Reactive Vaporization of Crude Glycerol in a Fluidized Bed Reactor

Aug 3, 2009 - B. Rafii Sereshki,† S.-J. Balan,† G. S. Patience,*,† and J.-L. Dubois‡. Department of Chemical Engineering, Ecole Polytechnique ...
0 downloads 0 Views 3MB Size
Download PDF
1050

Ind. Eng. Chem. Res. 2010, 49, 1050–1056

Reactive Vaporization of Crude Glycerol in a Fluidized Bed Reactor B. Rafii Sereshki,† S.-J. Balan,† G. S. Patience,*,† and J.-L. Dubois‡ Department of Chemical Engineering, Ecole Polytechnique de Montreal, C.P. 6079, Succ., Montre´al, Que´bec, Canada, H3C 3A7, and Arkema, Pierre-Be´nite, France

Crude glycerol is a coproduct of the trans-esterification of vegetable oil and animal fat to biodiesel that contains as much as 5% salt. Processing the glycerol to value added products often requires an expensive distillation step. In this work, we propose a new process in which crude liquid glycerol is fed directly to a fluidized bed, vaporizes, and then reacts to form acrolein over a tungsten doped zirconia catalyst. Salt in the crude glycerol crystallizes and accumulates in the fluid bed. In a commercial process, the salt crystals will be subject to significant mechanical stresses that will cause them to attrit. The attrited powder will be elutriated from the bed and collected by filters. The focus of this work has been to evaluate the effect of salt on catalyst activity as well as its attrition resistance. Thus far, there is little evidence of salt crystallization or migration to the interior of the catalyst particlessafter more than 1 day of continuous operation, catalyst activity remained unchanged. Tests on a standard air jet mill suggest that the mechanical stresses typical of a fluidized bed will be sufficient to attrit the salt and thus minimize its accumulation in the reactor with time. 1. Introduction Renewable feedstocks have taken center stage as a compliment to traditional fossil energy sources. Biodiesel is an interesting alternative energy vector, but glycerol and salt are major coproducts in the standard trans-esterification of vegetable oil and animal fat. The production of biodiesel is growing as is glycerol. Since new production of glycerol is increasing faster than the market need, the price for glycerol has dropped considerably due to the oversupply. Many new processes that use glycerol as a primary feedstock are still in the development stages, and a couple have proceeded to commercialization. Although crude glycerol is inexpensive, the processing costs to eliminate the salt and water through distillation are considerable since the normal boiling point of glycerol is 290 °C. New processes based on glycerol appear very attractive due to its low cost but may be unattractive due to the high cost of separation. Recent developments using glycerol as a primary feedstock include the selective oxidation to glyceric acid and tartronic acid,1,2 dihydroxyacetone,3 and ketomalonic acid.4 These oxygenates are used as valuable fuel additives since glycerol is nonvolatile and thus may not be added directly to fuel.5 The conversion of crude glycerol into propylene glycol based on hydrogenolysis is also an interesting opportunity.6 In the reforming process, glycerol, in the aqueous phase, is converted to hydrogen and carbon monoxide (syngas). Syngas is a primary feedstock for fuels and chemicals via Fischer-Tropsch or, alternatively, for methanol synthesis.7 Glycerol can be used as a feedstock for fermentation of one of the primary components of Sonora and Corterra polyester fibers.8 Glycerol carbonate can serve as a source of new polymeric materials such as glycidol, a component in the production of a number of polymers.9 Solvay recently commercialized a process to produce epichlorohydrin based on the catalytic reaction between glycerine and HCl followed by dehydrochlorination with Ca(OH)2. A recent study reported a route to produce the bioplastic PHB by fermentation * To whom correspondence should be addressed. Tel.: 1 514 3404711 ext. 3439. Fax: 1 514 340-4159. E-mail: gregory-s.patience@ polymtl.ca. † Ecole Polytechnique de Montreal. ‡ Arkema.

