CFD DEM simulations of a fluidized bed crystallizer Accepted Manuscript CFD DEM simulations of a fluidized bed crystallizer Kristin Kerst, Christoph Roloff, Luís G Medeiros de Souza, Antje Bartz, Andr[.]
Accepted Manuscript CFD-DEM simulations of a fluidized bed crystallizer Kristin Kerst, Christoph Roloff, Luís G Medeiros de Souza, Antje Bartz, Andreas Seidel-Morgenstern, Dominique Thévenin, Gábor Janiga PII: DOI: Reference: S0009-2509(17)30081-7 http://dx.doi.org/10.1016/j.ces.2017.01.068 CES 13415 To appear in: Chemical Engineering Science Received Date: Accepted Date: July 2016 29 January 2017 Please cite this article as: K Kerst, C Roloff, L.G Medeiros de Souza, A Bartz, A Seidel-Morgenstern, D Thévenin, G Janiga, CFD-DEM simulations of a fluidized bed crystallizer, Chemical Engineering Science (2017), doi: http://dx.doi.org/10.1016/j.ces.2017.01.068 This is a PDF file of an unedited manuscript that has been accepted for publication As a service to our customers we are providing this early version of the manuscript The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain CFD-DEM simulations of a uidized bed crystallizer Kristin Kerst , Christoph Rolo , Lus G Medeiros de Souza , Antje Bartz , Andreas Seidel-Morgenstern , Dominique Thevenin , Gabor Janiga a a b,c a a c a, a Institute of Fluid Dynamics and Thermodynamics, University of Magdeburg \Otto von Guericke", Magdeburg, Germany b Institute of Process Engineering, University of Magdeburg \Otto von Guericke", Magdeburg, Germany c Max Planck Institute for Dynamics of Complex Technical Systems (MPI), Magdeburg, Germany Abstract In the present study, important features of the two-phase ow in a uidized bed crystallizer are examined by numerical computations and companion experiments The simulations are carried out using a coupled CFD-DEM approach (CFD: Computational Fluid Dynamics; DEM: Discrete Element Method) After validating an open-source CFD-DEM software tool for this purpose, regions within the crystallizer with unfavorable hydrodynamic features and thus a negatively impacted process outcome have been identi ed This was rst accomplished by single-phase CFD simulations Then, the validated CFD-DEM model delivers valuable information that is dicult or even impossible to measure experimentally with sucient accuracy, such as the velocity and position of uidized crystals within the crystallizer Since the simulations are computationally challenging, a compromise between simulated process time and number of simulated particles must be found Hence, the CFD-DEM simulations are not utilized to simulate the whole crystallization process, but to examine a short time-window in detail Corresponding ndings rm proper uidization of the Corresponding author Email address: janiga@ovgu.de (Gabor Janiga) Preprint submitted to Chemical Engineering Science February 4, 2017 crystals support the model reduction carried out in a parallel project Keywords: 10 11 12 13 14 15 16 17 18 CFD, DEM, uidized bed, crystallizer, Shadowgraphy Introduction Crystallization is an important process in the chemical, pharmaceutical, and food industry and is the subject of many current research projects, e.g., [1, 2, 3] In solid/liquid crystallizers, crystal growth is greatly in uenced by the hydrodynamics [4, 5] In comparison with solid-gas uidized beds [6, 7], studies pertaining to solid-liquid uidization are relatively scarce in the scienti c literature [5, 8] A uidized bed crystallizer is examined in the present study, with the ultimate objective of optimizing set-up and process conditions The process under investigation aims to realize the synthesis of a desired enantiomer as well as the separation of racemic mixtures by selective crystallization [9, 10] Enantiomers are chemical molecules that spatially behave like mirror images of each other They have identical physicochemical properties, but di er in their behavior in a chiral environment (e.g., in living organisms) As such, the use of pure enantiomers plays a particularly important role in life-science industries Under certain conditions, crystallization can be used for enantiomer extraction It is particularly advantageous that crystallization is both highly selective, thus leading to high purities, while simultaneously allowing a targeted particle design To produce ne chemicals, batch processes are typically used However, in order to ensure constant product quality, but also to allow for larger production capacities (e.g., to produce aminoacids), continuous crystallization processes would be a very attractive alternative 19 20 The development and optimization of a continuous crystallization process is currently the subject of intense research activities in our group, as described in particular in [11, 12, 13, 14] Figure shows the basic structure of the novel operating process Figure 1: Principle of the novel continuous crystallization process for the synthesis of a desired enantiomer 21 22 23 24 25 26 27 28 