Article pubs.acs.org/IECR
3D Flow Reactors: Flow, Hydrodynamics, and Performance Mrityunjay Sharma, Venkateswara Reddy S., and Amol A. Kulkarni* Chemical Engineering and Process Development Division, CSIR-National Chemical Laboratory, Pune 411021, India ABSTRACT: A device comprised of a sequence of converging or diverging units aligned either in an axisymmetric or nonaxisymmetric manner can be used as a continuous flow reactor. Here we report the analysis of flow and hydrodynamics (pressure drop, residence time distribution, and mass transfer) for an axisymmetric geometry of a 3D flow reactor for single phase and two-phase flows. CFD simulations of the single phase flow have been used for identification of the precise geometrical configuration. The sequence of converging units as a flow reactor has been found to always be better than the sequence of diverging units. The residence time distribution analysis also favored the choice of converging flow as a better option. The performance of the device was verified by successfully carrying out a highly exothermic two-phase aromatic nitration of benzaldehyde (ΔHr ≈ −172 kJ/mol) with fuming nitric acid.
1. INTRODUCTION Continuous flow miniaturized process devices (microreactors or millireactors) are now well accepted as an important process intensification option. Although more emphasis has always been for continuous flow synthesis, continuous separation and integration are also practiced to some extent.1 Typically, microreactors or miniaturized flow reactors offer better performance than the conventional process equipment and are more efficient.2−6 Smaller dimensions and geometric features in a confined domain help achieve rapid mixing, a relatively high heat transfer area, an at least 1 order of magnitude higher mass transfer coefficient, and a narrow residence time distribution.7 These features allow the reactions to be conducted in a very different range of conditions than the conventional operation range, also known as novel process windows.8 For specific reactions (viz. sulfoxidation,9 halogenations,10 ozonolysis,11 nitration,12,13 catalytic hydrogenation,14 Grignard exchange reaction,15 etc.) that are fast and exothermic miniaturization helps to overcome the transport limitations. For a reaction under optimal conditions, further enhancements in the capacity or production rate are achieved by numbering up16 and also by increasing the throughput while retaining the miniaturized features. A vast body of literature is available on the microreactors, their advantages, and their performance for a variety of reactions for synthesis of chemicals and nanomaterials.8,17−19 Many reports exist on the use of miniaturized devices for pilot scale.15,20,21 For a single phase/homogeneous reaction, rapid mixing and high heat transfer rates are used as the basis for design. For a two phase reaction, the design objective, in addition to the above, aims to achieve very high heat and mass transfer rates and a narrow residence time distribution. A few recent studies on the hydrodynamics of multiphase flow reactors are useful for their application for production of chemicals.22−28 However, a large number of possibilities for the design of flow reactors do exist, and corresponding hydrodynamics and performance need to be evaluated to compare them with the known options. In view of this, here we report high throughput miniaturized flow reactors for carrying out exothermic chemical reactions. The proposed flow reactor is comprised of a sequence of © 2014 American Chemical Society
converging or diverging units aligned in an either axisymmetric or nonaxisymmetric manner. This configuration is among one of the MAGIC devices (modular, agile intensified, and continuous)29 that aim at developing high performance low cost devices for processing of fine chemicals and pharmaceutical intermediates. Here we bring out detailed analysis of hydrodynamics of a sequence of converging or diverging units that are used as three-dimensional (3D) flow reactors, referred hereafter as 3D Flow Reactors. Depending upon the geometry of the device, the flow pattern will have a three-dimensional nature and will help use more volume for processing than the conventional miniaturized devices and flow reactors that are after 2D. Initial analysis based on the CFD simulations has been used for finalizing the geometry, which is further explored for understanding the hydrodynamics (pressure drop, residence time distribution and mass transfer). Upon finalizing the geometry, a highly exothermic aromatic nitration30 (heat of reaction ∼−172 kJ/mol) involving fuming nitric acid was studied to produce the desired isomer at a scale of 10 g/min. The analysis of the nitration step clearly indicated that simple geometries and a detailed understanding of the flow can yield a very feasible and economically viable continuous flow reactor.
