Hybrid separations and adsorption/reaction processes: the case of isomerization/separation of xylenes

Hybrid Separations and Adsorption/Reaction Processes: The Case of Isomerization/Separation of Xylenes Dissertation presented to Faculdade de Engenharia da Universidade do Porto for the degree of PhD in Chemical and Biological Engineering by Jonathan Carlos Gonçalves da Silva Supervised by Professor Alírio Egídio Rodrigues Laboratory of Separation and Reaction Engineering, Associated Laboratory LSRE/LCM Department of Chemical Engineering, Faculty of Engineering, University of Porto April 2015 This thesis was financially supported by Fundação para a Ciência e a Tecnologia (Portugal) through the PhD grant SFRH/BD/74402/2010. The work was also co-financed by QREN, ON2 and FEDER (Project NORTE-07-0124-FEDER-000007 – Multifunctional Reactors/Process Intensification) and co-financed by FCT/MEC and FEDER under Programe PT2020 (Project UID/EQU/50020/2013). FEUP-LSRE/LCM - Universidade do Porto © Jonathan Carlos Gonçalves da Silva, 2011-2015 All rights reserved Acknowledgments In the first place, I would like to thank my supervisor Professor Alírio Rodrigues for giving me the opportunity to come to LSRE and continue my professional formation in chemical engineering. His invaluable guidance and vast experience were an essential keystone in the path for achieving my goals. I would also like to thank all the professors and colleagues at the LSRE for their patience and willingness to help me anyway they could, whether in terms of basic concepts or experimental work in the lab, they always found the time to assist me. Additionally, all their support outside the LSRE was imperative for me to adapt to a different environment and culture allowing me to grow as a human being as well. To all family I will always be grateful. Even though they are far away, all their love and support gave me the strength to keep pursuing my goals and dreams. Also to all my Venezuelan friends, here in Porto and those spread around the globe, who gave me the courage to face all the challenges that have presented in this period of my life. To all and each one of you… GRACIAS… TOTALES!!! Abstract Aromatics and specifically p-xylene are building blocks in the production of a variety of products used every day. Since worldwide population increases at exponential rate, the demand of this type of products will increase as well. The demand of p-xylene is expected to grow at a rate of 7.4% until 2022; moreover in Portugal the gap between demand and supply in p-xylene and benzene is considerably high. Industrial production of p-xylene consists of a cycle loop where p-xylene is separated and the other isomers are sent to an isomerization unit where p-xylene is produced and recycled back to said separation unit. Since isomerization of xylenes thermodynamically favors the production of m-xylene, the energy consumption within the recycle loop is significantly high to fulfill the p-xylene demands. In order to overcome the equilibrium constraints, a multifunctional reactor combining separation and isomerization for the production of p-xylene together with the modification of the existing aromatics complex in Portugal is proposed to increase the production of benzene and p-xylene. The existing simulated moving bed unit for separation of p-xylene is turned into a simulated moving bed reactor for separation and production of p-xylene at an intermediate concentration which is further purified in a single stage crystallization unit. More p-xylene and benzene are produced through selective toluene disproportionation to convert less valuable toluene into said products. The simulated moving bed reactor is analyzed to determine the appropriate arrangement of columns maintaining the actual dimensions of the unit and the optimal particle size according to the maximum pressure drop that leads to the higher production of p-xylene. Calculations based on the true moving bed reactor approach and the actual shifting of the ports leads to two columns, six columns, fourteen columns, and two columns in the first, second, third, and fourth zone respectively where 15% of each column comprises a bed of homogeneous mixture with adsorbent to adsorbent plus catalyst weight ratio of 0.4 followed by a second bed filled with adsorbents using a particle diameter of 0.62 mm. The dual-bed unit allows to obtain in the extract 175% of the p-xylene fed to the unit at 200 ºC. The proposed aromatics complex including the dual-bed simulated moving bed reactor, the selective toluene disproportionation, and the single stage crystallization unit results in an increase of 170% and 72% of production of benzene and p-xylene respectively. Finally, three zeolites are experimentally studied with the purpose to be used in the simulated moving bed reactor. Beta zeolite with Si/Al ratio of 35 exhibits the best performance due to its proper balance of acidity. Resumo Os aromáticos e especificamente o p-xileno atuam como precursores para a produção de uma variedade de produtos usados todos os dias. Uma vez que a população mundial aumenta em ritmo exponencial, a procura deste tipo de produtos também vai aumentar. A procura de p-xileno deverá crescer a uma taxa de 7,4% até 2022; além disso, em Portugal a diferencia entre a oferta e a procura de p-xileno e benzeno é consideravelmente alta. A produção industrial do p-xileno é constituída por um ciclo em que o p-xileno é separado dos outros isómeros os quais são enviados para uma unidade de isomerização onde o p-xileno é produzido e reciclado de volta para a referida unidade de separação. Já que o m-xileno é favorecido termodinamicamente na isomerização de xilenos o consumo de energia no ciclo é significativamente alto para atender a procura do p-xileno. A fim de superar as limitações de equilíbrio, propõe-se a utilização de um reator multifuncional que efetuará simultaneamente a separação e isomerização de p-xileno em conjunto com a alteração do complexo aromático existente em Portugal de forma a aumentar a produção do benzeno e p-xileno. A unidade de leito móvel simulado existente para a separação do p-xileno é substituída por um reator de leito móvel simulado para a separação e produção do p-xileno a uma concentração intermédia que é posteriormente purificado numa unidade de cristalização duma etapa. p-Xileno e benzeno são também produzidos através da dismutação seletiva de tolueno. O reator de leito móvel simulado é analisado para determinar a distribuição apropriada de colunas mantendo as dimensões reais da unidade e o tamanho de partícula ótimo de acordo com a queda de pressão máxima que conduz ao aumento da produção do p-xileno. Cálculos baseados no reator de leito móvel verdadeiro e no deslocamento real das linhas de entrada e saída conduz a duas, seis, catorze e duas colunas na primeira, segunda, terceira, e quarta zona respetivamente, onde 15% de cada coluna compreende uma primeira camada que consiste numa mistura homogénea com uma relação de adsorvente para adsorvente mais catalisador de 0.4 em peso seguida por uma segunda camada de adsorvente usando um diâmetro de partícula de 0.62 mm. A unidade de duas camadas permite a obtenção no extrato de 175% de p-xileno alimentado à unidade a 200 ºC. O complexo aromático proposto, incluindo o reator de leito móvel simulado, a dismutação seletiva de tolueno e a unidade de cristalização resulta num aumento de 170% e 72% na produção de benzeno e p-xileno. Finalmente, três zeólitos são experimentalmente estudados com a finalidade de serem usados no reator de leito móvel simulado. O zeólito Beta com uma relação Si/Al de 35 apresenta o melhor desempenho devido ao seu balaço adequado de acidez. vi List of Figures Figure 2.1 Molecular structures of xylene isomers and ethylbenzene .......................................... 8 Figure 2.2 Equilibrium product distribution for mixed xylenes at atmospheric pressure [1] ....... 8 Figure 2.3 Benzene derivatives [2] ............................................................................................... 9 Figure 2.4 Xylenes derivatives [2] ................................................................................................ 9 Figure 2.5 World p-xylene supply/demand [5] ........................................................................... 10 Figure 2.6 Galp aromatics complex [7] ...................................................................................... 11 Figure 2.7 Integrated UOP aromatics complex ........................................................................... 13 Figure 2.8 Catalyst selectivation in PxMax technology [25] ...................................................... 17 Figure 2.9 Block diagram of PxMax with crystallization [29] ................................................... 18 Figure 2.10 p-Xylene separation technologies [30] .................................................................... 19 Figure 2.11 Block diagram of ExxonMobil crystallization process [29] .................................... 20 Figure 2.12 Simplified flow diagram of a single crystallization stage [29] ................................ 20 Figure 2.13 Parex flow diagram [30] .......................................................................................... 24 Figure 2.14 Eluxyl stand-alone version [53] ............................................................................... 24 Figure 2.15 Schematic view of Eluxyl twin raffinate MX/OX splitter crystallization unit [54] 25 Figure 2.16 Schematic view of Eluxyl twin raffinate MX SMB unit [54] .................................. 26 Figure 2.17 UOP’s Isomar flow diagram [57] ............................................................................ 28 Figure 2.18 ExxonMobil’s dual bed catalyst system [25] ........................................................... 30 Figure 2.19 Simplified xylenes loop flowscheme including Oparis process [61] ...................... 31 Figure 3.1 Reaction scheme for xylene isomerization. PX = p-Xylene, MX = m-Xylene, OX = o-Xylene. K1=OX/MX, K2=MX/PX, K3=PX/OX. The triangular scheme adds to the mechanism the direct conversion between oand p-xylene in order to account for the influence of intracrystalline mass-transfer resistance [14]. ................................................. 48 Figure 3.2 Equilibrium constants Ki as a function of temperature according to equations (3.12) to (3.14). (♦) i = 1 (▲) i = 2 (■) i = 3. Error bars are larger for temperatures above 440 K due to the increase in the uncertainty of the saturation functions of m-xylene and o-xylene. ... 49 Figure 4.1 Simplified scheme of the aromatics complex under study. ....................................... 54 Figure 4.2 Isomerization unit ...................................................................................................... 55 Figure 4.3 Reactions for a) xylene isomerization, b) ethylbenzene isomerization, and c) ethylbenzene dealkylation from Silady [3] ......................................................................... 56 Figure 4.4 Simplified reactor scheme ......................................................................................... 58 Figure 4.5 Reaction system for ethylbenzene and xylene isomerization .................................... 61 vii Figure 4.6 Weight fraction of each species against dimensionless radial coordinate for a given day ....................................................................................................................................... 64 Figure 5.1 Current (left) and proposed (right) aromatics complex. ............................................ 73 Figure 5.2 Xylene isomerization reaction scheme ...................................................................... 75 Figure 5.3 Adsorption isotherms for each species. Isotherms for mand o-xylene are the same [12] ...................................................................................................................................... 76 Figure 5.4 Flow diagram to determine separation regions for different flow conditions ............ 80 Figure 5.5 Separation regions of 6-9-6-3 configuration and 69 s switching time for: a) several values of 4 and fixed 1 = 5.5 b) several values of 1 and fixed 4 = 0.4. Darker region indicates optimum value. ..................................................................................................... 81 Figure 5.6 Bulk concentration profiles for 1 = 5.5; 2 = 1.13; 3 = 1.36; 4 = 0.4 in 6-9-6-3 configuration and 69 s switching time................................................................................. 83 Figure 5.7 Separation regions for different configurations and 69 s switching time (1 = 5.5 and 4 = 0.4). .............................................................................................................................. 85 Figure 5.8 Bulk concentration profiles for 1 = 5.5; 2 = 1.12; 3 = 1.43; 4 = 0.4 in 2-6-14-2 configuration and 69 s switching time................................................................................. 86 Figure 5.9 Variation of productivity with desorbent consumption for different configurations . 88 Figure 6.1 Separation regions for 6-9-6-3 configuration and 69 s switching time with several values of 1 and fixed 4 (left) and several values of 4 and fixed 1 (right) for particle diameter: a) 0.5 mm (1=5.5 and 4=0.4) b) 0.7 mm (1=5.5 and 4=0.2) c) 0.8 mm (1=6.0 and4=0.4) d) 0.9 mm (1=6.5 and 4=0.4). Darker regions indicate the optimum values for 1 and 4. .............................................................................................................................. 96 Figure 6.2 Separation regions for different configurations and 69 s switching time for particle diameter: a) 0.5 mm (1=5.5 and 4=0.4) and b) 0.7 mm (1=5.5 and 4=0.2) c) 0.8 mm (1=6.0 and 4=0.4) d) 0.9 mm (1=6.5 and 4=0.4). ........................................................................ 98 Figure 6.3 Variation of productivity with desorbent consumption for different configurations for particle diameter: a) 0.5 mm b) 0.7 mm c) 0.8 mm d) 0.9 mm. ........................................ 100 Figure 6.4 Productivity (PR), pressure drop (ΔP), and desorbent consumption (DC) with 2-6-142 configuration for particle diameters: 0.5, 0.62, 0.7, 0.8, and 0.9 mm without maximum pressure drop constraint..................................................................................................... 100 Figure 6.5 Productivity (PR), pressure drop (ΔP), and desorbent consumption (DC) with 2-6-142 configuration for particle diameters: 0.5, 0.62, 0.7, 0.8, and 0.9 mm subject to maximum pressure drop constraint of 685 kPa. ................................................................................. 102 Figure 7.1 Distribution of adsorbents and catalysts within the columns of the simulated moving bed reactor. L1 represents the length of the first bed with homogeneous mixture of adsorbent and catalyst; L2 corresponds to the second bed with just adsorbents ................................ 109 viii Figure 7.2 Flow diagram to determine the optimum adsorbent to adsorbent plus catalyst weight ratio (𝜑) ............................................................................................................................. 112 Figure 7.3 Separation regions of the best configurations based on TMBR approach from Table 6.3 calculated using the SMBR approach. Flow rates in zones 1 and 4 (Q1 and Q4) and switching time (ts) are those equivalent to TMBR ............................................................ 115 Figure 7.4 Variation of productivity and desorbent consumption of configurations 2-6-14-2, 2-515-2, and 2-4-16-2 for several switching times. Desorbent consumption of 2-6-14-2 and 25-15-2 are overlapped........................................................................................................ 116 Figure 7.5 Concentration profile for a) SMBR under cyclic steady-state at the middle of the switching time and b) equivalent TMBR .......................................................................... 119 Figure 8.1 Block diagram and mass balance of the proposed aromatics complex .................... 128 Figure 9.1 Reaction schemes for xylene isomerization: a) linear b) triangular ........................ 139 Figure 9.2 Scanning electron microscopy images of H-BEA 25 .............................................. 141 Figure 9.3 Scanning electron microscopy images of H-MOR 30 ............................................. 142 Figure 9.4 Scanning electron microscopy images of H-BEA 35 .............................................. 143 Figure 9.5 X-ray diffraction (XRD) of H-BEA 25 and H-BEA 35 compared to reported powder pattern by International Zeolite Association (www.iza-structure.org) .............................. 144 Figure 9.6 X-ray diffraction (XRD) of H-MOR 30 compared to reported powder pattern by International Zeolite Association (www.iza-structure.org) ............................................... 144 Figure 9.7 Nitrogen adsorption of H-BEA 25 (●), H-MOR 30 (▲), and H-BEA 35 (■) at 77 K (P0 = 1 atmosphere) ........................................................................................................... 145 Figure 9.8 Experimental (symbol) and predicted (solid line) concentrations of o-xylene (■), mxylene (▲), p-xylene (♦), and toluene (●) for each reaction at temperatures: a) 473 K b) 493 K c) 513 K ......................................................................................................................... 148 ix List of Tables Table 2.1 Xylene isomerization units .......................................................................................... 31 Table 2.2 Recent patents in xylene production ........................................................................... 32 Table 3.1 Wagner parameters for equation (3.1) ........................................................................ 45 Table 3.2 Molar thermodynamic functions, enthalpy (H) and entropy (S) at saturation pressure (Ps) at temperatures from 250 to 550 Kd ............................................................................. 46 Table 3.3 Entropy (S0) and enthalpy (H0-H0(Tr)) of reference elements at reference temperature Tr = 298.15 K and standard pressure P0 = 100 kPa from Chase [13] ................................. 47 Table 3.4 Gibbs energy of formation (ΔfG0/RT) of xylene species in liquid phase .................... 48 Table 3.5 Equilibrium product distribution (mol %) based on the equilibrium constants from equations (3.12) to (3.14)a. .................................................................................................. 49 Table 4.1 Characteristics and dimensions of the catalyst within the reactor............................... 57 Table 4.2 Summarized reactor data used in the mathematical modeling .................................... 59 Table 4.3 Comparison of kinetic constantsa ................................................................................ 64 Table 4.4 Actual and calculated outlet weight fractions. Calculated values are predicted by the model with optimized kinetics ............................................................................................ 65 Table 5.1 Thermodynamic and physical parameters from Bergeot [12] ..................................... 76 Table 5.2 Optimum points for 1 values between 3.0 and 6.0 and 4 values between 0.3 and 1.0 for 6-9-6-3 configuration and 69 s switching time. ............................................................ 82 Table 5.3 Optimum points for several configurations and 69 s switching time (1 = 5.5 and 4 = 0.4)....................................................................................................................................... 84 Table 5.4 Optimization for several configurations at different desorbent consumption ............. 87 Table 6.1 Optimum points for 6-9-6-3 configuration and 69 s switching time for each particle diameter ............................................................................................................................... 97 Table 6.2 Optimization for 2-6-14-2 configuration with different particle sizes under the maximum pressure drop restriction (685 kPa). ................................................................. 101 Table 6.3 Optimization for several configurations with 0.62 mm particle size, 0.06 m3 kg-1 of desorbent consumption, and a maximum pressure drop of 685 kPa. ................................ 102 Table 7.1 Mass concentration (kg m-3) at the inlet of each column in zone 3 for configuration 26-14-2 with optimized flow conditions from Chapter 6 .................................................... 113 Table 7.2 Optimum adsorbent to adsorbent plus catalyst weight ratio (𝜑) in each column for several lengths of the first bed (L1) as percentage of column length (Lc) .......................... 114 Table 7.3 Performance of the best configurations based on TMBR approach from Table 6.3 calculated using the SMBR approach. Flow rates in zones 1 and 4 (Q1 and Q4) and switching x time (ts) are those equivalent to TMBR while flow rates in zones 2 and 3 (Q2 and Q3) corresponds to the peak within the separation region. ...................................................... 114 Table 7.4 Performance of configurations 2-6-14-2, 2-5-15-2, and 2-4-16-2 for several switching times (ts) and fixed flow rates in zones 1 and 4 (Q1 and Q4). Flow rates in zones 2 and 3 (Q2 and Q3) corresponds to the peak within the separation region .......................................... 116 Table 7.5 Performance of configuration 2-6-14-2 with 70 s switching time for several lengths of the first bed (L1) as percentage of column length (Lc) and fixed flow rates in zones 1 and 4 (Q1 and Q4). Flow rates in zones 2 and 3 (Q2 and Q3) corresponds to the peak within the separation region ............................................................................................................... 