- Email: [email protected]
Design and application of a millistructured heat exchanger reactor for an energy-efficient process
Chemical Engineering and Processing 108 (2016) 109–116
Contents lists available at ScienceDirect
Chemical Engineering and Processing: Process Intensification journal homepage: www.elsevier.com/locate/cep
Design and application of a millistructured heat exchanger reactor for an energy-efficient process Sebastian Schwolowa , Andreas Neumüllera , Lahbib Abahmaneb , Norbert Kockmannc , Thorsten Rödera,* a b c
Mannheim University of Applied Sciences, Institute of Chemical Process Engineering, Paul-Wittsack-Straße 10, 68163 Mannheim, Germany 3M Technical Ceramics, Max-Schaidhauf-Str. 25, 87437 Kempten, Germany TU Dortmund, Biochemical and Chemical Engineering, Equipment Design, Emil-Figge-Straße 68, 44227 Dortmund, Germany
A R T I C L E I N F O
Article history: Received 30 March 2016 Received in revised form 15 July 2016 Accepted 27 July 2016 Available online 9 August 2016 Keywords: Heat integration Flow reactor Milli- and microreactors Process intensification Energy efficiency Michael addition
A B S T R A C T
Flow chemistry in milli- and microstructured reactors exhibits great potential for process intensification. In the present work, this potential has been demonstrated through the process development for a solvent-free production process including a Michael addition and following product purification. Process simulation was used to maximize the material and energy efficiency of the overall process by recycling unconverted reactants and a catalyst (water) and by utilizing heat from the exothermic reaction to substitute external energy supply in the heat demanding process steps. As a proof-of-concept experiment, millistructured equipment was designed, manufactured and tested at a laboratory scale. A three-stream counter-current heat exchanger for reactant preheating and a plate heat exchanger reactor with zigzag reaction channels were investigated regarding the maximum transfer of reaction heat available for further process steps via a heat carrier cycle. The experiments showed stable reactor control in steady state operation and efficient heat transfer with a small driving temperature difference at the outlet of the heat exchanger reactor. ã 2016 Elsevier B.V. All rights reserved.
1. Introduction In pharmaceutical and fine-chemical production, fast and exothermic reactions can bring batch reactors to their technical limits and reduce productivity due to the need for safe and controlled reactor operation. Heat transfer limitation in vessels requires time-consuming semi-batch operation or low reactant concentrations. Thus, high potential exists for process intensification via flow chemistry, which can provide safer operating conditions, reduced waste, higher productivity, and higher energy efficiency. The transfer of batch processes to continuous production enables new chemical paths, and process conditions far from conventional practices can be considered. Often referred to as Novel Process Windows [8,19], these possibilities offer numerous benefits associated with process intensification [7]. Contrary to the optimization of an existing traditional process, the approach of process intensification [17,18] involves the utilization of entirely new equipment and process designs. The present paper focuses on a heat transfer limited process with an exothermic reaction; thus,
* Corresponding author. E-mail address: [email protected] (T. Röder). http://dx.doi.org/10.1016/j.cep.2016.07.017 0255-2701/ã 2016 Elsevier B.V. All rights reserved.
two major aspects of heat exchange intensification are emphasized: (1) process intensification by miniaturization [4] (milli/ micro process technology) and (2) the combination of a reactor and heat exchanger as one multi-functional piece of equipment (Heat exchanger (HEX) reactors) [1]. Using millistructured flow reactors for a small-scale production considerably improves mass and heat transfer. The large surfaceto-volume ratio and small length scales for heat transfer reduce the risk involved by handling fast and exothermic reactions. Better control of temperature and residence time distribution often results in higher yields and selectivity. Instead of the reaction time being determined by slow reactant dosing in semi-batch mode, the required residence time in the continuous reactor depends mostly on reaction kinetics. Furthermore, increased reactant concentrations or even solvent-free syntheses are possible. As a consequence, not only reaction time is reduced but also solvent waste and energy consumption for later solvent removal. Although miniaturization itself enhances heat removal, a further step for intensifying heat exchange is to utilize HEX reactors. Their design has to meet the requirements for both heat exchange and reaction control (e.g., residence time distribution) and is often based on the design of compact heat exchangers, i.e., plate heat exchangers.
110
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
Nomenclature aV Aw cEA,0 cp dh EA,obs,i h ΔHR k kobs,i L N Nu Q rEA rEA,0 Rth Re th tr T Tc u w XEA z
Surface-to-volume ratio (m1) Wetted surface area (m2) Theoretical, initial concentration of ethyl acrylate in the reaction mixture (mol m3) Heat capacity of the fluid (J kg1 K1) Hydraulic channel diameter (m) Observed activation energy of reaction i (J mol1) Channel height (m) Enthalpy of reaction (J mol1) Overall heat transfer coefficient (W m2 K1) Observed rate constant of reaction i, n-th order (Ln1 mol1n s1) Reactor length (m) Time ratio of reaction and cooling (dimensionless) Nusselt number (dimensionless) Volumetric flow rate (m3 s1) Rate of disappearance of ethyl acrylate due to reaction (mol m3 s1) Initial rate of disappearance of ethyl acrylate due to reaction (mol m3 s1) Thermal resistance (K W1) Reynolds number (dimensionless) Characteristic time scale of heat transfer (s) Characteristic time scale of reaction (s) Temperature (K) Heat carrier temperature (K) Axial flow velocity (m s1) Channel width (m) Conversion of ethyl acrylate (dimensionless) Axial distance from the reactor inlet (m)
Greek Symbols a Heat transfer coefficient (W m2 K1) l Heat conductivity of the fluid (W m1 K1) r Density of the fluid (kg m3)
Contrary to batch processes, continuous production in a HEX reactor enable the steady removal of reaction heat, almost as rapidly as it is generated by the reaction. Thus, the exothermic reaction can be considered as a potential heat source in the overall process. Consequently, continuous production processes with millistructured reactors enable new options for energy savings via heat recovery. Systematic approaches to heat integration, such as pinch analysis [9,11], are state-of-the-art in the continuous production of petrochemicals and bulk chemicals, where energy savings is a main target for process design. For the small scale production of pharmaceuticals or fine chemicals, the dominating aspect is the aim for higher material yield. However, direct energy savings (heat recovery) or indirect energy savings (less solvent to remove, higher raw material yield) can also decrease costs and significantly improve eco-efficiency (green processing) [6,7]. In the first part of this paper, the full potential for process intensification in a continuous production process is examined with respect to the addition of piperidine to ethyl acrylate (Michael addition). The overall process, including purification via distillation, is investigated and optimized via process simulation. To maximize heat recovery, the basic principles of pinch analysis are applied. In the second part of the paper, we present an integrated millistructured HEX reactor for controlling a fast and exothermic reaction while transferring the heat of the reaction to a heat carrier stream. Reaction conditions close to the simulated process are
chosen for a proof-of-concept laboratory experiment. With an additional capillary heat exchanger, reactant streams are heated to reaction temperature using the hot product stream in countercurrent operation. 2. Materials and methods The process was modeled using the SIMSCI Pro/II process simulator, version 9.3 (Invensys Systems Inc.). Steady-state temperature and conversion profiles in the HEX reactor were calculated by solving the following mass and heat balances for an ideal plug flow reactor. dXEA r ¼ EA dz cA;0 u
ð1Þ
dT rEA DHR 4kðT Tc Þ ¼ dz dh urcp urcp
ð2Þ
Herein, rEA is the rate of disappearance of ethyl acrylate and is defined by the kinetic model of the Michael addition with two parallel reactions, as described in Section 3.1. Calculation of the distillation column is based on the theory of separation stages. Experimental investigations were performed with a setup for continuous reactor operation, using a laboratory automation system (LabBox and LabVision Software, HiTec Zang GmbH, Germany) for control and permanent data logging. The feed and quench streams were supplied by three continuous syringe pumps (Syrdos 2, HiTec Zang GmbH) with 10 mL glass syringes. The heat carrier cycle was realized using a thermostat (Ministat 230, Huber GmbH, Germany) and silicon oil (M20.195/235.20, Huber GmbH). For measurement of the heat carrier flow rate, an ultrasonic flow sensor (Sonoflow IL.52, Sonotec Ultraschallsensorik Halle GmbH, Germany) was employed. A plate reactor designed for flow rates in the range between 10 and 30 mL/min was used for the continuous production experiment. Mixer, reactor channels and heat carrier channels were milled into an aluminum plate on opposite sides. Design criteria and channel dimensions are described in Section 4. Both channels were covered with additional aluminum plates, which were evenly pressed onto the reactor plate to avoid significant bypass flow. Five Pt100 sensors were inserted through the side of the reactor directly into the flow channel to measure the temperature of all inlet and outlet streams. The reactant preheater was manufactured by soldering three stainless steel capillaries (1/16 in. outer diameter, 1 mm inner diameter) of 1.2 m in length in direct contact. On the cold side of the heat exchanger, three further Pt100 sensors were used for temperature measurement. Both the reactor and reactant preheater were covered with Armaflex insulation. The synthesis of 3-piperidinopropionic acid ethyl was realized without any solvent. However, water was used as a catalyst and was premixed with piperidine. For the proof-of-concept experiments, the molar ratio of reactants was 1.1 (piperidine/ethyl acrylate) and the molar ratio of water was 0.5 (water/ethyl acrylate). To determine the reaction conversion, the reaction mixture at the reactor outlet was effectively quenched via continuous mixing with a solution of acetic acid in methanol (11 vol.%). Product samples were analyzed using an Agilent 7820 gas chromatograph equipped with a HP-5 column (30 m 0.32 mm 0.25 mm, Agilent) and a flame ionization detector. The inlet temperature was 275 C, the inlet pressure, 50 kPa, and the split ratio, 100:1. An injection volume of 0.5 mL was used in the automatic liquid sampler. The oven temperature (60 C) was kept constant for 2 min and then increased to 200 C with a ramp of 20 K/min.