2. GEOMETRICAL DESIGN PARAMETERS: FLOW SIMULATIONS In order to fix the geometry of the converging or diverging units, initially flow simulations of different geometries were carried out. Finite element based approach using Comsol Multiphysics was used for generating the geometry of the device. In order to capture the finest aspects of the flow and to avoid any issues related to the grid size, an extremely fine grid option was used with element size between 0.0082 and 0.0715 mm and with a maximum element growth rate of 1.08. 2D as well as 3D geometries were prepared to compare the nature of flow. Depending upon the geometry the grid for every Received: Revised: Accepted: Published: 1916
July 19, 2013 December 20, 2013 January 2, 2014 January 2, 2014 dx.doi.org/10.1021/ie402311y | Ind. Eng. Chem. Res. 2014, 53, 1916−1923
Industrial & Engineering Chemistry Research
Article
converging unit had 32 500 to 52 600 elements with over 91% triangular elements. The typical computational domain and the sectional view of the computational grids are shown in Figure 1.
Figure 2. Schematic of the experimental setup. (1) Flow reactor, (2 and 3) dosing pumps, (4) RTD probes, (5) data acquisition and transmission system, (6) digital manometer, and (7) laptop for data acquisition and processing.
both cases for exploring their respective hydrodynamics. The number of converging or diverging units in a sequence was varied in the range of 2−20. Milli-Q water (Millipore) was used for single phase experiments. A water−kerosene system was used as the immiscible nonreacting liquid−liquid system with the properties shown in Table 1. The two immiscible liquids, aqueous and organic, were pumped at constant flow rate using a dual syringe pump (Longer Pumps, China) with two syringes of 50 mL each on each pump. Experiments were conducted at room temperature (∼22 °C) at different water (0−80 mL/min) and kerosene flow rates (0 − 80 mL/min). The outlet was kept open to atmosphere. 3.2. Measurement Techniques. 3.2.1. Pressure Drop. Pressure drop across the device was measured using a digital manometer (AZ Instruments P8100) for a different flow rate range of 0.5−20 mL/min for both configurations. The data acquisition was done online so that the transient variations could be captured. 3.2.2. Measurement of Mass Transfer Coefficient. Propionic acid−water−kerosene system was chosen for the measurement of the overall mass transfer coefficient. Transfer of propionic acid from water to kerosene was monitored for different flow rates. The aqueous phase was prepared by mixing 20 mL of propionic acid in 10 L of water and this stock solution was used for the all of the experiments. After making the propionic acid stock solution conductivity of this solution was measured using a conductivity meter and the consistency in the conductivity of the solution was monitored to ensure the initial concentration. A dual syringe pump was used with syringes filled with the above stock solution and kerosene. Experiments were carried out over a range of 8−40 mL/min for both configurations (sequence of converging plates and diverging plates). At every flow rate, a 20 mL sample was collected at the outlet for conductivity measurement after the phases got separated by gravity. The conductivity data was converted in terms of concentration using a calibration chart. The mass transfer coefficient was estimated as
Figure 1. Typical device grid structure in 2D and 3D.
In order to take into account the boundary layer separation, eight grid layers were considered with a boundary layer stretching factor of 1.2. The laminar flow model for incompressible fluids was used. A standard set of Navier− Stokes’ equations was used for the steady state simulations of flow in the geometry. Water was used as the fluid for simulation purposes. Fluid velocity was used as the inlet boundary condition, while the outlet was given the pressure boundary condition (as atmospheric pressure). The simulations were carried out to estimate the pressure-drop across the entire system and also to see the variation in the steady state velocity contour plots to visualize the flow patterns. Detailed CFD modeling and study of the steady state and dynamic flow pattern in the sequence of such converging or diverging units will be discussed separately. A few simulations were also carried out in the three-dimensional grid for the reactor with four converging units for the above flow rate range, and there was a deviation of ±4.7% in the pressure drop.