from crude glycerol.10 Another investigation proposed crude glycerol for atmospheric autocatalytic organosolv pretreatment (AAOP) to enhance enzymatic hydrolysis.11 Other likely chemicals for development include tetrahydrofuran, acrolein, and acrylic acid. Acrolein is a major component of the chemical industry that is largely used as a feedstock for production of acrylic acid ester, superabsorber polymers, and detergents.12 Acrolein can be obtained from glycerol by a method based on glycerol dehydration on acidic solid catalysts.13 Figure 1 shows an overview of the reaction. Some laboratory-scale experiments of acrolein production from glycerol have been recently published in the literature.12,14-19 In these studies, good selectivity of acrolein has been reached for different catalysts and operating conditions. For example, Ott et al.16 reached a yield of 40% with zinc sulfate in supercritical water (SCW). However, the tests were performed using purified glycerol as the reactant while crude glycerol produced by the biodiesel industry is generally contaminated by salt and water. Studies in the open literature on the influence of salt impurities on the catalyst structure, which make it difficult to develop a process using crude glycerol as a feedstock, have yet to be reported. Different hypotheses can be formulated on the influence of salt impurities on the catalyst structure and performance. On the one hand, the salt impurity in crude glycerol can migrate into the catalyst and block the pores, which decreases the catalyst activity and perhaps even the overall selectivity. On the other hand, the salt may deposit on the catalyst surface while the glycerol goes into the catalyst pores to react. Also, due to the high mechanical stresses in fluidized bed, the salt is expected to attrit and to be captured in the filter downstream. The understanding of the influence of salt impurities is necessary to develop processes for the production of acrolein using crude glycerol as feedstock. For that purpose, the objective of this work is to characterize the influence of salt on the catalyst structure for the production of acrolein. More specifically, it consists of determining if the salt crystallizes on the catalyst surface and can then be removed by attrition or if the salt migrates into the catalyst pores and alters the catalyst structure in a reactive vaporization of crude glycerol process.

10.1021/ie9006968  2010 American Chemical Society Published on Web 08/03/2009

Ind. Eng. Chem. Res., Vol. 49, No. 3, 2010

1051

Figure 1. Acrolein from dehydration of glycerol.

Figure 2. Experimental setup.

Figure 4. SEM results for solids removed from the fluid bed.

Figure 3. Mechanical attrition mill setup.

2. Experimental Section 2.1. Crude Glycerol Dehydration. To study the vaporization of crude glycerol, three types of experiments were conducted. All catalytic testing was carried out in a Hastelloy C-276 vessel 41 mm in diameter and 790 mm tall. Its design pressure and temperature were 14 bar and -30/540 °C, respectively. The reactor was immersed in an electrically heated sand bath (30 A), and it was also equipped with two electrical band heaters on the wall to maintain near isothermal conditions. The

temperature of the bed was monitored with a thermocouple positioned in the bed. Synthetic crude glycerol (containing NaCl) and/or crude glycerol (containing Na2SO4) was introduced to the reactor through a tube from the top of the fluid bed or together with air and argon at the bottom below the distributor. We tested various superficial gas velocities, but most experiments were conducted in the bubbling fluidized-bed regime with a solids bed height around 110 mm. A flanged upper section minimized solids entrainment to the top of the reactor, and a sintered metal filter retained all catalyst in the reactor. Different feed configurations and compositions were explored. Salt would crystallize and block the inlet feed line when crude glycerol was fed together with nitrogen and air to the fluid bed below the grid. Blocking was also problematic when glycerol was fed through the top of the bed into the freeboard. For each experimental condition, a different solution of glycerol in water was fed from the top into the freeboard at different specified gas velocities. To minimize vaporization of glycerol in the feeding tube, which results in salt crystallization and blockage, the tube was insulated along its entire length. Often, the reactor was operated open to the atmosphere at the topsotherwise, the filter would become clogged requiring the experiment to be interrupted. The synthetic crude glycerol used in this work contained about 5 wt % salt, and we varied the amount of water. The crude glycerol contained impurities including free fatty acids and sulfur. A mixture of air and argon was preheated in the sand bath and introduced to the bottom of the reactor into a plenum and then through a sintered metal frit that distributed it evenly across the diameter (Figure 2). Their flow rates were metered with Tylan FC-2900 V-4S flow controllers. The effluents rose through

1052

Ind. Eng. Chem. Res., Vol. 49, No. 3, 2010

Figure 5. EDS results for solids removed from the fluid bed.