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
3. Process simulation 3.1. Kinetic model In the reaction of piperidine and ethyl acrylate, a zwitterionic intermediate is formed by a nucleophilic attack on the activated double bond. The subsequent proton transfer towards the neutral product can be considered as a rate determining step in the reaction mechanism. A water molecule or an amine molecule can act as a shuttle for the proton and therefore promotes the charge equalization of the zwitterionic intermediate, resulting in the product 3-piperidino propionic acid ethyl ester. A simplified reaction scheme with two parallel reactions can be used to describe the reaction progress with good accuracy (Scheme 1). In this simplified reaction scheme, fitting parameters are reduced to two observed third order rate constants, i.e., kobs, 2 2 1 s and kobs,B = 0.0170 L2 mol2 s1, and two A = 0.0080 L mol observed activation energies. Consequently, the addition of water to the reaction mixture can strongly enhance the reaction rate. The temperature dependencies of both observed rate constants were found to be surprisingly low, with activation energies EA,obs,A = 0 kJ/ mol and EA,obs,B = 14 kJ/mol. In general, this could indicate a mass transfer limitation of the reaction rate. However, the effect has been found in single-phase flow experiments and short mixing times compared to the reaction time were realized by using a highperformance micromixer (SIMM-V2, Fraunhofer ICT-IMM, Mainz, Germany). Thus, a mass transport limitation can be ruled out. Assuming the zwitterionic intermediate is formed via a preequilibrium step in the reaction mechanism, the observed activation energy would result from both the actual activation energy and the temperature dependency of the equilibrium step, which is given by the standard enthalpy of reaction. For an exothermic reaction, a low observed activation energy can result. Further details on the kinetic investigations can be found in a previous publication [14]. 3.2. Model of the production process As a basis for process simulation, reactor data of a commercially available plate HEX reactor were used. The production reactor (Flow reactor MR500, 3M Technical Ceramics, Kempten) consisted of combinable reactor modules manufactured from silicon carbide (Fig. 1a). Reactor channels with a rectangular cross section (3.5 3.5 mm) were arranged in a zigzag pattern. For a production of approximately 200 tons/year, a total feed rate of 31 L h1 was specified. By combining reactor modules with a total of 0.6 L inner volume (reaction side), a residence time of approximately 1 min can be achieved. With the MR500 reactor (198 L h1 max. throughput), overall heat transfer coefficients of up to 10 kW m2 K1 can be achieved in measurements with water (manufacturer’s data, personal communication). Considering significantly lower transfer coefficients for reduced throughputs and an organic medium, safe reaction control can still be ensured for k = 1 kW m2 K1. Temperature profiles for varying heat transfer coefficients were calculated and are given in the Supplementary information. Scale-up from the operation in microreactors to a similar silicon carbide flow reactor for small-scale production (2 2 mm zigzag channels) has been investigated in a previous publication [14]. Short mixing times, narrow residence time distribution (see
Scheme 1. Simplified reaction scheme used as a kinetic model for the Michael addition.
111
Supplementary material, Fig. S6) and effective heat transfer were found for the silicon carbide plate HEX reactor. The initial process design is shown in Fig. 1b. Subsequent to the reactor, water and unconverted reactants are separated from the product in a distillation column and led back to the reactor feed by a recycle stream. Thus, very high material yields can be obtained in the overall process, even if reactants are not fully converted in the reactor. The temperature increase from the reactor inlet (90 C) to the outlet can be controlled by the heat carrier flow (approx. 10 K). A distillation column is required to provide the bottom product in high purity, which is specified by a molar fraction of 0.995 mol/mol and determines the reboiler duty. Specified temperatures for simulation are given in Fig. 1b. To avoid exposure of the product to higher temperatures in the distillation column, a vacuum distillation at 100 mbar is applied. 3.3. Optimization of process conditions The production process was optimized with regard to energy efficiency. Thus, the column design was adjusted to obtain the specified product purity with minimum required reboiler duty. Because the head product is recycled and only the product purity of the bottom product is specified, a stripping column (10 theoretical stages) without an enriching section can be used. More details on the optimization of the distillation can be found in the Supplementary information. As described above, water is used as a catalyst for the Michael addition. Fig. 2 shows that the conversion in the reactor can be significantly increased by increasing the water ratio W (water/ ethyl acrylate, mol/mol) in the reactor feed. Due to the separation and recycling of unconverted reactants, the reactant conversion in the overall process is close to 100%, independent of the conversion that is achieved in the reactor. In Fig. 3, the transferred heat flow in the HEX reactor (heat source) and in the column reboiler (heat sink) is shown for a varying water ratio in the reactor feed. At low water ratio, more energy is required to separate unconverted reactants from the product. Contrarily, for high amounts of water, reboiler energy is mainly required to remove the water. Furthermore, a large water recycle stream lowers the reactant concentration and residence time in the reactor. An optimum water ratio for minimal reboiler duty exists at W = 0.18. This optimum is valid for the specified reactor data and can change with reactor volume or reactor temperature. Shorter residence time or lower temperature in the reactor would reduce the conversion at the reactor outlet and shift the optimum to a higher water (catalyst) amount in the reactor feed. As follows from Fig. 3, the reboiler duty is below the heat flow that is transferred to the heat carrier in the HEX reactor only for a small range around the optimum. 3.4. Pinch analysis and heat integration With the optimal process conditions, energy targets for heat integration can be determined via pinch analysis [9,11]. Composite curves in the T/H diagram (Fig. 4) represent all hot streams (or cold streams, respectively) and visualize the required enthalpy flow in the respective temperature intervals. The cold composite curve combines all streams that need to be heated (cold feed) or vaporized (i.e., reboiler duty), while the hot composite curve combines all streams with cooling demand (hot product, heat carrier, and condenser stream). As shown in Fig. 4, regarding the right side of the curves, a “nonutility end” at the hot end is feasible if a minimum temperature difference DTmin of 14 K or lower is acceptable for heat exchange. Thus, with an ideal heat exchanger network, no external heating is required. Energy demands for preheating reactants and for the column reboiler can be met by using hot streams leaving the HEX
112
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
Fig. 1. (a) HEX reactor MR500 (3M Technical Ceramics, Germany) with silicon carbide modules; (b) schematic flow sheet of the process with reactant preheating, HEX reactor and distillation column.
and prevents dynamic effects between distillation and reaction. As a result, heat recovery according to the energy targets (Fig. 4) is fully covered. Only external cooling is required for the condenser in the distillation column and to decrease the temperature of the separated product stream. By comparing both process designs in Fig. 5, it can be shown that the amount of required equipment has not been increased. Full heat recovery is realized mainly by using the multifunctional HEX reactor and a heat exchanger for reactant preheating. Both devices are the subject of the following experimental investigations. For practical realization, a multistream reactant preheater, rather than premixing reactor feeds, is advantageous. Thus, mixing inside the reactor results in a defined starting point of reaction. 4. Experimental proof-of-concept Fig. 2. Influence of the water ratio in the reactor feed on the ethyl acrylate conversion in the reactor and the overall process, determined by process simulation.