3. EXPERIMENTAL SECTION 3.1. Experimental Setup. A schematic view of the experimental setup is shown in Figure 2. The system consisted of a 3D flow reactor, two syringe pumps, a digital pressure manometer, a pair of RTD probes (made of copper wires with tip size of 0.2 mm inserted at the exit of the flow reactor), and a data acquisition system connected to a computer. A typical 3D flow reactor was comprised of a sequence of converging or diverging conical volumes having fixed convergence/diverging angles and inlet−outlet diameters, respectively. Each conical section has a volume of 0.24 mL. A schematic of the unit and the photographs of the device are shown in Figure 3. Experiments were carried out for both converging and diverging orientation of the sequence. The flow reactor used for the hydrodynamic studies was made of transparent PMMA. The visibility of the object was enhanced using back illumination and a suitable light diffuser. However the gradually varying thickness of the PMMA device did not allow a clear image of the two phase flow inside the device to be obtaining. An identical range of experimental conditions was adapted for
KLa =
(C − C*) 1 ln in t (C0 − C*)
where Cin and Co are the concentrations of propionic acid in aqueous phase at the inlet and outlet respectively, C* is the 1917
dx.doi.org/10.1021/ie402311y | Ind. Eng. Chem. Res. 2014, 53, 1916−1923
Industrial & Engineering Chemistry Research
Article
Figure 3. (A) Schematic of the sequence of miniaturized converging units with dimensions of each unit, (B) photograph of the assembly of four units in PMMA and in SS316, and (C) photograph of the assembly with 24 plates.
Table 1. Fluid Properties of Kerosene and Water at 20 °C33 fluid
density (kg/m3)
viscosity (Pa·s)
interfacial tension (kg/s2)
kerosene water
780 - 790 998.2
0.00162 0.001
0.05
ratio of the active zone to inactive zone was a function of the expansion angle as well as the inlet Re. The pressure drop was used for estimating the power consumption per unit volume (Pw = ΔPQ/V), where V = VC/DNC/D. The data plotted in Figure 3B showed that, although the absolute pressure drop for larger geometry ratio was higher, the power consumption per unit volume (P/V) showed an opposite trend. The comparison was made in terms of P/V as it gives the realistic measure of performance of such a device at a given inlet Re. Thus while lower power consumption may not yield the desired extent of mixing, an intermediate geometry that does not yield large dead zones was more suitable. It was also noted that the pressure drop for both the cases of sequence of converging or diverging units was almost identical. Based on the pressure drop per unit volume for a given geometry, the geometry with an expansion ratio of 5.5 was chosen and used for fabrication. 4.2. Hydrodynamics. 4.2.1. Pressure Drop. The pressure drop measured across a sequence of four units for single phase flow is shown in Figure 4A. It can be seen that the experimental and simulated data match beyond Re > 100. The simulations at low Re could not capture the developing flow properly. It is also possible that some part of the device might have trapped air in the corners, which will actually reduce the active volume and thereby enhance the pressure drop. The velocity contour plots in every third converging or diverging unit at different inlet flow rates (0.5, 5, 10, and 20 mL/min) are shown in Figure 5B,C. The plots show that, below Re < 100, the flow is in the developing region and streamlines are close to independent. With increasing Re, the flow circulation sets in and the intensity of the vortices continues to increase. In general, the pressure drop in diverging sections was always slightly higher than the converging sections. For both the configurations, the jet length increased from inlet to outlet with increasing flow rates. Increasing flow rates set strong three-dimensional circulatory flows in the confined geometry that lead to increasing intensity of mixing. Although the circulatory flows were stronger for the sequence of diverging units, much of the fluid actually rushed through the central core thereby yielding a shorter path. In the converging sections, a second vortex was seen, which actually rotates in the opposite direction thereby increasing the frictional resistance. Thus with increasing inlet flow rates, distinct vortices set-in and three distinct flow regions in a converging unit are observed. One smaller segment acts as a plug flow where a fraction of the fluid entering the unit directly
equilibrium concentration of the propionic acid in the aqueous phase, t is the residence time and kLa is the overall mass transfer coefficient. 3.2.3. Residence Time Distribution. RTD measurements were done with 4 and 20 plates (Figure 4B,C) for both configurations of converging and diverging sequences. A salt solution in water (NaCl 5 g/L) was used as tracer, and its concentration was measured in terms of conductivity using an inline conductivity meter at the reactor outlet. Step input approach was used for tracer injection and water flow rate was varied over a range (0.5−40 mL/min). Two different syringe pumps (Longer Pumps, China) were used for the tracer and water. Conductivity probes were attached at the outlet to measure the conductivity with respect to time. Upon studying the hydrodynamics of the sequence of converging units and diverging units separately, the best configuration was used for exploring the performance of this device for a fast and exothermic reaction. The details of the experiments and analysis are discussed in detail in section 4.3.
