Ind. Eng. Chem. Res. 2010, 49, 10297–10302
10297
Optimization Design for DTB Industrial Crystallizer of Potassium Chloride Xingfu Song,* Menghua Zhang, Jin Wang, Ping Li, and Jianguo Yu* State Key Laboratory of Chemical Engineering, National Engineering Research Center for Integrated Utilization of Salt Lake Resources, East China UniVersity of Science and Technology, Shanghai, China
A computational fluid dynamics (CFD) model was developed for the simulation and optimization of an existing continuous DTB crystallizer with KCl productivity of 0.1 million tons per year. The multiple reference frame (MRF) method was used in the CFD simulation. Both the hexagonal grid and the tetrahedral grid were adopted to divide meshes in this industrial DTB crystallizer, and in total 866 388 cells were used for CFD simulation. The fluid flow field in the DTB crystallizer was calculated using FLUENT6.3 software with the Reynoldsaveraged Navier-Stokes equation combined with the widely used κ-ε turbulence model. The crystal size distribution and the coefficient of variation of crystal product were studied by CFD simulation of two-phase flow model. The impeller shapes and various operating conditions were optimized to reduce the energy consumption of the crystallization process and to increase the KCl product quality. Based on the CFD optimization design, a new impeller was retrofitted into an existing continuous DTB crystallizer with the KCl productivity of 0.1 million tons per year located at Qinghai salt lake plant in China, and its excellent performance was confirmed against data collected using the original impeller. 1. Introduction Some pioneering studies have been done very well to understand the thermodynamics, kinetics, and processes of carnallite dissolution and KCl crystallization, but little work has been done to study the influence of hydrodynamics on the behavior of an industrial crystallizer of potassium chloride. The main challenge involved in the design of industrial crystallizers is to predict the influences of vessel geometry, configuration, operating conditions, and the effects of scale-up on the process behavior, product quality, and crystal size distribution. The design of the industrial crystallizer is hindered by the lack of rational scale-up rules and incorporation of hydrodynamic information and kinetics. Computational fluid dynamics (CFD) is a powerful simulation tool that has been successfully used to investigate mixing, turbulence, and shear stress in a crystallizer.1-7 CFD can give a qualitative engineering insight into the effects of the impeller configuration on the crystallization rates, crystal size distribution, and power consumption.8-16 Precipitation of particles with controlled morphology and crystal size distribution (CSD) is of considerable importance in the chemical industry. The energy consumption for subsequently drying and filtering after the crystallization process can be reduced by operating the crystallizer so as to obtain large crystals which can then be easily separated from the liquid phase. Many studies have been conducted on the optimal operation of crystallization processes to maximize the mean size of the product crystals or the production rate of large crystals.17-19 The Draft Tube Baffle (DTB) Crystallizer is a combination of a classifier and a crystallizer with mixed suspension, mixed particle removal unit, which has been used successfully where narrow crystal distribution and larger crystal size are required. 17-19 In this work, the operating conditions for KCl crystallization in a continuous DTB crystallizer are optimized by CFD simulation, where the crystallizer is an existing industrial DTB * Authors to whom correspondence should be addressed. E-mail:
[email protected] (X.S.),
[email protected] (J.Y.). Fax: 86-2164252170.
crystallizer located in Qinghai salt lake plant of China, with KCl productivity of 0.1 million tons per year. Emphasis is placed on hydrodynamics of liquid and crystal in this crystallizer with different shapes of impellers. The computational fluid dynamic (CFD) solver, FLUENT6.3 software, is used to simulate the velocity field in such units, to optimize both the structural design of impellers and the operating conditions to further reduce the energy consumption and increase the product quality for KCl production from carnallite at the industrial scale. 2. CFD Model and Grids In this work, the investigation by CFD simulation involves a continuous DTB industrial crystallizer with the external fines dissolution system, having the KCl productivity of 0.1 million tons per year. Figure 1 shows the schematic diagram of this DTB crystallizer with an axial-flow impeller, where the DTB crystallizer is operated with continuous input and output, and its geometrical dimensions are listed in Table 1. The mixing behavior of the three types of impellers will be analyzed by CFD simulation. The impellers include propeller, pitched blade turbine, and Rushton turbine; their shapes as shown in Figure 1.
Figure 1. Schematic diagram of the structure of a continuous DTB industrial crystallizer and the shapes of the impellers. Symbols: (1) overflow, (2) sedimented zone, (3) underflow, (4) draft tube, (5) impeller, (6) baffle, (7) mixed zone, (8) inlet.