Fig. 3. Influence of the water ratio in the reactor feed on the transferred heat in the HEX reactor and on the required reboiler duty.
reactor. In Fig. 5b, a heat exchanger network suitable for meeting the energy targets is shown in a schematic flowsheet and compared with the flowsheet of the process before heat integration (Fig. 5a). The hot reaction product stream is used to preheat the reactor feed. Heat integration between the reactor and separation (column reboiler) can be realized using a heat carrier cycle stream. As shown in Fig. 3, the heat transferred in the HEX reactor exceeds the column reboiler duty. However, an additional heat exchanger ensures the adjustment of the reactor temperature
4.1. Reactor design For the transfer from batch to continuous HEX reactors, the main criteria are thermal intensification, flow intensification, reactor dynamics and residence time [1]. For proof-of-concept experiments, a millistructured plate reactor was designed and manufactured via mechanical milling of aluminum (Fig. 6). An important improvement for flow intensification is the zigzag pattern of the channels on the reaction side [5,13]. As a result of repeatedly redirecting the flow by 90 , a secondary flow is generated, making the radial concentration profiles more uniform. Radial mixing of rapidly flowing fluid in the channel center and slowly flowing fluid in the channel wall region narrows the residence time distribution that is typically caused by the laminar flow velocity profile [3,12]. Measurements show that a Bodenstein number close to 100 can be obtained with this channel design at the selected flow rate (see Supplementary information), which indicates a narrow residence time distribution and a reactor behavior close to that of an ideal plug flow reactor. The reactants were mixed in 0.5 mm channels, which were arranged in an arrowshaped geometry. With this type of micromixer, a high mixing performance resulted from high energy input due to the high flow velocity and sharp flow redirection in the mixer [15,16]. The reaction channel is divided into two sections (Fig. 6b). In the entrance region, high reactant concentrations result in a fast heat release rate and the formation of local hot spots. Thus, a zigzag channel with a smaller cross-section (1 1 mm) is applied for enhanced heat transfer and effective temperature control (Channel A). With increasing conversion, the reaction rate of the third-order type reaction strongly decreases and a residence time channel with a larger cross-section (2 2 mm) is applied to obtain high reactant conversion (Channel B). Channel dimensions and heat transfer characteristics for the applied flow rates are given in Table 1.
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
113
Fig. 4. Composite curves for the simulated process after optimization of the process conditions.
of the reactor. Thermal resistance in laminar flow is given by Rth ¼
1
a Aw
¼
dh 1 wh ¼ NulAw NulLðw þ hÞ2
ð4Þ
Thus, low thermal resistance is obtained for long and narrow channels (maximum value at w/h = 1). A 10 3 mm heat carrier channel was realized to obtain fast heat transfer in laminar flow at moderate pressure drop. High flowrates of the heat carrier can increase heat transfer coefficients and decrease thermal resistance. However, a comparatively low flow rate was chosen in order to obtain an easily measureable temperature increase for the heat balance. In Table 1, maximum values for the thermal resistance are given, which were calculated for the theoretical case of a straight channel with the given channel dimensions. To evaluate the dynamic behavior of the reactor, time scales of heat transfer and reaction were compared by calculating the ratio of the characteristic reaction time to the characteristic cooling time N¼
Fig. 5. Schematic flowsheet of the heat exchanger network before (a) and after (b) heat integration.
Heat transfer in the laminar flow regime depends on the channel design and the flow velocity that results for each channel section. For calculation, heat transfer characteristics of a straight channel were chosen as a conservative estimation. Thus, a minimum constant Nusselt number Nu ¼
a dh l
ð3Þ
can be assumed, which is Nu = 2.98 for a quadratic cross-section, assuming a constant wall temperature. However, experimental and simulation results given in the literature [2,5] indicate significantly higher Nusselt numbers for the zigzag pattern compared to the straight channel. Formation of the secondary flow results in a heat transfer enhancement which is intensified with increasing Reynolds number. The heat carrier channel was designed to avoid major thermal resistance on the heat carrier side
dEA;0 4k tr ¼ th rEA;0 dh rcp
ð5Þ
Using Eqs. (1) and (2), the required length of the channel with a smaller cross-section can be estimated by calculating the total conversion in this channel section. At the applied process conditions, an ethyl acrylate conversion of approx. 0.6 at the transition from section A to section B results from the residence time in channel A. Thus, lower heat transfer in channel B is partly compensated for by a slower reaction rate. For both channels, the characteristic time for heat transfer is short compared to the reaction time. The minimum value of N for safe reactor operation depends on reaction order, adiabatic temperature rise and activation energy [10]. Due to the low activation energy of the Michael addition, no stability problems are expected for the obtained values of N. A scale-up of the HEX plate reactor concept requires consideration of the described mass and heat transfer characteristics. The total reactor volume and appropriate channel dimensions have to be chosen with regard to the required residence time and the resulting pressure drop in the channel at the aimed flow rate. Therefore, the laboratory-scale aluminum plate reactor and the production-scale reactor MR500 (Fig. 1a) differ in some design aspects. Although both reactors are based on the zig zag channel design, the MR500 reactor allows for significantly higher throughput, in a longer channel (4.9 m per reactor module) with a larger cross-sectional area (3.5 3.5 mm). However, direct up- or down-scaling with different channel designs is possible provided that plug flow behavior, fast reactant mixing and fast heat removal
114
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
Fig. 6. (a) Plate heat exchanger reactor without insulation; (b) channel design on the reaction side of the middle plate.
can be maintained. To fulfill these requirements, the described design with kinetic-adapted channel size variation can be an effective option. In a production scale reactor such as the MR500 reactor, a channel design with constant cross-section is applicable because significantly higher Reynolds numbers (Re > 1000) can be realized to intensify heat and mass transfer. 4.2. Heat balance To investigate the operability of the HEX reactor in the production process, a lab scale experiment was conducted in the previously described plate reactor. Under similar reaction conditions but at significantly reduced flow rates compared to the simulation (0.9 L/h), 3-piperidino propionic acid ethyl ester was continuously produced for an operating time of 50 min. Fig. 7a shows the flow sheet of the experimental setup, with temperature measuring points at inlets and outlets of the reactant preheater and HEX reactor. Calculated temperature profiles and measured steady-state temperatures are given in Fig. 7b. In the reactant
preheater, both feed streams were heated from 25 C (T1, T2) to a reactor entrance temperature of approximately 90 C (T3, T4). In counter-current flow, a mean logarithmic temperature difference of 7.4 K can be obtained. Due to the nearly identical heat capacity rates of the product and both reactant streams, the local temperature difference between hot and cold fluid can be assumed to be constant throughout the capillary length. However, heat losses to the surroundings and the reaction heat of unconverted reactants in the product stream can influence temperature profiles and have to be taken into consideration. In the HEX reactor, operation in a co-current manner was realized to reduce the hot spot temperature in the entrance region of the reaction channel. Due to the exothermic reaction, the heat carrier temperature increased from 88 C (T7) to 100 C (T8). The temperature increase was adjusted by using the heat carrier flow rate (approx. 5 L/h) to a value close to that of the process simulation (DT = 12 K). At the reactor outlet, the temperature difference between the reaction mixture and heat carrier reached a value of only 0.3 K. Thus, nearly the maximum amount of reaction heat was transferred to the heat carrier. A 93% conversion of ethyl acrylate in the reactor was determined by analyzing samples from the reactor outlet. With the determined temperatures and flow rates, the heat balance of the HEX reactor setup can be used to evaluate the efficiency of heat recovery. For better visualization, the heat balance is depicted in a Sankey type diagram (Fig. 8). Reference temperature for the thermal energy of the fluid streams was defined by the ambient temperature (25 C). In the reactant preheater, thermal energy of the hot product stream was transferred to the reactant streams with only 3% heat losses and leaving hot product. Thus, energy demanding preheating of the reactants in the subsequent reactor was not necessary and the majority of reaction heat could be transferred to the heat carrier for further use in the process. Heat losses in the plate HEX reactor amounted to approximately 20% of the transferred reaction heat. Because those losses mainly depended on insulation and the outer reactor surface, a smaller relative amount could be expected for a scale-up to a production scale reactor. As a result of the proof-of-concept experiment, 72% of the reaction heat is transferred to thermal oil at the temperature level of the reactor. Thus, the heat can be used in all further process steps that require thermal energy below the reactor outlet temperature. Transfer of reaction heat to low temperature reactant streams is avoided by the reactant preheater. In both devices, heat is transferred with a small temperature difference, which leads to a minimum amount of exergy loss. Approximately 100 kJ per liter of product could be saved using the presented experimental setup. 5. Conclusions In this work, process development for an energy-efficient continuous production process was presented for the synthesis of
Table 1 Flow channel dimensions and heat transfer characteristics for the flow channels in the aluminum plate heat exchanger reactor. Reaction channel A
Reaction channel B
Heat carrier channel
wh L av Rth,max
[mm2] [m] [m2/m3] [K/W]
1 1 1.68 4000 0.33
22 2.04 2000 0.28
3 10 2.90 867 0.13
Q Re th N
[mL/min] [] [s] []
15 140 0.6 4.6
15 70 2.4 2.1
82 88 12.7 –
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
115
Fig. 7. (a) Flow sheet of the experimental setup with reactant preheater, HEX reactor and measurement points; (b) calculated temperature profiles and measured inlet/outlet temperatures of the fluid streams in the reactant preheater and in the HEX reactor.