Figure 6. SEM and EDS results for solids collected in the mill after 18 h.

Figure 7. SEM and EDS results for solids collected in the filter after 18 h.

the bed and was subsequently condensed in distilled water, and the salt concentration was measured using an electrical conductivity meter with a four cell probe. The accumulated organics in the quench/absorber were sampled frequently and analyzed offline by high performance liquid chromatography (HPLC; Hewlett- Packard 1050) equipped with a variable-wavelength UV detector. A C18 column was used to separate acrolein, and it was detected at a UV wavelength of 195 nm. The absorber was connected to a vacuum to suck the products through the bed since the reactor top was typically open to the atmosphere.

To study the vaporization of crude glycerol, three types of experiments were conducted. In the first experiment, the reactor was charged with spherical glass beads to investigate the salt deposition and crystallization phenomena in the fluidized bed. A mass balance was carried out to confirm that the salt crystallizes in the reactor and remains with the solids. The second series of experiments involved VPO catalyst (vanadyl pyrophosphate (VO)2P2O7) to determine if the salt remaining in the reactor deposits on the exterior surface of the catalyst and if it could easily be removed by mechanical attrition. The

Ind. Eng. Chem. Res., Vol. 49, No. 3, 2010

1053

Figure 8. SEM photos from fresh catalyst Z-1044 before preparation by FIB.

Figure 9. SEM photos from fresh catalyst Z-1044 after preparation by FIB.

Figure 10. SEM photos from catalyst Z-1044 on stream for 24 h after preparation by FIB.

third series was conducted with a tungsten oxide doped zirconia catalyst ((WO3)0.056(ZrO2)0.944 with DIKK trade name Z-1044). This was carried out to verify if the salt remaining in the reactor altered the structural properties of the catalyst particles, which would affect the reactivity or selectivity, or if the salt would remain on the exterior surface of the catalyst. Furthermore, the acrolein yield achieved with this catalyst was estimated. The VPO is a commercial catalyst and has a

high attrition resistance whereas the tungsten oxide doped zirconia has been prepared in the laboratory, and thus, its attrition resistance is not representative of the expected performance at an industrial scale. In addition, after the second and third experiments the VPO and Z-1044 catalysts were characterized using SEM (scanning electron microscopy) and EDS (energy dispersive spectroscopy) techniques, allowing us to examine its composition.

1054

Ind. Eng. Chem. Res., Vol. 49, No. 3, 2010

Figure 11. EDS result for fresh catalyst Z-1044.

µm, and then spray-dried with polysilicic acid to form a porous silica shell containing up to 10% silica. The spray-dried catalyst is somewhat less attrition resistant than calcined-activated catalyst but sufficiently resistant to withstand 100 h in the fluid bed. A narrow fraction of the catalyst particles was analyzed, between about 50 and 120 µm, as measured by a Horiba particle size analyzer. The particle density was on the order of 1900 kg/m3, and the skeletal density, measured by helium pycnometry, was almost 3000 kg/m3. Tungsten-doped zirconia catalyst ((WO3)0.056(ZrO2)0.944 with DIKK trade name Z-1044) was purchased from Daiichi Kigenso Kagaku Kogyo. 3. Results and Discussion

Figure 12. EDS result for inside the catalyst particle on stream for 24 h.

Figure 13. EDS result for the surface of the catalyst particle on stream for 24 h.