117 Table 7.6 Performance of configuration 2-6-14-2 for several adsorbent to adsorbent plus catalyst weight ratio (𝜑) and fixed flow rates in zones 1 and 4 (Q1 and Q4). Flow rates in zones 2 and 3 (Q2 and Q3) corresponds to the peak within the separation region ................................. 118 Table 8.1 Selective toluene disproportionation unit parameters ............................................... 126 Table 8.2 Solubility and actual molar fractions of the mother liquor at several temperatures for cases I and II ...................................................................................................................... 130 Table 8.3 Configuration of the SMBR unit for cases I and II with 70 s switching time ........... 130 Table 9.1 Surface area and micropore area and volume of H-BEA 25, H-MOR 30, and H-BEA 35 ....................................................................................................................................... 145 Table 9.2 o-Xylene conversion after 8 hours for 10 g of catalyst and 250 ml .......................... 146 Table 9.3 Experimental concentration (mol L-1) of xylenes and toluene at different extent of the reaction for temperatures 473, 493, and 513 K ................................................................. 147 Table 9.4 Kinetic parameters for scheme 1, 2, and those obtained by Cappellazzo et al. [18] . 149 Table 9.5 Kinetic constants at each temperature for scheme 2 ................................................. 149 Table 9.6 Reactor outlet weight fractions based on the kinetics from Cappellazzo et al. [18] and this work. ........................................................................................................................... 150 Table A.1 Properties of the species involved throughout the thesis from Poling et al. [1] ....... 162 Table A.2 Viscosity of pure components [3] ............................................................................ 163 Table A.3 Molar volumes of each species at different conditions ............................................ 163 Table A.4 Heat of fusion and difference of solid and liquid heat capacity ............................... 164 Table A.5 Solid heat capacity of each species [3] .................................................................... 164 Table A.6 Liquid heat capacity of each species [3] .................................................................. 164 Table B.1 Vertex points of the separation regions for 1 values between 3.5 and 6.0 and 4 values between 0.1 and 0.8 for 6-9-6-3 configuration, 69 s switching time and 0.5 mm particle diameter ............................................................................................................................. 168 Table B.2 Vertex points of the separation regions for 1 values between 3.5 and 6.0 and 4 values between 0.1 and 0.8 for 6-9-6-3 configuration, 69 s switching time, and 0.7 mm particle diameter ............................................................................................................................. 169 xi Table B.3 Vertex points of the separation regions for 1 values between 4.0 and 6.5 and 4 values between 0.1 and 0.8 for 6-9-6-3 configuration, 69 s switching time, and 0.8 mm particle diameter ............................................................................................................................. 171 Table B.4 Vertex points of the separation regions for 1 values between 4.5 and 7.0 and 4 values between 0.1 and 0.8 for 6-9-6-3 configuration, 69 s switching time and 0.9 mm particle diameter ............................................................................................................................. 172 Table B.5 Optimum points for several configurations and 69 s switching time for each particle diameter. ............................................................................................................................ 174 Table B.6 Optimization for several configurations at different desorbent consumptions for each particle size ........................................................................................................................ 175 Table B.7 Optimization for the best three configurations at different productivity values for each particle diameter (Alternative approach). .......................................................................... 179 Chapter 1: Introduction This chapter presents the motivation, objectives, and outline of the research involving the development of a process that couples the separation and isomerization in p-xylene production. The motivation is based on three aspects: the promising p-xylene market behavior, the major importance of the development of sustainable processes in order to follow the environment guidelines that drive the world today, and the necessity of an integrated aromatics plant to match the required aromatics production. Furthermore, the main objective is stated and the rest of the chapters of the thesis are briefly described. Chapter 1 2 1.1. Relevance and motivation Xylenes are aromatic hydrocarbons used as raw material for the manufacture of a wide range of products used every day. Xylenes comprise four isomers: p-, m-, o-xylene, and ethylbenzene; the most important due to its application is p-xylene. The use of p-xylene is mainly the production of polyethylene terephthalate (PET), which is used as polyester fibers, films, and resins for a variety of applications [1]. Particularly PET bottles have received a lot of attention based on their recyclability. The demand is expected to grow at a rate of 7.4% from 2012 to 2022, driven mostly by the production of PTA (Purified Terephthalic Acid – precursor of PET) in China [2]. The production of p-xylene is currently performed based on two main operations units: Separation, which is basically the extraction of pure p-xylene; and Isomerization, where additional p-xylene is produced from the other isomers and recycled back to the separation unit. The most common process for p-xylene separation is selective adsorption where an adsorbent retains the p-xylene from a liquid mixture of xylenes and a desorbent is used to extract the product. Isomerization is carried out in a gas phase fixed bed catalytic reactor, where the p-xylene yield is limited by the thermodynamic equilibrium. Due to the equilibrium constraints a large cycle loop is often required to achieve the desired amount of p-xylene; a large loop along with gas phase conditions increase significantly the energy consumption within the process. The aforementioned could be minimized through the ensemble of both units based on the concept of process intensification. Process intensification, as part of the European Technology Platform on Sustainable Chemistry, indicates the ensemble of technologies that lead to substantially smaller, cleaner, safer, and more energy-efficient processes where lower consumption of raw materials and reduction of emissions of greenhouse gases and pollutants are achieved [3]. The proposed technology for coupling the processes of separation and isomerization in p-xylene production is the simulated moving bed reactor (SMBR). As mentioned before, the separation is a selective adsorption process carried out in liquid phase in fixed bed adsorption columns. The concept of simulated moving bed (SMB) is applied to mimic the counter-current flow of the solid adsorbent and liquid flow throughout the switching of inlets and outlets in the column. The SMBR uses the same principle and incorporates the reaction section, the catalyst, within the adsorption columns. In order to combine both processes, isomerization in liquid phase shall be studied. Although the conversion may be lower, it brings other advantages such as better thermal control and longer catalyst life, which allows for off-site catalyst regeneration and therefore easier control of pollution. Furthermore, since the p-xylene is withdrawn as it is formed, the equilibrium constraints in the isomerization can be minimized through the SMBR; thus, Introduction 3 reducing the cycle loop and the energy consumption within the process. Recent researches have given promising results, Minceva et al. [4] reported values as high as 1.75 for p-xylene deviation from equilibrium (actual p-xylene produced to equilibrium p-xylene produced ratio) under certain conditions. Bergeot et al. [5] claimed that with this type of technology the recycle feed (cycle loop) is reduced in 54%. In Portugal, the aromatics are produced within the Matosinhos Refinery located in the northern region of the country; the production capacities of benzene, toluene, oand p-xylene are about 43.5, 140, 10, and 92 thousand mtpy respectively [6]. However, start-up of new facilities for the production of nitrobenzene [7] and PTA [8] of 300 and 700 thousand mtpy respectively, establish a considerable difference between supply and demand of benzene and p-xylene in the local market. Situations like these shall be seen as an opportunity, a window is opened to improve the processes and increase the production. The high amount of toluene can be used to overcome the deficit, less-valuable toluene can be further processed to obtain valuable products such as benzene and p-xylene. Based on the aforementioned, a modified aromatics complex is proposed including Selective Toluene Disproportionation, Single Stage Crystallization, and Simulated Moving Bed Reactor units to increase the production of benzene and p-xylene. 1.2. Objectives and outline The main objective of this thesis is the development of a Simulated Moving Bed Reactor for the production of p-xylene in the framework of a proposal to modify the current aromatics complex which allows milder p-xylene purity constraints in the extract. The unit shall exhibit optimum arrangement of columns, flow rates, switching time, particle size, and efficient distribution of adsorbents and catalysts based on the existing Simulated Moving Bed facility. In addition, an extensive study on xylene isomerization in liquid phase shall be carried out. The thesis is divided in ten chapters to reach the aforesaid objectives. Chapter 2 consists of the discussion of the State-of-the-Art on p-xylene production. Starting with the uses, economics, and sources; and then focusing on the current technologies within the aromatics complex, specifically on the processes of p-xylene separation and isomerization of xylenes as well as toluene conversion. The study on xylene isomerization in liquid phase starts with the thermodynamic equilibrium in Chapter 3. Based on experiments performed by several researchers in the 1990s, three different expressions are developed to calculate the equilibrium constants as a function of temperature. Chapter 2 10 In addition to the previously mentioned applications as building blocks in the petrochemical industry, aromatics are also used as a high octane blending component in gasoline. Regarding the latter, restriction of total aromatic content in the gasoline pool is affecting the market. Moreover, leading economies such as North America and Europe are reducing the gasoline consumption by means of alternative fuels. However, the decrease of aromatics use in gasoline blending is smaller compared to the use as a petrochemical feedstock. A few decades ago the portion of aromatics used in gasoline was significantly higher than that for petrochemical derivatives. Nowadays the growth in petrochemical complexes is remarkable; in fact, China’s aromatic capacity has now exceeded the other regions. A substantial portion of China’s growth can be attributed to an increased demand to produce purified terephthalic acid (PTA) via p-xylene. Emerging regions along with an increasing demand of PET products worldwide have built a robust p-xylene market with an expected growth at a rate of 7.4% from 2012 to 2022 [4-6] (see Figure 2.5). Figure 2.5 World p-xylene supply/demand [5] 2.2.1. Aromatics production in Portugal Matosinhos is one of the main industrial complexes in Portugal owned by Galp Energia. Fuel production, aromatics and solvents, lube oils, paraffins, and sulfur are among the main process units within the complex. The aromatics and solvents unit has a capacity of 440 thousand mtpy, from which the main aromatics products are benzene, toluene, p-xylene, and o-xylene. A schematic representation of the aromatics plant with nominal capacities is presented in Figure 2.6. State-of-the-Art 11 From the flowsheet it can be seen that reformate is the source of aromatics in the complex. Liquid-liquid extraction is performed within the Arosolvan unit to separate the non-aromatics and through conventional distillation benzene and toluene are produced with a capacity around of 43.5 and 140 thousand mtpy respectively. A portion of o-xylene is separated based on the 5 ºC difference in the boiling points inside the Xylenes Splitter unit, approximately 10 thousand mtpy are produced. Finally the Parex unit produces 92 thousand mtpy of p-xylene [7]. The annual production is based on the nominal capacity with 335 days of operation. Figure 2.6 Galp aromatics complex [7] As mentioned before, benzene, toluene, and xylenes are basic intermediates for the petrochemical industry. In Portugal, a high amount of benzene is used in the production of nitrobenzene which is further processed to aniline. The nitrobenzene plant, owned by CUF – Químicos Industriais, started in 1991; after the last expansion project increased their capacity from 175 to 300 thousand mtpy [8]. p-Xylene is oxidized to produce PTA which is used as the raw material in the manufacture of polyester polymers. In March 2012 Artlant PTA, located at the south of the country, started the production of PTA with an installed capacity of 700 thousand mtpy [9]. From the aforesaid figures it can be seen that there is a considerable difference between the supply and demand of benzene and p-xylene in Portugal. Chapter 2 12 2.3. Sources of aromatics The only natural source is petroleum, the fraction varies according to location and geological age; even though the fraction can be as high as 35 wt%, the direct isolation is not economical [10]. Petroleum naphtha is the main feedstock for aromatics production. Reformed naphtha, or reformate, accounts for 70% of total world BTX supply. The pygas by-product from ethylene plants is the next-largest source at 23%. Coal liquids from coke ovens account for the remaining 7%. Pygas and coal liquids are important sources of benzene that may be used only for benzene production or may be combined with reformate and fed to an integrated aromatics complex. Pygas composition varies widely with the type of feedstock being cracked in an ethylene plant, light cracker feeds contain almost no C8-aromatics. Substantial amounts of C8-aromatics are found only in pygas from ethylene plants cracking naphtha and heavier feedstocks. Because reformate is much richer in xylenes than pygas, most p-xylene capacity is based on reforming petroleum naphtha. Straight-run naphtha is the material recovered directly from crude oil simple distillation. Hydrocracked naphtha, which is produced in the refinery by cracking heavier streams in the presence of hydrogen, is rich in naphthenes and makes an excellent reforming feedstock. Straight-run naphthas must be hydrotreated before being sent to the aromatics complex, but this pretreatment is not as severe as that required for pygas. Naphtha is characterized by its distillation curve and is defined by the initial boiling point (IBP) and endpoint (EP). A typical BTX cut has an IBP of 75ºC and an EP of 150ºC. However, many aromatics complexes tailor the cut of the naphtha to fit their particular processing requirements. An IBP of 75-80ºC maximizes benzene production by including all the precursors that form benzene in the reforming unit. Pre-fractionating the naphtha to an IBP of 100-105ºC minimizes the production of benzene by removing the benzene precursors. If heavy aromatics units are incorporated into the aromatics complex, C9-aromatics become a valuable source of additional xylenes. Heavier naphtha with an EP of 165-170 ºC maximizes the C9-aromatic precursors in the feed. A naphtha EP of 150-155 ºC minimizes the C9-aromatic precursors in the reforming unit feed [2]. 2.3.1. Alternative sources Aromatics can be produced from methane which through dehydroaromatization produces benzene at high temperatures. An alkylating agent is also produced from methane and HBr giving CH3Br leading to toluene and xylenes [11,12]. Methanol can also be used as raw material to State-of-the-Art 13 produce aromatics (MTA – methanol to aromatics), dimethyl ether, C1-C5 hydrocarbons, and finally aromatics are formed in fluidized bed reactors at high temperatures [13]. Biomass is also a potential source of aromatics. Catalytic fast pyrolysis (CFP) of lignocellulosic biomass produces oxygenates then converted to aromatics through cracking, deoxygenation, oligomerization, and aromatization. Selectivity to p-xylene can be achieved by silylation treatments and deposition of gallium, conversion is increased by employing mesoporous zeolites as catalysts at the expense of losing selectivity [14,15]. Furthermore, glucose can be converted to xylenes by two different paths. Isobutanol is obtained through fermentation, then dehydrated into isobutene followed by oligomerization to isooctane and dehydrocyclizated to xylenes. On the other hand, glucose is isomerized to fructose which is dehydrated to hydroxymethylfurfural (HMF) then hydrodeoxygenated to dimethylfuran (DMF), reaction with ethylene leads to p-xylene [16,17]. Acrolein, which is produced with biodiesel, can also be converted with DMF through oxidation, aromatization, and decarboxylation [18]. 2.4. Aromatics complex The following is the description of the process flow of an aromatics complex based on UOP’s technologies (see Figure 2.7); special attention is given to the units highlighted inside the rectangle. Other commercial available processes will be described later. Figure 2.7 Integrated UOP aromatics complex Chapter 2 14 The naphtha is first hydrotreated to remove sulfur and nitrogen compounds in order to protect the catalysts and then sent to a CCR Platforming unit, where paraffins and naphthenes are converted to aromatics. The CCR Platforming unit is designed to run at high severity, 104 to 106 research octane number clear (RONC), to maximize the production of aromatics. The reformate product from the CCR Platforming unit is sent to a reformate splitter column. The C7fraction from the overhead is sent to the Extraction unit for separation of benzene and toluene. The aromatic extract is clay-treated and then high-purity benzene and toluene products are recovered in the benzene-toluene (BT) fractionation section of the complex. The C8+ material from the bottom of the toluene column is sent to the xylene recovery section of the complex. The raffinate from the extraction unit may be further refined into paraffinic solvents, blended into gasoline, used as feedstock for an ethylene plant, or converted to additional benzene. Toluene is usually blended with C9and C10-aromatics (A9+) from the overhead of the A9 column and charged to a Tatoray unit for the production of additional xylenes and benzene. The effluent is sent to the BT fractionation section, where the benzene product is recovered and the xylenes are fractionated out and sent to the xylene recovery section. The overhead material from the stripper inside the unit is separated into gas and liquid products. The overhead gas is exported to the fuel gas system, and the overhead liquid is normally recycled to the CCR Platforming debutanizer for recovery of residual benzene. The C8+ fraction from the bottom of the reformate splitter is clay-treated and then charged to a xylene splitter column. The xylene splitter is designed to withdraw heavy aromatics from the mixed xylenes. The overhead from the xylene splitter is fed directly to the Parex unit, while the bottoms are sent to the A9 column where the A9 fraction is recycled to the Tatoray unit. If the complex has no Tatoray unit, the A9+ material is usually blended into gasoline or fuel oil. Furthermore, if o-xylene is to be produced in the complex, the xylene splitter is designed to make a split between mand o-xylene. The xylene splitter bottoms are then sent to an o-X column where high-purity o-xylene product is recovered overhead. The bottoms are sent to the A9 column. Even though it is possible to separate o-xylene due to the 5 ºC boiling point difference, 120-150 effective plates with high reflux ratio are required leading to excessive operating costs. Very often just a fraction of o-xylene is separated. The xylene splitter overhead is sent directly to the Parex unit, where 99.9 wt% pure p-xylene is recovered by adsorptive separation at 97 wt% recovery per pass. Any residual toluene in the Parex feed is extracted along with the p-xylene, fractionated out in the finishing column within the Parex unit, and then recycled to the Tatoray unit. The raffinate from the Parex unit is almost entirely depleted of p-xylene and is sent to the Isomar unit, where additional p-xylene is produced by re-establishing the equilibrium distribution of xylene isomers. Any ethylbenzene present is State-of-the-Art 15 either converted to additional xylenes or dealkylated to benzene, depending on the type of catalyst used. The effluent from the Isomar unit is sent to a deheptanizer column in order to continually recycle the C8-aromatics within the xylene recovery section until they exit the aromatics complex as p-xylene, o-xylene, or benzene. The overhead from the deheptanizer is split into gas and liquid products. The overhead gas is exported to the fuel gas system, and the overhead liquid is normally recycled to the CCR Platforming debutanizer for recovery of residual benzene [2,10]. 