Fig. 8. Sankey type depiction of the energy balance determined from the laboratory experiment.
3-piperidino propionic acid ethyl ester. Based on a proven kinetic model of the reaction, optimum process conditions in a commercially available plate HEX reactor with subsequent distillation have been determined via process simulation. With a defined, specified product purity, the required energy consumption was minimized by choosing an optimal water (catalyst) amount and recycling separated reactants and water after distillation. Process design was adapted to fully exploit the potential for heat recovery given by energy targets from pinch analysis. As a result, operation conditions could be found, where all heat consuming process steps, particularly reactant preheating and distillation, could be operated with heat from the hot outlet streams of the HEX reactor. The experimental part of this work focuses on the practical realization of continuous reaction control and heat recovery under laboratory conditions. For this purpose, two apparatus prototypes have been developed. (1) The reactant preheater, which transfers
heat from one hot process stream to two cold reactant streams in counter-current operation, has been demonstrated as an easy and effective tool for heat recovery and can be implemented in numerous flow chemistry applications at elevated reactor temperature. (2) The plate HEX reactor was manufactured with specific application-adapted flow channels. Channel design and dimensioning is a crucial aspect because temperature increase for heat transfer at the reaction temperature level has to be realized without losing thermal control in the reactor. To meet these requirements, zigzag channels for heat transfer intensification and reaction rate adapted channel widths were chosen. Stable steady state operability of both devices could be demonstrated by temperature and flow monitoring in a proof-of-concept experiment. In terms of energy efficiency, heat losses to the environment play a considerable part in the heat balance. Assuming the transfer to a scale of the simulated production process, transferred heat rates would increase significantly and thus lead to negligible percentages for heat loss. The example highlights how miniaturization and the design of a multifunctional HEX reactor, which considers both reaction kinetics and heat transfer requirements, can lead to a strongly intensified continuous process. Reactant preheating in a countercurrent heat exchanger can be generally applied for continuous syntheses at elevated reactor temperature to save energy and avoid additional equipment for reactant preheating and product cooling. The strategy of removing reaction heat in a HEX reactor for further utilization in the process is primarily applicable for fast and exothermic flow syntheses that have a high heat production potential, but can still be controlled in milli/microstructured reactors. Thus, the heat production potential r0 DHR should be in a range between 10 and 100 kW/L, which for example includes metalorganic reactions, nitrations or polymerization reactions [21]. Energy-demanding process steps for heat integration can be part of product purification, or endothermic reactions in continuous multistep syntheses [20]. In general, energy saving via heat recovery can be considered as a further advantage of continuous
116
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
production in milli/microstructured reactors, which could not be realized in discontinuous processes with reasonable effort. Acknowledgment This work was funded by the German Federal Ministry of Education and Research (BMBF, funding code 03FH012I2). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cep.2016.07.017. References [1] Z. Anxionnaz, M. Cabassud, C. Gourdon, P. Tochon, Heat exchanger/reactors (HEX reactors): concepts, technologies: state-of-the-art, Chem. Eng. Process. Process Intensif. 47 (12) (2008) 2029–2050, doi:http://dx.doi.org/10.1016/j. cep.2008.06.012. [2] Z. Anxionnaz-Minvielle, M. Cabassud, C. Gourdon, P. Tochon, Influence of the meandering channel geometry on the thermo-hydraulic performances of an intensified heat exchanger/reactor, Chem. Eng. Process. Process Intensif. 73 (2013) 67–80, doi:http://dx.doi.org/10.1016/j.cep.2013.06.012. [3] R. Aris, On the dispersion of a solute in a fluid flowing through a tube, Proc. R. Soc. A 235 (1200) (1956) 67–77, doi:http://dx.doi.org/10.1098/rspa.1956.0065. [4] J.-C. Charpentier, Process intensification by miniaturization, Chem. Eng. Technol. 28 (3) (2005) 255–258, doi:http://dx.doi.org/10.1002/ ceat.200407026. [5] A. Fazeli, M. Behnam, Hydrogen production in a zigzag and straight catalytic wall coated micro channel reactor by CFD modeling, Int. J. Hydrogen Energy 35 (17) (2010) 9496–9503, doi:http://dx.doi.org/10.1016/j.ijhydene.2010.05.052. [6] M. Hempel, Novel process windows—a contribution to more sustainable chemistry? Chem. Eng. Technol. 32 (11) (2009) 1651–1654, doi:http://dx.doi. org/10.1002/ceat.200900338. [7] V. Hessel, Novel process windows—Gate to maximizing process intensification via flow chemistry, Chem. Eng. Technol. 32 (11) (2009) 1655–1681, doi:http:// dx.doi.org/10.1002/ceat.200900474.
[8] V. Hessel, D. Kralisch, N. Kockmann, T. Noël, Q. Wang, Novel process windows for enabling, accelerating, and uplifting flow chemistry, ChemSusChem 6 (5) (2013) 746–789, doi:http://dx.doi.org/10.1002/cssc.201200766. [9] I.C. Kemp, Pinch analysis and process integration: A user guide on process integration for the efficient use of energy, 2nd ed., Butterworth-Heinemann, Amsterdam, Boston, 2007. [10] N. Kockmann, Micro process engineering: Fundamentals, devices, fabrication and applications, Wiley-VCH, Weinheim, 2006. [11] B. Linnhoff, The pinch design method for heat exchanger networks, Chem. Eng. Sci. 38 (5) (1983) 745–763. [12] O. Levenspiel, Chemical Reaction Engineering, 3rd ed., Wiley, New York, 1999. [13] V. Mengeaud, J. Josserand, H.H. Girault, Mixing processes in a zigzag microchannel: finite element simulations and optical study, Anal. Chem. 74 (16) (2002) 4279–4286, doi:http://dx.doi.org/10.1021/ac025642e. [14] S. Schwolow, B. Heikenwälder, L. Abahmane, N. Kockmann, T. Röder, Kinetic and scale-up investigations of a michael addition in microreactors, Org. Process Res. Dev. 18 (11) (2014) 1535–1544, doi:http://dx.doi.org/10.1021/acs. oprd.5b00184. [15] S. Schwolow, J. Hollmann, B. Schenkel, T. Röder, Application-oriented analysis of mixing performance in microreactors, Org. Process Res. Dev. 16 (9) (2012) 1513–1522, doi:http://dx.doi.org/10.1021/op300107z. [16] A. Soleymani, E. Kolehmainen, I. Turunen, Numerical and experimental investigations of liquid mixing in T-type micromixers, Chem. Eng. J. 135 (2008) S219–S228, doi:http://dx.doi.org/10.1016/j.cej.2007.07.048. [17] A.I. Stankiewicz, J.A. Moulijn, Process intensification, Ind. Eng. Chem. Res. 41 (8) (2002) 1920–1924, doi:http://dx.doi.org/10.1021/ie011025p. [18] A.I. Stankiewicz, J.A. Moulijn, Process intensification: transforming chemical engineering, Chem. Eng. Prog. 96 (2000) 22–34. [19] S. Taghavi-Moghadam, A. Kleemann, G. Golbig, Microreaction technology as a novel approach to drug design, process development and reliability, Org. Process Res. Dev. 5 (6) (2001) 652–658, doi:http://dx.doi.org/10.1021/ op010066u. [20] J. Wegner, S. Ceylan, A. Kirschning, Flow chemistry—a key enabling technology for (Multistep) organic synthesis, Adv. Synth. Catal. 354 (1) (2012) 17–57, doi: http://dx.doi.org/10.1002/adsc.201100584. [21] T. Westermann, L. Mleczko, Heat management in microreactors for fast exothermic organic syntheses—first design principles, Org. Process Res. Dev. 20 (2) (2016) 487–494, doi:http://dx.doi.org/10.1021/acs.oprd.5b00205.