4. RESULTS AND DISCUSSIONS 4.1. CFD Simulations of Single Phase Flow: Design Selection. Initially, the simulations were carried out to explore the flow inside a sequence of converging or diverging units and identify the geometry that performs better. Comsol Multiphysics (version 4.2) was used for these simulations. A two dimensional grid of the flow reactor was generated with the number of mesh elements varying in the range of 0.58−1.5 million cells. Initially simulations were carried out for five different geometries having different ratios of maximum to minimum (i.e., the inlet) diameter of the conical section in the range of 2−20. In all geometries the inlet diameter (1 mm) and height of each unit (7 mm) were maintained constant.31 This yields different angles of contraction/expansion. The typical geometries are shown in Figure 3A. Initially simulations were carried out to identify the suitable angle of expansion. Water was used as fluid and the pressure drop across a sequence of 4 units was monitored (not shown here). In general, the volume 1918
dx.doi.org/10.1021/ie402311y | Ind. Eng. Chem. Res. 2014, 53, 1916−1923
Industrial & Engineering Chemistry Research
Article
The pressure drop for the case of two phase flow (kerosene− water) was measured over a range of flow rates for sequence of 4 and 20 converging and diverging units independently (Figure 6). It was observed that the pressure drop was higher than the single phase flow and the predictions based on the Lockhart− Martinelli method32 (for Chisholm parameter C = 0.05) did not show a good match with the experimental data. This observation was consistent over a wider range of values of C. For the two phase flow, ΔPC was always lower than ΔPD, however at Re ≈ 1000 the values were similar. Similar to the trends observed for single phase flow, the two-phase pressure drop showed two different trends (for the Re below and above 100). In the absence of a flow visualization technique that would help us overcome the diffusion of light in the conical surface, it was difficult to measure the nature of two-phase flow inside the system to explore the reason for such trends. 4.2.2. Residence Time Distribution. The conductivity data in terms of output from the step change was used for the estimation of mean residence time, variance and the dispersion coefficient. Open vessel boundary condition approach was used for analyzing the data and the axial dispersion model was used for the estimation of dispersion coefficient. Typical E curves at three different flow rates for sequence of four diverging units are shown in Figure 7A, and there is a reasonably good comparison that can be used for estimation of dispersion coefficient for further analysis. The data in terms of vessel dispersion number (D/uL) for different Re (calculated based on the velocity at the smaller diameter of conical section) is shown in Figure 7B. Over the entire range of Re under analysis, the vessel dispersion number for the sequence of diverging units was always smaller than that of the sequence of converging units. This implies that at identical flow rates the sequence of diverging units shows lesser dispersion and hence more of a plug flow nature. For the case of a sequence of converging units in every single unit, there is excellent mixing thereby achieving a series of CSTRs in true sense. In the diverging units, in addition to the main stream, a confined single large vortex was observed which retained some tracer away from the main stream and this yielded the mean residence time much longer than the actual residence time. 4.2.3. Mass Transfer. The mass transfer coefficient for the case of two phase flow was also measured as discussed in section 2. Over a range of two-phase Re (100−1040), the overall mass transfer coefficient for a sequence of converging units and diverging units was seen to vary in the range of 2.4− 7.9 (±8%) 1/s and 2.2−5.34 (±9.6%) 1/s, respectively, for a range of power consumption per unit volume from 0.9 to 174.7 kW/m3. When compared with the typical microreactors and other flow reactors reported for liquid−liquid dispersion,22 these values are comparable for the range of power consumption per unit volume. These observations along with the analysis of pressure drop data and the RTD analysis thus support the use of sequence of converging units as a 3D flow reactor. On the basis of these observations from the measured hydrodynamic parameters, in the next section, we extend this study by actually performing a fast and exothermic reaction in a sequence of converging units (Figure 3D). 4.3. Case Study for Performance Evaluation of 3D Flow Reactor. Nitration of benzaldehyde is a two phase liquid−liquid reaction between benzaldehyde and a nitrating agent (usually the mixture of sulfuric acid and nitric acid (70%)/fuming nitric acid (>98%)). We chose this reaction for checking the efficiency of the 3D flow reactor. Conventionally,
Figure 4. (A) Geometry of the sequence of converging units used for simulations in Comsol. The numbers next to the geometry show the ratio of the largest diameter of a converging unit with that of the smaller diameter. (B) Variation in the power consumption per unit volume of the sequence of converging and diverging units over a range of Re. The legends Con and Div indicate converging and diverging sections, respectively.