2.2. Attrition Testing. An air jet attrition mill was designed to simulate the mechanical stresses inherent in a commercial fluid bed to examine whether the salt could be separated passively from the catalyst during normal operating conditions. The jet mill consists of an orifice 0.41 mm in diameter through which air is introduced at velocity of 230 m/s. The ensuing jet impinges on the particles causing the particles to collide at a high frequency and mechanical stress resulting in particle breakage (attrition). As the particles attrit, the fines that are created are carried upward through a 25.4 mm tube to an overhead filter as shown in Figure 3. The attrited particles and the bed were analyzed by SEM and EDS and particle size distribution was measured. In this work, the mill was loaded with 11 g of VPO catalyst which saw 15 h of operation in the fluidized bed reactor. After the unit was cleaned, it was run for another 18 h. The solids collected during this period represent accelerated aging that powder would be exposed to a commercial fluidized bed. The mass of attrited particles in the thimble filter and particles in the jet cup were measured. The vast majority of particles collected in the filters were salt crystals. 2.3. Catalyst Preparation. The VPO precursor was synthesized on a commercial scale in an organic medium with isobutanol and benzyl alcohol. It was micrometerized to 1-2

3.1. Evaporation over Glass Beads. This experiment was conducted to confirm that the salt impurity in crude glycerol remains in the bed while the glycerol evaporates. The reactor was loaded with 150 g of spherical glass beads. A 45% w/w solution of glycerol in water was fed to the bed from the top at a rate of 0.5 mL/min. At the same time argon was fed to the bottom at a rate of 500 mL/min, the reactor was operated for 2 h at 300 °C. The off gases were condensed in water, and electrical conductivity was measured. On the basis of conductivity measurements and a mass balance, we calculated that 99.8% of salt remained in the reactorsessentially all of the salt. Individual salt crystals were evident. There was little evidence to suggest that crystals formed or adhered to the surface of the glass beads. Clearly the experiments demonstrate the utility of employing a fluidized bed to separate salt from glycerol in a single step. Commercialization of this technology should be competitive with standard distillation. However, work is required to minimize the tendency of the salt to crystallize in the feed nozzle and lines. 3.2. Evaporation over VPO Catalyst. The fluidized bed was charged with 150 g of VPO catalyst. A 20% w/w solution of synthetic crude glycerol (with 5% NaCl) in water was fed at a rate of 3 mL/min from the top of the reactor into the bed of catalyst operating at 350 °C. A 50/50 mixture of air and argon entered below the frit at the bottom of the reactor at a rate of 440 mL/min. The experiment was run for 15 h, and four catalyst samples were collected after 4, 8, 12, and 15 h. SEM photos were taken of each sample, and EDS was employed to determine the composition of the catalyst surface. Figure 4 shows the results of SEM pictures of the VPO before and after use. Initially, the VPO particles appeared spherical. After 4 h, bridges form between particles and smaller particles were evident around these bridges after 8 h. After 12 h, large salt crystalssthe size of the VPO catalystswere observable. The results of EDS measurements in Figure 5 indicate that the concentration of the salt increased with increasing time in the fluid bed. This trend was confirmed by measuring the electrical conductivity of a 1 g sample of powder withdrawn from the fluid bed: after 4 h of operation, the conductivity was measured to be 1 mS/cm, and it was approximately 3 mS/cm after 12 h. 3.2.1. Mechanical Attrition Mill. A sample of 11 g of the VPO-salt mixture was loaded into the attrition mill. After 18 h, the particles collected in the mill and the thimble filter were analyzed by SEM, EDS, and a Horiba particle size analyzer. Figure 6 illustrates the unattrited particles that remain in the mill after the 18 h test with a scale of 200 µm. The large particles resemble the VPO microspheres and there are some cracked

Ind. Eng. Chem. Res., Vol. 49, No. 3, 2010 Table 1. Acrolein Yield As a Function of Temperature temperature in the bed (°C) duration of experiment (h) total salt fed to the fluidized bed (g) acrolein yield (%)