2.4.1. Naphtha reforming In the Platforming process, light petroleum distillate (naphtha) is contacted with a platinum-containing catalyst at elevated temperatures and low pressures. Semi-regenerative, fully regenerative, and continuously regenerative reformers normally operates between 510 and 540 ºC [10]. Platforming produces a high-octane liquid product that is rich in aromatic compounds. Chemical-grade hydrogen, light gas, and liquefied petroleum gas (LPG) are also produced as reaction by-products. The first UOP Platforming unit went on-stream in 1949. In 1971, Platforming with continuous regeneration, the CCR Platforming was commercialized. The CCR Platforming process has enabled ultralow-pressure operations at 345 kPa (50 psig) and produced product octane levels as high as 108. The continuous regeneration approach has been very successful with more than 95% of the new catalytic reformers being designed as CCR Platforming units. In 2011, UOP started to commercialize the new R-284 catalyst series [5,19]. Aromizing is Axens’ CCR reforming technology for aromatics production. The continuous catalyst regeneration system is fully automated, controlling all catalyst circulation and regeneration during start-up, shutdown, and normal operations. The process employs the AR series of catalysts designed to maximize aromatics yield and operates at low pressure and high severity, AR-501 is the latest generation of Aromizing. In the Axens’ aromatics complexes there is the Arofining reactor, located upstream of the Aromizing effluent stabilization, which hydrogenates undesirable olefin and diolefin compounds present in the high severity reformate [20]. 2.4.2. Aromatic extraction Axens’ Morphylane technology employs the concept of extractive distillation where a solvent is used to modify the relative vapor pressures of various hydrocarbons in such a way that aromatics can be separated from non-aromatics by simple distillation. High purity aromatics is Chapter 2 16 achieved owing to a carefully selected solvent, NFM is a non-corrosive material, thermally and chemically stable [20]. UOP’s Sulfolane process combines both liquid-liquid extraction and extractive distillation in the same process unit. Liquid-liquid extraction is more effective in separating aromatics from the heavy contaminants than from the light ones. Extractive distillation is more effective in separating aromatics from the light contaminants than from the heavy ones [21]. 2.4.3. Olefin removal Olefinic material can interfere with the performance of downstream equipment, adsorbents, and catalysts. The problems associated with the high olefin content in the feeds to adsorbent-type p-xylene recovery units, xylene isomerization units, disproportionation units, and others are accelerated catalyst and sieve aging. Clay treaters reduce the olefins content by an acid catalyzed reaction whereby the olefins react with aromatics to form heavy molecules; therefore, they are usually located upstream of fractionation in order to remove the heavier reaction products from the treated streams. ExxonMobil’s Olgone is a new technology that is an alternative to clay treating that is now used to remove olefinic material from, hence reducing the Bromine Index (BI), heavy reformate, aromatic extract, and other streams commonly found in aromatics facilities. Olgone provides higher performance and longer catalyst life [22]. 2.4.4. Toluene disproportionation and transalkylation The two major reactions in the UOP’s Tatoray process are disproportionation and transalkylation. Disproportionation is the conversion of toluene alone to an equilibrium mixture of benzene and xylenes. Transalkylation is the conversion of a blend of toluene and heavier aromatics to xylenes through the migration of methyl groups between methyl-substituted aromatics. The Tatoray process effectively converts the ethyl, propyl, and higher alkyl group substituted to lighter single-ring aromatics via dealkylation, while preserving the methyl groups [23]. TransPlus is ExxonMobil’s toluene/C9+ aromatics transalkylation technology which was co-developed with the Chinese Petroleum Corporation (CPC) of Taiwan. The proprietary catalyst is carefully designed to maximize desirable reactions such as disproportionation, transalkylation, and dealkylation. TransPlus technology has the flexibility to process up to 100 wt% of C9+ aromatics in the fresh feed while maintaining long cycle lengths [24]. State-of-the-Art 17 2.4.5. Selective toluene disproportionation The first toluene disproportionation process was introduced by Mobil in 1975 and had a selectivity of about 24% as limited by chemical equilibrium. It was later discovered that pretreatment or selectivation of the catalyst could improve p-xylene selectivity beyond the equilibrium limitation [25]. Catalysts with reduced pores favor the transfer of the methyl group to the least hindered position and the p-xylene formed diffuses out of the pores faster than the other isomers which isomerize to form p-xylene to re-establish the equilibrium [26,27]. Between the zeolites used as catalysts, ZSM-5 is preferred due to high selectivity and slow aging; larger-pore Mordenite is used to convert bulky trimethylbenzenes in disproportionation and transalkylation processes [3,28]. Among the selectivation procedures, chemical liquid deposition (CLD) and chemical vapor deposition (CVD) are the most industrially used methods; the purpose is pore blockage, increase tortuosity, and deactivation of external acid sites [26,28]. See Figure 2.8. Figure 2.8 Catalyst selectivation in PxMax technology [25] ExxonMobil’s selective toluene disproportionation process based on this development is known as PxMax, which allows p-xylene selectivity to be improved to over 90%. Two catalysts technologies are being offered for licensing as part of PxMax process: EM-2200 with in-situ coke selectivation and MTPX with ex-situ permanent selectivation. The process flow scheme (see Figure 2.9) includes a reactor section, fractionation to separate the aromatics products, and a p-xylene recovery unit [25]. Chapter 2 18 Figure 2.9 Block diagram of PxMax with crystallization [29] In the UOP aromatics complex, instead of feeding the toluene to Tatoray, another processing strategy for toluene is to feed it to a p-xylene catalytic process such as PX-Plus, where the p-xylene in the xylene product is enriched to >85%. The concentrated p-xylene product could then be easily recovered in a single stage crystallization unit [2]. 2.5. p-Xylene separation Ethylbenzene, p-xylene, m-xylene, and o-xylene boil so closely together that separating them by conventional distillation is not practical. Just o-xylene can be separated through distillation since there is a 5ºC difference with the next isomer, which is m-xylene, although it constitutes an expensive separation as discussed in Section 2.4. Based on the aforementioned, p-xylene recovery is typically accomplished either by fractional crystallization or through adsorptive-type processes. 2.5.1. Crystallization Prior to the 1970’s, the typical method for recovering p-xylene from mixed xylene streams was low temperature crystallization. These units typically operated at cold stage temperatures of -60 to -65ºC and provided only 60-70% recovery. The situation changed completely with the commercialization of UOP’s Parex, a more efficient adsorption-based technology that provides >97% recovery of p-xylene (See Figure 2.10). State-of-the-Art 19 Figure 2.10 p-Xylene separation technologies [30] The crystallization process takes advantage of the large differences between the freezing points of the C8-aromatic components to separate p-xylene from the mixture. C8-aromatics are known to behave as a eutectic mixture where, as temperature is reduced, one of the components will begin to drop out of the solution as a pure solid phase. The initial composition of the solution determines which component will drop out first. Although p-xylene freezing point is the highest, normally it is not present at the highest concentration, resulting in the need of several stages and a recovery of only 60-70%; in recent years, there has been a renewed interest in p-xylene crystallization technology in combination with selective toluene disproportionation processes due to the high p-xylene content of the C8-aromatic product, thus, eliminating the eutectic constraints. [29]. Typical process consists of crystallization stages at different temperatures, liquid-solid separation equipment, melting, heat exchangers, and washing streams for purification [31]. Liquid-solid separation depends on the size of the crystals which is governed by crystallization kinetics, the average size of the crystals is set to above 0.5 mm to guarantee good separation [32,33]. Crystallizer may be the suspension or layer type; direct or indirect refrigeration systems can also be used [10]. ExxonMobil’s p-xylene crystallization process is conducted at three temperature levels: a purity stage at 3 ºC, a scavenger stage at -4 ºC and a recovery stage at -29 ºC to facilitate p-xylene crystal separation and washing operations and to maximize p-xylene recovery (See Figure 2.11). In suspension type crystallization, the feed is chilled below the equilibrium temperature and held at this temperature for several hours to allow the crystals to grow. Chapter 2 26 The second process is especially dedicated to m-xylene production. Raffinate 2 coming from the Eluxyl Twin Raffinate is sent to another SMB unit dedicated to the production of m-xylene at high purity (99.5%) and high recovery (see Figure 2.16). It can be noted that raffinate 2 is a better feed for the second SMB unit than a conventional single raffinate stream, as it is more concentrated in m-xylene and as it is completely depleted in ethylbenzene which allows the reduction in the amount of adsorbent and desorbent needed in the SMB unit [54]. Figure 2.16 Schematic view of Eluxyl twin raffinate MX SMB unit [54] 2.5.3. Crystallization/adsorption hybrid process The alliance between UOP, Washington Group International, and Niro Process Technology, introduced the HySorb XP process, a simplified, single-chamber, light desorbent adsorption process coupled with single stage crystallization and Niro wash column technology. This combination of technologies when integrated into existing multistage crystallization facilities can increase p-xylene production by as much as 500%. The HySorb process produces a 95 wt% p-xylene concentrate, eliminating eutectic constraints and enabling single stage crystallization recoveries above 90% [30]. The Axens’ hybrid version produces intermediate purity product, ideally suited for a second stage crystallization. Owing to the lower p-xylene purity requirement, a smaller amount of sieve is required and only one adsorber is necessary. The hybrid configuration is the most effective way to drastically debottleneck existing crystallization plants [53]. State-of-the-Art 27 Chevron’s hybrid process is currently used in its Pascagoula plant, it is characterized by the use of benzene as desorbent which results in greater boiling point difference leading to easier distillation in the extract and raffinate columns [55]. GTC also offers a hybrid process with a single stage crystallizer as mentioned in Section 2.5.1 and an adsorption process to concentrate the feed to about 90% using Zeosorb PX-200 from Clariant as adsorbent [56]. 2.6. Xylene isomerization The xylene isomerization process is used to maximize the recovery of a particular xylene isomer from a mixture of C8-aromatic isomers, although is more often applied to p-xylene recovery. UOP’s Isomar, ExxonMobil’s XyMax and Advanced MHAI, and Axens’ Oparis are one of the most used technologies worldwide. The feed is first combined with hydrogen-rich recycle gas and makeup gas to replace the small amount of hydrogen consumed in the reactor (see Figure 2.17). The combined feed is then preheated by exchange with the reactor effluent, vaporized in a fired heater, and raised to reactor operating temperature. The heater is normally a radiant convection-type heater. The process stream is heated in the radiant section, and the convection section is used for a hot-oil system or steam generation. The hot feed gas stream is then sent to the reactor. The reactor effluent is cooled by exchange with the combined feed and then sent to the product separator. The purpose of the product separator is to split the condensed reactor effluent into liquid product and hydrogen-rich recycle gas. The pressure in the product separator determines the pressure in the reactor and is regulated by controlling the rate of hydrogen makeup flow. Hydrogen purity in the recycle gas is monitored by a hydrogen analyzer at the recycle-gas compressor suction. When hydrogen purity gets too low, a small purge is taken from the recycle gas. Liquid from the bottom of the product separator is charged to the deheptanizer column. The C7overhead from the deheptanizer is cooled and separated into gas and liquid products. The deheptanizer overhead gas is exported to the fuel gas system. The overhead liquid is recycled to the reforming unit so that any benzene in this stream may be recovered. The C8+ fraction from the bottom of the deheptanizer is clay-treated, combined with fresh mixed-xylenes feed and recycled [57]. The difference between the technologies lies mainly on the reactor and catalysts used, which will be discussed in the next sections. 2.6.1. Xylene isomerization catalysts In the reactor two main categories of xylene isomerization catalysts are used, ethylbenzene dealkylation catalysts and ethylbenzene isomerization catalysts. The former converts Chapter 2 28 ethylbenzene to benzene and the latter to additional xylenes. Since ethylbenzene isomerization is an equilibrium-limited reaction, the conversion of ethylbenzene is usually limited to about 30-35 wt% per pass. Ethylbenzene dealkylation catalyst allows conversion of up to 70 wt% or greater. For a new aromatics complex design, using an ethylbenzene dealkylation catalyst minimizes the size of the xylene column and downstream units required to produce a given amount of p-xylene. However, this reduction in size of the xylene loop comes at the expense of lower p-xylene yields, because all the ethylbenzene in the feed is being converted to benzene rather than to additional p-xylene. Lower p-xylene yield means that more feedstock will be required [57]. The reaction is further studied in Chapter 4. Figure 2.17 UOP’s Isomar flow diagram [57] 2.6.2. UOP’s Isomar The Isomar process normally uses a radial-flow reactor. The gas stream enters the top of the reactor and is directed to the sidewall. The fluid then travels radially through the fixed bed and into a center pipe. The reactor effluent then flows down through the center pipe to the reactor outlet. The advantage of the radial-flow reactor is low pressure drop, which is important due to the influence of hydrogen partial pressure on the reaction rates. UOP offers both types of commercial catalysts; the last catalyst released is I-500 which offers higher selectivity at lower temperatures in dealkylation of ethylbenzene [5]. All xylene isomerization catalysts exhibit some by-product formation across the reactor. The precise level of expected by-product formation varies with catalyst type and operating severity, but it is normally in the range of 1.0 to 4.0 wt% per pass of the feed. By-products are predominantly aromatics, such that overall ring retention is greater than 99%. Moreover, non-aromatic State-of-the-Art 29 compounds in the feed to the Isomar unit are primarily cracked to light ends and removed from the Parex-Isomar loop [57]. 2.6.3. ExxonMobil’s XyMax and Advanced MHAI Soon after the discovery of ZSM-5 in the early 1970’s, Mobil introduced MVPI, the Mobil Vapor Phase Isomerization process. MVPI utilized the first high activity, zeolite based xylene isomerization catalyst. In 1978, Mobil introduced MLPI, the Mobil Low Pressure Isomerization process, which was capable of operating without H2 recirculation while achieving low xylene losses and long cycle lengths. In 1981, Mobil introduced MHTI, the Mobil High Temperature Isomerization process; and then in 1990 MHAI, the Mobil High Activity Isomerization process. The advances were mainly the higher ethylbenzene conversions with lower xylenes losses [58]. Nowadays ExxonMobil offers XyMax and Advanced MHAI technologies. For sites with lower reactor temperature limitations, and for those that utilize crystallization for p-xylene separation, the Advanced MHAI process offers optimum operation. For sites with higher reactor temperature capability, and for those that utilize adsorption-based p-xylene separation technology, the XyMax Process is the best choice. Both processes incorporate the latest advances in ExxonMobil’s zeolite catalyst technology and are the ethylbenzene dealkylation type [59]. The primary chemical reactions are the conversion of ethylbenzene to benzene and ethylene, cracking of non-aromatics, and isomerization of the p-xylene depleted feedstock to an equilibrium mixture of xylenes. These reactions take place in a fixed-bed reactor with two distinct zeolite catalysts. In the top bed, the catalyst is designed to convert ethylbenzene to benzene and ethylene, via dealkylation, and to crack non-aromatics. The ethylene produced is largely hydrogenated to ethane in the top bed also, reducing the likelihood of xylene loss through alkylation to heavy aromatics. The catalyst in the bottom bed is optimized for complete xylene isomerization to near-equilibrium levels of p-xylene (see Figure 2.18) [60]. 2.6.4. Axens’ Oparis Oparis (OPtimized ARomatics ISomerization) is Axens’ new generation catalyst for ethylbenzene and xylenes isomerization. The unique feature of Oparis is its ultra-high selectivity, which allows conversion of xylenes and ethylbenzene to an equilibrium mixture of xylenes with maximum yields at milder operating conditions [61]. The xylenes isomerization reaction is favored by temperature, as are the undesired side reactions. Since ethylbenzene isomerization implies the formulation of hydrogenated, naphthenic intermediate, the reaction is favored by milder temperature and the presence of hydrogen partial pressure. The adjustment of a Chapter 2 30 C8-aromatics isomerization unit requires an optimized balance between temperature that drives xylenes isomerization reaction, but results in a higher amount of side reactions, and hydrogen partial pressure, which drives the ethylbenzene isomerization reactions [62]. Figure 2.18 ExxonMobil’s dual bed catalyst system [25] The reaction takes place in a conventional gas phase reactor. The liquid effluent from the reaction section is a near-equilibrium mixture of C8-aromatics which also contains lighter and heavier components resulting from minor side reactions. C8-naphthenes are also present in equilibrium-related amounts, these are separated and recycled to the isomerization reactor in order to maintain optimum performance. The naphthenes recycle is fine-tuned by employing a small fractionation tower and sent directly to the Oparis isomerization section (see Figure 2.19). In other technologies, the naphthenes are recycled to the xylenes column and p-xylene separation section before re-entering the isomerization section [61]. Several technologies are summed up in Table 2.1. 2.7. New trends in xylene production According to the patent literature of the last few years the new trends in the xylene production, especially in xylene isomerization, are focused on the use of several reactors under different conditions tailored for specific purposes, as well as the use of different catalysts, zeolitic and non-zeolitic. Moreover, productions of other products along with p-xylene and process intensification have also attracted attention among the researches worldwide. Some of these patents are presented in Table 2.2. State-of-the-Art 31 Figure 2.19 Simplified xylenes loop flowscheme including Oparis process [61] Table 2.1 Xylene isomerization units Process Catalyst Reaction Typea T, ºC P, MPa Octafining Pt on alumina combined with Hmordeniteb Ethylbenzene isomerization in the presence of H2 425-480 1.14-2.51 Isomarc Pt on alumina Ethylbenzene isomerization in the presence of H2 388 1.68 MVPI NiHZSM-5 with an alumina binder Ethylbenzene disproportionation in the presence of H2 315-370 1.48 MLPI HZSM-5 with an alumina binder Ethylbenzene disproportionation in the absence of H2 290-380 0.27 MHTI Pt on acidic ZSM-5 Ethylbenzene dealkylation in the presence of H2 427-460 1.48-1.83 XyMaxd Noble metal ZSM-5 with amorphous binder Ethylbenzene dealkylation in the presence of H2 400-482 0.45-2.87 Oparise EUO-structural-type zeolite with Pt. Ethylbenzene isomerization in the presence of H2 350-420 0.6-1.5 a All reactions are in gas phase. b Corresponding to O-750 catalyst. c Corresponding to I-9 catalyst. d From Patent 7,247,762 B2 [63]. e From Patent 6,376,734 B1 [64]. 