leaves through the exit. The second zone corresponds to a larger vortex which is in direct contact with the central core and in the third zone a smaller vortex is seen close to the corner away from the main stream. The pressure drop for four units and two units was multiplied by 6 and 12, respectively and compared with the pressure drop for 24 plates. The pressure drop for 24 plates was always higher (and the difference increased with increasing Re) and showed a nonlinear dependence on the number of units (N). For Re > 1000, the ΔP was seen to follow a simple correlation as :ΔP = (AN2 + BN)Re, with A = 0.019 and 0.022 for the sequence of converging and diverging units, respectively, and B = 0.26 for both the sets. Thus, the pressure drop was seen to scale nonlinearly with N and linearly with the Re. 1919
dx.doi.org/10.1021/ie402311y | Ind. Eng. Chem. Res. 2014, 53, 1916−1923
Industrial & Engineering Chemistry Research
Article
Figure 5. (A) Pressure drop across a sequence of four converging and diverging units. Symbols indicate the experimental data while the lines correspond to the simulated values. Dotted line corresponds to the pressure drop in a straight tube (i.d. = 1 mm) estimated using the Hagen− Poiseuille equation. The streamlines shown for half geometry are for the flow from left to right. Velocity contour plots (axisymmetric) for the (B) converging and (C) diverging units at the smallest and largest flow rates. Regions of maximum velocity (Vmax = 0.014, 0.158, 0.316, and 0.633 m/s from top to bottom, respectively) are shown in red. In all cases the flow direction is from left to right. (D) Parity plot based on four units and two units to verify linear increase in pressure drop.
In view of this, here the experiments were carried out using fuming nitric acid alone. Our previous experience of this reaction helped decide the experimental setup and the conditions.12 Experiments were carried out with different device configurations (viz. the converging flow reactor alone, the flow reactor followed by a residence time tube, a simple T followed by residence time tube, etc.). For the later two cases, care was taken to have identical volumes of the individual assemblies so that the velocity (and hence the Re and the residence time) remained constant. The whole assembly was kept in a constant temperature bath. The performance of the
nitration of benzalhdehyde is carried out in batch mode by very slow addition of the nitrating agent to the substrate under stirring. The reaction is carried out in the temperature range of −5 to +15 °C with an addition time of about 6 h and a reaction time of 2−6 h.33 The product is comprised of mononitro derivatives with 4:1 mol ratio of meta to ortho isomers and trace quantities of benzoic acid. The isolation procedure involves neutralizing the spent acid which is not an environmentally friendly situation. Different nitrating agents can help change the product composition due to variation in the concentration of nitronium ions and the prevailing mechanism. 1920
dx.doi.org/10.1021/ie402311y | Ind. Eng. Chem. Res. 2014, 53, 1916−1923
Industrial & Engineering Chemistry Research
Article
Figure 8. Variation in the % yield of m-nitro benzaldehyde and the isomer ratio at different flow rates at 40 °C. Filled symbols show the mole ratio (m:o) of the mononitro isomers, and the open symbols show the % yield of the desired m-nitro benzaldehyde.