300 4 7 4

320 8 14 15

350 42 30 21

spheres and small angular particles. The attrited particles collected in the filter are shown in Figure 7 with a four times higher magnification. On the basis of EDS measurements, there is a significantly higher concentration of NaCl in the thimble filter compared to the mill, which indicates that the salt is preferentially attriting. On the basis of the Horiba analysis of the particle size distribution, the average diameter of the powder remaining in the mill was 79 µm while it was only 2 µm in the thimble filter. Note that the gas velocity in the fluid bed was about 0.10 m/s. This low gas velocity together with the sintered metal grid plate results in very low attrition rates, and thus, elutriation of salt from the bed is negligible. Solids were sampled from the bed while in a commercial process they would attrit and be collected in the filters. Attrition rates will be orders of magnitude higher in a commercial reactor. Orifice plates are commonly installed to distribute gas across the diameter of turbulent fluidized beds, and the gas velocities may be as high as 30-50 m/s. Besides the mechanical stresses at the orifice due to particle-particle collisions, particles elutriated from the bed will experience shear stresses in cyclones. The highest attrition zone is at the grid and spargers where high gas pressure drop is required to evenly distribute the gas across the reactor. The attrition mill simulates the best mechanical stresses occurring at the grids and spargers. The results of this test indicate that the mechanical stresses typical of a fluid bed would be sufficient to separate the salt from the catalyst. 3.3. Reactive Evaporation over Z-1044 Catalyst Bed. In this series of experiments, the reactor was loaded with 150 g of Z-1044 catalyst and the temperature was set at temperatures of 300, 320, and 350 °C. A 20% w/w solution of crude glycerol in water (3 mL/min) was fed from the top to the bottom of the reactor, and air (220 mL/min) and argon (220 mL/min) were preheated in the sand bath and entered the bottom of the reactor to fluidize the bed. Crude glycerol was evaporated reactively over Z-1044 catalyst, and the acrolein yield was estimated based on the accumulated liquid volume collected in an ice bath. The outlet of the reactor was condensed in water and analyzed by Varian HPLC. The C18 column was maintained at 28 °C, and water was used as mobile phase for analysis. The acrolein yield was calculated 21% at 350 °C. It was approximately constant throughout the course of the 34 h. Samples were taken on an hourly basis, and the yield remained unchanged. The results are summarized in Table 1. At the end of 42 h, a total of approximately 75 g of salt was fed to the reactor, which represents 50% of the total mass of catalyst. The catalyst particles were characterized by using FIB (focus ion beam), EDS, and SEM techniques. In order to analyze inside the catalyst particles, the particles were cut with FIB and EDS was employed to measure the composition inside the particle. Figure 8 shows a fresh catalyst particle before FIB preparation, and Figures 9 and 10 demonstrate the particle after preparation by FIB for fresh and for the catalyst on stream for 24 h, respectively. Figure 11 illustrates the results from EDS for fresh catalyst. The fresh catalyst contains tungsten oxide and zirconium oxide. Figures 12 and 13 show the EDS results for used