29 30 31 Two identically-built crystallizers constitute the main part of the process The two crystallizers are connected in parallel and a reaction runs synchronously in both crystallizers This allows producing simultaneously two outlet streams; each contains just a single enantiomer Although typically only one of the enantiomers is the target, the continuous production of the second one is also very attractive The counter-enantiomer can be applied for other applications (e.g., as a building block for further chiral molecules) or it can be racemized and recycled The crystallizers are continuously supplied with a saturated solution of a racemic solid via a feed tank The inlet is located at the bottom of the crystallizer To start the crystallization process, rst, neglecting the particles This corresponds to the start-up of the real crystallization process, before adding the seed crystals to the crystallizer For all of the ow simulations discussed in this paper, the open-source software OpenFOAM 2.2.x [35] was used Two similar gurations are considered Setup corresponds to the complete threedimensional geometry of the uidized bed crystallizer employed in reality The utilized block-structure computational mesh is illustrated for Setup in Fig The total crystallizer height is 1:18 m and the diameter of the central crystallizer at the product outlet nozzle is 0:03 m Figure 4: Employed geometry and block-structure computational grid Left: full view Right: details 191 192 193 194 The second geometry, denoted Setup 2, is similar to Setup but slightly simpli ed (no connecting tubes) and completely cylindrical, instead of conical It was used to validate the computational models by comparison with experimental data, and will be described in more detail in section 13 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 The uid (continuous) phase is water, hence incompressible Additionally, the ow is always considered to be laminar, since the maximum diameter-based Reynolds number in both setups, found at the smallest cross-section near the inlet of Setup (bottom part of Fig 4), is 1710, and thus far below commonly accepted values characterizing the onset of transition to turbulence in pipe ows; further above, the Reynolds number is obviously much lower, due to the increased crystallizer diameter and to the extraction of part of the liquid to the ultrasonic bath In order to check grid independency for CFD, a systematic simulation of Setup with an injected volume ow rate of 25:1 L/h was performed on successively re ned grids involving 30 080, 101 250, 240 640 and 470 000 nite-volume cells, respectively Compared to the reference solution (that on the nest grid), the obtained volumetric ow rate with only 30 080 cells shows a relative error of 2:6 % concerning the ow-rate obtained in the cross-section at a column height of 0:5 m (column mid-height), at the level where particles will later be injected in companion experiments This di erence is still too large for accurate simulations On the other hand, the grid with 240 640 cells leads to a relative error below 0:2 % compared to the reference, which is fully acceptable A resolution with at least 240 640 cells has thus been retained for the later simulations The obtained cell size was then kept identical when simulating the full-scale apparatus, Setup Note that the number of grid cells needed for Setup is slightly less than for Setup 2, due to the conical arrangement 14 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 2.3.2 CFD-DEM Simulations In order to save computational time, each CFD-DEM simulation is systematically initialized with the pressure and velocity elds obtained from a previous CFD simulation at steady-state for the continuous (liquid) phase, without particles The CFD-DEM simulations are performed using the open-source software CFDEMcoupling [36] The software CFDEMcoupling was developed by combining two well-known existing, open-source software (written in C++): OpenFOAM (for CFD) and LIGGGHTS (for DEM) The solution is obtained by combining the uid (CFD) and particle (DEM) calculations using these two separate codes The interaction is realized by exchanging relevant information with a prede ned time step, as described in [37] For the CFD simulation the transient solver pisoFoam of OpenFOAM is used As its name states, this solver relies on the PISO algorithm (Pressure Implicit with Splitting of Operator) for pressure-velocity coupling The real process in Setup involves asparagine monohydrate crystals in the crystallizer The cumulative crystal size distribution (CSD) function Q of the asparagine crystals during process time when crystals are grown to the desired crystal size distribution was determined by optical measurement with a CAMSIZER XT (Retsch GmbH, Haan, Germany) and is depicted in Fig This CSD will be implemented in Setup 1, considering 200 000 crystals For the validation experiments in Setup 2, 100 000 particles are injected