14 th European onference on Mixing Warszawa, 10-13 September 2012 EXPERIMENTAL INVESTIGATION AND FD MODELING OF MIROMIXING OF A SINGLE-FEED SEMI-BATH PREIPITATION PROESS IN A LIQUID-LIQUID STIRRED REATOR Dang heng, Jingcai heng, Yumei Yong, hao Yang*, Zai-Sha Mao Key Laboratory of Green Process and Engineering, Institute of Process Engineering, hinese Academy of Sciences, Beijing 100190, hina E-mail: chaoyang@home.ipe.ac.cn, Tel.: +86-10-62554558 Abstract. In this work, the BaSO 4 precipitation process in a single-feed semi-batch liquid-liquid stirred reactor is experimentally investigated and numerically simulated using the single-phase FM-PDF model. Both feed addition time and kerosene volume fraction effects have been shown to be significant in the process. The kerosene is observed to have no impact on the morphology of BaSO 4 particles. The single-phase FM-PDF model calls for improvement in the description of the precipitation process in the immiscible liquid-liquid system. Keywords: Micromixing, precipitation, liquid-liquid, FD simulation 1. INTRODUTION Precipitation is an important unit operation in chemical industry for a very wide range of chemical and pharmaceutical products. As rapid precipitation is a fast reaction by nature, the process is indirectly influenced by macromixing and mesomixing, and more importantly, directly influenced by micromixing (i.e., mixing on the molecular scale). The macromixing and mesomixing determine the environment in which micromixing takes place. Mixing effects are significant in determining the distribution of supersaturation and subsequently the particle size distribution and morphology of the product. The influence of mixing conditions on the process of precipitation of barium sulphate in single-phase stirred reactors has received much attention both experimentally and by computational fluid dynamics (FD) simulation. To the authors best knowledge, the relevant investigation with respect to how the micromixing impacts on the BaSO 4 precipitation process in an immiscible liquid-liquid stirred reactor has not been found in the open literature, in spite of its importance. Therefore, the aim of this work is to fill this gap experimentally and by FD simulation. 55
2. EXPERIMENTAL 2.1 Experimental Setup Experiments are carried out in a Rushton turbine driven transparent cylindrical stirred reactor with a flat bottom. The diameter of the stirred tank is 0.24 m and the total height is 0.428 m. Four baffles with width of 0.024 m are installed equally-spaced at the wall. The liquid height for all experiments is set at H=T. The impeller and reactor diameter are D=0.08 m and T=0.24 m, respectively. The impeller off-bottom clearance (measured from the impeller disk) is T/5 in order to prevent surface aeration at relatively higher impeller speeds which could not be avoided at a standard clearance of =T/3. Figure 1 shows the experimental setup for the precipitation system. The model system is oil-in-water dispersion in the stirred reactor, in which the barium chloride aqueous solution serves as the continuous phase and kerosene is used as the dispersed phase. The holdup of kerosene ranges from 0% to 20% by volume. Two feed locations are selected midway between two baffles. Figure 1. Experimental setup (Feeding position P1: r= 84 mm, z=144 mm; P2: r=84 mm, z=80 mm. The origin is at the centre of tank bottom). 2.2 Precipitation Procedure In experiments, the volume ratio ( α v ) and the stoichiometric ratio (η) of reactants are important parameters as defined by α = (1) v V / V η = (2) where V and V are the initial volumes of reactants A (barium chloride, Bal 2 ) and B (sodium sulfate, Na 2 SO 4 ), respectively. The volume ratio, mean initial concentrations ( and ), and the actual initial concentrations of A and B ( and ) are related as = ( + ), ( α ) 1 1/ αv 56 = + (3) 1 v