10.1021/ie100786f 2010 American Chemical Society Published on Web 09/09/2010
10298
Ind. Eng. Chem. Res., Vol. 49, No. 21, 2010
Table 1. Dimensions of the Industrial DTB Crystallizer Used in CFD Simulation main body, including a columniform upper part and a coniform under part draft tube inner baffle baffle (near the overflow port) impeller
diameter D ) 12.2 m, height H ) 14.8 m height H ) 10.4 m, diameter D ) 2.7 m; off-bottom 2.7 m height H ) 7.5 m; diameter D ) 3 m; off-bottom 7 m height H ) 1.05 m; diameter D ) 11.47 m diameter D ) 2.5 m; shaft diameter 0.35 m
The multiple reference frame (MRF) method is used in CFD simulation. A hexagonal numerical grid is used to establish meshes in the main body of the DTB industrial crystallizer, and near the impeller blade and the inlet and outlet ports a fine grid with tetrahedral element is used. The GAMBIT software release 2.2 is used to generate these structured and unstructured meshes. The total meshes divided in the DTB crystallizer consist of 866 338 cells for CFD simulation, the details as shown in Figure 2. With these fine structured and unstructured meshes, the steady numerical solution for the fluid flow field can be obtained using FLUENT6.3 software with the converging precision 10-6. In CFD simulation, an averaged form of the Navier-Stokes equation of motion (referred to as the Reynolds-averaged Navier-Stokes equation or RANS) is used for the calculation of fluid flow field in the DTB crystallizer, which requires an estimation of the effect of the turbulence fluctuations on the mean flow, here adopting the widely used κ-ε model, assuming that turbulence is locally isotropic and is fully characterized by its time-averaged kinetic energy and time-averaged rate of kinetic energy dissipation. The optimization of crystal size distribution in the DTB crystallizer is done by CFD simulation of two-phase flow, where the dispersed phase model is used to describe the crystal motion, and feedstock with the suspension crystals is assumed to have a given crystal size distribution. The numerical simulation on the flow field is done using FLUENT6.3 software, and all the necessary equations for fluid and crystal simulation are available in that commercial software. It should be explained that, for simplicity, the CFD simulation and the numerical analysis are limited on the hydrodynamics of liquid and crystal in an existing continuous DTB industrial crystallizer, while the kinetics of nucleation, growth, dissolution, and attrition of crystal are ignored. Therefore, some improvements based on the work presented in this paper are necessary in the future research.
Figure 3. Positions of z and x coordinates referenced in Figures 4 and 5.
3. Results and Discussion 3.1. Optimal Design for an Existing Continuous DTB Industrial Crystallizer. Figure 4 demonstrates the typical fluid flow field in the existing continuous DTB industrial crystallizer with propeller, calculated by the CFD solver with single phase flow model. As mentioned above, the DTB crystallizer with an external fines dissolution vessel has the KCl productivity of 0.1 million tons per year, and it has two inlets and two outlets: one inputting the fresh fines dissolution stream with 0.41 m/s velocity, the other imputing the feed and recirculation stream with 1.59 m/s velocity (800 m3/h), underflow for the crystal product output with 120 m3/h flow rate, and overflow to removal the fines to the external dissolution vessel. The propeller is installed 5.7 m height off-bottom of crystallizer, with 70 rpm rotational speed, and temperature in KCl solution is 10 °C. Because of the strong agitation of propeller, there exists a negative pressure drop between upside and downside of propeller, which results in a partial amount of fluid circulating directly from the draft tube to the baffle tube. The most amount of fluid from draft tube will go down the bottom of the crystallizer, and circulate to the clarification zone due to the block of a coniform under-part of the crystallizer, and then circulate to the baffle tube, as shown in Figure 4b. At the bottom of the crystallizer, two small circulations are formed, which can be designed as outflow ports for the crystal product output.
Figure 2. Hexagonal and tetrahedral meshes divided in the continuous DTB industrial crystallizer used for CFD simulation.
Ind. Eng. Chem. Res., Vol. 49, No. 21, 2010
10299
Figure 4. Fluid flow field in an existing continuous DTB crystallizer with propeller.
Figure 5. Comparison of the axial and the radial temporal mean velocity distributions in the existing DTB crystallizer among three impellers (propeller, pitched blade turbine, and Rushton turbine).