Contents lists available at ScienceDirect
Chemical Engineering and Processing: Process Intensification journal homepage: www.elsevier.com/locate/cep
Design and application of a millistructured heat exchanger reactor for an energy-efficient process Sebastian Schwolowa , Andreas Neumüllera , Lahbib Abahmaneb , Norbert Kockmannc , Thorsten Rödera,* a b c
Mannheim University of Applied Sciences, Institute of Chemical Process Engineering, Paul-Wittsack-Straße 10, 68163 Mannheim, Germany 3M Technical Ceramics, Max-Schaidhauf-Str. 25, 87437 Kempten, Germany TU Dortmund, Biochemical and Chemical Engineering, Equipment Design, Emil-Figge-Straße 68, 44227 Dortmund, Germany
A R T I C L E I N F O
Article history: Received 30 March 2016 Received in revised form 15 July 2016 Accepted 27 July 2016 Available online 9 August 2016 Keywords: Heat integration Flow reactor Milli- and microreactors Process intensification Energy efficiency Michael addition
A B S T R A C T
Flow chemistry in milli- and microstructured reactors exhibits great potential for process intensification. In the present work, this potential has been demonstrated through the process development for a solvent-free production process including a Michael addition and following product purification. Process simulation was used to maximize the material and energy efficiency of the overall process by recycling unconverted reactants and a catalyst (water) and by utilizing heat from the exothermic reaction to substitute external energy supply in the heat demanding process steps. As a proof-of-concept experiment, millistructured equipment was designed, manufactured and tested at a laboratory scale. A three-stream counter-current heat exchanger for reactant preheating and a plate heat exchanger reactor with zigzag reaction channels were investigated regarding the maximum transfer of reaction heat available for further process steps via a heat carrier cycle. The experiments showed stable reactor control in steady state operation and efficient heat transfer with a small driving temperature difference at the outlet of the heat exchanger reactor. ã 2016 Elsevier B.V. All rights reserved.
1. Introduction In pharmaceutical and fine-chemical production, fast and exothermic reactions can bring batch reactors to their technical limits and reduce productivity due to the need for safe and controlled reactor operation. Heat transfer limitation in vessels requires time-consuming semi-batch operation or low reactant concentrations. Thus, high potential exists for process intensification via flow chemistry, which can provide safer operating conditions, reduced waste, higher productivity, and higher energy efficiency. The transfer of batch processes to continuous production enables new chemical paths, and process conditions far from conventional practices can be considered. Often referred to as Novel Process Windows [8,19], these possibilities offer numerous benefits associated with process intensification [7]. Contrary to the optimization of an existing traditional process, the approach of process intensification [17,18] involves the utilization of entirely new equipment and process designs. The present paper focuses on a heat transfer limited process with an exothermic reaction; thus,
* Corresponding author. E-mail address: [email protected] (T. Röder). http://dx.doi.org/10.1016/j.cep.2016.07.017 0255-2701/ã 2016 Elsevier B.V. All rights reserved.
two major aspects of heat exchange intensification are emphasized: (1) process intensification by miniaturization [4] (milli/ micro process technology) and (2) the combination of a reactor and heat exchanger as one multi-functional piece of equipment (Heat exchanger (HEX) reactors) [1]. Using millistructured flow reactors for a small-scale production considerably improves mass and heat transfer. The large surfaceto-volume ratio and small length scales for heat transfer reduce the risk involved by handling fast and exothermic reactions. Better control of temperature and residence time distribution often results in higher yields and selectivity. Instead of the reaction time being determined by slow reactant dosing in semi-batch mode, the required residence time in the continuous reactor depends mostly on reaction kinetics. Furthermore, increased reactant concentrations or even solvent-free syntheses are possible. As a consequence, not only reaction time is reduced but also solvent waste and energy consumption for later solvent removal. Although miniaturization itself enhances heat removal, a further step for intensifying heat exchange is to utilize HEX reactors. Their design has to meet the requirements for both heat exchange and reaction control (e.g., residence time distribution) and is often based on the design of compact heat exchangers, i.e., plate heat exchangers.
110
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
Nomenclature aV Aw cEA,0 cp dh EA,obs,i h ΔHR k kobs,i L N Nu Q rEA rEA,0 Rth Re th tr T Tc u w XEA z
Surface-to-volume ratio (m1) Wetted surface area (m2) Theoretical, initial concentration of ethyl acrylate in the reaction mixture (mol m3) Heat capacity of the fluid (J kg1 K1) Hydraulic channel diameter (m) Observed activation energy of reaction i (J mol1) Channel height (m) Enthalpy of reaction (J mol1) Overall heat transfer coefficient (W m2 K1) Observed rate constant of reaction i, n-th order (Ln1 mol1n s1) Reactor length (m) Time ratio of reaction and cooling (dimensionless) Nusselt number (dimensionless) Volumetric flow rate (m3 s1) Rate of disappearance of ethyl acrylate due to reaction (mol m3 s1) Initial rate of disappearance of ethyl acrylate due to reaction (mol m3 s1) Thermal resistance (K W1) Reynolds number (dimensionless) Characteristic time scale of heat transfer (s) Characteristic time scale of reaction (s) Temperature (K) Heat carrier temperature (K) Axial flow velocity (m s1) Channel width (m) Conversion of ethyl acrylate (dimensionless) Axial distance from the reactor inlet (m)
Greek Symbols a Heat transfer coefficient (W m2 K1) l Heat conductivity of the fluid (W m1 K1) r Density of the fluid (kg m3)
Contrary to batch processes, continuous production in a HEX reactor enable the steady removal of reaction heat, almost as rapidly as it is generated by the reaction. Thus, the exothermic reaction can be considered as a potential heat source in the overall process. Consequently, continuous production processes with millistructured reactors enable new options for energy savings via heat recovery. Systematic approaches to heat integration, such as pinch analysis [9,11], are state-of-the-art in the continuous production of petrochemicals and bulk chemicals, where energy savings is a main target for process design. For the small scale production of pharmaceuticals or fine chemicals, the dominating aspect is the aim for higher material yield. However, direct energy savings (heat recovery) or indirect energy savings (less solvent to remove, higher raw material yield) can also decrease costs and significantly improve eco-efficiency (green processing) [6,7]. In the first part of this paper, the full potential for process intensification in a continuous production process is examined with respect to the addition of piperidine to ethyl acrylate (Michael addition). The overall process, including purification via distillation, is investigated and optimized via process simulation. To maximize heat recovery, the basic principles of pinch analysis are applied. In the second part of the paper, we present an integrated millistructured HEX reactor for controlling a fast and exothermic reaction while transferring the heat of the reaction to a heat carrier stream. Reaction conditions close to the simulated process are
chosen for a proof-of-concept laboratory experiment. With an additional capillary heat exchanger, reactant streams are heated to reaction temperature using the hot product stream in countercurrent operation. 2. Materials and methods The process was modeled using the SIMSCI Pro/II process simulator, version 9.3 (Invensys Systems Inc.). Steady-state temperature and conversion profiles in the HEX reactor were calculated by solving the following mass and heat balances for an ideal plug flow reactor. dXEA r ¼ EA dz cA;0 u
ð1Þ
dT rEA DHR 4kðT Tc Þ ¼ dz dh urcp urcp
ð2Þ
Herein, rEA is the rate of disappearance of ethyl acrylate and is defined by the kinetic model of the Michael addition with two parallel reactions, as described in Section 3.1. Calculation of the distillation column is based on the theory of separation stages. Experimental investigations were performed with a setup for continuous reactor operation, using a laboratory automation system (LabBox and LabVision Software, HiTec Zang GmbH, Germany) for control and permanent data logging. The feed and quench streams were supplied by three continuous syringe pumps (Syrdos 2, HiTec Zang GmbH) with 10 mL glass syringes. The heat carrier cycle was realized using a thermostat (Ministat 230, Huber GmbH, Germany) and silicon oil (M20.195/235.20, Huber GmbH). For measurement of the heat carrier flow rate, an ultrasonic flow sensor (Sonoflow IL.52, Sonotec Ultraschallsensorik Halle GmbH, Germany) was employed. A plate reactor designed for flow rates in the range between 10 and 30 mL/min was used for the continuous production experiment. Mixer, reactor channels and heat carrier channels were milled into an aluminum plate on opposite sides. Design criteria and channel dimensions are described in Section 4. Both channels were covered with additional aluminum plates, which were evenly pressed onto the reactor plate to avoid significant bypass flow. Five Pt100 sensors were inserted through the side of the reactor directly into the flow channel to measure the temperature of all inlet and outlet streams. The reactant preheater was manufactured by soldering three stainless steel capillaries (1/16 in. outer diameter, 1 mm inner diameter) of 1.2 m in length in direct contact. On the cold side of the heat exchanger, three further Pt100 sensors were used for temperature measurement. Both the reactor and reactant preheater were covered with Armaflex insulation. The synthesis of 3-piperidinopropionic acid ethyl was realized without any solvent. However, water was used as a catalyst and was premixed with piperidine. For the proof-of-concept experiments, the molar ratio of reactants was 1.1 (piperidine/ethyl acrylate) and the molar ratio of water was 0.5 (water/ethyl acrylate). To determine the reaction conversion, the reaction mixture at the reactor outlet was effectively quenched via continuous mixing with a solution of acetic acid in methanol (11 vol.%). Product samples were analyzed using an Agilent 7820 gas chromatograph equipped with a HP-5 column (30 m 0.32 mm 0.25 mm, Agilent) and a flame ionization detector. The inlet temperature was 275 C, the inlet pressure, 50 kPa, and the split ratio, 100:1. An injection volume of 0.5 mL was used in the automatic liquid sampler. The oven temperature (60 C) was kept constant for 2 min and then increased to 200 C with a ramp of 20 K/min.
