Search

A drop size segregation technique leads to a direct measurement of the interdrop coalescence frequencies of equisized droplet swarms in liquid-liquid fluidized beds. Mass transfer between the phases and the use of a tracer dye are avoided. The study includes the effects of highly viscous continuous as well as dispersed phases. Binary liquid solutions of ethylene glycol and distilled water provided aqueous phases of increasing viscosity, and toluene was the second liquid phase. Twenty two liquid-liquid systems including other dispersed liquids, namely, benzene, ethyl acetate, cyclohexane, and carbon tetrachloride, were studied by this technique. In the case of toluene continuous, the drops of water dispersed in toluene were large in size, and they coalesced at unusually high rates up to about 100 times faster than those of toluene dispersed in water. Similarly, anhydrous ethylene glycol dispersed in toluene showed somewhat high coalescence frequencies. All the data excluding these two anomalous systems were analyzed by a stepwise regression technique to yield the best fitting correlation in terms of the most significant dimensionless ratios. For the present, the proposed empirical correlation for the dimensionless parameter [ω D × 105/Vs] containing the coalescence frequency (ω, time−1) as a function of the coalescence number (σ/μcVs) and the viscosity ratio (μd/μc) is as follows: .

Mixing effects for the homogeneous, liquid-phase, second-order, competitive-consecutive reaction of iodine (B) with L-tyrosine (A) to form 3-iodo-L-tyrosine (R) and 3,5-diiodo-L-tyrosine (S) were determined for the following conditions: vessel volume, 5 and 36 liters (baffled and unbaffled); turbine diameter (2 to 6 in.) and speed (95 to 1,600 r.p.m.); feed inlet locations (3), addition rate (0.25 to 18 min.), and distribution; temperature (11° to 43°C.); initial A concentration (0.1 to 0.4 g.-mole/liter); and kinematic viscosity (0.765 to 6.35 centistokes). A was initially charged to the reactor and an equimolar quantity of feed B was added over a time period. Yields of R are less than that expected for perfect mixing owing to local regions of excess B concentration that exist for time periods during which R over-reacts to S. Agitation power for a given yield is less in unbaffled vessels without an air-liquid interface than for baffled vessels. The local fluctuating velocity u′ where feed is introduced correlates the mixing variables and predicts mixing requirements for maintaining yields of R on scale-up. Regions of excess B concentration are related to a concept of partial segregation. The extent of reaction occurring under this condition is correlated by the dimensionless group (k1bτ) (a0/b), where τ is a microtime scale of mixing related to u′ and the characteristic length of a microscale eddy. The magnitude of this group provides a criterion for predicting the importance of mixing effects on other reaction systems.

The coalescence rates of equisized drops of organic liquids fluidized by water under conditions of mutual saturation were studied. A drop-size separation technique permitted the direct count of the coalesced drops. Six organic liquids (benzene, cyclohexane, n-butylacetate, methyl isobutyl ketone, diisobutyl carbinol, and tetrahydrobenzaldehyde) were used, and the coalescence was in all cases total, that is, without formation of small satellite drops. A new hydrodynamic model is presented to calculate the parameters governing the drainage of liquid between two deformable drops in head-on collision. A parameter from this model permits correlation of the coalescence rates with only two arbitrary constants, provided the ratio of dispersed to continuous liquid viscosities lies between 0.5 and 2. High viscosity ratios or the presence of trace quantities of surfactant greatly depress the coalescence rate, while continued reuse of the dispersed liquid leads to a statistically significant reduction. The presence of a fog of very small droplets increases the coalescence rate.

Axial mixing in the liquid, with and without simultaneous countercurrent flow of air, was measured in a twelve-compartment, six-inch diameter, mechanically agitated tower for gas rates 9 to 94 and liquid rates 1300 to 19,600 lb/hr (sq ft), agitator speeds 200 to 2000 rpm. The corresponding eddy diffusivities and gas holdup were correlated in terms of the variables studied. Attempts to make similar measurements for the gas phase failed to give meaningful data, but the indications are that axial mixing in the gas is of minor importance. The tower was also used to measure the mass-transfer rates during absorption of ammonia from air into water. At 110 lb gas/hr (sq ft), the performance corresponds roughly to that of towers packed with 1- to 2-in conventional packings. The mass-transfer coefficients corrected for axial mixing are roughly 7 to 25 per cent larger than those before correction.