2.7.1. Simulated moving bed reactor Although not industrially proven, simulated moving bed reactor (SMBR) for p-xylene production has dragged plenty of attention among researchers. Based on the concept of process intensification, it combines p-xylene separation and xylene isomerization in one single unit. Reduction of operating and capital costs, energy consumption, environment impact; are among Chapter 2 32 Table 2.2 Recent patents in xylene production Reference [65] [66] [67] Application It consists in a combination of stages for isomerization in liquid phase of a fraction that is high in m-xylene and o-xylene and for isomerization in gas phase of a fraction that is high in ethylbenzene. Ethylbenzene is previously separated by distillation or by adsorption and isomerized at 350-450ºC, 5-20 bar, and with a catalyst that contains EUO-structural-type zeolite. The other fraction with a limited quantity of ethylbenzene is sent to a simulated moving bed with toluene as desorbent which serves as diluent in the isomerization in liquid phase which is carried out at 200-260ºC, 20-30 bar, and with a zeolitic catalyst such as ZSM-5. Any ethylbenzene in the feed is removed, either by dealkylation or isomerization, in a separate reactor upstream of the xylene isomerization reactor. The ethylbenzene depleted product is sent to a second reactor, typically the clay treater, as that used to accommodate the olefin removal catalyst. Xylene isomerization is carried out under mild conditions, 160-232ºC and 100-500 psig (liquid phase). This second reactor contains a zeolite catalyst for the xylene isomerization and also another catalyst effective to remove olefins. In certain cases, a single catalyst may be used to effect both xylene isomerization and olefin removal. The C8 isomerization zone may include a first isomerization stage and a second isomerization stage. At the first isomerization stage, a non-equilibrium alkylaromatic feed mixture is contacted with a zeolite catalyst in the absence of hydrogen at 250-350ºC and enough pressure to keep the mixture in liquid phase for isomerizing at least one isomer. At the second isomerization stage, the intermediate stream from the first stage and a stream rich in one naphthene are contacted with a zeolite catalyst in the presence of hydrogen at 300-500ºC and a pressure less than 2 MPa. The purpose of the second stage is the isomerization of ethylbenzene. Often, C8 naphthenes can be created in the second isomerization stage; therefore, it is desirable to recycle excess C8 naphthenes back to the second isomerization stage. Inventor/Assignee Julia MagneDrisch, Fabio Alario, JeanFrançois Joly, Ari Minkkinen and Elisabeth Merlen / Institut Français du Pétrole Gary David Mohr / ExxonMobil Chemical Company John E. Bauer / Honeywell UOP Patent No., Date and Title US 6,448,459 B1 Sept 10, 2002 Process for the Production of Paraxylene that comprises an Adsorption Stage, a Liquid Phase Isomerization Stage and a Gas Phase Isomerization Stage with an EUO-type zeolite US 2004/0236166 A1 Nov 25, 2004 Xylene Isomerization US 2009/0182182 A1 Jul 16, 2009 Process for Isomerizing a Non-Equilibrium Alkylaromatic Feed Mixture and an Aromatic Production Facility State-of-the-Art 33 Reference [68] [69] [70] Application The feed of hydrocarbons containing xylenes and ethylbenzene is separated in a simulated moving bed from which an extract rich in p-xylene and a raffinate are withdrawn. The ethylbenzene contained in the raffinate is dehydrogenated to styrene with conversion of at least 50 wt%. A stream rich in styrene is separated by means of adsorption, distillation, liquid-liquid extraction, membrane or any combination of said techniques. The unconverted ethylbenzene, m-xylene and o-xylene is isomerized, preferably in liquid phase (e.g. 200-260ºC and 2-3 MPa), and recycled back to the simulated moving bed unit. Method for producing high-purity p-xylene by coupling a step of selective p-xylene adsorption and a step of xylene isomerization from an aromatic C8 mixture optionally containing ethylbenzene. The adsorbent for p-xylene separation is BaLSX with a Si/Al ratio of about one, the desorbent is toluene, and the isomerization catalyst is ZSM-5 type. The adsorption is carried out at preferably 200-220ºC and less than 4 MPa. The reactors are located in zone 3 of the SMB unit and the conditions are preferably 180-300ºC and 2-3 MPa at liquid phase. Isomerization of a non-equilibrium C8 aromatic mixture of xylenes and ethylbenzene with a catalytic composition comprising at least one platinum-group metal component and MgAPSO-31. MgAPSO-31 is a magnesium-containing nonzeolite molecular sieve which has a narrowly defined composition and is particularly useful for isomerization. It contains from about 0.003 to 0.011 mol fraction of magnesium in the microporous crystalline framework structure and comprises crystals with a median diameter of no more than about 1.0 micron. Two isomerization zones can be used. First to isomerize xylenes in the absence of hydrogen over a zeolitic aluminosilicate catalyst; and the second with MgAPSO-31 to isomerize ethylbenzene in the presence of hydrogen to increase the p-xylene content of the product. The conditions for the proprietary catalyst comprise a temperature preferably in the range of 300-500ºC and pressure less than about 50 atm. Inventor/Assignee Luc Wolff, Philibert Leflaive and Alain Methivier / Institut Français du Pétrole Ghislain Bergeot, Catherine Laroche, Philibert Leflaive, Damien Leinekugel Le Cocq and Luc Wolff / Institut Français du Pétrole Hayim Abrevaya, Julio C. Marte, Stephen T. Wilson, Susan C. Koster, John E. Bauer, Wharton Sinkler, Ben A. Wilson and Lance L. Jacobsen / Honeywell UOP Patent No., Date and Title US 7,592,499 B2 Sept 22, 2009 Process for co-Producing para-Xylene and Styrene WO 2009/130402 A1 Oct 29, 2009 Reactive Simulated Moving Bed for Producing paraXylene US 2010/0152511 A1 Jun 17, 2010 Hydrocarbon Conversion using an Improved Molecular Sieve Chapter 2 34 the main advantages of these types of process. However, flexibility in operation and control, as well as the lack of past experiences of these type of units constitutes a barrier for being used in the industry [71]. For p-xylene separation, SMB is the most employed technique in adsorption-type separations as seen in Section 2.5. The SMBR uses the same principle and incorporates the reaction section within the adsorption columns. There are two possible scenarios: adsorbent and catalyst are two different materials or both are present in the same pellet. For the reversible reaction of the type A↔B, reaction cannot occur near the extract point if high purity is required (otherwise maximum purity obtained is below 99%). To overcome this situation, reactors are inserted between the adsorption columns far from the extract point [72,73] Minceva et al. [72] proposed a SMBR unit for p-xylene production from a mixture of xylene free of ethylbenzene operating in liquid phase. Ba exchanged faujasite type of adsorbent and ZSM-5 catalyst for the reaction section were used in the study. The configuration proposed consisted in six adsorbers in zone 1, nine adsorbers in zone 2, six adsorbers and five reactors in zone 3, and three adsorbers in zone 4. The authors simulated the unit at temperatures between 453 and 573 K for two types of feed: p-xylene composition higher than equilibrium (similar to that fed to Parex unit) and one with p-xylene composition lower than equilibrium (similar to the raffinate of Parex unit). They found out that for the first type of feed, 1.75 was reached for the p-xylene deviation from equilibrium, which is significantly higher than conventional processes. They also concluded that best SMBR performance was achieved with shorter reactors and lower temperatures. Bergeot et al. [73] followed a similar configuration with six reactors and seven adsorbers between the feed injection and the raffinate withdrawal. They studied the xylene isomerization in liquid phase over HZSM-5 at 523-573 K and came to the conclusion that ethylbenzene cannot be converted. Simulations reported a decrease of more than half within the cycle loop. 2.8. Conclusions p-Xylene world market is expected to grow in the upcoming years driven by the increasing demand of its derivatives. In Portugal, there is a significant difference between the production of benzene and p-xylene and the capacity of recently installed downstream plants that uses said species as raw material; hence, an opportunity to increase the production is identified. Based on the bibliographic research presented in this chapter, several technologies can be used to achieve this goal. In the first place, the conversion of excess toluene to more valuable benzene and p-xylene seems to be an understandable option to be considered. In addition, the successful of State-of-the-Art 35 hybrid processes combining adsorption and crystallization together with the promising results of recent studies on simulated moving bed reactor for p-xylene production constitute a potential alternative to increase the production of benzene and p-xylene within the existing aromatics complex of the country. 2.9. Nomenclature Abbreviations BTX = Benzene, Toluene, and Xylenes CCR = Continuous Catalytic Regeneration CPC = Chinese Petroleum Corporation CUF = Companhia União Fabril (Factory Union Company) EB = Ethylbenzene EP = End Point GTC = Global Technology Licensor IBP = Initial Boiling Point IFP = Institut Français du Pétrole (French Institute of Petroleum) MHAI = Mobil High Activity Isomerization MHTI = Mobil High Temperature Isomerization MLPI = Mobil Low Pressure Isomerization MOF = Metal Organic Framework MVPI = Mobil Vapor Phase Isomerization MX = m-Xylene OX = o-Xylene PET = Polyethylene Terephthalate PTA = Purified Terephthalic Acid PX = p-Xylene This chapter is based on: Gonçalves, J. C., and A. E. Rodrigues. 2013. "Thermodynamic equilibrium of xylene isomerization in the liquid phase." Journal of Chemical and Engineering Data no. 58 (6):1425-1428. Chapter 3: Thermodynamic equilibrium of xylene isomerization in liquid phase This chapter deals with the thermodynamic equilibrium for xylene isomerization. Experiments performed by several researchers to calculate the equilibrium in gas phase in the 1990s led to the conclusion that the earlier available thermodynamic data for xylenes, which were mainly based on experimental work performed in the 1940s, were in error. In this work a similar procedure is followed to determine the thermodynamic equilibrium for xylene isomerization in liquid phase. By means of the thermodynamic functions at saturated conditions presented by the previously mentioned studies, the standard free energies of formation are calculated between 250 and 550 K. Three different expressions are developed to calculate the equilibrium constants as a function of temperature. Chapter 3 44 3.1. Introduction As previously mentioned, the isomerization reaction involved in the production of p-xylene (PX) is limited by the thermodynamic equilibrium which results in a large recycle loop to achieve the desired amount of p-xylene. Due to the direct influence of the thermodynamic equilibrium in the process, accurate equilibrium values are of the most importance. Amelse [1] carried out isomerization experiments over non-shape-selective and shapeselective catalysts and calculated the thermodynamic equilibrium of xylenes at 623 and 673 K. Based on the results obtained, Amelse [1] concluded that xylenes thermodynamic data were erroneous. According to Chirico et al. [2], available standard thermodynamic properties of formation (e.g. standard Gibbs free energy of formation) for the xylenes were the result of experimental work performed in the 1940s. Those experimental results were obtained only at one temperature, 298.15 K. p-Xylene has been studied extensively in the past in order to expand and improve the available data, but o-xylene (OX) and m-xylene (MX) have been left aside [2]; this was probably due to the higher economic importance of p-xylene. Chirico and co-workers [2-5] carried out experimental studies to measure calorimetric and physical properties used to determine standard Gibbs free energies of formation between 250 and 500 K for the three xylenes and ethylbenzene. Furthermore, they developed expressions to evaluate the thermodynamic equilibrium which were in excellent agreement with Amelse [1] results. Chirico and Steele [6] concluded that the largest error was associated with the entropy of o-xylene in the liquid and gas phases. Both, Amelse [1] and Chirico et al. [4], highlighted the influence of the rotation of the methyl groups. Unfortunately, the aforementioned studies presented expressions for the equilibrium constants as a function of temperature only for gas phase; this was probably due to the fact that xylene isomerization occurs industrially under gas phase conditions. However, in the last few years new trends in xylene production are focused on the use of xylene isomerization in liquid phase driven by the environmental benefits of reduction of energy and pollution. Namely, the isomerization is separated in two stages: one in liquid phase for xylenes and one in gas phase for ethylbenzene [7,8]. Moreover, research efforts are being made on process intensification by coupling xylenes isomerization and xylenes separation (both in liquid phase) in a single unit using the Simulated Moving Bed Reactor (SMBR) technology [9,10] and, as previously stated, one of the objectives of this thesis is a complete study on xylene isomerization in liquid phase. This chapter is intended to develop similar expressions of the equilibrium for xylene isomerization in liquid phase based on the thermodynamic functions at saturated conditions presented by Chirico and co-workers [2-4]. Thermodynamic equilibrium of xylene isomerization in liquid phase 45 3.2. Equilibrium in liquid phase In order to obtain the standard Gibbs energy of formation to estimate the equilibrium constants in liquid phase, the molar thermodynamic functions at saturation pressure (Ps) are extracted from the studies cited before, for temperatures between 250 and 550 K. The saturation pressure is obtained from the Wagner equation [11] modified to four coefficients into the form (2,4) by Duschek et al. [12]: ln(𝑃𝑠 𝑃𝑐)=1 𝑇𝑟{𝐴(1−𝑇𝑟)+𝐵(1−𝑇𝑟)1.5+𝐶(1−𝑇𝑟)2+𝐷(1−𝑇𝑟)4} (3.1) where Tr = T/Tc, Pc is the critical pressure, and Tc is the critical temperature. The required equation parameters (A, B, C, and D) of each species are presented in Table 3.1. Critical properties and some other intrinsic properties are shown in Table A.1 (see Annex A). The molar thermodynamic functions at saturation pressure are presented in Table 3.2; based on those values, the enthalpy (H) and entropy (S) at standard state (i.e., at P0 = 100 kPa) are calculated at each temperature using Maxwell relations: (𝜕𝑆 𝜕𝑃)𝑇=−(𝜕𝑉 𝜕𝑇)𝑃 (3.2) (𝜕𝐻 𝜕𝑃)𝑇=𝑉−𝑇(𝜕𝑉 𝜕𝑇)𝑃 (3.3) Table 3.1 Wagner parameters for equation (3.1) p-Xylene m-Xylene o-Xylene A -7.59306 -7.564368 -7.457432 B 1.77964 1.623819 1.519744 C -1.24526 -1.139601 -1.030912 D -3.93248 -4.004004 -3.997287 Integrating within the pressure range and assuming no variation of molar volume (V) in liquid phase due to pressure, equations (3.4) and (3.5) are obtained: ∆𝑆= −𝑑𝑉 𝑑𝑇∆𝑃 (3.4) ∆𝐻=(𝑉−𝑇𝑑𝑉 𝑑𝑇)∆𝑃 (3.5) where ΔP = P0-Ps. Following the procedure of Chirico and co-workers [2-4], molar volumes are obtained by means of the molecular weight (MW) and the densities (ρ) calculated with a form of the corresponding-states equation of Riedel: Chapter 3 46 Table 3.2 Molar thermodynamic functions, enthalpy (H) and entropy (S) at saturation pressure (Ps) at temperatures from 250 to 550 Kd o-Xylenec ∆0𝑇S/R 25.690 26.517 28.114 29.511 29.651 31.138 32.585 33.997 35.380 36.738 38.074 39.39 40.70 41.99 43.27 44.54 45.82 46.45 ∆0𝑇H/RT 16.129 16.319 16.694 17.031 17.066 17.440 17.818 18.201 18.589 18.984 19.384 19.79 20.21 20.63 21.06 21.50 21.95 22.18 Ps, kPa 0.02716 0.06361 0.2827 0.8945 0.9967 2.922 7.379 16.50 33.39 62.20 108.1 177.3 276.8 414.7 599.5 840.8 1149 1332 m-Xyleneb ∆0𝑇S/R 26.616 27.419 28.958 30.307 30.442 31.884 33.290 34.667 36.020 37.351 38.665 39.96 41.25 42.52 43.79 45.06 46.33 46.96 ∆0𝑇H/RT 15.450 15.635 16.001 16.335 16.369 16.743 17.122 17.510 17.905 18.308 18.719 19.14 19.56 20.00 20.44 20.89 21.36 21.61 Ps, kPa 0.03593 0.08300 0.3597 1.116 1.241 3.572 8.878 19.58 39.14 72.15 124.3 202.2 313.6 467.1 672.0 938.8 1279 1480 p-Xylenea ∆0𝑇S/R 26.038 26.829 28.368 29.721 29.857 31.304 32.715 34.098 35.454 36.789 38.107 39.409 40.699 41.979 43.249 44.512 45.771 46.401 ∆0𝑇H/RT 17.448 17.553 17.783 18.012 18.037 18.311 18.603 18.913 19.239 19.578 19.932 20.299 20.678 21.069 21.472 21.885 22.310 22.528 Ps, kPa 0.03907 0.08961 0.3836 1.179 1.310 3.739 9.225 20.22 40.21 73.78 126.6 205.3 317.5 471.6 677.0 943.9 1284 1485 T, K 250 260 280 298.15 300 320 340 360 380 400 420 440 460 480 500 520 540 550 a From Chirico et al. [2]. b From Chirico et al. [3]. c From Chirico et al. [4]. d Values are presented with one digit more than justified. Thermodynamic equilibrium of xylene isomerization in liquid phase 47 𝜌 𝜌𝑐=1+0.85(1−𝑇 𝑇𝑐)+(1.6916+0.9845𝜔)(1−𝑇 𝑇𝑐)13 ⁄ (3.6) Once the thermodynamic functions are obtained in the standard state for each temperature, the formation functions are calculated. Enthalpies of formation include the enthalpies of the reference elements [13]: 𝛥𝑓𝐻0(𝑇)=𝛥𝑓𝐻0(298.15)+[𝐻0(𝑇)−𝐻0(298.15)]𝑐𝑜𝑚𝑝𝑜𝑢𝑛𝑑 −∑[𝐻0(𝑇)−𝐻0(298.15)]𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 (3.7) where ΔfH0 at 298.15 K is -24.39±0.63, -25.38±0.37, and -24.39±0.40 kJ kmol-1 for p-, m-, and o-xylene respectively [6]. The Gibbs energy of formation is calculated with the enthalpy of formation and the entropies of the reference elements [13]: 𝛥𝑓𝐺0(𝑇)=𝛥𝑓𝐻0(𝑇)−𝑇{𝑆0(𝑇)𝑐𝑜𝑚𝑝𝑜𝑢𝑛𝑑−∑𝑆0(𝑇)𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠} (3.8) The enthalpies and entropies of the elements are obtained from Chase [13] and shown in Table 3.3. Table 3.4 presents the Gibbs energy of formation for each xylene species with the corresponding uncertainty; errors associated to pressure and molar volume are not taken into account since they are negligible compared to that of enthalpy and entropy. Table 3.3 Entropy (S0) and enthalpy (H0-H0(Tr)) of reference elements at reference temperature Tr = 298.15 K and standard pressure P0 = 100 kPa from Chase [13] Graphite (C) Hydrogen (H2) T, K S0, J mol-1∙K-1 (H0-H0(Tr)), kJ mol-1 S0, J mol-1K-1 (H0-H0(Tr)), kJ mol-1 200 3.082 -0.665 119.412 -2.774 250 4.394 -0.369 125.640 -1.378 298.15 5.740 0 130.680 0 300 5.793 0.016 130.858 0.053 350 7.242 0.487 135.325 1.502 400 8.713 1.039 139.216 2.959 450 10.191 1.667 142.656 4.42 500 11.662 2.365 145.737 5.882 600 14.533 3.943 151.077 8.811 Ethylbenzene cannot be converted in liquid phase because its isomerization to xylenes goes through naphthenes intermediates, which requires the presence of hydrogen. Nevertheless isomerization of xylenes can be carried out in liquid phase over acid catalysts [9,10]. Due to the aforementioned fact, the ethylbenzene is not taken into account in the thermodynamic equilibrium in liquid phase. The equilibrium constants are defined for each isomer pair, according to the reaction scheme in Figure 3.1, as seen in equations (3.9) to (3.11). Chapter 3 48 Table 3.4 Gibbs energy of formation (ΔfG0/RT) of xylene species in liquid phase T, K p-Xylene m-Xylene o-Xylene 250 42.672 ± 0.019 41.615 ± 0.019 42.866 ± 0.019 260 43.098 ± 0.018 42.050 ± 0.018 43.296 ± 0.018 280 43.882 ± 0.018 42.867 ± 0.018 44.088 ± 0.018 298.15 44.526 ± 0.017 43.537 ± 0.017 44.732 ± 0.017 300 44.589 ± 0.017 43.601 ± 0.017 44.795 ± 0.017 320 45.225 ± 0.017 44.264 ± 0.017 45.430 ± 0.017 340 45.801 ± 0.017 44.861 ± 0.017 46.002 ± 0.017 360 46.320 ± 0.016 45.403 ± 0.016 46.518 ± 0.016 380 46.792 ± 0.016 45.892 ± 0.016 46.983 ± 0.016 400 47.218 ± 0.016 46.336 ± 0.016 47.404 ± 0.016 420 47.604 ± 0.016 46.737 ± 0.016 47.783 ± 0.016 440 47.954 ± 0.016 47.11 ± 0.14 48.13 ± 0.14 460 48.271 ± 0.016 47.43 ± 0.14 48.44 ± 0.14 480 48.559 ± 0.015 47.74 ± 0.14 48.72 ± 0.14 500 48.822 ± 0.015 48.01 ± 0.14 48.98 ± 0.14 520 49.062 ± 0.015 48.25 ± 0.14 49.21 ± 0.14 540 49.282 ± 0.015 48.48 ± 0.14 49.42 ± 0.14 550 49.384 ± 0.015 48.60 ± 0.14 49.53 ± 0.14 Deviations are larger for mand o-xylene above 440 K due to the uncertainty of their saturation functions 𝑋OX 𝑋MX ⁄=𝐾1=exp(−∆𝑅1𝐺0𝑅𝑇 ⁄ ) (3.9) 𝑋MX 𝑋PX ⁄=𝐾2=exp(−∆𝑅2𝐺0𝑅𝑇 ⁄ ) (3.10) 𝑋PX 𝑋OX ⁄=𝐾3=exp(−∆𝑅3𝐺0𝑅𝑇 ⁄ ) (3.11) Figure 3.1 Reaction scheme for xylene isomerization. PX = p-Xylene, MX = m-Xylene, OX = o-Xylene. K1=OX/MX, K2=MX/PX, K3=PX/OX. The triangular scheme adds to the mechanism the direct conversion between oand p-xylene in order to account for the influence of intracrystalline masstransfer resistance [14]. Three expression of the form lnK = f(1/T) are obtained for each equilibrium constant through weighted least squares regression; F-test is used in order to determine the order of the polynomial and the significance of each parameter [15]: ln𝐾1=4190000±130000(𝑇K ⁄)−3−259±4(𝑇K ⁄)−1−0.486±0.008; 𝑅adj 2=0.9996 (3.12) ln𝐾2=−8700±1800(𝑇K ⁄)−2+175±12(𝑇K ⁄)−1+0.500±0.018; 𝑅adj 2=0.9989 (3.13) ln𝐾3=−29500±600(𝑇K ⁄)−2+197±4(𝑇K ⁄)−1−0.122±0.005; 𝑅adj 2=0.9964 (3.14) Thermodynamic equilibrium of xylene isomerization in liquid phase 49 The curves obtained by equations (3.12) to (3.14) are depicted in Figure 3.2. The product distribution in thermodynamic equilibrium for the three xylenes is calculated by linear combination of two of the equilibrium constants defined by equations (3.9) to (3.11), and the material balance (∑Xi = 1). The obtained product distribution for several temperatures is presented in Table 3.5. Figure 3.2 Equilibrium constants Ki as a function of temperature according to equations (3.12) to (3.14). (♦) i = 1 (▲) i = 2 (■) i = 3. Error bars are larger for temperatures above 440 K due to the increase in the uncertainty of the saturation functions of m-xylene and o-xylene. Table 3.5 Equilibrium product distribution (mol %) based on the equilibrium constants from equations (3.12) to (3.14)a. T, K p-Xylene m-Xylene o-Xylene 250 21.2 ± 1.1 61.3 ± 2.3 17.5 ± 1.0 300 22.2 ± 1.2 59.7 ± 2.4 18.1 ± 1.0 350 23.0 ± 1.2 58.2 ± 2.5 18.8 ± 1.1 400 23.5 ± 1.2 56.9 ± 2.6 19.6 ± 1.2 450 24.0 ± 1.3 55.8 ± 2.7 20.2 ± 1.2 500 24.3 ± 1.3 54.9 ± 2.8 20.8 ± 1.3 550 24.5 ± 1.3 54.2 ± 2.8 21.3 ± 1.3 a Uncertainties of equilibrium constants from equations (3.12) to (3.14) are calculated based on prediction of new values of the fitted curves and combined in quadrature to obtain the uncertainties within the product distribution Unfortunately, there are very few references of thermodynamic equilibrium for xylene isomerization in liquid phase in the literature. For instance, Cappellazzo et al. [14] and Norman Chapter 3 50 et al. [16] carried out experiments on xylene isomerization in liquid phase; they used the equilibrium constants within the kinetic parameters, however they do not show the actual values. Chirico and Steele [6] only reported isomerization equilibrium in the liquid phase at T = 323 K: (58.9 ± 2.9) % of m-xylene, (18.3 ± 1.7) % of o-xylene, and (22.8 ± 2.4) % of p-xylene. Using the expressions obtained in this study the following equilibrium distribution is obtained: (59.0 ± 2.4) % of m-xylene, (18.4 ± 1.1) % of o-xylene, and (22.6 ± 1.2) % of p-xylene. The aforementioned values show excellent agreement between themselves. 