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
3. Process simulation 3.1. Kinetic model In the reaction of piperidine and ethyl acrylate, a zwitterionic intermediate is formed by a nucleophilic attack on the activated double bond. The subsequent proton transfer towards the neutral product can be considered as a rate determining step in the reaction mechanism. A water molecule or an amine molecule can act as a shuttle for the proton and therefore promotes the charge equalization of the zwitterionic intermediate, resulting in the product 3-piperidino propionic acid ethyl ester. A simplified reaction scheme with two parallel reactions can be used to describe the reaction progress with good accuracy (Scheme 1). In this simplified reaction scheme, fitting parameters are reduced to two observed third order rate constants, i.e., kobs, 2 2 1 s and kobs,B = 0.0170 L2 mol2 s1, and two A = 0.0080 L mol observed activation energies. Consequently, the addition of water to the reaction mixture can strongly enhance the reaction rate. The temperature dependencies of both observed rate constants were found to be surprisingly low, with activation energies EA,obs,A = 0 kJ/ mol and EA,obs,B = 14 kJ/mol. In general, this could indicate a mass transfer limitation of the reaction rate. However, the effect has been found in single-phase flow experiments and short mixing times compared to the reaction time were realized by using a highperformance micromixer (SIMM-V2, Fraunhofer ICT-IMM, Mainz, Germany). Thus, a mass transport limitation can be ruled out. Assuming the zwitterionic intermediate is formed via a preequilibrium step in the reaction mechanism, the observed activation energy would result from both the actual activation energy and the temperature dependency of the equilibrium step, which is given by the standard enthalpy of reaction. For an exothermic reaction, a low observed activation energy can result. Further details on the kinetic investigations can be found in a previous publication [14]. 3.2. Model of the production process As a basis for process simulation, reactor data of a commercially available plate HEX reactor were used. The production reactor (Flow reactor MR500, 3M Technical Ceramics, Kempten) consisted of combinable reactor modules manufactured from silicon carbide (Fig. 1a). Reactor channels with a rectangular cross section (3.5 3.5 mm) were arranged in a zigzag pattern. For a production of approximately 200 tons/year, a total feed rate of 31 L h1 was specified. By combining reactor modules with a total of 0.6 L inner volume (reaction side), a residence time of approximately 1 min can be achieved. With the MR500 reactor (198 L h1 max. throughput), overall heat transfer coefficients of up to 10 kW m2 K1 can be achieved in measurements with water (manufacturer’s data, personal communication). Considering significantly lower transfer coefficients for reduced throughputs and an organic medium, safe reaction control can still be ensured for k = 1 kW m2 K1. Temperature profiles for varying heat transfer coefficients were calculated and are given in the Supplementary information. Scale-up from the operation in microreactors to a similar silicon carbide flow reactor for small-scale production (2 2 mm zigzag channels) has been investigated in a previous publication [14]. Short mixing times, narrow residence time distribution (see
Scheme 1. Simplified reaction scheme used as a kinetic model for the Michael addition.
111
Supplementary material, Fig. S6) and effective heat transfer were found for the silicon carbide plate HEX reactor. The initial process design is shown in Fig. 1b. Subsequent to the reactor, water and unconverted reactants are separated from the product in a distillation column and led back to the reactor feed by a recycle stream. Thus, very high material yields can be obtained in the overall process, even if reactants are not fully converted in the reactor. The temperature increase from the reactor inlet (90 C) to the outlet can be controlled by the heat carrier flow (approx. 10 K). A distillation column is required to provide the bottom product in high purity, which is specified by a molar fraction of 0.995 mol/mol and determines the reboiler duty. Specified temperatures for simulation are given in Fig. 1b. To avoid exposure of the product to higher temperatures in the distillation column, a vacuum distillation at 100 mbar is applied. 3.3. Optimization of process conditions The production process was optimized with regard to energy efficiency. Thus, the column design was adjusted to obtain the specified product purity with minimum required reboiler duty. Because the head product is recycled and only the product purity of the bottom product is specified, a stripping column (10 theoretical stages) without an enriching section can be used. More details on the optimization of the distillation can be found in the Supplementary information. As described above, water is used as a catalyst for the Michael addition. Fig. 2 shows that the conversion in the reactor can be significantly increased by increasing the water ratio W (water/ ethyl acrylate, mol/mol) in the reactor feed. Due to the separation and recycling of unconverted reactants, the reactant conversion in the overall process is close to 100%, independent of the conversion that is achieved in the reactor. In Fig. 3, the transferred heat flow in the HEX reactor (heat source) and in the column reboiler (heat sink) is shown for a varying water ratio in the reactor feed. At low water ratio, more energy is required to separate unconverted reactants from the product. Contrarily, for high amounts of water, reboiler energy is mainly required to remove the water. Furthermore, a large water recycle stream lowers the reactant concentration and residence time in the reactor. An optimum water ratio for minimal reboiler duty exists at W = 0.18. This optimum is valid for the specified reactor data and can change with reactor volume or reactor temperature. Shorter residence time or lower temperature in the reactor would reduce the conversion at the reactor outlet and shift the optimum to a higher water (catalyst) amount in the reactor feed. As follows from Fig. 3, the reboiler duty is below the heat flow that is transferred to the heat carrier in the HEX reactor only for a small range around the optimum. 3.4. Pinch analysis and heat integration With the optimal process conditions, energy targets for heat integration can be determined via pinch analysis [9,11]. Composite curves in the T/H diagram (Fig. 4) represent all hot streams (or cold streams, respectively) and visualize the required enthalpy flow in the respective temperature intervals. The cold composite curve combines all streams that need to be heated (cold feed) or vaporized (i.e., reboiler duty), while the hot composite curve combines all streams with cooling demand (hot product, heat carrier, and condenser stream). As shown in Fig. 4, regarding the right side of the curves, a “nonutility end” at the hot end is feasible if a minimum temperature difference DTmin of 14 K or lower is acceptable for heat exchange. Thus, with an ideal heat exchanger network, no external heating is required. Energy demands for preheating reactants and for the column reboiler can be met by using hot streams leaving the HEX
112
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
Fig. 1. (a) HEX reactor MR500 (3M Technical Ceramics, Germany) with silicon carbide modules; (b) schematic flow sheet of the process with reactant preheating, HEX reactor and distillation column.
and prevents dynamic effects between distillation and reaction. As a result, heat recovery according to the energy targets (Fig. 4) is fully covered. Only external cooling is required for the condenser in the distillation column and to decrease the temperature of the separated product stream. By comparing both process designs in Fig. 5, it can be shown that the amount of required equipment has not been increased. Full heat recovery is realized mainly by using the multifunctional HEX reactor and a heat exchanger for reactant preheating. Both devices are the subject of the following experimental investigations. For practical realization, a multistream reactant preheater, rather than premixing reactor feeds, is advantageous. Thus, mixing inside the reactor results in a defined starting point of reaction. 4. Experimental proof-of-concept Fig. 2. Influence of the water ratio in the reactor feed on the ethyl acrylate conversion in the reactor and the overall process, determined by process simulation.
