the results obtained in this work, yield stress (and consistency, in the power law model) increases with phase ratio and the rate of prestirring (energy input to the emulsion). Data obtained is not sufficient to assess the trends in flow index, n. Viscoelasticity, and the related index of thixotropy bear no apparent relation with the phase ratio but increase with the rate of prestirring. It can be inferred that a reorientation in drop shape and size distribution does have an effect on the viscoelasticity and relaxation time of the emulsions.

Future Meetings | Past Meetings | Proceedings on Sale | Related Links

ICHMT World Wide Web Administrator