( ) V = V + V, i = A, B (4) i0 i0 i0 Feed (sodium sulfate solution, B) is added to barium chloride aqueous solution (A)-kerosene dispersion using a tube of 1.2 mm ID to avoid backmixing. The feed addition is made after about one hour of stirring as Zhao et al. [1] suggested. Feed rate is well controlled by a constant flow feed pump. Barium chloride, sodium sulfate (both AR grade) and deionized water are used to prepare the aqueous solutions. No temperature changes occur during the course of the reaction. Samples are collected at the end of the experiments without delay to eliminate settling and aging effects. A separating funnel is used to separate oil from water solution quickly and carefully. The samples are then filtered through a 0.22 μm membrane filter and washed with deionized water and then ethanol. The filter is dried in an oven. The crystal morphology is observed with a JEM-2100 TEM. To prevent agglomeration, the experimental operation conditions are defined at = =0.0045 mol/l with α v =50. The stirrer speeds are in the range of 390 rpm to 500 rpm. It should be noted that the kerosene phase is well dispersed into the continuous phase at the selected impeller speeds under corresponding oil holdups with no surface aeration occurring. All the parameter values used in the analysis are averaged from at least four sets of measurements. 3. RESULTS AND DISUSSION The effects of operating conditions on the mean particle size are based on the relative competition between the micromixing of fluids and the nucleation and growth of the precipitated particles. 3.1 Effect of Feed Addition Time The effect of addition time at φ d =0% is shown in Figure 2. When the feed is positioned at P2 with N=300 rpm in Figure 2(a), the mass weighted mean size (d 43 ) generally increases with feed time (t f ), and becomes independent of it for t f >30 min. When the impeller speed is increased to 390 rpm at φ d =10% with P1 feed, as shown in Figure 2(b), the mesomixing effects are still observable but disappeared at a lower t f value (27 min). This is consistent with the hypothesis that increasing the level of turbulence decreases the importance of mesomixing relative to micromixing [2]. The effect of mesomixing deceases when increasing the feed addition time at a constant impeller speed, and thus the intensity of mixing increases and the local supersaturation and the nucleation rate decrease. onsequently, a smaller number of larger crystals are produced. In order to ensure that micromixing is always controlling, subsequent experiments are carried out at 4.7 ml/min (the corresponding feed addition time is 45.65 min). 57
8 8.5 6 P2 Feed, =T/5, N=300 rpm, ϕ d =0% 8.0 d 43 (μm) 4 2 d 43 (μm) 7.5 7.0 6.5 P1 Feed, =T/5, N=390 rpm, ϕ d =10% 0 10 15 20 25 30 35 40 45 50 Feed time (min) 6.0 10 15 20 25 30 35 40 45 50 Feed time (min) (a) φ d =0% (b) φ d =10% Figure 2. Effect of feed time on d 43 of the precipitate. 3.2 Effect of Kerosene Volume Fraction It is of interest to know how the dispersed kerosene volume fraction impacts on the nucleation and growth process of BaSO 4 precipitation. As shown in Figure 3, the mean crystal size first increases with the increase of kerosene volume fraction from 0% to 5% (to a maximum at 5%), and then decreases with the further increase of kerosene from 7 vol.% to 20 vol.% in the liquid-liquid systems. According to the engulfment model [3], increasing turbulence also increases the engulfment rate, which dilutes the local supersaturation level and correspondingly larger mean crystal sizes are produced and vice versa. Therefore, the non-monotonic results might reveal that the dispersed oil phase increases the turbulence at relatively lower dispersed oil phase holdups, but decreases the turbulence at higher dispersed oil phase holdups. This inference is consistent with Zhao et al. [1], who measured the macro-mixing time at different kerosene volume fractions, and found that the mixing time decreased first with oil increased from 0% to 7% by volume and increased further with oil holdup ranged from 7% to 20%. 7 6 d 43 (μm) 5 4 3 2 =T/5, N=470 rpm, P2 Feed Feed rate=4.7 ml/min 1 0.00 0.04 0.08 0.12 0.16 0.20 Kerosene volume fraction Figure 3. Effect of kerosene volume fraction on d 43 of the precipitate. The TEM photographs in Figure 4 reveal that kerosene has nearly no influence on the morphology of the barium sulfate particles. 58