Figure 6. Pressure drop (ΔP) for the kerosene−water system for sequence of 4 and 20 converging (Con) and diverging (Div) units.
much higher than the conventionally observed mole ratio of 4. In general the yield of meta isomer increased with residence time, and it was always higher where the mixing was efficient. With the 3D flow reactor followed by a small tubular reactor complete conversion of the substrate was found in 15 min and the yield of meta isomer was higher than the simple T-mixer followed by tubular reactor having same volume. Thus a simple geometry such as a 3D flow reactor shown here helps to achieve a smaller foot print, lower capital cost, and yet an excellent performance due to the nature of flow that helps create good dispersion. Moreover the device can withstand high pressure and temperatures up to 200 °C, which brings out a robust option for process development and further capacity enhancement by numbering-up. 4.4. Design Options for 3D Flow Reactor. The flow in the sequence of converging units is largely periodic, and the
reaction was studied at two different mole ratios of fuming nitric acid to benzaldehyde (i.e., 3.1:1 and 7:1) over a temperature range from 10 to 50 °C. The residence time was varied over a range of 180−930 s. A 5 mL sample was collected at the outlet for every set of conditions, which was quenched in 25 g of ice. After quenching, the product was extracted using 20 mL of toluene in three parts 10, 6, and 4 mL. Traces of nitric acid from the extracted sample were removed by water washing (two times) and then by a salt solution to make samples suitable for the GC analysis. An internal standard was used for confirmation of consistency in the analysis. Every sample was run at least twice to know and quantify the extent of variation. The results at 40 °C are shown in Figure 8. The % yield of the meta isomer was always higher than that of the ortho isomer. Importantly, the mole ratio of m:o isomers in the product was found to vary between 7.41 and 10.24, which is
Figure 7. (A) Typical E(θ) curves from experiments (open symbols) and axial dispersion model (filled symbols) for a sequence of four diverging units for three flow rates. (B) Variation in the vessel dispersion number (D/uL) for a sequence of 20 units. 1921
dx.doi.org/10.1021/ie402311y | Ind. Eng. Chem. Res. 2014, 53, 1916−1923
Industrial & Engineering Chemistry Research
Article
Figure 9. Streamlines for various alternatives of arrangement of converging and diverging units. In all cases the flow is from top to bottom. All simulations are done for a flow rate of 20 mL/min for water as fluid. (A and B) Sequence of axisymmetric converging and diverging sections with center-line connections, (C and D) sequence of axisymmetric converging and diverging sections with off-center connections, (E) sequence of skewed converging sections with all connections along the same line, and (D) sequence of skewed converging sections with cross-diagonal connections.
increase the yield of the desired isomer, while use of fuming nitric acid alone helped increase the selectivity of the metaisomer. The demonstration of the nitration reaction clearly indicated that the devices with simple geometries and a good understanding of the flow can yield a very feasible and economically viable continuous flow reactor option.
mixing can be further improved by inducting perturbations in the flow through geometrical variations. A few such design variations and the simulated streamlines for single phase flow at a constant inlet velocity are shown in Figure 9. More work on a detailed analysis and comparison with the axisymmetric sequence of converging units is in progress. It is necessary to highlight that while for exothermic reactions 2D geometries of flow reactors offer high heat transfer area, the scale-up through numbering-up using 2D geometries is lot more costly. In the 3D flow reactor (e.g., considered in this study with a maximum to minimum diameter ratio of 5.5), while the heat transfer area is 20% of that of a 2D flow reactor (having identical crosssection and aspect ratio), the volume is 6 times larger. This means that while the heat transfer coefficient can be enhanced by increasing the flow rates (to compensate for smaller heat transfer area) simple devices can process more volume. In reality, the present geometry offers a heat transfer area >770 m2/m3, which is sufficient to conduct exothermic reactions as demonstrated here. Moreover the individual converging or diverging unit in an assembly, if required, can be replaced, which thus offers better flexibility. Pilot scale studies using a scaled-up configuration of such systems are in progress and the details will be reported separately.
■
AUTHOR INFORMATION
Corresponding Author
*Phone: 00-91-20-25902153. Fax: 00-91-20-25902621. E-mail:
[email protected]. Notes
The authors declare the following competing financial interest(s): The said device and its variants have been disclosed as 2011-NF-0164.
■
ACKNOWLEDGMENTS V.S.R. thanks DST for financial support (DST/SR/S3/CE/ 0032/2010). All authors thankfully acknowledge the financial support from Indus Magic Program of CSIR’s 12th FYP.