3.3. Conclusions Three expressions are developed to determine the thermodynamic equilibrium constants for xylene isomerization in liquid phase between 250 and 550 K. A simple procedure is followed based on published thermodynamic functions at saturation pressure. 3.4. Nomenclature 𝐺 = Gibbs free energy, J mol-1 𝐻 = Enthalpy, J mol-1 𝐾𝑖 = Equilibrium constant 𝑖 𝑀𝑊 = Molecular weight, g mol-1 𝑃 = Pressure, kPa 𝑅 = Universal gas constant, J mol-1K-1 𝑅adj 2 = Coefficient of determination adjusted R-squared 𝑆 = Entropy, J mol-1K-1 𝑇 = Temperature, K 𝑉 = Molar volume, m3 mol-1 𝑋𝑖 = Molar fraction of component 𝑖 Greek letters 𝜌 = Density, kg m-3 Thermodynamic equilibrium of xylene isomerization in liquid phase 51 𝜔 = Acentric factor Superscripts and subscripts 0 = Standard state conditions 𝑐 = Critical 𝑓 = Formation property 𝑚 = Property on molar basis 𝑟 = Reduced property 𝑅𝑖 = Reaction 𝑖 𝑠 = Saturation conditions Abbreviations MX = m-Xylene OX = o-Xylene PX = p-Xylene 3.5. References [1] Amelse, J. A. 1992. The influence of diffusion limitations on xylene isomerization. Paper read at Proceedings from the Ninth International Zeolite Conference, at Montreal. [2] Chirico, R. D., S. E. Knipmeyer, A. Nguyen, and W. V. Steele. 1997. "Thermodynamic Equilibria in Xylene Isomerization. 1. The Thermodynamic Properties of p-Xylene." J. Chem. Eng. Data no. 42 (2):248-261. [3] Chirico, R. D., S. E. Knipmeyer, A. Nguyen, J. W. Reynolds, and W. V. Steele. 1997. "Thermodynamic Equilibria in Xylene Isomerization. 2. The Thermodynamic Properties of mXylene." J. Chem. Eng. Data no. 42 (3):475-487. [4] Chirico, R. D., S. E. Knipmeyer, A. Nguyen, A. B. Cowell, J. W. Reynolds, and W. V. Steele. 1997. "Thermodynamic Equilibria in Xylene Isomerization. 3. The Thermodynamic Properties of o-Xylene." J. Chem. Eng. Data no. 42 (4):758-771. [5] Chirico, R. D., S. E. Knipmeyer, A. Nguyen, and W. V. Steele. 1997. "Thermodynamic Equilibria in Xylene Isomerization. 4. The Thermodynamic Properties of Ethylbenzene." J. Chem. Eng. Data no. 42 (4):772-783. Chapter 4 58 Figure 4.4 Simplified reactor scheme 4.2.3. Reaction system Reactor data, supplied by GALP personnel, regarding about two months of continuous operation are used in the study; however, data exhibiting deviations from normal operation are ruled out. Reactions involving hydrogen consumption are not considered in the mathematical modeling since hydrogen consumption is less than 2%. These reactions are non-aromatic cracking with the production of light hydrocarbons and hydrodealkylation where xylenes are converted to toluene, which may undergo further hydrodealkylation to benzene over high acidic catalyst, and methane. In normal operation, the naphthene concentration is kept constant for the isomerization of ethylbenzene; loss of naphthenes in the top of the deheptanizer or through cracking leads to a loss of aromatics in order to synthesize more naphthenes. Furthermore, side reactions such as disproportionation and transalkylation are not considered since benzene, toluene, and C9-aromatic concentrations are below 1%. Table 4.2 summarizes the reactor data to be used in the mathematical modeling. According to the aforementioned, two main reactions are considered: ethylbenzene isomerization and xylene isomerization. For ethylbenzene isomerization, a bifunctional model is followed according to Roebschlaeger and Christoffel [13]. Ethylbenzene is hydrogenated on platinum sites, then the Gas phase isomerization unit 59 Table 4.2 Summarized reactor data used in the mathematical modeling OXout, wt% 14.96 15.53 15.29 14.63 15.01 15.41 15.46 15.13 15.15 14.83 15.11 15.16 14.86 15.04 15.22 14.66 14.29 MXout, wt% 34.74 36.05 35.35 34.03 34.75 35.58 35.58 34.89 35.33 34.55 35.24 35.34 34.90 34.92 35.48 34.16 33.28 PXout, wt% 12.94 13.41 13.39 13.60 13.86 14.19 14.34 15.67 14.99 14.48 14.79 14.93 14.78 14.75 14.99 14.45 14.09 EBout, wt% 6.97 7.18 7.08 6.93 7.19 7.43 7.40 7.33 7.35 7.71 7.83 7.70 7.27 7.27 7.48 7.20 6.98 OXin, wt% 14.14 14.88 14.34 14.11 14.38 14.63 14.50 14.65 14.38 14.22 14.59 14.55 14.32 14.27 14.41 14.00 13.56 MXin, wt% 43.93 45.31 44.75 43.56 44.68 46.10 46.39 46.36 46.17 45.18 46.16 46.31 45.78 45.89 46.55 44.80 43.77 PXin, wt% 0.90 0.92 1.06 0.92 1.00 1.04 1.11 1.02 1.10 1.21 1.25 1.18 1.08 1.03 1.10 1.05 0.95 EBin, wt% 10.63 11.06 10.95 10.60 10.76 10.84 10.78 10.99 11.15 10.95 10.97 11.10 10.63 10.78 11.10 10.62 10.35 Inert, wt% 24.61 23.38 23.42 25.86 23.83 22.29 22.10 21.92 21.54 23.39 21.80 21.84 23.59 21.77 22.40 23.21 25.85 H2, wt% 5.79 4.45 5.47 4.95 5.35 5.10 5.12 5.06 5.65 5.04 5.22 5.02 4.60 6.26 4.45 6.32 5.52 Pressure, kPa 814 814 804 824 814 804 814 794 794 804 804 804 804 804 804 814 834 Flow, ton h-1 90.31 88.21 88.83 91.52 90.22 87.86 87.79 87.77 88.19 88.57 87.09 85.81 87.37 87.09 85.97 88.66 91.03 Day 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Chapter 4 60 OXout, wt% 14.24 14.32 14.42 14.95 14.81 14.48 14.62 13.43 13.52 13.72 14.12 13.42 13.38 13.58 14.32 14.57 14.76 MXout, wt% 33.23 33.40 33.65 35.00 34.63 34.00 34.35 31.57 31.77 32.67 33.53 31.47 31.44 31.96 33.51 34.21 34.78 PXout, wt% 14.02 14.09 14.17 14.76 14.64 14.43 14.59 13.34 13.39 13.65 14.01 13.43 13.41 13.69 14.24 14.63 14.82 EBout, wt% 7.00 7.01 7.09 7.38 7.22 6.99 7.03 6.64 6.68 6.88 7.01 6.62 6.71 6.88 7.52 7.48 7.77 OXin, wt% 13.64 13.57 13.54 14.03 13.61 13.50 13.64 12.59 12.74 12.79 13.30 12.15 12.50 13.02 13.44 13.54 13.53 MXin, wt% 43.52 43.66 44.19 45.97 45.70 44.75 44.94 41.28 41.56 42.57 43.54 40.91 40.63 41.10 42.81 44.10 44.85 PXin, wt% 0.98 1.03 1.16 1.21 1.22 1.22 1.31 1.37 1.32 1.42 1.57 1.67 1.67 1.78 2.27 2.22 2.68 EBin, wt% 10.34 10.55 10.45 10.88 10.77 10.44 10.69 9.74 9.74 10.14 10.27 10.22 10.13 10.22 11.07 11.02 11.07 Inert, wt% 27.58 26.78 24.26 23.00 24.06 25.08 25.17 31.17 29.29 28.45 27.15 30.76 30.55 29.67 25.48 23.37 21.79 H2, wt% 3.94 4.40 6.40 4.91 4.65 5.02 4.24 3.85 5.36 4.64 4.18 4.29 4.51 4.22 4.93 5.74 6.08 Pressure, kPa 843 834 824 814 814 814 814 873 863 843 843 863 863 863 853 834 824 Flow, ton h-1 91.30 92.39 90.54 86.41 86.64 88.40 88.84 96.05 95.52 92.14 89.99 93.31 93.58 96.40 93.33 88.28 88.85 Day 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Gas phase isomerization unit 61 intermediate is isomerized in the rate controlling step on the acidic sites and very rapidly dehydrogenated producing a mixture of xylene based on the high acidity of the catalyst. According to the authors, there was evidence of competition for adsorption on the acidic sites. Since thermodynamic equilibrium between aromatics and the related naphthenes is rapidly established, naphthene concentrations can be replaced by concentrations of aromatics and hydrogen. Regarding the xylene isomerization, Corma and Cortes [4] presented a consecutive 1,2-methyl shift with single site surface reaction controlling mechanism. The reaction system to be used in the model is the result of combining both ethylbenzene isomerization and xylene isomerization (see Figure 4.5). According to Bhatia et al. [14], from data on ethylbenzene isomerization, the selectivity of o-, m-, and p-xylene is 0.32, 0.47, and 0.21 respectively. Based on that, the rates are as follows: 𝑅EB=𝑘2𝑃H2 2(𝑃OX+𝑃MX+𝑃PX)−𝑘1𝑃H2 2𝑃EB 1+(𝑃𝐸𝐵+𝑃𝑂𝑋+𝑃𝑀𝑋+𝑃𝑃𝑋)(𝐾X+𝐾H𝑃H2 2) (4.1) 𝑅OX=𝑘3𝑃MX−𝑘4𝑃OX 1+𝐾OX𝑃OX+𝐾PX𝑃PX+𝐾MX𝑃MX+0.32𝑘1𝑃H2 2𝑃EB−𝑘2𝑃H2 2𝑃OX 1+(𝑃EB+𝑃OX+𝑃MX+𝑃PX)(𝐾X+𝐾H𝑃H2 2) (4.2) 𝑅MX=𝑘4𝑃OX+𝑘5𝑃PX−𝑘6𝑃MX−𝑘3𝑃MX 1+𝐾OX𝑃OX+𝐾PX𝑃PX+𝐾MX𝑃MX+0.47𝑘1𝑃H2 2𝑃EB−𝑘2𝑃H2 2𝑃MX 1+(𝑃EB+𝑃OX+𝑃MX+𝑃PX)(𝐾X+𝐾H𝑃H2 2) (4.3) 𝑅PX=𝑘6𝑃MX−𝑘5𝑃PX 1+𝐾OX𝑃OX+𝐾PX𝑃PX+𝐾MX𝑃MX+0.21𝑘1𝑃H2 2𝑃EB−𝑘2𝑃H2 2𝑃PX 1+(𝑃EB+𝑃OX+𝑃MX+𝑃PX)(𝐾X+𝐾H𝑃H2 2) (4.4) Figure 4.5 Reaction system for ethylbenzene and xylene isomerization Chapter 4 62 4.2.4. Reactor modeling The reactor model is based on the following assumptions: 1. Steady state 2. Isothermal conditions. 3. Constant volume. 4. Boundary conditions are of Danckwerts’ type. 5. Mass transfer in radial direction can be described by means of the diffusion model. 6. Channeling or shortcut effects do not occur. 7. Absence of gradients in axial direction. 8. Pressure drop is neglected. Mass balance of component i in the volume element for inward-type radial flow: (𝑁𝑖𝐴𝑟)𝑟+∆𝑟−(𝑁𝑖𝐴𝑟)𝑟+𝑅𝑖𝜌𝐵𝐴𝑟∆𝑟=0 (4.5) 𝑑 𝑑𝑟(𝑁𝑖𝐴𝑟)+𝑅𝑖𝜌𝐵𝐴𝑟=0 (4.6) where Ni is the molar flux of component i, Ri is the reaction rate defined previously for each species, and ρB is the bed density (total catalyst mass / catalyst-bed volume). Convection and dispersion in the radial direction leads to: 𝑁𝑖=𝑣𝐶𝑖+𝐷𝑑𝐶𝑖 𝑑𝑟 (4.7) Combining equations (4.6) and (4.7): 𝑑 𝑑𝑟(2𝜋𝑟𝐿𝐷𝑑𝐶𝑖 𝑑𝑟)+𝑑 𝑑𝑟(2𝜋𝑟𝐿𝑣𝐶𝑖)+2𝜋𝑟𝐿𝜌𝐵𝑅𝑖=0 (4.8) where D is the coefficient of turbulent mixing in the radial direction, u is the local velocity which depends on the radial coordinate r, and Ci is the concentration of component i. After rearranging: 𝐷(𝑑2𝐶𝑖 𝑑𝑟2+1𝑟𝑑𝐶𝑖 𝑑𝑟)+𝐶𝑖 𝑟𝑑 𝑑𝑟(𝑣𝑟)+𝑣𝑑𝐶𝑖 𝑑𝑟+𝜌𝐵𝑅𝑖=0 (4.9) Following assumption (3), the continuity equation is in the form: 2𝜋𝑅1𝑣1=2𝜋𝑅0𝑣0=2𝜋𝑟𝑣=𝑐𝑜𝑛𝑠𝑡.→𝑑 𝑑𝑟(𝑣𝑟)=0 (4.10) Boundary conditions based on assumption (4): 𝑟=𝑅1 𝑑𝐶𝑖 𝑑𝑟=0; 𝑟=𝑅0 𝑣0(𝐶𝑖𝑖𝑛−𝐶𝑖)=𝐷𝑑𝐶𝑖 𝑑𝑟 (4.11) Using dimensionless variables Yi = Ci/C; ξ = r/R0; and introducing the Peclet number Pe = R0v0/D, model equations become: Gas phase isomerization unit 63 1 Pe(𝑑2𝑌𝑖 𝑑𝜉2+1𝜉𝑑𝑌𝑖 𝑑𝜉)+1𝜉𝑑𝑌𝑖 𝑑𝜉+𝑅0𝜌𝐵 𝑣0𝐶𝑅𝑖=0 (4.12) 𝑑𝑌𝑖 𝑑𝜉=0 at 𝜉=𝑅1 𝑅0; Pe(𝑌𝑖𝑖𝑛−𝑌𝑖)=𝑑𝑌𝑖 𝑑𝜉 𝑎𝑡 𝜉=1 (4.13) The parameters and properties of the species involved in the simulation are taken from Green and Perry [15]. The molar volume (V) for the aromatics and naphthenes (inert) is calculated using the virial equation truncated after the second virial coefficient (see Section A.2 in Annex A). At an average temperature 365 ºC (638.15 K) and pressure 8.4 bar (823.8 kPa) the molar volumes of o-, m-, p-xylene, ethylbenzene, and ethylcyclohexane (used for naphthene) are 5.926, 5.938, 5.935, 5.948, and 5.876 m3 kmol-1 respectively. For hydrogen, ideal gas equation is used since temperature is above critical conditions (V = 6.441 m3 kmol-1). For the mixture, the molar volume and molecular weight are calculated by means of a weighted mean using the mole fractions. Mole fractions at the reactor inlet are used since naphthene and hydrogen fractions are constant within the reactor and properties of ethylbenzene and xylenes are very similar. According to Levenspiel [16] and Balakotaiah and Luss [17], for high Reynolds numbers D/vdp ≈ 1/2. Based on that, the Peclet number is determined as follows: Pe=𝑅0𝑣0 𝐷=𝑅0𝑣0 12 ⁄𝑣0𝑑𝑝=2𝑅0 𝑑𝑝 (4.14) 4.3. Results and discussion The second order differential equations are solved through the commercial software gPROMS v.3.7.1 from Process Systems Enterprise (www.psenterprise.com) by the numeric solver DASOLV with a second order orthogonal collocation in finite elements method discretization of the radial domain using 30 uniform intervals with 10-5 as tolerance. The kinetic and adsorption constants used, as a first attempt, are obtained from the authors from which the kinetic model is based on. Deviations between the calculated and the actual data are expected since neither the catalysts nor the conditions are the same, Roebschlaeger and Christoffel [13] used Pt/zeolite at 422 ºC and Corma and Cortes [4] worked with Ni/Silica-Alumina at 400 ºC. Based on the aforementioned, a trust-region reflective least squares procedure [18,19] is carried out with the purpose to obtain new values of kinetic constants in order to minimize the error. The adsorption constants are not included in the optimization (i.e. the adsorption constants are not optimized) based on their small variation among C8-aromatics and low influence of temperature. The values are taken directly from Roebschlaeger and Christoffel [13], KX = 6.3×10-3 kPa-1 and KH = 5×10-7 kPa-3; and from Corma and Cortes [4], KOX = 7.9×10-3 Chapter 4 64 kPa-1, KPX = 1.43×10-2 kPa-1, and KMX = 1.14×10-2 kPa-1. The function to be minimized is the following: 𝑆=√∑∑(𝑌𝑗,𝑖𝑎𝑐𝑡𝑢𝑎𝑙−𝑌𝑗,𝑖𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒𝑑)2 4 𝑖=1 34 𝑗=1 (4.15) The new kinetic constants reduce the objective function (S) in about 50%. Table 4.3 presents the kinetic constants for both simulations and the value of the objective function. Moreover, Table 4.4 presents the actual and calculated weight fractions for the species using the kinetics obtained from the optimization for each day. Hydrogen and Inert are not shown since the fractions are constant within the reactor. The gradients for a given day can be seen in Figure 4.6. Table 4.3 Comparison of kinetic constantsa Source k1×108 k2×109 k3×104 k4×104 k5×104 k6×104 S Literature 1.8 1.6 2.80 6.76 9.27 3.92 0.066 Optimizationb 1.62 1.41 2.24 5.82 7.4 3.48 0.033 a Units correspond to reaction rate kmol kg-1h-1. b Values are presented with one digit more than justified Figure 4.6 Weight fraction of each species against dimensionless radial coordinate for a given day Moreover, according to Hlavacek [12] the effect of mixing in the flow direction can be disregarded when Pe(1−𝜉)>50. Based on this assumption the reactor model is as follows: 1𝜉𝑑𝑌𝑖 𝑑𝜉+𝑅0𝜌𝐵 𝑢0𝐶𝑅𝑖=0 (4.16) 𝜉=1 𝑌𝑖=𝑌𝑖𝑖𝑛 (4.17) Gas phase isomerization unit 65 Table 4.4 Actual and calculated outlet weight fractions. Calculated values are predicted by the model with optimized kinetics Day Ethylbenzene p-Xylene m-Xylene o-Xylene actual calculated actual calculated actual calculated actual calculated 1 6.97 7.15 12.94 13.87 34.74 33.90 14.96 14.67 2 7.18 7.45 13.41 14.61 36.05 34.99 15.53 15.12 3 7.08 7.35 13.39 14.24 35.35 34.59 15.29 14.93 4 6.93 7.13 13.60 13.92 34.03 33.62 14.63 14.52 5 7.19 7.29 13.86 14.17 34.75 34.47 15.01 14.89 6 7.43 7.41 14.19 14.62 35.58 35.36 15.41 15.22 7 7.40 7.40 14.34 14.69 35.58 35.47 15.46 15.22 8 7.33 7.49 15.67 14.68 34.89 35.55 15.13 15.30 9 7.35 7.54 14.99 14.54 35.33 35.48 15.15 15.25 10 7.71 7.38 14.48 14.44 34.55 34.80 14.83 14.94 11 7.83 7.47 14.79 14.72 35.24 35.53 15.11 15.27 12 7.70 7.49 14.93 14.83 35.34 35.56 15.16 15.26 13 7.27 7.28 14.78 14.62 34.90 34.95 14.86 14.97 14 7.27 7.32 14.75 14.36 34.92 35.16 15.04 15.13 15 7.48 7.50 14.99 14.92 35.48 35.54 15.22 15.19 16 7.20 7.18 14.45 14.05 34.16 34.41 14.66 14.82 17 6.98 6.99 14.09 13.81 33.28 33.46 14.29 14.37 18 7.00 6.99 14.02 14.04 33.23 33.24 14.24 14.21 19 7.01 7.10 14.09 13.98 33.40 33.42 14.32 14.31 20 7.09 7.07 14.17 13.83 33.65 33.89 14.42 14.55 21 7.38 7.35 14.76 14.69 35.00 35.08 14.95 14.97 22 7.22 7.26 14.64 14.61 34.63 34.68 14.81 14.74 23 6.99 7.08 14.43 14.24 34.00 34.07 14.48 14.52 24 7.03 7.23 14.59 14.46 34.35 34.30 14.62 14.60 25 6.64 6.60 13.34 13.40 31.57 31.57 13.43 13.41 26 6.68 6.62 13.39 13.23 31.77 31.88 13.52 13.63 27 6.88 6.81 13.65 13.71 32.67 32.55 13.72 13.83 28 7.01 6.96 14.01 14.17 33.53 33.37 14.12 14.17 29 6.62 6.69 13.43 13.42 31.47 31.50 13.42 13.33 30 6.71 6.67 13.41 13.36 31.44 31.50 13.38 13.40 31 6.88 6.85 13.69 13.52 31.96 32.06 13.58 13.69 32 7.52 7.34 14.24 14.13 33.51 33.73 14.32 14.39 33 7.48 7.33 14.63 14.41 34.21 34.48 14.57 14.68 34 7.77 7.46 14.82 14.61 34.78 35.16 14.76 14.90 The results obtained with this simpler model are exactly the same to those obtained with equations (4.12) and (4.13). The optimized kinetic constants are the following: k1=(1.62±1.08)×10-8 kmol kg-1kPa-3h-1; k2=(1.43±1.41)×10-9 kmol kg-1kPa-3h-1; k3=(2.24±0.31)×10-4 kmol kg-1kPa-1h-1; k4=(5.82±0.70)×10-4 kmol kg-1kPa-1h-1; k5=(7.4±3.8)×10-4 Chapter 4 66 kmol kg-1kPa-1h-1; and k6=(3.48±1.36)×10-4 kmol kg-1kPa-1h-1. The standard error of each kinetic constant is calculated following the deleted-one Jackknife method as described by Kinsella [20]. According to Al Khattaf [7] shape selectivity in xylene isomerization is observed in medium-pore zeolites such as ZSM-5; this is not the case for large-pore zeolites. The shape selectivity is based on the steric hindrance due to the sizes of the molecules involved. The p-xylene critical size is lower than the other xylenes, which enables it to diffuse faster through the catalyst pores. For larger pores, the steric hindrance decreases and the three xylenes diffuse at the same rate. It can be seen in Table 4.3 that k3, k4, k5, and k6 are slightly reduced by the optimization, probably due to the lower temperature used in the unit. Furthermore, k5 is reduced at a higher degree than k6, favoring the formation of p-xylene from m-xylene; similarly, k3 is more decreased than k4, also favoring the formation of p-xylene indirectly through m-xylene as an intermediate. This may indicate shape selectivity of the catalyst currently used in the isomerization unit. The EU-1 zeolite has a 10-membered ring structure very similar to that of ZSM-5, whereas the catalyst used by Corma and Cortes [4] has large pores. Moreover, the particle size used was very small in order to eliminate the intraparticle diffusion constraints. For larger particles, such as those used in industry applications, the aforementioned diffusion constraints start to appear and increase the reaction rate of p-xylene compared to oand m-xylene. Isomerization units normally operate between 380 and 480 ºC [21]; the main advantage of EU-1 zeolite is the possibility to operate at milder conditions, thus reducing C8-aromatic losses through transalkylation and cracking, which is confirmed by the lower contents of side products in the data used in the modeling. Furthermore, it can be seen in Table 4.3 that the reduction on k1 and k2 is low compared to the almost 60 ºC difference between the unit and the work of Roebschlaeger and Christoffel [13]. The higher ethylbenzene isomerization activity observed in this catalyst is due to the easy access by the intermediates to the protonic sites located at the side pockets at the surface of the crystallites and close to the hydrogenation sites (i.e., platinum). This situation is more advantageous than that of the normally used mordenite [10]. The competitive adsorption of product species within the catalyst is evaluated by using a simplified linear kinetic model eliminating the effect of the adsorption constants. In this case the value of S is higher, from which it is validated the use of the Hougen-Watson model for the C8-aromatic isomerization. Furthermore, a triangular scheme for xylene isomerization (considering the direct conversion between oand p-xylene) is also evaluated to account for shape selectivity in the catalyst. Although there are signs that may indicate selectivity towards p-xylene, S is similar for this case; since there is no improvement, the simpler model with less parameters is preferred. Laboratory experiments where intrinsic kinetic data can be obtained are required to choose the model that better fits the data and reduce the uncertainty of the parameters estimated. Gas phase isomerization unit 67 4.4. Conclusions The gas phase isomerization unit with a radial-type reactor and EU-1 zeolite with platinum as catalyst is analyzed. It was confirmed that side reactions and C8-aromatic losses are significantly reduced by operating at milder conditions due to the higher activity of the catalyst. A mathematical model that effectively simulates the operation of the reactor within the isomerization unit is developed; the model is intended to be used within the simulation of the proposed aromatics complex. 