Drop size and dispersed-phase holdup were measured for liquid dispersions in covered, unbaffled agitated vessels with no gas-liquid interface and hence no vortex for batch and continous flow. Vessel diameters of 0.245 and 0.372 m, turbine impellers of diameters 0.0762 and 0.127 m in two locations, four organic- and three water-dispersed systems of a wide range of properties, and a wide range of operating conditions were studied. The data for average drop diameter and dispersed-phase holdup, from which specific interfacial area may be computed, were successfully correlated through modification of the theories of maximum drop size in an isotropically turbulent fluid.

A method of calcuation is presented by which estmates may be made of the stage efficiency of continuously operated, agitated, baffled vessels used in mixer-settler extractors. The calculations are limited to cases where the agitatiing impeller is a flat-blade turbine, and do not include estimates of the entrance and exit effects. The method has been tested with all the available experimental data, which include three different sizes of vessels, systems, and impeller sizes, and a variety of operating conditions including speeds of agitation, rates of flow, and ratios of contacted liquids. Because of limitations of the author's knowledge, the calculations are necessarily approximate, but they nevertheless correctly indicate the nature of the variations in stage efficiency to be expected with all of the design and operating variables for which tests could be applied.

Axial mixing was measured in both phases for a countercurrently operated, mechanically agitated extractor of the Oldshue-Rushton design for water continuous and toluene and kerosene dispersed. Results in terms of an eddy axial diffusivity were correlated in terms of the variables studied. These correlations were then applied to column performance during extraction of n-butyl amine from kerosene into water, and the true mass transfer coefficients, corrected for axial mixing, were determined. These are shown to be predictable for droplets of dispersed phase considered to be rigid spheres.

The dynamics of agitating single- and two-phase liquid mixtures in the absence of an air-liquid interface was compared with that of similar systems with an interface. Two geometrically similar cylindrical vessels, 12 and 18 in. in diameter respectively, were used, each fitted with a similar six-bladed, disk-turbine impeller. New general correlation curves of power number with Reynolds number are presented which show that elimination of the air-liquid interface makes it possible to attain dynamic similarity in scale-up in unbaffled vessels but makes little difference in baffled vessels.

A theory is developed to enable prediction of interfacial tension in unsymmetrical, regular, binary liquid systems from solubility data. This is also shown to be applicable to nonregular aqueous binary systems by using solubility parameters and molar volumes calculated from mutual solubilities as substitutes for true molar volumes and the solubility parameters obtained from energies of vaporization. The theory is extended to include unsymmetrical, regular, ternary liquid systems, and is shown to be applicable to systems consisting of perfluoro-n-hexane and other nonpolar organic liquids. The theory is not adequate for aqueous ternary systems.

Area-free continuous-phase mass-transfer coefficients for continuous flow of ethyl acetate (dispersed) and water (continuous) were measured in a 9.5 in. diameter baffled vessel, agitated with 3 and 5 in. diameter flat-blade turbines, and with the 3 in. turbine in the absence of baffles. Specific interfacial area was determined by a light-transmission technique which permitted a vertical traverse of the vessel. The specific area is greatest, and mean drop diameter smallest, near the impeller. The mean area is lower than that predicted from existing correlations for batch operation in the absence of mass transfer. Continuous-phase mass-transfer coefficients increase with increased impeller Reynolds number and dispersed-phase holdup, and are about 2.5 times larger in the unbaffled vessel at the same impeller power. The coefficients are generally larger than those predicted by existing correlations based on suspended solids, and it is suggested that this and the effect of holdup is the result of drop coalescence and redispersion, which causes a renewal of the continuous phase surrounding each drop. Rough estimates of the minimum coalescence frequency in the baffled vessel and application of the renewal concept to the data appear to confirm this view. Lack of information on circulation rates prevents immediate test of the data from the unbaffled vessel.