Figure 5a and b shows the comparisons of the axial and the radial temporal averaged velocity distributions in the DTB crystallizer among the three types of impellers (propeller, pitched
blade turbine, and Rushton turbine), with the calculation conditions as above-mentioned. The comparisons are presented in two representative positions, x ) 1.3 m (inside of the draft
10300
Ind. Eng. Chem. Res., Vol. 49, No. 21, 2010
Table 2. Effects of Operating Conditions on the Power Consumption of Impeller effect of impeller shape on power consumption impeller power/kw
propeller pitched blade Rushton turbine 87.3 187.3 570.0 effect of propeller rotational speed on power consumption
rotational speed/rpm power/kw
60 57.4
65 73.0
70 87.3
75 80 112.2 136.3
effect of propeller installed height on power consumption H position/m power/kw
3.5 90.3
5.7 87.3
7.6 85.4
9.7 85.1
11.8 82.6
effect of feed charging rate on propeller power consumption feed flux/m · h-1 power/kw
700 87.34
800 87.34
900 87.33
effect of fluid viscosity on propeller power consumption fluid viscosity/mPa · s power/kw
8.9 87.2
8.45 87.2
7.10 85.4
5.89 5.22 85.5 82.5
tube), and z ) 1.3 (at the bottom of the crystallizer), respectively (Figure 3). As shown in Figure 5, when three impellers with the same rotational speed (70 rpm) are simulated, there exist significant differences of the axial and the radial temporal mean velocity distributions among these three cases (i.e., propeller, pitched blade turbine, and Rushton turbine), respectively, which will have significant effect on the crystal growth and the fines dissolution in the DTB crystallizer. There are two apparent fluctuations of the axial and the radial temporal mean velocities for the case at x ) 1.3 m, one taking place near the impeller installed height (z ) 5.7 m) due to the strong agitation of impeller, and the other taking place in the outlet of the draft tube (z ) 2.7 m). With the same rotation speed of impeller, the fluctuation of the velocity for propeller is smaller than the others (pitched blade turbine and Rushton turbine), which can reduce the energy consumption. At the bottom of the DTB crystallizer (z ) 1.3 m), the fluctuations of the velocities are weaker, which can favor the crystal precipitation and the crystal production output with underflow. Since the power consumption of the impeller is a very important design parameter in the DTB industrial crystallizer, the effects of various operating conditions on the power consumption of the impeller are investigated by CFD simulation. The simulation results are summarized in Table 2, calculated for an existing continuous DTB industrial crystallizer with 0.1 million tons KCl productivity per year, and the other operating conditions as above-mentioned. The effects of impeller shape, its rotational speed, its installed position, feed charging rate (feed and recirculation stream), fluid viscosity, temperature, etc., on
the power consumption of the impeller, are presented. According to the simulation results, we found that the propeller is the best stirred vessel among these three types of impellers, and its power consumption is far from lower than that of pitched blade turbine and Rushton turbine. The other operating conditions, such as the rotational speed of the impeller, installed height, fluid viscosity, temperature, and feed flow rate, have little effect on the power consumption. The energy cost for the subsequent drying and filtering in a practical crystallization process can be reduced by operating the crystallizer so as to obtain large crystals which can then be easily separated from the liquid phase.18 The optimal operation in the DTB crystallizer is very important to maximize the mean size of the product crystals or the production rate of large crystals. Here, the two phase flow model in CFD solver is used to optimize the operating conditions to obtain the large size crystals. In the simulation, the KCl crystal size distribution in the feedstock stream is described with Rosin-Rammler model.18 The Rosin-Rammler equation is defined as j
Yd ) e-(d/d)
n
where d is crystal size µm, dj means crystal size, n is formula exponent, Yd is crystal volume fraction more than the averaged crystal size. Based on the KCl sample provided by Qinghai salt lake plant in China, the model parameters are obtained as dj ) 750 µm, n ) 1.49. Figure 6 demonstrates the typical crystal trajectories and the crystal concentration distribution in the continuous DTB crystallizer, where propeller is installed at 5.7 m height with 70 rpm rotational speed; two fluid input ports are set, one with 0.41 m/s and the other with 1.59 m/s; temperature in KCl solution is 10 °C; two output ports are set, overflow to removal the fines to the fines dissolution vessel and underflow with the large size crystals for the final KCl product (120 m3/h flow rate from the bottom of the crystallizer). As shown in Figure 6, with the befittingly installed height of propeller, the crystals can be suspended uniformly in the fluid phase over the whole DTB crystallizer, which will favor the fines removal to the external fines dissolution vessel with the overflow, and favor the large size crystals precipitation to the bottom of the DTB crystallizer and output with underflow. The simulation results of the crystal size distributions at both overflow and underflow are show in Figure 7a and b, respectively, and the fines are effectively removed with overflow and the large size crystals are outputted with underflow from the bottom of the DTB crystallizer.
Figure 6. Typical crystal trajectories and crystal concentration distributions in the continuous DTB crystallizer with propeller.