4.5. Nomenclature 𝐴𝑟 = Cross sectional area, m2 𝐵 = Virial coefficient 𝐶 = Concentration, kmol m-3 𝐷 = Dispersion coefficient in the flow direction, m2 h-1 𝑑𝑝 = Particle diameter, m 𝑘𝑙 = Kinetic constant on reaction 𝑙 = 1,2, kmol kg-1kPa-3h-1 𝑘𝑙 = Kinetic constant on reaction 𝑙 = 3-6, kmol kg-1kPa-1h-1 𝐾X = Adsorption constant of C8-aromatics, kPa-1 𝐾H = Product of adsorption of naphthenes and hydrogenation, kPa-3 𝐾𝑚 = Adsorption constant of species 𝑚 = OX, MX, PX, kPa-1 𝐿 = Catalyst-bed length, m 𝑁𝑖 = Surface molar flow, kmol m-2h-1 𝑃 = Pressure, kPa Pe = Peclet number 𝑟 = Radial coordinate, m 𝑅 = Universal gas constant, m3kPa kmol-1K-1 𝑅0 = Catalyst-bed outer radius, m Chapter 5 74 5.3. Simulated moving bed reactor Xylene isomerization is a reaction of the type 𝐴↔𝐵. In this case, reaction cannot occur near the extract point if high purity is required, otherwise the reverse reaction will pollute the product and purity will always be below 99%. To overcome this situation, reactors are inserted between the adsorption columns far from the extract point [6,7]. However, since the minimum concentration required in the extract for this new configuration is about 70 wt%, a much simpler configuration can be employed. Keeping the catalyst and adsorbent mixed inside the columns, it may produce a high enough p-xylene concentration stream to be further processed by the crystallization unit. This approach involves simpler operation and allows the direct contact between catalyst and adsorbent resulting in more efficient p-xylene withdraw as it is formed to overcome the thermodynamic equilibrium constraints. One of the most employed SMB based technologies for p-xylene separation is UOP’s Parex. The studied aromatics complex uses this technology consisting of 24 adsorbent beds with length and diameter of 1.14 and 4.12 m respectively, p-diethylbenzene as desorbent, particle diameter of 0.62 mm, and a switching time of 1.15 min [2]. The SMBR unit will keep the geometric characteristics of the Parex unit, i.e., 24 adsorbent beds, with the possibility to modify the location of inlets and outlets in order to use the appropriate number of columns in each zone since column configuration plays an important role when dealing with different product concentrations [8]. Moreover, p-diethylbenzene cannot be used since it isomerizes into oand m-diethylbenzene over acid catalysts; toluene, which has been used in the industry, is used as desorbent [7,9]. Generally, the feed to the Parex unit contains a naphthenic fraction which is involved in the ethylbenzene isomerization in the Isomar unit. These non-aromatic compounds increase the utility consumption of the unit; however, they do not affect the xylene adsorption [10]. The feed used, as a first attempt, is that used by Minceva et al. [6]: 23.6 wt% p-xylene; 49.7 wt% m-xylene; 12.7 wt% o-xylene; 14 wt% ethylbenzene. This work is foreseen as a modification of the current aromatics complex; that is the main reason to maintain the physical dimensions of the equipment. In case that the resulting flow rates are below or above the downstream units, a second train with the same characteristics could be installed to guarantee optimal operation of said units. It is strongly recommended, whenever possible, to use similar units to the original ones while installing second trains in revamp and/or expansion projects in order to keep operation simplicity. Simulated moving bed reactor: Adsorbent and catalyst homogeneous mixture 75 5.3.1. Adsorption and reaction data The normal operating conditions for the Parex process is around 180 ºC and 9 bar [11]. Pressure shall be high enough to maintain the operation in liquid phase and to avoid failure of associated equipment (e.g., pump cavitation) due to pressure drop in lines and columns; in other words, the influence on adsorption and reaction data is neglected. Temperature, on the other hand, definitely affects the reaction and adsorption data. According to Minceva et al. [6], increase in temperature leads to lower adsorption capacity and faster isomerization, which means that a compromise shall exist somewhere above normal Parex operation. Bergeot [12] carried out adsorption and reaction experiments of xylenes at 200 ºC in liquid phase. The adsorbent used was low silica X zeolite exchanged with barium (BaLSX); the author claimed that the adsorbent presented better selectivity compared to that of BaX, specifically on ethylbenzene. The adsorption equilibrium is described with the generalized Langmuir isotherm: 𝑞𝑖=𝑞𝑠𝑎𝑡 𝑏𝑖𝐶𝑖 1+∑𝑏𝑗𝐶𝑗𝑗 (5.1) The catalyst used in the isomerization tests was HZSM-5, which is industrially used in xylene isomerization in gas phase. The reaction scheme followed is presented in Figure 5.2. According to Cappellazzo et al. [13], the triangular scheme adds to the mechanism the direct conversion between oand p-xylene, which actually does not occur, to account for the influence of intracrystalline mass-transfer resistance. Following the triangular scheme, the reaction rates for each species are given by equations (5.2) to (5.4). Figure 5.2 Xylene isomerization reaction scheme 𝑅PX=𝑘5𝐶OX+𝑘3𝐶MX−𝑘6𝐶PX−𝑘4𝐶PX (5.2) 𝑅MX=𝑘1𝐶OX+𝑘4𝐶PX−𝑘2𝐶MX−𝑘3𝐶MX (5.3) 𝑅OX=𝑘2𝐶MX+𝑘6𝐶PX−𝑘5𝐶OX−𝑘1𝐶OX (5.4) Table 5.1 presents adsorption, reaction, and physical data used in the mathematical modeling of the SMBR unit. Figure 5.3 presents the adsorption isotherms for each species; it can be seen that toluene is a suitable desorbent since it is adsorbed less strongly that p-xylene but more than the rest of the isomers [9]. Chapter 5 76 Table 5.1 Thermodynamic and physical parameters from Bergeot [12] Adsorption Data Reaction Data Physical Data qsat 0.148 kg kg-1 k1 2.1859×10-8 m3 kg-1s-1 ρads 2013 kg m-3 bPX 5.1 m3 kg-1 k2 7.7491×10-9 m3 kg-1s-1 ρcat 1150 kg m-3 bMX 1.5883 m3 kg-1 k3 1.1544×10-4 m3 kg-1s-1 ε 0.32 bOX 1.5883 m3 kg-1 k4 2.5742×10-4 m3 kg-1s-1 εp 0.352 bEB 1.7647 m3 kg-1 k5 3.6647×10-7 m3 kg-1s-1 Rpa 3.1×10-4 m bTol 3.4 m3 kg-1 k6 2.8973×10-7 m3 kg-1s-1 a Particle size currently used within the SMB unit [2]. Figure 5.3 Adsorption isotherms for each species. Isotherms for mand o-xylene are the same [12] 5.3.2. Mathematical model The SMB process (and analogously the SMBR process) can be modeled by means of a continuous TMB (True Moving Bed) model or by simulating the actual shifting of the inlet and outlet ports along the unit. The equivalent TMB approaches the SMB when a large number of columns are involved (e.g., 24 columns) and provides a fast way to obtain product yields for different configurations and parameters at steady-state [14]. Regardless of the methodology used to model the SMBR, the assumptions are the following: 1. Isothermal conditions. 2. Axial dispersed plug flow for the fluid phase. 3. Plug flow for the solid phase (just for TMBR). 4. Constant flow rate in each zone. 5. Mass transfer described by linear driving force (LDF) approximation. 6. Adsorbents and catalysts are mixed homogeneously and possess similar physical characteristics. Simulated moving bed reactor: Adsorbent and catalyst homogeneous mixture 77 7. Pressure drop is not considered. Mass balances for species 𝑖 in the bulk phase, adsorbent particle phase, and catalyst particle phase are given by equations (5.5) to (5.7) respectively: 𝜕𝐶𝑖 𝜕𝑡=𝐷𝑎𝑥𝜕2𝐶𝑖 𝜕𝑧2−𝑢𝜕𝐶𝑖 𝜕𝑧−1−𝜀 𝜀3 𝑅𝑝𝐾𝑙[𝜑(𝐶𝑖−𝐶𝑝𝑖     𝑎𝑑𝑠)+(1−𝜑)(𝐶𝑖−𝐶𝑝𝑖     𝑐𝑎𝑡)] (5.5) 𝜀𝑝𝜕𝐶𝑝𝑖     𝑎𝑑𝑠 𝜕𝑡 +(1−𝜀𝑝)𝜌𝑎𝑑𝑠𝜕𝑞𝑖  𝜕𝑡 =𝑢𝑠[𝜀𝑝𝜕𝐶𝑝𝑖     𝑎𝑑𝑠 𝜕𝑧 +(1−𝜀𝑝)𝜌𝑎𝑑𝑠𝜕𝑞𝑖  𝜕𝑧]+3 𝑅𝑝𝐾𝑙(𝐶𝑖−𝐶𝑝𝑖     𝑎𝑑𝑠) (5.6) 𝜀𝑝𝜕𝐶𝑝𝑖     𝑐𝑎𝑡 𝜕𝑡 =𝑢𝑠𝜀𝑝𝜕𝐶𝑝𝑖     𝑐𝑎𝑡 𝜕𝑧 +3 𝑅𝑝𝐾𝑙(𝐶𝑖−𝐶𝑝𝑖     𝑐𝑎𝑡)+𝜌𝑐𝑎𝑡𝑅𝑖 (5.7) where 𝐶𝑖 is the concentration of each species in the bulk phase, the axial dispersion coefficient 𝐷𝑎𝑥 is estimated through the particle Peclet number (Pe=2𝑢𝑅𝑝𝐷𝑎𝑥 ⁄), which is about 0.45 for low Reynolds in liquids [15], with constant interstitial velocity 𝑢 throughout each zone. Physical properties such as bed porosity 𝜀, particle porosity 𝜀𝑝, particle radius 𝑅𝑝, density of adsorbent 𝜌𝑎𝑑𝑠 and catalyst 𝜌𝑐𝑎𝑡 are given in Table 5.1. The average mass adsorbed 𝑞𝑖  is described by the equilibrium isotherm as a function of the average particle concentration in the adsorbents 𝐶𝑝𝑖     𝑎𝑑𝑠; the reaction rate 𝑅𝑖 is a function of the average particle concentration in the catalysts 𝐶𝑝𝑖     𝑐𝑎𝑡, and 𝜑 represents the adsorbent to adsorbent plus catalyst weight ratio (𝜑=𝑚𝑎𝑑𝑠 𝑚𝑎𝑑𝑠+𝑚𝑐𝑎𝑡 ⁄). The axial derivative terms in the particle balances represent the movement of the solid, these terms do not exist when using the SMBR method. Moreover, all time derivative terms are set to zero for steady state using the TMBR method resulting in a simpler and faster model; this reduction cannot be done while using SMBR method since the model is intrinsically dynamic leading towards a cyclic steady state after a certain number of cycles. The boundary condition for the particle phase, both adsorbent and catalyst, is given by setting the outlet concentration within the particles equal to that entering the previous zone due to the countercurrent movement of the solid. For the bulk phase, Danckwerts boundary conditions for species 𝑖 are used: 𝑧=0 → 𝐷𝑎𝑥𝜕𝐶𝑖 𝜕𝑧=𝑢(𝐶𝑖−𝐶𝑖𝑖𝑛); 𝑧=𝐿 → 𝜕𝐶𝑖 𝜕𝑧=0 (5.8) where 𝐶𝑖𝑖𝑛 depends on the specific zone within the unit. The inlet concentration is determined by mass balances in each inlet and outlet port: Chapter 5 78 Desorbent (D) port: 𝑄4+𝑄𝐷=𝑄1; 𝐶𝑖,4 𝑜𝑢𝑡𝑄4+𝐶𝑖,𝐷𝑄𝐷=𝐶𝑖,1 𝑖𝑛𝑄1 (5.9) Extract (X) port: 𝑄1=𝑄2+𝑄𝑋; 𝐶𝑖,1 𝑜𝑢𝑡=𝐶𝑖,2 𝑖𝑛=𝐶𝑖,𝑋 (5.10) Feed (F) port: 𝑄2+𝑄𝐹=𝑄3; 𝐶𝑖,2 𝑜𝑢𝑡𝑄2+𝐶𝑖,𝐹𝑄𝐹=𝐶𝑖,3 𝑖𝑛𝑄3 (5.11) Raffinate (R) port: 𝑄3=𝑄3+𝑄𝑅; 𝐶𝑖,3 𝑜𝑢𝑡=𝐶𝑖,4 𝑖𝑛=𝐶𝑖,𝑅 (5.12) where Q1, Q2, Q3, and Q4 are the flow rates for each zone. Normally, the value of an SMB facility is measured through the following performance parameters: purity (desorbent free), recovery, desorbent consumption, and productivity. These parameters can also be calculated for the SMBR; although, deviation from the equilibrium is commonly used instead of recovery since the reaction involved is limited by the thermodynamic equilibrium. The performance parameters, considering p-xylene as the desired product, are defined as follows: Extract Purity: PurX= 𝐶PX,𝑋 𝐶PX,𝑋+𝐶MX,𝑋+𝐶OX,𝑋+𝐶EB,𝑋 (5.13) Raffinate Purity: PurR= 𝐶MX,𝑋+𝐶OX,𝑋+𝐶EB,𝑋 𝐶PX,𝑋+𝐶MX,𝑋+𝐶OX,𝑋+𝐶EB,𝑋 (5.14) Deviation from the Equilibrium: DE=𝐶PX,𝑋𝑄𝑋+𝐶PX,𝑅𝑄𝑅 𝐶PX,𝑒𝑞𝑄𝐹 (5.15) Desorbent Consumption: DC=𝑄𝐷 𝐶PX,𝑋𝑄𝑋 (5.16) Productivity: PR=𝐶PX,𝑋𝑄𝑋 𝑉𝑎𝑑𝑠+𝑐𝑎𝑡 (5.17) where 𝐶PX,𝑒𝑞 is the p-xylene concentration in equilibrium at operating conditions estimated with the expressions developed in Chapter 3. As stated before, LDF is used for the mass transfer resistance. Instead of calculating the gradient of the particle concentration, an average is used; the transfer between phases will then be proportional to the concentration difference. External and internal mass transfer resistances are coupled in a single global mass transfer coefficient [16]: 1 𝐾𝑙=1 𝑘𝑒𝑥𝑡+1 𝜀𝑝𝑘𝑖𝑛𝑡 (5.18) where for consistency with the LDF approximation, the internal coefficient is calculated as 𝑘𝑖𝑛𝑡=5𝐷𝑚𝜏𝑅𝑝 ⁄, for which the correlation proposed by Wakao and Smith is used to estimate Simulated moving bed reactor: Adsorbent and catalyst homogeneous mixture 79 the tortuosity factor as 𝜏=1𝜀𝑝 ⁄ [17]. The external mass transfer coefficient is estimated by the Wilson and Geankoplis correlation valid for 0.0016 < Re < 55 [18]: Sh=1.09 𝜀(ReSc)0.33 with Sh=2𝑘𝑒𝑥𝑡𝑅𝑝 𝐷𝑚; Re=2𝜌𝑅𝑝𝑣 𝜇; Sc=𝜇 𝜌𝐷𝑚 (5.19) where the molecular diffusivity 𝐷𝑚 (cm2 s-1) is calculated through the Wilke-Chang method modified to include the mixed solvent case by Perkins and Geankoplis [19]. Properties are presented in Section A.3 in Annex A. 5.4. Results and discussion The SMBR unit is modeled through the continuous TMBR approach since a large number of simulations are required and some optimization are foreseen to adjust several operation parameters. The simulation comprises a numerical solution using the commercial software gPROMS v3.7.1 from Process Systems Enterprise (www.psenterprise.com). The numerical method involves the discretization of the axial domain using second-order orthogonal collocation on 50 finite elements with 10-5 as tolerance. In the first simulations it was noticed that the adsorbent to adsorbent plus catalyst ratio shall be at least 0.9 in order to reach the desired purity in the extract (0.70); it was also noticed that the maximum purity in the raffinate is slightly above 0.95; therefore, those values are fixed in the entire study. 5.4.1. Separation regions and separation volumes Normally in this type of unit, the flow rates of each zone are expressed as velocity ratios (𝛾𝑗=𝑢𝑗𝑢𝑠 ⁄) using the interstitial velocity in zone 𝑗 and the solid velocity (column length / switching time). In pure separation systems, the flow rates in zone 1 and 4 shall guarantee the regeneration of the adsorbent and desorbent respectively, while the actual separation occurs in zones 2 and 3. In the absence of mass-transfer resistances, the region of separation in the plane 2×3 is constant regardless the value of 1 and 4 (given the previous constraints fulfilled); in the presence of mass-transfer effects, the region is expected to become smaller and somehow dependent on 1 and 4 and shall be evaluated through successive simulations [20]. The steady-state TMBR model is successively solved for several values of 2 and 3 for given 1 and 4. Starting from a small enough value for the feed flow rate, flow rate in zone 2 is increased until the purity constraints are no longer satisfied; afterwards, the feed flow rate is increased and the procedure is repeated as long as the said constraints are not violated. The whole process is Chapter 5 80 repeated for different values of 1 and 4. A simplified diagram to determine the separation regions is presented in Figure 5.4. Figure 5.4 Flow diagram to determine separation regions for different flow conditions Simulated moving bed reactor: Adsorbent and catalyst homogeneous mixture 81 Since the reverse reaction is present in every zone of the unit, a larger amount of desorbent (i.e., higher 1 - 4) is expected comparing to that of TMB. In each separation region in the plane 2×3, the best flow rate in zones 2 and 3 is that which provides higher feed flow rate (i.e., higher 3 - 2). Figure 5.5 shows the influence of 4 and 1 in the separation regions with fixed 1 and 4 and 69 s switching time. Table 5.2 presents the optimum point for each separation region with the corresponding productivity, desorbent consumption, and deviation from the equilibrium for the Parex configuration (6-9-6-3) for 1 values between 3.0 and 6.0 and 4 values between 0.3 and 1.0. Figure 5.5 Separation regions of 6-9-6-3 configuration and 69 s switching time for: a) several values of 4 and fixed 1 = 5.5 b) several values of 1 and fixed 4 = 0.4. Darker region indicates optimum value. It can be seen that the separation region increases as flow rate in zone 1 increases and flow rate in zone 4 decreases. This is referred to as transition region by Azevedo and Rodrigues [20] and is dependent on the mass-transfer effects. Outside the transition region (i.e., higher 1 and lower 4) the separation region is fairly constant, from which the optimum point is chosen as the boundaries of said transition region that provides higher productivity. For this SMBR unit, the a) b) Chapter 5 82 optimum point corresponds to a value of 1 and 4 of 5.5 and 0.4 respectively (see Table 5.2); although there is an increase in productivity for a 1 value from 5.5 to 6.0, it is very small (around 1%) compared to the 10% higher desorbent consumption. Figure 5.6 depicts the profiles within the columns at those operating conditions. Table 5.2 Optimum points for 1 values between 3.0 and 6.0 and 4 values between 0.3 and 1.0 for 6-9-6-3 configuration and 69 s switching time. 4 2 3 PR, kg m-3h-1 DC, m3 kg-1 DE 1 = 3.0 0.3 1.20 1.30 59.73 0.145 1.52 0.4 1.20 1.30 65.89 0.126 1.58 0.5 1.19 1.29 64.18 0.125 1.61 0.6 1.19 1.28 62.30 0.123 1.65 0.7 1.19 1.26 44.60 0.165 1.63 1 = 3.5 0.3 1.17 1.34 121.55 0.084 1.74 0.4 1.17 1.34 127.71 0.078 1.77 0.5 1.16 1.33 126.80 0.076 1.80 0.6 1.16 1.32 121.19 0.076 1.83 0.7 1.16 1.31 116.81 0.077 1.88 0.8 1.16 1.29 107.20 0.081 1.92 0.9 1.16 1.27 89.43 0.093 1.97 1 = 4.0 0.3 1.15 1.35 153.28 0.077 1.80 0.4 1.15 1.35 159.41 0.072 1.82 0.5 1.15 1.35 158.86 0.070 1.85 0.6 1.15 1.34 153.75 0.071 1.88 0.7 1.15 1.32 145.47 0.073 1.93 0.8 0.14 1.31 141.10 0.073 1.98 0.9 1.14 1.28 124.44 0.080 2.04 1.0 1.16 1.24 72.34 0.133 2.08 1 = 4.5 0.3 1.14 1.36 173.23 0.078 1.84 0.4 1.14 1.36 179.35 0.073 1.86 0.5 1.14 1.35 174.67 0.073 1.89 0.6 1.13 1.34 171.48 0.073 1.94 0.7 1.13 1.33 167.79 0.072 1.98 0.8 1.13 1.31 157.38 0.075 2.02 0.9 1.13 1.29 144.43 0.080 2.07 1.0 1.14 1.25 102.51 0.109 2.13 1 = 5.0 0.3 1.13 1.37 186.59 0.081 1.85 0.4 1.13 1.36 186.33 0.079 1.89 0.5 1.13 1.35 185.93 0.077 1.92 0.6 1.13 1.34 181.03 0.078 1.95 Simulated moving bed reactor: Adsorbent and catalyst homogeneous mixture 83 4 2 3 PR, kg m-3h-1 DC, m3 kg-1 DE 0.7 1.13 1.33 177.54 0.077 2.00 0.8 1.13 1.32 168.91 0.079 2.04 0.9 1.13 1.30 156.04 0.084 2.10 1.0 1.13 1.26 123.14 0.104 2.16 1 = 5.5 0.3 1.13 1.37 193.35 0.086 1.88 0.4 1.13 1.36 193.02 0.084 1.91 0.5 1.13 1.36 190.73 0.084 1.93 0.6 1.12 1.34 187.68 0.083 1.97 0.7 1.12 1.33 182.53 0.084 2.01 0.8 1.12 1.32 175.63 0.086 2.07 0.9 1.12 1.30 162.80 0.090 2.13 1.0 1.13 1.27 133.59 0.108 2.18 1 = 6.0 0.3 1.13 1.37 195.74 0.093 1.90 0.4 1.12 1.36 195.25 0.092 1.93 0.5 1.12 1.35 192.94 0.091 1.94 0.6 1.12 1.35 192.18 0.090 1.98 0.7 1.12 1.33 184.57 0.092 2.02 0.8 1.12 1.31 177.50 0.094 2.09 0.9 1.12 1.30 168.01 0.097 2.14 1.0 1.12 1.27 140.10 0.114 2.20 Bold indicates optimum Figure 5.6 Bulk concentration profiles for 1 = 5.5; 2 = 1.13; 3 = 1.36; 4 = 0.4 in 6-9-6-3 configuration and 69 s switching time. Chapter 5 90 𝜌𝑐𝑎𝑡= Density of catalyst, kg m-3 𝜏 = Tortuosity factor 𝜑 = Adsorbent to adsorbent plus catalyst weight ratio Abbreviations EB = Ethylbenzene D = Desorbent F = Feed LDF = Linear Driving Force MX = m-Xylene OX = o-Xylene PX = p-Xylene R = Raffinate SMB = Simulated Moving Bed SMBR = Simulated Moving Bed Reactor TMB = True Moving Bed TMBR = True Moving Bed Reactor Tol = Toluene X = Extract Superscripts and subscripts 𝑖𝑛 = Inlet 𝑜𝑢𝑡 = Outlet 5.7. References [1] Cavani, F., and G. Centi. 2000. "Sustainable Development and Chemistry." In Kirk-Othmer Encyclopedia of Chemical Technology. John Wiley & Sons, Inc. Simulated moving bed reactor: Adsorbent and catalyst homogeneous mixture 91 [2] Minceva, M., and A. E. Rodrigues. 2005. "Two-level optimization of an existing SMB for pxylene separation." Computers and Chemical Engineering no. 29 (10):2215-2228. [3] Starkey, D. R., J. L. Andrews, S. L. Luo, and K. J. Knob. 2009. PxMax with Crystallization. An Integrated Process for High Purity Paraxylene Production. edited by ExxonMobil: ExxonMobil Chemical Technology Licensing LLC. [4] Bass, G., and T. Kinn. 2005. Enhancing Para-xylene Production by Utilizing ExxonMobil's PxMax Process. edited by ExxonMobil: ExxonMobil Chemical Company. [5] GTC Technology. 2001. "Paraxylene production." Hydrocarbon Engineering no. 6 (9):54-55. [6] Minceva, M., P. S. Gomes, V. Meshko, and A. E. Rodrigues. 2008. "Simulated moving bed reactor for isomerization and separation of p-xylene." Chemical Engineering Journal no. 140 (13):305-323. [7] Bergeot, G., D. Leinekugel-Le-Cocq, L. Wolff, L. Muhr, and M. Bailly. 2010. "Intensification of Paraxylene Production using a Simulated Moving Bed Reactor." OGST – Revue d’IFP Energies nouvelles no. 65 (5):721-733. [8] Mun, S. 2012. "Consideration of a target product concentration level in the optimal design of a four-zone simulated moving bed process for binary separation." Chemical Engineering Journal no. 204–206 (0):179-187. [9] Ruthven, D. M., and C. B. Ching. 1989. "Counter-current and simulated counter-current adsorption separation processes." Chemical Engineering Science no. 44 (5):1011-1038. [10] Silva, M. S. P., J. P. B. Mota, and A. E. Rodrigues. 2012. "Fixed-bed adsorption of aromatic C 8 isomers: Breakthrough experiments, modeling and simulation." Separation and Purification Technology no. 90:246-256. [11] Minceva, M. 2004. Separation/Isomerization of Xylenes by Simulated Moving Bed Technology, Departamento de Engenharia Química, Universidade do Porto, Portugal. [12] Bergeot, G. 2010. Extension du concept "One-column" au lit mobile simulé réactif. Application à la séparation réactive des C8 aromatiques, École doctorale Sciences et Ingénierie des Ressources, Procédés, Produits, Environnement, Institut National Polytechnique de Lorraine, France. [13] Cappellazzo, O., G. Cao, G. Messina, and M. Morbidelli. 1991. "Kinetics of Shape-Selective Xylene Isomerization over a ZSM-5 Catalyst." Industrial & Engineering Chemistry Research no. 30 (10):2280-2287. [14] Sá Gomes, P., N. Lamia, and A. E. Rodrigues. 2009. "Design of a gas phase simulated moving bed for propane/propylene separation." Chemical Engineering Science no. 64 (6):13361357. [15] Knaebel, K. S. 2008. "Adsorption." In Albright's Chemical Engineering Handbook, edited by L. Albright. Boca Raton: Taylor & Francis Group, LLC. Chapter 5 92 [16] Santacesaria, E., M. Morbidelli, P. Danise, M. Mercenari, and S. Carrà. 1982. "Separation of xylenes on Y zeolites. 1. Determination of the adsorption equilibrium parameters, selectivities, and mass transfer coefficients through finite bath experiments." Industrial and Engineering Chemistry Process Design and Development no. 21 (3):440-445. [17] Do, D. D. 1998. Adsorption Analysis: Equilibria and Kinetics. London: Imperial College Press. [18] Ruthven, D. M. 1984. Principles of Adsorption and Adsorption Processes. First ed. New York: John Wiley & Sons, cop. [19] Poling, B. E., J. M. Prausnitz, and J. J. P. O'Connell. 2001. The Properties of Gases and Liquids. 5th ed. New York: The McGraw-Hill Companies. [20] Azevedo, D. C. S., and A. E. Rodrigues. 1999. "Design of a simulated moving bed in the presence of mass-transfer resistances." AIChE Journal no. 45 (5):956-966. This chapter is based on: Gonçalves, J. C., and A. E. Rodrigues. 2015. "Simulated Moving Bed Reactor for p-xylene production: Optimal particle size." Canadian Journal of Chemical Engineering. In press. Chapter 6: Simulated moving bed reactor: Optimal particle size In this chapter a similar study as in the previous one is carried out with four particle diameters: 0.5, 0.7, 0.8, and 0.9 mm, maintaining the extract and raffinate purity in 0.70 and 0.95 respectively, and adsorbent to adsorbent plus catalyst weight ratio of 0.9. After performing simulations using the true moving bed approach, it is verified that the high amount of desorbent is mainly caused by the reverse reaction in the isomerization of xylenes. Furthermore, the highest productivity is offered by the 2-6-14-2 configuration for every particle size studied. The system is then analyzed with that arrangement of columns and the aforesaid particle diameters together with the currently used 0.62 mm under the maximum pressure drop of the existing Simulated Moving Bed unit (685 kPa). The optimal particle diameter is 0.62 mm exhibiting the highest productivity. The results also show that a single study with a small particle size is sufficient to accurately determine the best configuration of the system. Chapter 6 94 6.1. Introduction The core of this thesis is the development of a Simulated Moving Bed Reactor (SMBR) for the production of p-xylene in the framework of a proposal to modify the aromatics complex. In the previous chapter, a complete analysis with the currently used particle diameter (i.e., 0.62 mm) was carried out at 200 ºC and milder extract purity constraint (i.e., 0.70) provided by further purification through a crystallization unit. The purpose of this chapter is to follow a similar methodology for different particle diameters (i.e., 0.5, 0.7, 0.8, and 0.9 mm) to determine the optimal size of adsorbents and catalysts with their corresponding flow rates and switching time to be used in the SMBR unit operating at the same temperature subject to the maximum pressure drop constraint of the existing Simulated Moving Bed (SMB) facility. 6.2. Mathematical model As in the previous chapter, the system is modelled by means of a continuous True Moving Bed Reactor (TMBR). This model is simpler, less time-consuming, and suitable for optimizations [1]. The assumptions, mass balances, and boundary conditions are the same; the performance parameters to assess the SMBR unit – productivity (PR), desorbent consumption (DC), and deviation from the equilibrium (DE) – are defined in the same manner, considering p-xylene in the extract point as the desired product. The adsorption isotherms and reaction kinetics are again taken from Bergeot [2] and assumed to be constant for different particle diameters. Based on the aforementioned, the particle size only affects the dispersion, pressure drop, and the mass-transfer resistance which is described by the linear driving force (LDF) approximation. 