■
REFERENCES
(1) McMullen, J. P.; Jensen, K. F. Integrated Microreactors for Reaction Automation: New Approaches to Reaction Development. Ann. Rev. Anal. Chem. 2010, 3, 19−42. (2) Ehrfeld, W.; Hessel, V.; Kiesewalter, S.; Lowe, H.; Richter, T.; Schiewe, J. Implementation of microreaction technology in process engineering. Microreact. Technol.: Ind. Prospects 2000, 14−34. (3) Hessel, V. From microreactor design to microreactor process design. Chem. Eng. Technol. 2005, 28 (3), 243−243. (4) Jensen, K. F. Microreaction engineering - is small better? Chem. Eng. Sci. 2001, 56 (2), 293−303. (5) Kockmann, N.; Roberge, D. M. Harsh Reaction Conditions in Continuous-Flow Microreactors for Pharmaceutical Production. Chem. Eng. Technol. 2009, 32 (11), 1682−1694. (6) Mason, B. P.; Price, K. E.; Steinbacher, J. L.; Bogdan, A. R.; McQuade, D. T. Greener approaches to organic synthesis using microreactor technology. Chem. Rev. 2007, 107 (6), 2300−2318.
5. CONCLUSIONS A sequence of converging or diverging units aligned in axisymmetric manner has been used as a 3D flow reactor. CFD simulations of the single phase flow have been used for identification of the right geometrical configuration. The hydrodynamics (pressure drop, residence time distribution, and mass transfer) of this configuration are studied for single phase and liquid−liquid multiphase flows. The pressure drop for two phase flow was always higher than that for single phase flow. The sequence of converging units as a flow reactor has been found to be always better than the sequence of diverging units. The nitration of benzaldehyde could be successfully carried out using a 3D flow reactor. Enhanced mixing helped 1922
dx.doi.org/10.1021/ie402311y | Ind. Eng. Chem. Res. 2014, 53, 1916−1923
Industrial & Engineering Chemistry Research
Article
film reactor using laser scanning confocal microscopy. Exp. Therm. Fluid Sci. 2006, 30 (5), 463−472. (29) Olah, G. A.; Lin, H. C. Aromatic-Substitution 0.35. BoronTrifluoride Catalyzed Nitration of Benzene, Alkylbenzenes, and Halobenzenes with Methyl Nitrate in Nitromethane Solution. J. Am. Chem. Soc. 1974, 96 (9), 2892−2898. (30) Olah, G. A.; Narang, S. C.; Olah, J. A.; Lammertsma, K. Recent Aspects of Nitration - New Preparative Methods and Mechanistic Studies (a Review). Proc. Natl. Acad. Sci. U.S.A. 1982, 79 (14), 4487− 4494. (31) This height was retained due to the availability of PMMA sheets of this thickness so that the suitable geometry can be machined. In. (32) Lockhart, R. W.; Martinelli, R. C. Proposed Correlation of Data for Isothermal 2-Phase, 2-Component Flow in Pipes. Chem. Eng. Progress 1949, 45 (1), 39−48. (33) Icke, R. N.; Redemann, C. E.; Wisegarver, B. B.; Alles, G. A. mNitrobenzaldehyde Dimethylacetal. Org. Synth. Collect. 1955, 3, 644.