Ind. Eng. Chem. Res., Vol. 49, No. 21, 2010
10301
Figure 7. Crystal size distributions at overflow and underflow from the continuous DTB crystallizer with propeller. Table 3. Effect of Various Operatiing Conditions on CV Value in Underflow effect of propeller rotational speed on CV value in underflow rotational speed/rpm CV%
60 40.5
65 26.2
70 27.4
75 30.0
80 47.5
effect of propeller position off-bottom on CV value in underflow H position/m CV%
3.7 39.9
4.7 32.3
5.7 27.4
6.7 52.0
effect of feed charging rate on CV value in underflow feed flux/m3 · h-1 CV%
700 30.0
800 27.4
Figure 8. The old impeller and the newly designed impeller used in the existing continuous DTB crystallizer with 0.1 million tons of KCl productivity per year at Qinghai salt lake plant in China.
900 27.1
effect of fluid viscosity on CV value in underflow fluid viscosity/mPa · s CV%
8.9 54.6
8.45 46.5
7.10 27.4
5.89 34.8
5.22 38.3
Furthermore, the coefficient of variation (CV) of crystal in the underflow can be calculated by the two-phase flow CFD solver, and the operating conditions in the continuous DTB crystallizer can be optimized to improve the crystal product quality. The coefficient of variation18 is calculated by CV )
(d84 - d16) × 100% 2d50
where d84, d50, and d16 represent the sieve port size when sieve fraction up to 84%, 50%, 16%, respectively. The effects of various operating conditions on the CV values are presented in Table 3. Based on the KCl sample provided by Qinghai salt lake plant in China, the mean crystal size of the sample is estimated as dj ) 750 µm, the small CV value obtained from the underflow means the product with the more amounts of the large size crystals and having a narrow crystal size distribution (CSD). The optimal operating conditions can be found based on the simulation results listed in Table 3. 3.2. Experimental Validation for the Existing Continuous DTB Industrial Crystallizer. As mentioned above, there exists a significant effect of the impeller shape on the hydrodynamics in the DTB crystallizer, so the impeller structure design is very important. In the last part, an industrial application example of the optimal continuous DTB crystallizer with 0.1 million tons of KCl productivity per year is demonstrated at Qinghai salt lake plant in China, in which the original pitched blade turbine is substituted by the new designed propeller, as shown in Figure 8. Based on the aforementioned simulation results, the operating parameters are optimized further by CFD simulation, as the fresh fines dissolution stream with 70 m3/h, the feed and recirculation stream with 800 m3/h, underflow for the crystal product output with 120 m3/h, and
the propeller installed height 5.7 m with 60 rpm rotational speed, and temperature in KCl solution at 10 °C. With the improved impeller and the optimal operating conditions, the product quality is increased, that is the percentage of product with crystal size larger than 2 mm is raised from 50% to 90%; the amount of natural gas consumption is decreased from 22 to 15 m3 per ton of crystal product. Moreover, the power consumption used by the new impeller also is reduced by 15% in contrast with the original pitched blade turbine. Conclusion Computational fluid dynamics (CFD) is a powerful simulation tool that can be successfully used to investigate mixing, turbulence, shear stress, and crystal size distribution in an industrial DTB crystallizer. CFD can give a qualitative engineering insight into the effects of the impeller configuration on the crystallization rates and crystal size distribution and power consumption. According to the CFD optimization design, the newly designed propeller was retrofitted into an existing continuous DTB crystallizer with the 0.1 million tons of KCl productivity per year at Qinghai salt lake plant in China, and its performance is significantly improved against data collected using the original pitched blades turbine. For example, the percentage of product with crystal size more than 2 mm is raised from 50% to 90%; the amount of natural gas consumption is decreased from 22 to 15 m3 per ton of product; and power consumption of the impeller is reduced by 15% in contrast with the original pitched blade turbine. Acknowledgment This work was financially supported by Program for New Century Excellent Talents in University (NCET-08-0776), the