6.2.1. Pressure drop The pressure drop in SMBR units is normally estimated by the sum of the pressure drop in each zone of the unit [3-5]. Ergun [6] developed an equation to calculate the pressure drop (∆𝑃) in fixed beds that covers laminar and turbulent flow conditions: ∆𝑃=150𝜇𝑢𝐿𝑐 𝑑𝑝2(1−𝜀 𝜀)2+1.75𝜌𝑢2𝐿𝑐 𝑑𝑝(1−𝜀 𝜀) (6.1) where 𝜇 and 𝜌 are the mixture viscosity and density flowing through the fixed bed with interstitial velocity 𝑢, 𝐿𝑐 is the length of a single column with bed porosity 𝜀 and particle diameter 𝑑𝑝. Simulated moving bed reactor: Optimal particle size 95 The properties of the mixture are estimated by the same methods (see Annex A) and assumed constant since they are very similar among the isomers and do not change significantly with pressure in liquid phase; consequently, the interstitial velocity is taken as constant in each zone. The bed porosity is also assumed constant since the particles are rigid and spherical and so deformation due to process conditions is not expected; moreover, the column to particle diameter ratio is large enough (i.e. >> 10) for the particles sizes studied [7]. For lower ratios the bed porosity becomes a function of said ratio, particularly below 10 when shortcutting may occur at the wall [8]. Additionally, the size of the adsorbent and catalyst are equal for every case in order to prevent the smaller particles of being dragged to the bottom by the liquid displacing the larger particles to the top leading to variable porosity throughout the column and, more importantly, losing of homogeneity in the adsorbent-catalyst mixture. It is important to highlight that the value from equation (6.1) must be multiplied by the number of columns to obtain the pressure drop of a certain zone. Moreover, since the expression was developed for fixed beds, the interstitial velocity shall be the one that corresponds to the actual SMBR unit although the model is solved using the TMBR approach. Both velocities are related through the velocity of solid phase based on the concept of relative velocity (𝑢SMBR=𝑢TMBR+𝑢𝑠). 6.3. Results and discussion The system is solved numerically using the commercial software gPROMS v3.7.1 from Process Systems Enterprise (www.psenterprise.com). The axial domain is discretized by the second-order orthogonal collocation method on 50 finite elements with 10-5 as tolerance. The values of extract and raffinate purity and adsorbent to adsorbent plus catalyst weight ratio are not modified (0.70; 0.95; and 0.9 respectively). 6.3.1. Separation regions and separation volumes For each particle diameter, the steady-state TMBR model is solved for several values of 2 and 3 to evaluate the influence of 1 and 4 for the configuration and switching time currently used in the SMB unit (6-9-6-3 and 69 s respectively) as described in Chapter 5. Figure 6.1 shows the influence of 4 and 1 in the separation regions for each particle size. For the four diameters Chapter 6 96 the transition region described by Azevedo and Rodrigues [9], where the separation region increases as flow rate in zone 1 increases and flow rate in zone 4 decreases, is clearly identified. Figure 6.1 Separation regions for 6-9-6-3 configuration and 69 s switching time with several values of 1 and fixed 4 (left) and several values of 4 and fixed 1 (right) for particle diameter: a) 0.5 mm (1=5.5 and 4=0.4) b) 0.7 mm (1=5.5 and 4=0.2) c) 0.8 mm (1=6.0 and4=0.4) d) 0.9 mm (1=6.5 and 4=0.4). Darker regions indicate the optimum values for 1 and 4. Simulated moving bed reactor: Optimal particle size 97 Moreover, from smaller to larger particles it can be seen a significant reduction in the separation volume due to the greater influence of the particle diameter in the mass-transfer resistance. For instance the intraparticle mass-transfer coefficient, calculated by 𝑘𝑝=15𝐷eff (𝑑𝑝2 ⁄)2 ⁄ to keep consistency with the LDF approximation, presents values of 0.320, 0.163, 0.125, and 0.099 s-1 for particle diameter 0.5, 0.7, 0.8, and 0.9 mm respectively. Table 6.1 presents the optimum point (i.e., peak of the darker separation region in Figure 6.1) for each particle size with the corresponding productivity, desorbent consumption, and deviation from the equilibrium. The optimum point is chosen at the boundary of the previously mentioned transition region (i.e., region fairly constant for higher values of 1 and/or lower values of 4); in other words, the point at which higher desorbent consumption does not lead to a significant increase in productivity is selected as optimum for each size (see Tables B.1 to B.4 in Annex B). The optimum point for diameters 0.5, 0.62 (see Table 5.2 in Chapter 5), 0.7, 0.8, and 0.9 mm corresponds to a value of 1 and 4 of 5.5 and 0.4, 5.5 and 0.4, 5.5 and 0.2, 6.0 and 0.4, and 6.5 and 0.4 respectively. As expected for larger particles higher flow rates of desorbent (i.e., higher 1 - 4) are needed to compensate the increase in the mass-transfer resistance. On the other hand, for smaller particles the amount of desorbent tends to a constant yet still high value, this is due to the reverse reaction rather than the mass-transfer. Table 6.1 Optimum points for 6-9-6-3 configuration and 69 s switching time for each particle diameter Size, mm 1 2 3 4 PR, kg m-3h-1 DC, m3 kg-1 DE 0.5 5.5 1.13 1.41 0.4 209.99 0.078 1.76 0.7 5.5 1.13 1.36 0.2 187.52 0.090 1.90 0.8 6.0 1.13 1.31 0.4 151.10 0.118 1.95 0.9 6.5 1.13 1.28 0.4 126.75 0.154 1.96 6.3.2. Arrangement of columns Normally, higher desorbent flow rates are needed to regenerate the solid and desorbent when section lengths of zones 1 and 4 are shorter [10], a trade-off analysis between the amount of desorbent and the number of columns determines the proper configuration of the unit [11]. However, as indicated in the previous chapter and verified in the last section, the higher desorbent flow rates are mainly due to the presence of the reverse reaction in each section of the SMBR unit. Based on the aforesaid, separation regions for different configurations with fixed flow rates in zone 1 and 4 and 69 s switching time corresponding to each diameter are estimated and presented in Figure 6.2; as usual, the best configurations provide larger separation regions. The vertices of separation regions for several configurations with the corresponding productivity, desorbent consumption, and deviation from the equilibrium for each particle diameter are presented in Table B.5 in Annex B. Chapter 6 98 Figure 6.2 Separation regions for different configurations and 69 s switching time for particle diameter: a) 0.5 mm (1=5.5 and 4=0.4) and b) 0.7 mm (1=5.5 and 4=0.2) c) 0.8 mm (1=6.0 and 4=0.4) d) 0.9 mm (1=6.5 and 4=0.4). Simulated moving bed reactor: Optimal particle size 99 As expected, fewer columns in zones 1 and 4 provide higher productivities. Regardless the particle diameter, configurations with larger zone 3 exhibit better performances since in this zone p-xylene is produced in the isomerization reaction due to its lower concentration. However, the difference between the regions of the analyzed columns arrangements is smaller for larger particles; as the particle size increases the mass-transfer resistance becomes more important reducing the influence of the length of each zone in the performance of the SMBR [12]. 6.3.3. Optimization without maximum pressure drop constraint The best six configurations for each particle size are included in an optimization procedure using the solver (CVP_SS) of the commercial software gPROMS. The flow rates for each zone and the switching times are optimized to maximize the productivity through a single-objective optimization procedure. Several values of desorbent consumption are used as constraints along with the purity in the extract and raffinate port (i.e., 0.70 and 0.95 respectively) as the procedure followed in Chapter 5. Table B.6 in Annex B presents the optimization results for each particle diameter. It can be seen that 2-6-14-2 is the best configuration for every size, as for 0.62 mm. As expected, for larger particles the feasible DC values for the unit increases due to the mass-transfer resistance. In addition, the switching time also increases with higher desorbent consumption constraints; as pointed out by Sá Gomes et al. [1], longer switching times allow to process more feed by increasing contact time with the disadvantage of higher mass-transfer resistance. The productivity as function of desorbent consumption for the four particle diameters studied can be seen in Figure 6.3. Similarly to 0.62 mm, a change in the profile can be pinpointed at a specific value of desorbent consumption: 0.05; 0.07; 0.09; and 0.10 m3 kg-1 for 0.5; 0.7; 0.8; and 0.9 mm respectively. It is recommended to operate around said values since the gain in productivity for desorbent consumption is the highest. Moreover, the difference between the configurations is shorter for larger particles as noted in Section 6.3.2; some overlapping is even observed for 0.7 mm and larger. Based on the aforementioned, it is recommended to use smaller particles when choosing the appropriate configuration for an SMBR unit. Based on the previous results, the configuration chosen for further studies considering the pressure drop within the unit is 2-6-14-2. Figure 6.4 presents the productivity, pressure drop, and desorbent consumption for the particle diameters studied including the currently size used (0.62 mm). On the absence of pressure drop constraints it can be seen that below 0.62 mm the increase in pressure drop outweighs that of productivity, while above 0.7 mm a faster increase in desorbent consumption is noticed. In case of a new unit, the proper particle size would be around 0.6 – 0.7 This chapter is based on: Gonçalves, J. C., and A. E. Rodrigues. 2015. “Simulated moving bed reactor for p-xylene production: Dual-bed column.” Submitted to Computers & Chemical Engineering Chapter 7: Simulated moving bed reactor: Dual-bed column A dual-bed Simulated Moving Bed Reactor comprising an adsorbent/catalyst homogeneous mixture bed followed by just adsorbents within the columns is developed under the framework of the proposed aromatics complex. A method comprising dynamic optimizations of a single column is followed to estimate the optimum proportion of adsorbents and catalyst within the first bed in such a way that the rate of production of p-xylene is equal to its rate of adsorption. Afterwards, the switching time and the first bed length are optimized through successive simulations using the simulated moving bed reactor approach. Finally, an integrated method that combines less time-consuming true moving bed reactor results with rigorous simulated moving bed reactor calculations to accurate develop a dual-bed Simulated Moving Bed Reactor unit is proposed. Chapter 7 108 7.1. Introduction In Chapter 5 and 6 it was used the True Moving Bed Reactor (TMBR) simplified model to determine the optimum arrangement of columns, flow rates, particle size, and switching time of the Simulated Moving Bed Reactor (SMBR); however, the results were not obtained using the SMBR actual model. Generally, True Moving Bed (TMB) models provide fast and precise results of the real Simulated Moving Bed (SMB) unit, especially when a large number of columns are involved [1-3]. However, small deviations may cause large errors with very steep profiles at the outlet ports [4]; moreover, large oscillations in the products of SMB, due to high mass-transfer coefficients, are not properly reproduced by the TMB approach [5,6]. In the SMBR, the presence of catalysts throughout the columns increases said oscillations preventing the TMBR model to match the results of the actual SMBR unit. The aforesaid can be overcome by using a different distribution of adsorbents and catalysts that reduces the oscillations in the outlet ports and at the same time provides better interaction between both solids enhancing the performance of the unit. Multilayer configurations alternating catalysts and adsorbents have been used in steam reforming of methane [7] and ethanol [8]; even layers of different adsorbents have been used in pressure swing adsorption units [9]. In the same spirit, an SMBR comprising dual-bed columns as shown in Figure 7.1 is proposed. The purpose of this chapter is then to study this innovative dual-bed SMBR for p-xylene production taking into account previous results obtained using the TMBR approach. Evidently, 24 adsorbent beds with length and diameter of 1.14 and 4.12 m are taken from the existing SMB facility; the unit operates at 200 ºC with a particle size of 0.62 mm as calculated in the previous chapter. 7.2. Mathematical model 7.2.1. Dual-bed column system The system is modeled by means of the actual SMBR instead of a continuous TMBR as in the previous chapters. Since there are two different beds in each column, the counter-current movement of the solid phase does not provide constant product yields in the outlet ports at steady-state. The system must be simulated by the actual shifting of the inlet and outlet ports along the unit; nevertheless, the assumptions followed are those used in Chapter 5: isothermal operation, axial dispersed plug flow, constant flow rate in each zone, mass transfer described by linear driving force approximation, and similar physical characteristics for adsorbents and catalysts. Simulated moving bed reactor: Dual-bed column 109 Figure 7.1 Distribution of adsorbents and catalysts within the columns of the simulated moving bed reactor. L1 represents the length of the first bed with homogeneous mixture of adsorbent and catalyst; L2 corresponds to the second bed with just adsorbents Mass balances in the first bed (i.e., 0 < z < L1) for species 𝑖 in the bulk phase, adsorbent particle phase, and catalyst particle phase are given by equations (7.1) to (7.3) respectively: 𝜕𝐶𝑖 𝜕𝑡=𝐷𝑎𝑥𝜕2𝐶𝑖 𝜕𝑧2−𝑢𝜕𝐶𝑖 𝜕𝑧−1−𝜀 𝜀3 𝑅𝑝𝐾𝑙[𝜑(𝐶𝑖−𝐶𝑝𝑖     𝑎𝑑𝑠)+(1−𝜑)(𝐶𝑖−𝐶𝑝𝑖     𝑐𝑎𝑡)] (7.1) 𝜀𝑝𝜕𝐶𝑝𝑖     𝑎𝑑𝑠 𝜕𝑡 +(1−𝜀𝑝)𝜌𝑎𝑑𝑠𝜕𝑞𝑖  𝜕𝑡=3 𝑅𝑝𝐾𝑙(𝐶𝑖−𝐶𝑝𝑖     𝑎𝑑𝑠) (7.2) 𝜀𝑝𝜕𝐶𝑝𝑖     𝑐𝑎𝑡 𝜕𝑡 =3 𝑅𝑝𝐾𝑙(𝐶𝑖−𝐶𝑝𝑖     𝑐𝑎𝑡)+𝜌𝑐𝑎𝑡𝑅𝑖 (7.3) while in the second bed (i.e., L1 < z < Lc) the catalyst is not present, which leads to mass balances for species 𝑖 in just the bulk phase and adsorbent particle phase: 𝜕𝐶𝑖 𝜕𝑡=𝐷𝑎𝑥𝜕2𝐶𝑖 𝜕𝑧2−𝑢𝜕𝐶𝑖 𝜕𝑧−1−𝜀 𝜀3 𝑅𝑝𝐾𝑙(𝐶𝑖−𝐶𝑝𝑖     𝑎𝑑𝑠) (7.4) 𝜀𝑝𝜕𝐶𝑝𝑖     𝑎𝑑𝑠 𝜕𝑡 +(1−𝜀𝑝)𝜌𝑎𝑑𝑠𝜕𝑞𝑖  𝜕𝑡=3 𝑅𝑝𝐾𝑙(𝐶𝑖−𝐶𝑝𝑖     𝑎𝑑𝑠) (7.5) where 𝐶𝑖 is the concentration of each species in the bulk phase, 𝐷𝑎𝑥 is the axial dispersion coefficient, and 𝑢 is the interstitial velocity. Physical properties such as bed porosity 𝜀, particle porosity 𝜀𝑝, particle radius 𝑅𝑝, density of adsorbent 𝜌𝑎𝑑𝑠 and catalyst 𝜌𝑐𝑎𝑡 are given in Chapter Chapter 7 110 5. The average mass adsorbed 𝑞𝑖  and the reaction rate 𝑅𝑖 are obtained from Bergeot [10] and used as function of the average particle concentration in the adsorbents (𝐶𝑝𝑖     𝑎𝑑𝑠) and catalysts (𝐶𝑝𝑖     𝑐𝑎𝑡) respectively as in the previous chapters. The adsorbent to adsorbent plus catalyst weight ratio (𝜑=𝑚𝑎𝑑𝑠 𝑚𝑎𝑑𝑠+𝑚𝑐𝑎𝑡 ⁄) is calculated in the next section. The properties are calculated in the same manner and presented in Annex A; similarly, the internal mass-transfer coefficient is calculated using molecular diffusivity estimated by the Wilke-Chang method modified to include the mixed solvent case by Perkins and Geankoplis [11], and the external mass-transfer coefficient is estimated by the Wilson and Geankoplis correlation [12]. Danckwerts boundary conditions for species 𝑖 are used for the bulk phase in equation (7.6) at the inlet and outlet of the column, while equation (7.7) guarantees continuity between the two beds providing the other boundary conditions in the bulk phase: 𝑧=0 𝐷𝑎𝑥𝜕𝐶𝑖 𝜕𝑧=𝑢(𝐶𝑖−𝐶𝑖𝑖𝑛); 𝑧=𝐿𝑐 𝜕𝐶𝑖 𝜕𝑧=0 (7.6) 𝐶𝑖|𝐿1−=𝐶𝑖|𝐿1+; 𝜕𝐶𝑖 𝜕𝑧|𝐿1−=𝜕𝐶𝑖 𝜕𝑧|𝐿1 + (7.7) where the inlet concentration (𝐶𝑖𝑖𝑛) depends on the specific zone within the unit and is determined by mass balances in each inlet and outlet port as in the previous studies (see Chapter 5). Furthermore, the performance parameters to assess the SMBR unit – productivity (PR), desorbent consumption (DC), and deviation from the equilibrium (DE) – are defined in the same manner, considering p-xylene in the extract point as the desired product. However, since the system is intrinsically dynamic, an average of the concentrations is calculated over the last cycle when the system has reached the cyclic steady-state: Extract Purity: PurX= ∫𝐶PX,𝑋 𝑡+𝑁𝑐𝑡𝑠 𝑡𝑑𝑡 ∫(𝐶PX,𝑋+𝐶MX,𝑋+𝐶OX,𝑋+𝐶EB,𝑋)𝑑𝑡 𝑡+𝑁𝑐𝑡𝑠 𝑡 (7.8) Raffinate Purity: PurR= ∫(𝐶MX,𝑋+𝐶OX,𝑋+𝐶EB,𝑋)𝑑𝑡 𝑡+𝑁𝑐𝑡𝑠 𝑡 ∫(𝐶PX,𝑋+𝐶MX,𝑋+𝐶OX,𝑋+𝐶EB,𝑋)𝑑𝑡 𝑡+𝑁𝑐𝑡𝑠 𝑡 (7.9) Deviation from the Equilibrium: DE=𝑄𝑋∫𝐶PX,𝑋𝑑𝑡 𝑡+𝑁𝑐𝑡𝑠 𝑡+𝑄𝑅∫𝐶PX,𝑅𝑑𝑡 𝑡+𝑁𝑐𝑡𝑠 𝑡 𝑁𝑐𝑡𝑠𝐶PX,𝑒𝑞𝑄𝐹 (7.10) Desorbent Consumption: DC=𝑄𝐷𝑁𝑐𝑡𝑠 𝑄𝑋∫𝐶PX,𝑋𝑑𝑡 𝑡+𝑁𝑐𝑡𝑠 𝑡 (7.11) Productivity: PR=𝑄𝑋∫𝐶PX,𝑋𝑑𝑡 𝑡+𝑁𝑐𝑡𝑠 𝑡 𝑉𝑎𝑑𝑠+𝑐𝑎𝑡𝑁𝑐𝑡𝑠 (7.12) where a cycle length is given by the number of columns (Nc) times the switching time (ts). The SMBR unit is assumed to operate under cyclic steady state when the average concentrations of each species do not differ from those in the preceding cycle for more than 1 %. Simulated moving bed reactor: Dual-bed column 111 7.2.2. Equivalence between TMBR and SMBR As discussed in Chapter 5 and 6, in TMBR the flow rates are expressed as ratios of interstitial velocity to solid velocity (𝑢TMBR 𝑢𝑠 ⁄) where column length divided by the switching times gives the velocity of solid (𝑢𝑠=𝐿𝑐𝑡𝑠 ⁄). Moreover, in TMBR the solid moves counter-currently while it is actually fixed in SMBR; hence, both models are related through the solid velocity: 𝑢SMBR= 𝑢TMBR+𝑢𝑠. The equivalent SMBR flow rate is then calculated using the corresponding interstitial velocity. 7.2.3. Optimum adsorbent to adsorbent plus catalyst weight ratio The adsorbent to adsorbent plus catalyst weight ratio to be used in the model is determined by minimizing the following objective function: min 𝜑{∫(𝑅𝑎𝑑𝑠,PX−𝑅𝑐𝑎𝑡,PX)2𝑑𝑡 𝑡𝑠 0} (7.13) where 𝑅𝑎𝑑𝑠,PX and 𝑅𝑐𝑎𝑡,PX represent the average rate of adsorption and production of p-xylene as shown in equations (7.14) and (7.15). In other words, the optimal ratio is that where the amount of p-xylene entering into the adsorbents equals that leaving the catalysts: 𝑅𝑎𝑑𝑠,PX=𝜑∫ (𝐶PX−𝐶𝑝,PX        𝑎𝑑𝑠)𝑑𝑧 𝐿1 𝑜 (7.14) 𝑅𝑐𝑎𝑡,PX=−(1−𝜑)∫ (𝐶PX−𝐶𝑝,PX        𝑐𝑎𝑡)𝑑𝑧 𝐿1 0 (7.15) The optimizations are conducted in a single column using equations (7.1) to (7.3). Due to the dynamic behavior of the system an average over time must be used, in this case the switching time of the TMBR from Chapter 6 is used (i.e., 57 s). The flow and concentrations are also taken from said TMBR; the values of zone 3 are used since this zone favors the production of p-xylene. The inlet flow and concentrations are those entering to one of the columns in zone 3 while the initial concentration corresponds to the concentration of the next column which is what actually occurs within the columns of the simulated moving bed reactor; the process is repeated for each column in zone 3. Unfortunately up to this point L1 is not known, the procedure must then be repeated for several values provided L1 < Lc. Figure 7.2 presents the diagram to determine the optimum adsorbent to adsorbent plus catalyst weight ratio. Chapter 7 112 Figure 7.2 Flow diagram to determine the optimum adsorbent to adsorbent plus catalyst weight ratio (𝜑) 7.3. Results and discussion All simulations are conducted with the commercial software gPROMS v.3.7.1 from Process Systems Enterprise (www.psenterprise.com). The numerical method involves the discretization of the axial domain of each column by the second-order orthogonal collocation method on 10 and 20 finite elements in the first and second bed respectively with 10-5 as tolerance. The extract and raffinate purity are set to 0.70 and 0.95 throughout the study as in the previous simulations. Simulated moving bed reactor: Dual-bed column 113 7.3.1. Optimum adsorbent to adsorbent plus catalyst weight ratio The dynamic optimizations are carried out using the solver (CVP_SS) of the commercial software gPROMS. The concentrations of each component in zone 3 to be used in the estimation of the optimum adsorbent to adsorbent plus catalyst weight ratio are presented in Table 7.1. The SMBR equivalent flow rate of zone 3 of the TMBR, which corresponds to configuration 2-6-14-2 with optimized flow and switching time (57 s) from Chapter 6, is used as inlet flow for each column (i.e., 747 m3 h-1). Table 7.1 Mass concentration (kg m-3) at the inlet of each column in zone 3 for configuration 2-6-14-2 with optimized flow conditions from Chapter 6 Column p-Xylene m-Xylene o-Xylene Ethylbenzene Toluene 1 157.44 346.95 86.67 97.08 9.28 2 146.59 334.06 86.48 96.10 33.95 3 141.65 321.38 86.17 95.05 52.84 4 137.83 311.98 86.00 94.30 66.94 5 134.40 303.86 85.86 93.67 79.23 6 131.15 296.41 85.73 93.10 90.63 7 127.66 288.65 85.57 92.51 102.63 8 124.22 281.21 85.37 91.96 114.29 9 120.37 273.14 85.09 91.36 127.12 10 115.43 263.12 84.68 90.63 143.29 11 109.47 251.50 84.13 89.80 162.37 12 101.11 235.92 83.32 88.72 188.40 13 87.94 212.79 82.08 87.21 227.90 14 61.18 168.99 79.34 83.87 305.83 15a 5.98 37.76 24.57 23.85 594.62 a To be used only for initial concentration of column 14 Table 7.2 presents the optimized 𝜑 for each column and the average value at several lengths values of L1. Due to the feed concentration to the unit, there is not an optimum value that balances adsorption and reaction rates in the first column; thus, the value of 𝜑 is zero. A similar phenomena is observed for the last column at high L1. The total average corresponds to a ratio of 0.35; however, the first bed is not expected to occupy a significant portion of the column bed since the reverse reaction may prevent the system from satisfying the purity requirements. The first bed fraction can be quickly estimated using the ratios of TMBR (𝜑TMBR) and SMBR (𝜑SMBR) by keeping the same amount of catalyst within the bed column: 𝐿1=100(1−𝜑TMBR)(1−𝜑SMBR) ⁄. The length of the first bed, using 0.9 and 0.35 for the ratios of TMBR and SMBR respectively, is about 15% of the length of the column; based on this value it would be more reasonable to use the average of just 10 and 20 %Lc giving a ratio of 0.4 corresponding to L1 of 17 %Lc (rounded up to 20 %Lc). Chapter 7 114 Table 7.2 Optimum adsorbent to adsorbent plus catalyst weight ratio (𝜑) in each column for several lengths of the first bed (L1) as percentage of column length (Lc) Column 10 %Lc 20 %Lc 30 %Lc 40 %Lc 50 %Lc 60 %Lc 70 %Lc 80 %Lc 90 %Lc 1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 