Fig. 3. Influence of the water ratio in the reactor feed on the transferred heat in the HEX reactor and on the required reboiler duty.
reactor. In Fig. 5b, a heat exchanger network suitable for meeting the energy targets is shown in a schematic flowsheet and compared with the flowsheet of the process before heat integration (Fig. 5a). The hot reaction product stream is used to preheat the reactor feed. Heat integration between the reactor and separation (column reboiler) can be realized using a heat carrier cycle stream. As shown in Fig. 3, the heat transferred in the HEX reactor exceeds the column reboiler duty. However, an additional heat exchanger ensures the adjustment of the reactor temperature
4.1. Reactor design For the transfer from batch to continuous HEX reactors, the main criteria are thermal intensification, flow intensification, reactor dynamics and residence time [1]. For proof-of-concept experiments, a millistructured plate reactor was designed and manufactured via mechanical milling of aluminum (Fig. 6). An important improvement for flow intensification is the zigzag pattern of the channels on the reaction side [5,13]. As a result of repeatedly redirecting the flow by 90 , a secondary flow is generated, making the radial concentration profiles more uniform. Radial mixing of rapidly flowing fluid in the channel center and slowly flowing fluid in the channel wall region narrows the residence time distribution that is typically caused by the laminar flow velocity profile [3,12]. Measurements show that a Bodenstein number close to 100 can be obtained with this channel design at the selected flow rate (see Supplementary information), which indicates a narrow residence time distribution and a reactor behavior close to that of an ideal plug flow reactor. The reactants were mixed in 0.5 mm channels, which were arranged in an arrowshaped geometry. With this type of micromixer, a high mixing performance resulted from high energy input due to the high flow velocity and sharp flow redirection in the mixer [15,16]. The reaction channel is divided into two sections (Fig. 6b). In the entrance region, high reactant concentrations result in a fast heat release rate and the formation of local hot spots. Thus, a zigzag channel with a smaller cross-section (1 1 mm) is applied for enhanced heat transfer and effective temperature control (Channel A). With increasing conversion, the reaction rate of the third-order type reaction strongly decreases and a residence time channel with a larger cross-section (2 2 mm) is applied to obtain high reactant conversion (Channel B). Channel dimensions and heat transfer characteristics for the applied flow rates are given in Table 1.
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
113
Fig. 4. Composite curves for the simulated process after optimization of the process conditions.
of the reactor. Thermal resistance in laminar flow is given by Rth ¼
1
a Aw
¼
dh 1 wh ¼ NulAw NulLðw þ hÞ2
ð4Þ
Thus, low thermal resistance is obtained for long and narrow channels (maximum value at w/h = 1). A 10 3 mm heat carrier channel was realized to obtain fast heat transfer in laminar flow at moderate pressure drop. High flowrates of the heat carrier can increase heat transfer coefficients and decrease thermal resistance. However, a comparatively low flow rate was chosen in order to obtain an easily measureable temperature increase for the heat balance. In Table 1, maximum values for the thermal resistance are given, which were calculated for the theoretical case of a straight channel with the given channel dimensions. To evaluate the dynamic behavior of the reactor, time scales of heat transfer and reaction were compared by calculating the ratio of the characteristic reaction time to the characteristic cooling time N¼
Fig. 5. Schematic flowsheet of the heat exchanger network before (a) and after (b) heat integration.
Heat transfer in the laminar flow regime depends on the channel design and the flow velocity that results for each channel section. For calculation, heat transfer characteristics of a straight channel were chosen as a conservative estimation. Thus, a minimum constant Nusselt number Nu ¼
a dh l
ð3Þ
can be assumed, which is Nu = 2.98 for a quadratic cross-section, assuming a constant wall temperature. However, experimental and simulation results given in the literature [2,5] indicate significantly higher Nusselt numbers for the zigzag pattern compared to the straight channel. Formation of the secondary flow results in a heat transfer enhancement which is intensified with increasing Reynolds number. The heat carrier channel was designed to avoid major thermal resistance on the heat carrier side
dEA;0 4k tr ¼ th rEA;0 dh rcp
ð5Þ
Using Eqs. (1) and (2), the required length of the channel with a smaller cross-section can be estimated by calculating the total conversion in this channel section. At the applied process conditions, an ethyl acrylate conversion of approx. 0.6 at the transition from section A to section B results from the residence time in channel A. Thus, lower heat transfer in channel B is partly compensated for by a slower reaction rate. For both channels, the characteristic time for heat transfer is short compared to the reaction time. The minimum value of N for safe reactor operation depends on reaction order, adiabatic temperature rise and activation energy [10]. Due to the low activation energy of the Michael addition, no stability problems are expected for the obtained values of N. A scale-up of the HEX plate reactor concept requires consideration of the described mass and heat transfer characteristics. The total reactor volume and appropriate channel dimensions have to be chosen with regard to the required residence time and the resulting pressure drop in the channel at the aimed flow rate. Therefore, the laboratory-scale aluminum plate reactor and the production-scale reactor MR500 (Fig. 1a) differ in some design aspects. Although both reactors are based on the zig zag channel design, the MR500 reactor allows for significantly higher throughput, in a longer channel (4.9 m per reactor module) with a larger cross-sectional area (3.5 3.5 mm). However, direct up- or down-scaling with different channel designs is possible provided that plug flow behavior, fast reactant mixing and fast heat removal
114
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
Fig. 6. (a) Plate heat exchanger reactor without insulation; (b) channel design on the reaction side of the middle plate.
can be maintained. To fulfill these requirements, the described design with kinetic-adapted channel size variation can be an effective option. In a production scale reactor such as the MR500 reactor, a channel design with constant cross-section is applicable because significantly higher Reynolds numbers (Re > 1000) can be realized to intensify heat and mass transfer. 4.2. Heat balance To investigate the operability of the HEX reactor in the production process, a lab scale experiment was conducted in the previously described plate reactor. Under similar reaction conditions but at significantly reduced flow rates compared to the simulation (0.9 L/h), 3-piperidino propionic acid ethyl ester was continuously produced for an operating time of 50 min. Fig. 7a shows the flow sheet of the experimental setup, with temperature measuring points at inlets and outlets of the reactant preheater and HEX reactor. Calculated temperature profiles and measured steady-state temperatures are given in Fig. 7b. In the reactant
preheater, both feed streams were heated from 25 C (T1, T2) to a reactor entrance temperature of approximately 90 C (T3, T4). In counter-current flow, a mean logarithmic temperature difference of 7.4 K can be obtained. Due to the nearly identical heat capacity rates of the product and both reactant streams, the local temperature difference between hot and cold fluid can be assumed to be constant throughout the capillary length. However, heat losses to the surroundings and the reaction heat of unconverted reactants in the product stream can influence temperature profiles and have to be taken into consideration. In the HEX reactor, operation in a co-current manner was realized to reduce the hot spot temperature in the entrance region of the reaction channel. Due to the exothermic reaction, the heat carrier temperature increased from 88 C (T7) to 100 C (T8). The temperature increase was adjusted by using the heat carrier flow rate (approx. 5 L/h) to a value close to that of the process simulation (DT = 12 K). At the reactor outlet, the temperature difference between the reaction mixture and heat carrier reached a value of only 0.3 K. Thus, nearly the maximum amount of reaction heat was transferred to the heat carrier. A 93% conversion of ethyl acrylate in the reactor was determined by analyzing samples from the reactor outlet. With the determined temperatures and flow rates, the heat balance of the HEX reactor setup can be used to evaluate the efficiency of heat recovery. For better visualization, the heat balance is depicted in a Sankey type diagram (Fig. 8). Reference temperature for the thermal energy of the fluid streams was defined by the ambient temperature (25 C). In the reactant preheater, thermal energy of the hot product stream was transferred to the reactant streams with only 3% heat losses and leaving hot product. Thus, energy demanding preheating of the reactants in the subsequent reactor was not necessary and the majority of reaction heat could be transferred to the heat carrier for further use in the process. Heat losses in the plate HEX reactor amounted to approximately 20% of the transferred reaction heat. Because those losses mainly depended on insulation and the outer reactor surface, a smaller relative amount could be expected for a scale-up to a production scale reactor. As a result of the proof-of-concept experiment, 72% of the reaction heat is transferred to thermal oil at the temperature level of the reactor. Thus, the heat can be used in all further process steps that require thermal energy below the reactor outlet temperature. Transfer of reaction heat to low temperature reactant streams is avoided by the reactant preheater. In both devices, heat is transferred with a small temperature difference, which leads to a minimum amount of exergy loss. Approximately 100 kJ per liter of product could be saved using the presented experimental setup. 5. Conclusions In this work, process development for an energy-efficient continuous production process was presented for the synthesis of
Table 1 Flow channel dimensions and heat transfer characteristics for the flow channels in the aluminum plate heat exchanger reactor. Reaction channel A
Reaction channel B
Heat carrier channel
wh L av Rth,max
[mm2] [m] [m2/m3] [K/W]
1 1 1.68 4000 0.33
22 2.04 2000 0.28
3 10 2.90 867 0.13
Q Re th N
[mL/min] [] [s] []
15 140 0.6 4.6
15 70 2.4 2.1
82 88 12.7 –
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
115
Fig. 7. (a) Flow sheet of the experimental setup with reactant preheater, HEX reactor and measurement points; (b) calculated temperature profiles and measured inlet/outlet temperatures of the fluid streams in the reactant preheater and in the HEX reactor.