in reality, and this number is retained in the CFD-DEM simulations All further parameters for the CFD-DEM simulations of both setups are shown in Table (left column for Setup 2, right column for Setup 1) 15 CSD will be implemented in Setup 1, considering 200 000 crystals For the validation experiments in Setup 2, 100 000 particles are injected in reality, and this number is retained in the CFD-DEM 195 simulations, as well All further parameters for the CFD-DEM simulations of both setups are shown in Table ?? (left column for Setup 2, right column for Setup 1) 100 Q3 (%) 80 60 40 20 10 100 1,000 d (µm) Figure 5: Measured cumulative particle size distribution function Q3 of the asparagine crystals in Setup Figure 5: Measured cumulative particle size distribution function Q3 of the asparagine crystals in Setup 235 236 237 238 239 240 241 242 243 244 245 246 247 10 Validation of the CFD-DEM Simulation Model Due to the complexity of the resulting simulation model, a proper validation of the developed approach is absolutely necessary Here, this is performed with dedicated experiments in Setup For this purpose, inert glass particles were injected into the ow Obviously, such particles will not grow, but this is sucient to review particle trajectories and verify that the numerical model is able to reproduce the correct uidization behavior The employed glass particles particle size distribution (PSD), Q , obtained by the CAMSIZER XT (Retsch GmbH, Haan, Germany) is depicted in Fig For the CFD-DEM simulation in this case, the same PSD has been discretized by 22 equidistributed classes In the simulation all particles are initially injected at mid-height of the column using volume injection, which is similar to the experiments The identi cation of a suitable measurement technique to inspect the particle velocities proved to be a major challenge in the end All available in-line measurement probes often 16 Figure 6: Measured cumulative particle size distribution function Q3 of the glass beads in Setup 248 249 250 251 252 253 254 255 256 257 258 259 260 used in real crystallizers (e.g., Endoscopy, Focused Beam Re ectance) were found to noticeably impact the hydrodynamics inside the crystallizer, and were thus rejected As the walls are transparent (glass), using non-intrusive, image-based measurement techniques from outside would, in principle, be possible Particularly, the Shadowgraphy technique, which is able to track suspended particles by recording their shadow images, seemed promising in this case However, with the real crystallizer design in Setup 1, employing this method with sucient quality proved impossible due to severe light refractions from the crystallizer jacket, since the crystallizer uses a thick double-jacket for temperature control Light refraction taking place at four strongly curved interfaces (from outside to inside: air - glass water - glass - water) leads to an unacceptable distortion of the light beams To solve this problem, a second column was built (see Setup shown in Fig 7, left), with dimensions slightly smaller but similar to the real crystallizer in Setup It consists of a single-jacket column made of Makrolon material with an inner diameter of 26 mm and 17 261 262 263 264 265 266 267 268 a height of 1050 mm This cylindrical tube is embedded into a surrounding, transparent square-shaped duct with a side width of 100 mm The space between the cylindrical column and the surrounding duct is lled with water Hence, the change in refractive index is now limited to the small di erence between water and the thin Makrolon wall, leading only to minor refraction, whereas the large di erence between refraction index of air and Makrolon is no longer of practical relevance, since the camera acquires images perpendicular to the at outer boundary of the square-shaped jacket Overall, an excellent signal quality is obtained in this manner in Setup Figure 7: Left: Model crystallizer column for validation; right: target for camera calibration 269 270 271 272 The Shadowgraphy measurement setup is depicted in Fig The column (1) is connected to a gear-pump (ISMATEC ISM405A) (5), which drives the ow in a controlled and constant manner from the lower (7) to the upper reservoir (8) The continuous ow rate is monitored by an ultrasonic ow meter (Sonotec, Halle, Germany; number 6) Homogeneous background 18 ... uidized bed crystallizer are examined by numerical computations and companion experiments The simulations are carried out using a coupled CFD- DEM approach (CFD: Computational Fluid Dynamics; DEM: ... back into the crystallizer and act as seeds Signi cant advantages of the process are the recycling of both 1) the undesired enantiomer, and of 2) all crystals that are not in the target size area... corresponding momentum balances serve as the basis in computing the translational and angular accelerations of each particle Summing up all forces acting on a particle, one obtains: F total 100 101 102