(a) φ d =0% (b) φ d =20% Figure 4. TEM photographs of the precipitate for N=470 rpm and P2 feed. 4. NUMERIAL SIMULATION The governing equations of RANS version for the liquid-liquid flow in a stirred tank are solved using an in-house FD code [4]. Experimental conditions are replicated as closely as possible in the simulations in order to compare the simulation results with the corresponding experimental data. The simulated flow field has been validated with the experimental data of velocity field and the dispersed-phase holdup distribution. For details of the calculation of flow field, refer to heng et al. [4]. Macro- and meso-scale mixing are modeled using FD with the turbulence model. However, a micromixing model must be coupled with the population balance equation (PBE) in order to take into account the mixing on the molecular level when describing the fast precipitation process. The population balance takes into account the nucleation of the solid phase and the growth of the crystals. The popular single-phase FM-PDF (3-mode) model is used in this work. The FM-PDF model and the moment method for the PBE details of this approach are referred to Marchisio and Barresi [5]. The PBE is solved based on the work of heng et al. [6]. In-house codes have been developed, and the present single-phase FM-PDF model produces mean particle size (d 43 ) only of the order of magnitude of the measured data in liquid-liquid systems, indicating that the presence of oil droplets in each computational cell exerts much influence on the micromixing rate between mode 1 and mode 2 containing unmixed reactants A and B, respectively. The more accurate description of the micromixing effect in liquid-liquid systems is thus expected by modifying/improving the single-phase FM-PDF model. Extension of the present single-phase FM-PDF model to multiphase systems taking the impact of the dispersed phase into consideration is underway. Moreover, the effects of feed location, agitation speed, and the dispersed phase physical properties on the mean particle size and coefficient of variation (V) are to be examined. 59
5. ONLUSION Both feed addition time and kerosene volume fraction effects have been shown to be significant in single-feed semi-batch BaSO 4 precipitation processes. The kerosene is observed to have no impact on the morphology of BaSO 4 particles. The single-phase FM-PDF model calls for improvement in the description of the precipitation process in immiscible liquid-liquid systems. Acknowledgements: The financial supports from 973 Program (2012B224806), the National Natural Science Foundation of hina (21106154, 20990224) and the National Science Fund for Distinguished Young Scholars (21025627) are gratefully acknowledged. 6. NOTATION impeller clearance off the tank bottom, m, the actual initial concentrations of A, B, mol/l, the initial concentrations of A, B, mol/l D diameter of impeller, m H liquid height in the tank, m N agitation speed, s -1 T diameter of the tank, m t f feed addition time, s V, V initial volumes of reactants A, B, L R radial coordinate from shaft, m z axial coordinate from tank bottom, m volume fraction of the dispersed oil phase φ d 7. REFERENES [1] Zhao Y.., Li X.Y., heng J.., Yang., Mao Z.-S., 2011. Experimental Study on Liquid-Liquid Macro-mixing in a Stirred Tank, Ind. Eng. hem. Res., 50(10), 5952 5958. [2] Bourne J.R., Thoma S.A., 1991. Some Factors Determining the ritical Feed Time of a Semi-Batch Reactor, hem. Eng. Res. Des., 69, 321-323. [3] Baldyga J., Pohorecki R., 1995. Turbulent micromixing in chemical reactors a reivew, hem. Eng. J., 58, 183-195. [4] heng D., heng J.., Yong Y.M., Yang., Mao Z.-S., 2011. FD Prediction of ritical Agitation Speed for omplete Dispersion in Liquid-Liquid Stirred Reactors, hem. Eng. Technol., 34(12), 2005-2015. [5] Marchisio D.L., Barresi A.A., Fox R.O., 2001. Simulation of Turbulent Precipitation in a Semi-Batch Taylor-ouette Reactor using FD, AIhE J., 47(3), 664-676. [6] heng J.., Yang., Mao Z.-S., Zhao.J., 2009. FD Modeling of Nucleation, Growth, Aggregation, and Breakage in ontinuous Precipitation of Barium Sulfate in a Stirred Tank, Ind. Eng. hem. Res., 48, 6992-7003. 60