(7) Roberge, D. M.; Zimmermann, B.; Rainone, F.; Gottsponer, M.; Eyholzer, M.; Kockmann, N. Microreactor technology and continuous processes in the fine chemical and pharmaceutical industry: Is the revolution underway? Org. Process Res. Dev. 2008, 12 (5), 905−910. (8) Hessel, V. Novel Process Windows - Gate to Maximizing Process Intensification via Flow Chemistry. Chem. Eng. Technol. 2009, 32 (11), 1655−1681. (9) Noguchi, T.; Hirai, Y.; Kirihara, M. Highly selective 30% hydrogen peroxide oxidation of sulfides to sulfoxides using micromixing. Chem. Commun. 2008, No. 26, 3040−3042. (10) Pelleter, J.; Renaud, F. Facile, fast and safe process development of nitration and bromination reactions using continuous flow reactors. Org. Process Res. Dev. 2009, 13 (4), 698−705. (11) Wada, Y.; Schmidt, M. A.; Jensen, K. F. Flow distribution and ozonolysis in gas-liquid multichannel microreactors. Ind. Eng. Chem. Res. 2006, 45 (24), 8036−8042. (12) Kulkarni, A. A.; Kalyani, V. S.; Joshi, R. A.; Joshi, R. R. Continuous Flow Nitration of Benzaldehyde. Org. Process Res. Dev. 2009, 13 (5), 999−1002. (13) Yu, Z.; Lv, Y.; Yu, C.; Su, W., A High-Output Continuous Selective and Heterogeneous Nitration of p-Difluorobenzene. Org. Process Res. Dev.. (14) de Bellefon, C.; Lamouille, T.; Pestre, N.; Bornette, F.; Pennemann, H.; Neumann, F.; Hessel, V. Asymmetric catalytic hydrogenations at micro-litre scale in a helicoidal single channel falling film micro-reactor. Catal. Today 2005, 110 (1−2), 179−187. (15) Wakami, H.; Yoshida, J. Grignard exchange reaction using a microflow system: From bench to pilot plant. Org. Process Res. Dev. 2005, 9 (6), 787−791. (16) Iwasaki, T.; Kawano, N.; Yoshida, I. T. Radical Polymerization Using Microflow System: Numbering-up of Microreactors and Continuous Operation. Org. Process Res. Dev. 2006, 10 (6), 1126− 1131. (17) Ehrfeld, W. Design guidelines and manufacturing methods for microreaction devices. Chimia 2002, 56 (11), 598−604. (18) Gunther, A.; Jensen, K. F. Multiphase microfluidics: from flow characteristics to chemical and materials synthesis. Lab Chip 2006, 6 (12), 1487−1503. (19) Hartman, R. L.; McMullen, J. P.; Jensen, K. E. Deciding Whether To Go with the Flow: Evaluating the Merits of Flow Reactors for Synthesis. Angew. Chem., Int. Ed. 2011, 50 (33), 7502−7519. (20) Calabrese, G. S.; Pissavini, S. From Batch to Continuous Flow Processing in Chemicals Manufacturing. AIChE J. 2011, 57 (4), 828− 834. (21) Togashi, S.; Miyamoto, T.; Asano, Y.; Endo, Y. Yield Improvement of Chemical Reactions by Using a Microreactor and Development of a Pilot Plant Using the Numbering-Up of Microreactors. J. Chem. Eng. Jpn. 2009, 42 (7), 512−519. (22) Nieves-Remacha, M. J.; Kulkarni, A. A.; Jensen, K. F. Hydrodynamics of Liquid-Liquid Dispersion in an Advanced-Flow Reactor. Ind. Eng. Chem. Res. 2012, 51 (50), 16251−16262. (23) Wu, G. H.; Zhou, H. Z.; Zhu, S. L. Preparation of ultrafine barium sulfate in impinging streams microreactor. J. Inorg. Mater. 2006, 21 (5), 1079−1084. (24) Amador, C.; Wenn, D.; Shaw, J.; Gavriilidis, A.; Angeli, P. Design of a mesh microreactor for even flow distribution and narrow residence time distribution. Chem. Eng. J. 2008, 135S, S259−S269. (25) Hessel, V.; Angeli, P.; Gavriilidis, A.; Lowe, H. Gas-liquid and gas-liquid-solid microstructured reactors: Contacting principles and applications. Ind. Eng. Chem. Res. 2005, 44 (25), 9750−9769. (26) Kulkarni, A. A.; Gorasia, A. K.; Ranade, V. V. Hydrodynamics and liquid phase residence time distribution in mesh microreactor. Chem. Eng. Sci. 2007, 62 (24), 7484−7493. (27) Commenge, J. M.; Obein, T.; Genin, G.; Framboisier, X.; Rode, S.; Schanen, V.; Pitiot, R.; Matlosz, M. Gas-phase residence time distribution in a falling-film microreactor. Chem. Eng. Sci. 2006, 61 (2), 597−604. (28) Yeong, K. K.; Gavriilidis, A.; Zapf, R.; Kost, H. J.; Hessel, V.; Boyde, A. Characterisation of liquid film in a microstructured falling 1923
dx.doi.org/10.1021/ie402311y | Ind. Eng. Chem. Res. 2014, 53, 1916−1923