1055

catalyst. The concentrations of Na and Cl were very low in the central part of the catalyst, and they was much higher on the exterior surface indicating that migration of salt toward the center was negligible. 4. Conclusions Evaporation of glycerol over an inert bed was experimentally investigated in this work. It was demonstrated that glycerol will evaporate in a fluidized bed reactor leaving behind salt crystals. This technology helps avoid the expensive processing cost to remove the salt through distillation before treating crude glycerol. This experiments demonstrate the potential to eliminate the salt content in crude glycerol from the products. The salt either will form a discontinuous layer on the surface of the catalyst particles or crystals independent of the catalyst. On the basis of attrition testing in a jet mill, we showed that the salt was only loosely bound to the surface and detached from the catalyst with mechanical agitation typical of commercial fluid bed reactors. On the basis of an extended experiment with a Z-1044 catalyst, glycerol appears to react with the catalyst unimpeded by the saltsthat is, salt does not appear to block the pores of the catalyst or migrate to the interior to any great extent. Acknowledgment Financing for the experimental equipment was provided by the Natural Sciences and Engineering Research Council of Canada as well as the Canada Foundation for Innovation. Literature Cited (1) Garcia, R.; Besson, M.; Gallezot, P. Chemoselective Catalytic Oxidation of Glycerol with Air on Platinum Metals. Appl. Catal. A: Gen. 1995, 127 (1-2), 165. (2) Dimitratos, N.; Lopez-Sanchez, J. A.; Lennon, D.; Porta, T.; Prati, L.; Villa, A. Effect of Particle Size on Monometallic and Bimetallic (Au, Pd)/C on the Liquid Phase Oxidation of Glycerol. Catal. Lett. 2006, 108, 147. (3) Ciriminna, R.; Palmisano, G.; Della Pina, C.; Rossi, M.; Pagliaro, M. One-Pot Electrocatalytic Oxidation of Glycerol to DHA. Tetrahedron Lett. 2006, 45, 6993. (4) Ciriminna, R.; Pagliaro, M. The Effects of Material Properties on the Activity of Sol-Gel Entrapped Perruthenate Supercritical Conditions. AdV. Synth. Catal. 2003, 345 (3), 383. (5) Liotta, F. J.; Karas, L. J.; Kesling, H. Diesel fuel. US Patent 5,308,365, 1994. (6) Dasari, M.; Kiatsimkul, P.; Sutterlin, W.; Suppes, G. J. Low-Pressure Hydrogenolysis of Glycerol to Propylene Glycol. Appl. Catal. A: Gen. 2005, 281, 225. (7) Soares, R. R.; Simonetti, D. A.; Dumesic, J. A. Glycerol as a Source for Fuels and Chemicals by Low-Temperature Catalytic Processing. Angew. Chem., Int. Ed. 2006, 45 (24), 3982. (8) Lin, R.; Liu, H.; Hao, J.; Cheng, K.; Liu, D. Enhancement of 1,3Propanediol Production by Klebsiella Pneumoniae with Fumarate Addition. Biotechnol. Lett. 2005, 27 (22), 1755. (9) Rokicki, G.; Rakoczy, P.; Parzuchowski, P.; Sobiecki, M. Hyperbranched Aliphatic Polyethers Obtained from Environmentally Benign Monomer: Glycerol Carbonate. Green Chem. 2005, 7, 529. (10) Mothes, G.; Schnorpfeil, C.; Ackermann, J. U. Production of PHB from Crude Glycerol. Eng. Life Sci. 2007, 7 (5), 475. (11) Sun, F.; Chen, H. Organosolv Pretreatment by Crude Glycerol from Oleochemicals Industry for Enzymatic Hydrolysis of Wheat Straw. Bioresour. Technol. 2008, 99 (13), 5474. (12) Corma, A.; Huber, G. W.; Sauvanaud, L.; O’Connor, P. Biomass to Chemicals: Catalytic Conversion of Glycerol/Water Mixtures into Acrolein, Reaction Network. J. Catal. 2008, 257 (1), 163. (13) Girke, W.; Klenk, H.; Arntz, D.; Hass, T.; Neher A. Process for the production of acrolein. US patent 5,387,720, 1995.

1056

Ind. Eng. Chem. Res., Vol. 49, No. 3, 2010

(14) Chai, S.-H.; Wang, H.-P.; Liang, Y.; Xu, B.-Q. Sustainable Production of Acrolein: Gas-Phase Dehydration of Glycerol over Nb2O5 Catalyst. J. Catal. 2007, 250 (2), 342. (15) Ning, L.; Ding, Y.; Chen, W.; Gong, L.; Lin, R.; Yuan, L.; Xin, Q. Glycerol Dehydration to Acrolein over Activated Carbon-Supported Silicotungstic Acids. Chin. J. Catal. 2008, 29 (3), 212. (16) Ott, L.; Bicker, M.; Vogel, H. Catalytic Dehydration of Glycerol in Sub- and Supercritical Water: A New Chemical Process for Acrolein Production. Green Chem. 2005, 8, 113. (17) Pagliaro, M.; Ciriminna, R.; Kimura, H.; Rossi, M.; Della Pina, C. From Glycerol to Value-Added Products. Angew. Chem., Int. Ed. 2007, 46 (24), 4434.

(18) Tsukuda, E.; Sato, S.; Takahashi, R.; Sodesawa, T. Production of acrolein from glycerol over silica-supported heteropoly acids. Catal. Commun. 2007, 8 (9), 1349. (19) Watanabe, M.; Iida, T.; Aizawa, Y.; Aida, T. M.; Inomata, H. Acrolein Synthesis from Glycerol in Hot-Compressed Water. Bioresour. Technol. 2007, 98 (6), 1285.

ReceiVed for reView April 30, 2009 ReVised manuscript receiVed July 16, 2009 Accepted July 17, 2009 IE9006968