Fig. 8. Sankey type depiction of the energy balance determined from the laboratory experiment.
3-piperidino propionic acid ethyl ester. Based on a proven kinetic model of the reaction, optimum process conditions in a commercially available plate HEX reactor with subsequent distillation have been determined via process simulation. With a defined, specified product purity, the required energy consumption was minimized by choosing an optimal water (catalyst) amount and recycling separated reactants and water after distillation. Process design was adapted to fully exploit the potential for heat recovery given by energy targets from pinch analysis. As a result, operation conditions could be found, where all heat consuming process steps, particularly reactant preheating and distillation, could be operated with heat from the hot outlet streams of the HEX reactor. The experimental part of this work focuses on the practical realization of continuous reaction control and heat recovery under laboratory conditions. For this purpose, two apparatus prototypes have been developed. (1) The reactant preheater, which transfers
heat from one hot process stream to two cold reactant streams in counter-current operation, has been demonstrated as an easy and effective tool for heat recovery and can be implemented in numerous flow chemistry applications at elevated reactor temperature. (2) The plate HEX reactor was manufactured with specific application-adapted flow channels. Channel design and dimensioning is a crucial aspect because temperature increase for heat transfer at the reaction temperature level has to be realized without losing thermal control in the reactor. To meet these requirements, zigzag channels for heat transfer intensification and reaction rate adapted channel widths were chosen. Stable steady state operability of both devices could be demonstrated by temperature and flow monitoring in a proof-of-concept experiment. In terms of energy efficiency, heat losses to the environment play a considerable part in the heat balance. Assuming the transfer to a scale of the simulated production process, transferred heat rates would increase significantly and thus lead to negligible percentages for heat loss. The example highlights how miniaturization and the design of a multifunctional HEX reactor, which considers both reaction kinetics and heat transfer requirements, can lead to a strongly intensified continuous process. Reactant preheating in a countercurrent heat exchanger can be generally applied for continuous syntheses at elevated reactor temperature to save energy and avoid additional equipment for reactant preheating and product cooling. The strategy of removing reaction heat in a HEX reactor for further utilization in the process is primarily applicable for fast and exothermic flow syntheses that have a high heat production potential, but can still be controlled in milli/microstructured reactors. Thus, the heat production potential r0 DHR should be in a range between 10 and 100 kW/L, which for example includes metalorganic reactions, nitrations or polymerization reactions [21]. Energy-demanding process steps for heat integration can be part of product purification, or endothermic reactions in continuous multistep syntheses [20]. In general, energy saving via heat recovery can be considered as a further advantage of continuous
116
S. Schwolow et al. / Chemical Engineering and Processing 108 (2016) 109–116
production in milli/microstructured reactors, which could not be realized in discontinuous processes with reasonable effort. Acknowledgment This work was funded by the German Federal Ministry of Education and Research (BMBF, funding code 03FH012I2). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cep.2016.07.017. References [1] Z. Anxionnaz, M. Cabassud, C. Gourdon, P. Tochon, Heat exchanger/reactors (HEX reactors): concepts, technologies: state-of-the-art, Chem. Eng. Process. Process Intensif. 47 (12) (2008) 2029–2050, doi:http://dx.doi.org/10.1016/j. cep.2008.06.012. [2] Z. Anxionnaz-Minvielle, M. Cabassud, C. Gourdon, P. Tochon, Influence of the meandering channel geometry on the thermo-hydraulic performances of an intensified heat exchanger/reactor, Chem. Eng. Process. Process Intensif. 73 (2013) 67–80, doi:http://dx.doi.org/10.1016/j.cep.2013.06.012. [3] R. Aris, On the dispersion of a solute in a fluid flowing through a tube, Proc. R. Soc. A 235 (1200) (1956) 67–77, doi:http://dx.doi.org/10.1098/rspa.1956.0065. [4] J.-C. Charpentier, Process intensification by miniaturization, Chem. Eng. Technol. 28 (3) (2005) 255–258, doi:http://dx.doi.org/10.1002/ ceat.200407026. [5] A. Fazeli, M. Behnam, Hydrogen production in a zigzag and straight catalytic wall coated micro channel reactor by CFD modeling, Int. J. Hydrogen Energy 35 (17) (2010) 9496–9503, doi:http://dx.doi.org/10.1016/j.ijhydene.2010.05.052. [6] M. Hempel, Novel process windows—a contribution to more sustainable chemistry? Chem. Eng. Technol. 32 (11) (2009) 1651–1654, doi:http://dx.doi. org/10.1002/ceat.200900338. [7] V. Hessel, Novel process windows—Gate to maximizing process intensification via flow chemistry, Chem. Eng. Technol. 32 (11) (2009) 1655–1681, doi:http:// dx.doi.org/10.1002/ceat.200900474.
[8] V. Hessel, D. Kralisch, N. Kockmann, T. Noël, Q. Wang, Novel process windows for enabling, accelerating, and uplifting flow chemistry, ChemSusChem 6 (5) (2013) 746–789, doi:http://dx.doi.org/10.1002/cssc.201200766. [9] I.C. Kemp, Pinch analysis and process integration: A user guide on process integration for the efficient use of energy, 2nd ed., Butterworth-Heinemann, Amsterdam, Boston, 2007. [10] N. Kockmann, Micro process engineering: Fundamentals, devices, fabrication and applications, Wiley-VCH, Weinheim, 2006. [11] B. Linnhoff, The pinch design method for heat exchanger networks, Chem. Eng. Sci. 38 (5) (1983) 745–763. [12] O. Levenspiel, Chemical Reaction Engineering, 3rd ed., Wiley, New York, 1999. [13] V. Mengeaud, J. Josserand, H.H. Girault, Mixing processes in a zigzag microchannel: finite element simulations and optical study, Anal. Chem. 74 (16) (2002) 4279–4286, doi:http://dx.doi.org/10.1021/ac025642e. [14] S. Schwolow, B. Heikenwälder, L. Abahmane, N. Kockmann, T. Röder, Kinetic and scale-up investigations of a michael addition in microreactors, Org. Process Res. Dev. 18 (11) (2014) 1535–1544, doi:http://dx.doi.org/10.1021/acs. oprd.5b00184. [15] S. Schwolow, J. Hollmann, B. Schenkel, T. Röder, Application-oriented analysis of mixing performance in microreactors, Org. Process Res. Dev. 16 (9) (2012) 1513–1522, doi:http://dx.doi.org/10.1021/op300107z. [16] A. Soleymani, E. Kolehmainen, I. Turunen, Numerical and experimental investigations of liquid mixing in T-type micromixers, Chem. Eng. J. 135 (2008) S219–S228, doi:http://dx.doi.org/10.1016/j.cej.2007.07.048. [17] A.I. Stankiewicz, J.A. Moulijn, Process intensification, Ind. Eng. Chem. Res. 41 (8) (2002) 1920–1924, doi:http://dx.doi.org/10.1021/ie011025p. [18] A.I. Stankiewicz, J.A. Moulijn, Process intensification: transforming chemical engineering, Chem. Eng. Prog. 96 (2000) 22–34. [19] S. Taghavi-Moghadam, A. Kleemann, G. Golbig, Microreaction technology as a novel approach to drug design, process development and reliability, Org. Process Res. Dev. 5 (6) (2001) 652–658, doi:http://dx.doi.org/10.1021/ op010066u. [20] J. Wegner, S. Ceylan, A. Kirschning, Flow chemistry—a key enabling technology for (Multistep) organic synthesis, Adv. Synth. Catal. 354 (1) (2012) 17–57, doi: http://dx.doi.org/10.1002/adsc.201100584. [21] T. Westermann, L. Mleczko, Heat management in microreactors for fast exothermic organic syntheses—first design principles, Org. Process Res. Dev. 20 (2) (2016) 487–494, doi:http://dx.doi.org/10.1021/acs.oprd.5b00205.