10302
Ind. Eng. Chem. Res., Vol. 49, No. 21, 2010
Key Project of Qinghai Science and Technology R&D (2006G-161), and Shanghai Leading Academic Discipline Project (no. B506). Literature Cited (1) Liu, X. M.; Hatziavramidis, D.; Arastoopour, H.; Myerson, A. S. CFD Simulations for Analysis and Scale-up of Anti-Solvent Crystallization. AIChE J. 2006, 52, 3621–3625. (2) Jaworski, Z.; Dyster, K. N.; Nienow, A. W. The Effect of Size, Location and Pumping Direction of Pitched Blade Turbine Impellers on Flow Patterns: LDA Measurements and CFD Predictions. Chem. Eng. Res. Des. 2001, 79, 887–894. (3) Kelly, W.; Gigas, B. Using CFD to Predict the Behavior of Power Law Fluids Near Axial-Flow Impellers Operating in the Transitional Flow Regime. Chem. Eng. Sci. 2003, 58, 2141–2152. (4) Vakili, M. H.; Esfahany, M. N. CFD Analysis of Turbulence in a Baffled Stirred Tank: A Three-Compartment Model. Chem. Eng. Sci. 2009, 64, 351–362. (5) Alexopoulosa, A. H.; Maggiorisa, D.; Kiparissides, C. CFD Analysis of Turbulence Non-Homogeneity in Mixing Vessels: A Two-Compartment Model. Chem. Eng. Sci. 2002, 57, 1735–1752. (6) Sahu, A. K.; Kumar, P.; Patwardhan, A. W.; Joshi, J. B. CFD Modelling and Mixing in Stirred Tanks. Chem. Eng. Sci. 1999, 54, 2285– 2293. (7) Rousseaux, J.-M.; Vial, C.; Muhr, H.; Plasari, E. CFD Simulation of Precipitation in the Sliding-Surface Mixing Device. Chem. Eng. Sci. 2001, 56, 1677–1685. (8) Guo, Sh. Ch.; Evans, D. G.; Li, D. Q.; Duan, X. Experimental and Numerical Investigation of the Precipitation of Barium Sulfate in a Rotating Liquid Film Reactor. AIChE J. 2009, 55, 2024–2034. (9) Bermingham, S. K.; Verheijen, P. J. T.; Kramer, H. J. M. Optimal Design of Solution Crystallization Processes with Rigorous Models. Chem. Eng. Res. Des. 2003, 81, 893–903. (10) Kougoulos, E.; Jones, A. G.; Wood-Kaczmar, M. W. Process Modelling Tools for Continuous and Batch Organic Crystallization Processes Including Application to Scale-Up. Org. Process Res. DeV. 2006, 10, 739– 750.
(11) Derksen, J. J.; Kontomaris, K.; McLaughlin, J. B.; Van den Akker, H. E. A. Large-Eddy Simulation of Singe-Phase Flow Dynamics and Mixing in an Industrial Crystallizer. Chem. Eng. Res. Des. 2007, 85, 169–179. (12) Yapici, K.; Karasozen, B.; Schafer, M.; Uludag, Y. Numerical Investigation of the Effect of the Rushton Type Turbine Design Factors on Agitated Tank Flow Characteristics. Chem. Eng. Process. 2008, 47, 1340– 1349. (13) Brucato, A.; Ciofalo, M.; Grisafi, F.; Micale, G. Numerical Prediction of Flow Fields in Baffled Stirred Vessels---A Comparison of Alternative Modelling Approaches. Chem. Eng. Sci. 1998, 53, 3653– 3684. (14) Pinelli, D.; Montante, G.; Magelli, F. Dispersion Coefficients and Settling Velocities of Solids in Slurry Vessels Stirred with Different Types of Multiple Impellers. Chem. Eng. Sci. 2004, 59, 3081–3089. (15) Montante, G.; Lee, K. C.; Brucato, A.; Yianneskis, M. Numerical Simulations of the Dependency of Flow Pattern on Impeller Clearance in Stirred Vessels. Chem. Eng. Sci. 2001, 56, 3751–3770. (16) Galletti, C.; Brunazzi, E. On the Main Flow Features and Instabilities in an Unbaffled Vessel Agitated With an Eccentrically Located Impeller. Chem. Eng. Sci. 2008, 63, 4494–4505. (17) Stoller, M.; Bravi, M.; Chianese, A. The Effect of Scaling on the Performances of a DTB Crystallizer by means of a Compartmental Simulation Model. Chem. Eng. Res. Des. 2005, 83, 126–132. (18) Sotowa, K.-I.; Hamamura, T.; Taniguchi, H.; Kano, M.; Hasebe, Sh.; Hashimoto, I. Influence of the Incomplete Dissolution of Fines in the Fines Dissolver on the Optimal Operation of Industrial Continuous DTB Crystallizers. Powder Technol. 2001, 121, 93–98. (19) Menon, A. R.; Kramer, H. J. M.; Grievink, J.; Jansens, P. J. Modelling the Cyclic Behaviour in a DTB Crystallizer---a Two-Population Balance Model Approach. J. Cryst. Growth 2005, 275, 1373–1381.
ReceiVed for reView April 1, 2010 ReVised manuscript receiVed August 10, 2010 Accepted August 26, 2010 IE100786F