Measurements of the terminal velocity of liquid drops immersed in an insoluble liquid were made for eleven liquid systems, covering a wide range of physical properties: continuous-phase density 0.960 to 1.145 g./cc., viscosity 0.93 to 1.56 centiposes; dispersed-phase density 0.807 to 1.674 g./cc., viscosity 0.59 to 72.1 centipoises; interfacial tension 0.3 to 42.4 dynes/cm. A correlation of the data was developed for each of the two velocity-drop-diameter regions which permits easy computation of the drop velocities. The correlations are shown to reproduce the present data and much of the published data very satisfactorily. Photographs permitted measurement of the drop eccentricities, which are shown to be a function of the drop diameter, density difference, and interfacial tension.

The mass transfer coefficient in covered, right-cylindrical tanks full of liquid, turbulently agitated at various speeds by turbines with six flat blades, was measured by the rate of solution suspended solids in water and in 45% sucrose solutions. Screened crystals in the following U. S. mesh sizes were used: boric acid: 18/20, 16/18, 16/20, 14/16, 12/14, 10/12, 8/10, 6/8; rock salt: 6/8, 4/6. Pellets were benzoic acid: 0.126 in. long by 0.218-in. diam.; salt: 0.565-in. diam. by 0.531-in. long (over rounded ends). Tanks were 6, 12, 18, and 30 in. Turbines were 2, 3, 4, 6, 9, and 12 in. in diameter, centrally located. Four full-length baffles 10% of the tank diameter wide were spaced at 90 deg. A few runs were made without baffles. The coefficient of mass transfer was found to be independent of particle size and Schmidt member (NSc = 735 to 62,000) and could be correlated with turbine Reynolds number in each tank, with larger tanks yielding smaller coefficients at the same NRe. An empirical equation which fits all the data from the baffled tanks within about 4% (in the range 0.1 < k < 2) is where l2=0.8235−1.544V1/3+0.115V2/3The variance of estimate for this expression i s 0.0383, in units of [ln(10 k)]2.For extrapolation outside the experimental range of vessel sizes it is recommended that l1 = 0.676 − 1.266 V1/3 be used in place of I2. NRe = T2n/v. The results indicate that power per unit volume for a given k goes through a maximum, with the following relative values for the 6-, 12-, 18- and 30-in. tanks: 1, 1.73, 1.78, 0.62.A treatment of the data according to dimensionless groups provides another correlation:kd/D=0.02NNt is shown that for the systems used 1/D is essentially proportional to N, and so the effect of diffusivity here is only apparent.

Solution rates of 1/2-, 3/4-, and 1-in. cast benzoic acid spheres were measured under natural-and forced-convection conditions. Upward-flowing streams of water and aqueous propylene glycol, in laminar and turbulent flow, were used to contact single spheres in either a 1.50- or 4.00-in.- diameter column. Sphere Reynolds numbers ranged from 10 to 16,920; Schmidt numbers from 987 to 69,680; sphere- to column-diameter ratios, ds/dc, from 0.123 to 0.497; and Grashof numbers from 5,130 to 125,200. The Sherwood number was found independent of the laminar-turbulent transition for pipe flow and the sphere- to column-diameter ratio when the Reynolds number is based on the average fluid velocity and sphere diameter. A correlation of the data based on the additivity of the natural and forced convection processes is proposed, and statistical analyses of the new experimental data result in equations which correlate other heat and mass transfer data for single spheres immersed in bounded and free-jet streams of gases and liquids for NRe, S from 1 to 30,000 and NSe from 0.6 to 3,000, within an average deviation of about 20%.

Publisher Summary Liquid extraction is a means of separating the components of a solution and it is accomplished by bringing the solution into contact with another insoluble phase. In the case of liquid extraction, the two phases in question are both liquids, but, just as in the other mass transfer operations, intimacy of contact and large interfacial area between the phases are required for rapid diffusional transfer of substance from one phase to the other. Skillful, dynamic, and multistage equipment is required as the potentialities of liquid extraction are developing continuously. Therefore, devices such as the bubble-tray tower and the packed tower of distillation are used. This chapter describes some of the devices in use, focusing on those types that have been found to be of industrial importance. Among the devices, the ratio of throughput to volume of a theoretical stage is a characterizing device, which is also the reason for the disparity in performance of the devices.