2 0.58 0.57 0.55 0.50 0.47 0.44 0.42 0.41 0.41 3 0.59 0.58 0.55 0.51 0.48 0.46 0.45 0.45 0.45 4 0.58 0.57 0.53 0.50 0.47 0.45 0.44 0.44 0.44 5 0.55 0.53 0.50 0.47 0.44 0.42 0.42 0.42 0.43 6 0.49 0.46 0.41 0.37 0.34 0.34 0.35 0.36 0.37 7 0.49 0.47 0.43 0.39 0.37 0.37 0.39 0.40 0.41 8 0.49 0.46 0.43 0.39 0.38 0.38 0.39 0.41 0.42 9 0.41 0.38 0.33 0.30 0.30 0.32 0.34 0.36 0.38 10 0.42 0.40 0.36 0.34 0.34 0.36 0.38 0.40 0.42 11 0.39 0.37 0.33 0.32 0.33 0.35 0.38 0.40 0.43 12 0.35 0.31 0.28 0.29 0.31 0.34 0.37 0.40 0.43 13 0.22 0.18 0.17 0.20 0.24 0.28 0.32 0.36 0.39 14 0.22 0.04 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Average 0.41 0.38 0.35 0.33 0.32 0.32 0.33 0.34 0.36 7.3.2. Configurations from TMBR to SMBR Once the adsorbent to adsorbent plus catalyst weight ratio and the length of the first bed have been defined (i.e., 𝜑 = 0.4 and L1 = 20 %Lc), the SMBR unit can be simulated based on the results using the TMBR approach from the previous chapter (see Table 6.3). Six configurations are taken with the corresponding flow rates in zones 1 and 4 and switching time from the optimization conducted using the TMBR approach. Table 7.3 presents the performance of said configurations at the peak of their separation regions which are depicted in Figure 7.3. Table 7.3 Performance of the best configurations based on TMBR approach from Table 6.3 calculated using the SMBR approach. Flow rates in zones 1 and 4 (Q1 and Q4) and switching time (ts) are those equivalent to TMBR while flow rates in zones 2 and 3 (Q2 and Q3) corresponds to the peak within the separation region. Config Q1, m3 h-1 Q2, m3 h-1 Q3, m3 h-1 Q4, m3 h-1 PR, kg m-3h-1 DC, m3 kg-1 ts, s 2-4-16-2 1640 641 682 380 136.53 0.12 59 2-5-15-2 1650 653 694 370 133.80 0.12 58 2-6-14-2 1650 665 707 370 133.58 0.12 57 3-4-15-2 1540 610 649 350 129.28 0.12 62 3-5-14-2 1550 621 661 350 128.93 0.12 61 3-4-14-3 1570 632 668 380 118.72 0.13 60 In the previous chapter, 2-6-14-2 was the optimum configuration as opposed to the results presented in Table 7.3. Nevertheless, the higher switching time of 2-4-16-2 shifts the separation Simulated moving bed reactor: Dual-bed column 115 region (see Figure 7.3) to the left resulting in lower flow rate in zone 2 which in turns increases the flow rate in the extract leading to a higher productivity despite of the lower peak. Based on the aforementioned, the switching time must be adjusted. Figure 7.3 Separation regions of the best configurations based on TMBR approach from Table 6.3 calculated using the SMBR approach. Flow rates in zones 1 and 4 (Q1 and Q4) and switching time (ts) are those equivalent to TMBR 7.3.3. Optimization of switching time As stated before, in TMBR the flow rates are expressed as interstitial velocity to solid velocity ratio where the solid velocity depends on the switching time. The switching times presented in Table 7.3 are the result of simultaneous optimizations of said ratios and switching time; in other words, they affected the flow rates in each zone of the unit. In this case, several switching times are studied in the best three configurations maintaining the flow rates in zone 1 and 4 constant. Table 7.4 presents the performance at the peak of the separation region for switching times from 55 to 75 s. Configuration 2-6-14-2 exhibits the best performance at 70 s switching time as it can be seen in Figure 7.4, which is higher than the 57 s switching time from the TMBR studies. As indicated by Sá Gomes et al. [1] and verified for this type of system in prior chapters, longer switching times result in higher feed flow rate due to the increased contact time up to a certain point where mass-transfer resistance limits the capacity of the unit causing a reduction in productivity. The flow rates in zones 2 and 3 for these units are slightly higher than those obtained with TMBR at the same switching time; higher rates pushes the optimum switching times to higher values in order to balance the contact time and mass-transfer resistance as previously indicated. Based on the aforementioned, it is expected to obtain longer switching times in SMBR compared to those obtained using the TMBR approach. Chapter 7 122 SMB = Simulated Moving Bed SMBR = Simulated Moving Bed Reactor TMB = True Moving Bed TMBR = True Moving Bed Reactor X = Extract Superscripts and subscripts 𝑖𝑛 = Inlet 7.6. References [1] Sá Gomes, P., N. Lamia, and A. E. Rodrigues. 2009. "Design of a gas phase simulated moving bed for propane/propylene separation." Chemical Engineering Science no. 64 (6):1336-1357. [2] Pais, L. S., J. M. Loureiro, and A. E. Rodrigues. 1998. "Modeling Strategies for Enantiomers Separation by SMB Chromatography." AIChE Journal no. 44 (3):561-569. [3] Ruthven, D. M., and C. B. Ching. 1989. "Counter-current and simulated counter-current adsorption separation processes." Chemical Engineering Science no. 44 (5):1011-1038. [4] Beste, Y. A., M. Lisso, G. Wozny, and W. Arlt. 2000. "Optimization of simulated moving bed plants with low efficient stationary phases: Separation of fructose and glucose." Journal of Chromatography A no. 868 (2):169-188. [5] Grosfils, V., C. Levrie, M. Kinnaert, and A. Vande Wouwer. 2007. "On simplified modelling approaches to SMB processes." Computers & Chemical Engineering no. 31 (3):196-205. [6] Zhong, G., and G. Guiochon. 1997. "Simulated moving bed chromatography. Effects of axial dispersion and mass transfer under linear conditions." Chemical Engineering Science no. 52 (18):3117-3132. [7] Oliveira, E. L. G., C. A. Grande, and A. E. Rodrigues. 2011. "Effect of catalyst activity in SMR-SERP for hydrogen production: Commercial vs. large-pore catalyst." Chemical Engineering Science no. 66 (3):342-354. [8] Cunha, A. F., Y. J. Wu, F. A. Díaz Alvarado, J. C. Santos, P. D. Vaidya, and A. E. Rodrigues. 2012. "Steam reforming of ethanol on a Ni/Al 2O 3 catalyst coupled with a hydrotalcite-like sorbent in a multilayer pattern for co 2 uptake." Canadian Journal of Chemical Engineering no. 90 (6):1514-1526. [9] Bárcia, P. S., J. A. C. Silva, and A. E. Rodrigues. 2010. "Octane Upgrading of C5/C6 Light Naphtha by Layered Pressure Swing Adsorption." Energy & Fuels no. 24 (9):5116-5130. Simulated moving bed reactor: Dual-bed column 123 [10] Bergeot, G. 2010. Extension du concept "One-column" au lit mobile simulé réactif. Application à la séparation réactive des C8 aromatiques, École doctorale Sciences et Ingénierie des Ressources, Procédés, Produits, Environnement, Institut National Polytechnique de Lorraine, France. [11] Poling, B. E., J. M. Prausnitz, and J. J. P. O'Connell. 2001. The Properties of Gases and Liquids. 5th ed. New York: The McGraw-Hill Companies. [12] Ruthven, D. M. 1984. Principles of Adsorption and Adsorption Processes. First ed. New York: John Wiley & Sons, cop. [13] Alario, F., and J. Rault. 2002. Boost you Xylene Loop Performance with OPARIS. edited by Axens: Axens. [14] Haag, J., A. Vande Wouwer, S. Lehoucq, and P. Saucez. 2001. "Modeling and simulation of a SMB chromatographic process designed for enantioseparation." Control Engineering Practice no. 9 (8):921-928. Chapter 8: Proposed aromatics complex The proposed aromatics complex is analyzed quantitatively in this chapter. Two cases are studied with different flow rates of reformate fed to the complex to obtain the increase in the production of p-xylene and benzene. The mass balance is calculated considering complete separation within the distillation columns and more rigorous models for the isomerization and simulated moving bed reactor unit developed in previous chapters. The performance of the selective toluene disproportionation unit is estimated based on conversion and selectivity reported in the literature while solid-liquid equilibria is used to determine the theoretical recovery of the crystallization unit. Chapter 8 126 8.1. Introduction The main objective of this thesis is the development of a simulated moving bed reactor (SMBR) unit for the production of p-xylene in the framework of a modified aromatics complex leading to an increase in the production of p-xylene and benzene. The proposed aromatics complex is described in Chapter 5 and is analyzed in this Chapter. Models and results obtained in previous chapters along with simplified models for the rest of the units are used with the purpose of calculating the mass balance of the complex and determining the increase in the production of p-xylene and benzene. 8.2. Mathematical modeling The most employed separation unit within the complex is the distillation column, throughout the calculations the outlet streams of said units are estimating assuming complete separation of the corresponding key components. The SMBR unit is taken from Chapter 7 with a first bed length of 15% and adsorbent to adsorbent plus catalyst weight ratio of 0.4 and the isomerization unit corresponds to that of Chapter 4 assuming 5 and 25 wt% of hydrogen and naphthenes respectively. The models for the selective toluene disproportionation (STDP) and the crystallization unit are described in detail further on. Moreover, the non-aromatic components and heavy aromatics (i.e., C9+) are not taken into account. 8.2.1. Selective toluene disproportionation This unit is intended to convert toluene in more valuable p-xylene and benzene through disproportionation of two molecules of toluene. Normally, the reaction is carried out in the presence of hydrogen to extend the catalyst life by suppressing cracking; therefore, benzene is also produced by the dealkylation of toluene [1,2]. The parameters used for this unit are presented in Table 8.1: Table 8.1 Selective toluene disproportionation unit parameters Toluene conversion, % 25a Benzene / Xylenes, molar 1.5a Xylene distribution, % p-Xylene 89.8b m-Xylene 9.4b o-Xylene 0.8b a from Beck et al. [3]. b from Ji et al. [4] Proposed aromatics complex 127 8.2.2. Single stage crystallization Crystallization processes are based on solid-liquid equilibria [5]; as the temperature is reduced the solute solubility decreases resulting in saturation and, subsequently, precipitation of pure component crystals for eutectic mixtures [6,7]. The solubility of xylenes is calculated by equation (8.1) from Hildebrand et al. [8] assuming ideal behavior, i.e., the influence of activity coefficients is neglected [9,10]: ln𝑥𝑖=∆𝐻𝑖𝑓𝑢𝑠𝑖𝑜𝑛 𝑅(1 𝑇𝑚,𝑖−1𝑇)+∆𝐶𝑝,𝑖 𝑅(ln 𝑇 𝑇𝑚,𝑖+𝑇𝑚,𝑖 𝑇−1) (8.1) where xi is the molar fraction of component i in liquid phase at temperature T, and ∆𝐻𝑖𝑓𝑢𝑠𝑖𝑜𝑛 and ∆𝐶𝑝,𝑖 are the heat of fusion and the difference of the liquid and solid capacities for pure component i at its melting point Tm,i; the values are reported in Section A.4 in Annex A. The procedure consists of reducing the temperature until a second component reaches saturation (i.e., eutectic point), which is calculated through equation (8.2) based on a simple mass balance: 𝑥𝑖,𝑇−1=𝑥𝑖,𝑇1−𝑥PX,𝑇−1 1−𝑥PX,𝑇 (8.2) where 𝑥𝑖,𝑇 is the molar fraction of component i in the mother liquor at temperature T and 𝑥𝑖,𝑇−1 is its molar fraction at a lower temperature; 𝑥PX is obtained from equation (8.1) at the corresponding temperature. The eutectic point corresponds to the temperature when the molar fraction from equation (8.2) is higher than that calculated with equation (8.1). Normally, p-xylene is the first component to precipitate and the eutectic point is determined by m-xylene [9,11]. 8.3. Results and discussion The block diagram of the proposed aromatics complex is presented in Figure 8.1 along with the mass balance for two cases. Case I corresponds to the current feed of reformate (stream 1) to the aromatics complex based on the 90% of the nominal capacity by Galp Energia [12], the double is used in Case II. The reformate is sent to the fractionation column where benzene, toluene, and non-aromatics are separated through the top and sent to the aromatics extraction unit where non-aromatics are separated and benzene and toluene (stream 2) is mixed with the outlet of the STDP unit and sent to the benzene column to obtain benzene as final product (stream 11). The bottom of the fractionation is mixed with the recycle from the isomerization unit and fed to the xylene splitter. Composition of xylenes in stream 1 is estimated assuming equilibrium in gas phase at 525 ºC Chapter 8 128 Figure 8.1 Block diagram and mass balance of the proposed aromatics complex (operating condition in naphtha reforming) using the expressions developed by Chirico and Steele [13]. A portion of o-xylene is separated in the splitter and sent to the o-xylene column to obtain the final product (stream 13). The top of the splitter is sent to the SMBR unit where a high p-xylene stream with desorbent toluene (stream 6) is obtained and mixed with also a high p-xylene mixture of xylene with non-reacted toluene from the STDP unit. Toluene is separated in the toluene column and recycled back to be used as desorbent and the rest (stream 3) is sent to disproportionation to be converted into benzene and xylenes. Heavy aromatics formed in the reactions are withdrawn in the xylene column and the high p-xylene stream (stream 7) is sent to Fractionation Aromatics Extraction Simulated Moving Bed Reactor Gas phase Isomerization Xylene Splitter Reformate Xylenes o-Xylene Selective Toluene Disproportionation Single Stage Crystallization Benzene p-Xylene Benzene Column Toluene Column Xylene Column Toluene Recovery o-Xylene Column C6/C7 5 1 2 3 7 12 9 8 10 13 4 11 6 Heavy Aromatics Stream 1 2 3 4 5 6 7 8 9 10 11 12 13 Flow, ton h-1 32.00 20.60 64.60 23.95 64.60 567.03 24.53 6.84 6.81 13.65 13.18 17.72 1.10 Ben,wt% 15.31 23.79 0.00 0.00 12.81 0.00 0.00 0.00 0.00 0.00 100.00 0.00 0.00 Tol, wt% 49.06 76.21 100.00 0.00 75.58 97.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 PX, wt% 7.81 0.00 0.00 23.53 10.43 2.10 76.04 2.83 13.74 22.98 0.00 100.00 0.00 MX, wt% 16.52 0.00 0.00 50.16 1.09 0.88 23.24 14.06 83.66 49.28 0.00 0.00 0.00 OX, wt% 7.49 0.00 0.00 16.60 0.09 0.01 0.54 50.36 1.95 19.63 0.00 0.00 100.00 EB, wt% 3.81 0.00 0.00 9.71 0.00 0.01 0.18 32.75 0.65 8.11 0.00 0.00 0.00 Flow, ton h-1 64.00 41.20 131.47 58.16 131.47 757.95 47.99 24.72 12.84 37.56 26.65 35.15 2.20 Ben,wt% 15.31 23.79 0.00 0.00 12.81 0.00 0.00 0.00 0.00 0.00 100.00 0.00 0.00 Tol, wt% 49.06 76.21 100.00 0.00 75.58 95.68 0.00 0.00 0.00 0.00 0.00 0.00 0.00 PX, wt% 7.81 0.00 0.00 23.43 10.43 3.07 76.93 5.12 13.74 22.98 0.00 100.00 0.00 MX, wt% 16.52 0.00 0.00 50.01 1.09 1.09 20.19 39.34 75.47 49.28 0.00 0.00 0.00 OX, wt% 7.49 0.00 0.00 17.14 0.09 0.06 1.23 37.01 4.61 19.63 0.00 0.00 100.00 EB, wt% 3.81 0.00 0.00 9.42 0.00 0.10 1.65 18.53 6.18 8.11 0.00 0.00 0.00 Case I Case II Proposed aromatics complex 129 the single stage crystallization unit where p-xylene is obtained as final product (stream 12). Toluene in the low p-xylene content stream from the SMBR is separated in the toluene recovery column, the xylene mixture is then sent to the gas phase isomerization unit (stream 8) along with the outlet from the crystallization (stream 9) where thermodynamic equilibrium is re-established by producing p-xylene out of oand m-xylene and ethylbenzene and recycled back (stream 10) to the xylene splitter. The gas phase isomerization unit includes a dedicated fractionation tower to separate the naphthenes required for the isomerization of ethylbenzene as presented in Section 2.6.4 in order to prevent the unnecessarily circulation of said species in the SMBR and crystallization units. 8.3.1. Case I The production of benzene corresponds to 117.5 thousand mtpy (based on 335 days of operation per year) which represents an improvement of 170% compared to the current production. In the case of p-xylene, 158 thousand mtpy is obtained representing 72% more p-xylene within the proposed aromatics complex. The production of o-xylene is not modified. The p-xylene concentration fed to the crystallization unit is above the minimum set in Chapter 5 (75%). The temperature at which m-xylene starts to drop out of the mother liquor is 219 K, the operating temperature is therefore fixed at 220 K. It can be seen in Table 8.2 that m-xylene fraction in the mother liquor exceeds its solubility at 219 K, therefore the operation must be carried out above this point. o-Xylene has the highest melting point after p-xylene, however it is not the second species to crystallize since its molar fraction is significantly lower than that of m-xylene. Moreover, ethylbenzene presents unrealistic values of solubility at these temperatures because its melting point is very distant. Even though the temperature is higher than the colder stage of the conventional crystallization (213 K), is not high enough to use only one refrigeration system. Practical limitations of propylene refrigeration systems limit the temperature to 244 K [14]; a two stage crystallization unit might be more energetically efficient. Nevertheless, the recovery in the crystallization unit is 95%. The increment in the production of p-xylene in the SMBR compared to the current separation unit is only 15% while the rest is provided by selective toluene disproportionation. This is due to a considerably reduction in the outlet stream from the isomerization unit leading to lower feed to the SMBR unit. It has been seen that for the SMBR the extract flow rate is significantly higher than the raffinate due to the continuous production of p-xylene, this results in lesser xylenes sent to the isomerization unit. Minceva and Rodrigues [15] used a feed of 87 m3 h-1 in their optimization studies of the existing separation unit; in this case the feed is reduced to about 35 m3 h-1. Furthermore, since the feed is considerably lower less amount of desorbent can be used; Chapter 8 130 the equivalent true moving bed approach model is used to calculate the corrected flow rates in zones 1 and 4 of the SMBR unit maintaining the same switching time, Table 8.3 presents the parameters and performances of the SMBR unit for each case. Table 8.2 Solubility and actual molar fractions of the mother liquor at several temperatures for cases I and II T, K 222 221 220 219 218 217 216 Solubility from equation (8.1) OX 0.474 0.460 0.446 0.432 0.418 0.405 0.392 MX 0.913 0.888 0.863 0.839 0.816 0.793 0.771 PX 0.148 0.143 0.138 0.133 0.128 0.124 0.119 EB 3.998 3.884 3.773 3.665 3.559 3.456 3.356 Case I Molar fraction in the mother liquor from equation (8.2) OX 0.019 0.019 0.019 0.020 0.020 0.020 0.020 MX 0.826 0.831 0.836 0.841 0.846 0.850 0.854 PX 0.148 0.143 0.138 0.133 0.128 0.124 0.119 EB 0.006 0.006 0.006 0.006 0.007 0.007 0.007 Case II Molar fraction in the mother liquor from equation (8.2) OX 0.046 0.046 0.046 0.046 0.047 0.047 0.047 MX 0.746 0.750 0.754 0.759 0.763 0.767 0.771 PX 0.148 0.143 0.138 0.133 0.128 0.124 0.119 EB 0.061 0.061 0.062 0.062 0.062 0.063 0.063 The aforementioned indicates that the SMBR and isomerization units are now oversized for their corresponding services, the reformate supplied to the aromatics complex must be increased in order to provide more efficient use of the existing said units assuming appropriate upstream equipment to handle the increased rates. Table 8.3 Configuration of the SMBR unit for cases I and II with 70 s switching time Case Q1, m3 h-1 Q2, m3 h-1 Q3, m3 h-1 Q4, m3 h-1 PR, kg m-3h-1 DC, m3 kg-1 DE L1, %Lc I 1370 529 564 480 150.68 0.07 2.32 15 II 1650 526 610 370 293.35 0.06 1.94 10 8.3.2. Case II The production now corresponds to 238 thousand mtpy which represents an improvement of 447% in benzene and 314 thousand mtpy representing 241% in p-xylene using the double of the current feed within the aromatics complex. The production of o-xylene in this case is also doubled. Even though 300 and 700 thousand mtpy of benzene and p-xylene are not achieved, the increase in production of said products is still very remarkable. Proposed aromatics complex 131 The ratio of benzene to xylene from STDP can be modified through the catalyst employed. Low Si/Al ratio increases dealkylation of toluene leading to more production of benzene; if an equimolar mixture is desired high Si/Al ratio should be used [1]. Even more p-xylene can be produced by alkylation of toluene with methanol where also a high p-xylene content stream can be sent to a crystallization unit [16]. In case II the eutectic point in the crystallization unit is 216 K due to the slightly higher concentration of p-xylene with consequently lower amount of m-xylene which is the second compound to crystallize (see Table 8.2). However, the operating temperature is maintained at 220 K corresponding to a recovery of 95% since no significant improvement is observed between the two temperatures. The feed to the SMBR is higher than that in case I as expected, around 84 m3 h-1 are sent to the unit due to the doubled rate from the fractionation bottom and the increased recycle from the isomerization unit. Moreover, the unit configuration used in case I shall be modified to handle the larger feed; a first bed of 10% is now used as it was discussed in Section 7.3.4 in Chapter 7. It can be seen in Table 8.3 that the productivity in case II is considerably higher due to the larger feed involved; however, case I SMBR seems to be more efficient regarding the process intensification since it possesses higher deviation from the equilibrium, in fact, case II SMBR doubles the production of p-xylene with a feed higher by 2.4 times. 8.4. Conclusions The proposed aromatics complex is analyzed and mass balances for the currently feed to the system and a hypothetically two fold feed are presented. In both cases the production of benzene and p-xylene is significantly enhanced, a 170% and 72% respectively for the first case and 447% and 241% respectively for the second case compared to the current production. It is verified that an extract purity of 0.70 in the SMBR unit guarantees p-xylene content higher than 75% leading to a 95% recovery within the single stage crystallization unit. Moreover the SMBR unit proves to be flexible to handle different feed flow rates and the flow in the gas phase isomerization unit is significantly reduced. 8.5. Nomenclature 𝐶𝑝 = Heat capacity, J mol-1K-1 DC = Desorbent consumption, m3 kg-1