### Abstract:

The dispersion behaviour of cyclohexane (dispersed phase) into water (continuous phase) in a stirred unbaffled batch vessel is studied experimentally and with computational fluid dynamic (CFD) simulation. The important aspects and fundamentals in liquid-liquid systems including mass transfer are described. The dispersion and distribution of cyclohexane into water is visualized by a red colour tracer and registered with a video camera. Different types of stirrers are used such as anchor impeller (tangential flow), 3-bladed propeller (axial flow), 4-bladed pitched blade turbine PBT (combined flow) and 6-bladed rushton turbine impeller RTI (radial flow). The minimum stirrer velocities for complete and uniform dispersion are determined by tracking the red layer of cyclohexane. The shape and size of the cyclohexane droplets are measured by using the particle vision microscope (PVM). The axial and radial profiles of cyclohexane volume fractions at different stirrer velocities are measured by using sampling withdrawal method. The minimum stirrer velocities are also determined from this method. Mass transfer of the solute (benzoic acid) from water to cyclohexane is used as an indirect method to calculate the volumetric mass transfer coefficient kcA between the two phases –from the measured transient concentrations of benzoic acid in water– by using the chemical acid-base titration method. The flow velocity fields produced by RTI and propeller impeller are measured by using photographic-light cut method with the help of tracer particles and MathCAD software. For the 3D steady and transient CFD simulations (using ANSYS CFX-11 and ICEM CFD-11), the geometry and the unstructured mesh generation for the anchor and RTI as well as all used submodels, the initial and boundary conditions and also the solution algorithm, are described. The submodels are: Eulerian-Eulerian multi-fluid model, the k-ε turbulent model, algebraic slip (ASM) and Ishii-Zuber models, buoyancy model, the standard free surface flow model, particle model and the sliding mesh model. It is found many vortices and symmetric flow around the impellers. With anchor impeller, large axial vortex is formed in the water phase and large symmetrical vortex of cyclohexane is concentrated around the shaft of the anchor impeller. Whereas a radial flow with two circulations above and below the RTI is formed. The shape of the cyclohexane droplets is found to be spherical. Mean Sauter droplet diameter d32 is ordered referring to the type of the impeller as d32,RTI (409 µm) > d32,anchor (388 µm) > d32,propeller (376 µm) > d32,PBT (343 µm), referring to the percentage volume of cyclohexane with RTI as d32,20 vol% (399 µm) > d32,10 vol% (348 µm) and referring to the RTI bottom clearance as d32,4.1 cm (409 µm) > d32,3.2 cm (399 µm). For better understanding, a droplet size distribution is also analyzed by the number frequency diagram. The anchor impeller gives a bi-modal distribution, whereas the other stirrers give a uni-modal one. The CFD predicted axial and radial profiles of cyclohexane volumetric fractions are in a good agreement with that measured from the sampling method. The ASM is able to predict the axial and radial profiles of cyclohexane volume fractions in case of the RTI, whereas Ishii-Zuber model is suitable in case of the anchor impeller. The grid size should not exceed 0.001 m for good predictions. Increasing the bottom clearance of RTI reduces the minimum velocity for complete dispersion in the order of 5.5 cm (400 rpm) < 4.1 cm (500 rpm) < 3.2 cm (550 rpm). The minimum RTI velocity to get a uniform dispersion of cyclohexane is 750 rpm, it is independent on the RTI clearance. The minimum velocity for complete dispersion referring to the stirrer type is ordered as anchor (350 rpm) < propeller (400 rpm) < PBT (450) < RTI (500 rpm). Non uniform distribution of cyclohexane in the axial and radial distances is found with anchor impeller. The dispersion of cyclohexane becomes uniform at PBT velocity of 550 rpm. For propeller impeller, the dispersion becomes nearly uniform at a minimum velocity of 650 rpm, especially in the region of the axial circulation. The level of the total liquid phases remains nearly constant with all types of the stirrers except that for anchor impeller. The dispersion behaviour obtained by propeller and anchor impellers is similar. Another similarity exists between RTI and PBT impeller. The dispersion of cyclohexane/water and the minimum stirrer velocities from the visualisation with red tracer coincide with those obtained from the measured axial and radial profiles of cyclohexane volume fractions, and show a good agreement with the CFD predictions. The interfacial area A between the two phases is calculated from the measured droplet diameters and the CFD predicted averaged volumetric fractions of cyclohexane. From the measured kc A, the kc is then determined. It is found that kc increases approximately as the 4.23 power of the RTI velocity and 2.95 of the anchor velocity, whereas A increases as the 2.62 power of RTI velocity and 0.75 power of anchor velocity. The kc A is influenced by the stirrer type (flow pattern) in the order of (kcA)anchor > (kcA)propeller > (kcA)PBT > (kcA)RTI. The propeller impeller gives the maximum volumetric mass transfer coefficient of 0.00003 m3/s at a minimum velocity of 500 rpm, whereas the PBT impeller gives the same value at 600 rpm. The RTI gives the maximum value of 0.00002 m3/s at velocity of 550 rpm. Increasing RTI bottom clearance enhances the mass transfer of benzoic acid for RTI velocities 450 rpm, but it has no significant effect on kc A. High mass transfer occurs when cyclohexane is completely dispersed and enhanced when a uniform dispersion is achieved. The surface renewal of the dispersed phase by diffusion and turbulence is responsible for mass transfer of the solute. Increasing the initial volume (percentage) of cyclohexane in case of anchor impeller increases kc A. The equilibrium concentrations are similar for all types of stirrers. The anchor impeller gives the minimum equilibrium time of 30 s at a minimum velocity of 350 rpm, whereas the RTI and PBT both reach the equilibrium concentration at the same equilibrium time of 300 s at velocities of 500 and 400 rpm, respectively. The RTI and propeller impeller reach the equilibrium concentration at the same velocity of 500 rpm but at different equilibrium times of 300 s and 60 s, respectively. The predicted radial flow velocities are higher than the axial one in case of the RTI, whereas the axial velocities are higher than the radial one in case of the anchor impeller. The turbulence is high near the impeller, minimum at the free surface and bottom of the vessel and intermediate between these locations. The CFD predicted turbulent flow velocity distributions are in agreement to that obtained by photographic light-cut method and with the cyclohexane volumetric fraction distribution from sampling and visualization methods.