(288b) Simulated Moving Bed- Learning from the Past and Shaping the Future

Simulated Moving Bed (SMB) technology developed by UOP inventors Broughton, Gerhold (1) and Carson (2) under the name of Sorbex processes is used in many sectors from the petrochemical industry (PAREX process for separating p-xylene from C8 isomers), sugar industry (SAREX process for production of HFCS) to more recent applications in the pharmaceutical industry for chiral separations (3).

I first learned about SMB during my PhD studies in Nancy back in early seventies and I found the concept of simulating the solid movement by shifting the inlet/outlet ports periodically very bright. When going from “old” applications to “new” applications one can see differences on number of columns used and particle size but also similarities in terms of aspect ratio L/D around 10 (4). These new applications include the production of drugs as Zyrtec, Prozac, insulin, etc. New operating modes have emerged and we should mention VARICOL (5) where asynchronous shits are introduced which allows reducing the number of columns in many pharma applications from 6 to 5. These non-conventional methods are easily implemented in the FlexSMB unit built in our lab and shown in Figure 1b)

Figure 1 SMB Licosep 12x26 from Novasep (a) and FlexSMB (b)

Design methods for SMB started with triangle theory (6), expanding to separation volume (7) and standing wave theory (8). In the classical application of p-xylene separation the nature of desorbent changed along the years from heavy desorbent as para diethylbenzene (PDEB) to low affinity desorbent as toluene (LD Parex process). Detailed modeling of industrial Parex unit has been presented elsewhere (9).

An interesting problem in chiral separations is the separation of nadolol isomers since there are two pairs of racemates and various strategies can be envisaged to address that problem (10). All these applications are related to liquid-solid systems. Applications for vapor phase and gas phase systems are scarce. We have been working at LSRE-LCM on the separations of light olefins/paraffins (one of the “seven separations to change the world”) using different adsorbents (binderless 13X zeolie, ZIF-8, MOFs) (11).

In our lab we are involved in Process Intensification coupling reaction and adsorptive separation by using Simulated Moving Bed Reactors (SMBR) and even coupling SMBR with membrane permeation in PermSMBR (12). These technologies have been tested in the production of acetals (DEE, DME, DBE, DEB, GEA), esters (ethyl lactate, butylacrylate (13) ).Here we address the use of SMBR for p-xylene production.

The current production of p-xylene is based on a separation-isomerization loop with limited yield due to the thermodynamic equilibrium, which results in large cycle loops to achieve the desired amount of p-xylene. A large loop along with gas phase conditions within the isomerization unit increases significantly the energy consumption within the process. The energy, together with raw materials, required in the process can be significantly reduced by the use of multifunctional reactors. By combining reaction and separation in the same unit, the thermodynamic equilibrium can be shifted towards the target product; in this case, in-situ removal of p-xylene shifts the isomerization of xylene towards the para-isomer. The most employed process for the separation of p-xylene is based on the SMB technology. The proposed SMBR uses the same principle as the SMB and incorporates the reaction section, the catalyst, within the adsorption columns. In order to combine both processes, isomerization of xylenes must be conducted in liquid phase. 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. The presence of the catalyst throughout the columns contaminates the p-xylene in the extract port due to the reverse reaction. Therefore, the SMBR must be combined with a crystallization unit to obtain pure p-xylene. Moreover, since p-diethylbenzene (the most employed desorbent in the SMB) isomerizes in the presence of acid catalysts, other aromatic compounds (e.g., toluene) must be considered as desorbent for the SMBR.

The successful implementation of any reactive-adsorptive process depends on the determination of the fundamental reaction and adsorption data governing those processes. Even though adsorption of xylenes is well-known at industrial conditions (about 180 °C), studies at higher temperatures are not found in the literature. Since the optimum temperature for the SMBR is expected to be above 200 °C, batch experiments above 180 °C were carried out. The adsorbent was faujasite-type zeolite exchanged with Ba, similar to what is used in industry. Since catalysts and adsorbents are mixed throughout the unit, the isotherms were determined in the absence of water to prevent the catalyst from going through dealumination, which irreversibly affects its acid sites. p-Xylene was still the more adsorbed compound although the selectivity was lower and decreased at higher temperatures.

The industrial isomerization of xylenes is carried out in gas phase over zeolite ZSM-5 with a noble metal such as Pt. In liquid phase, diffusion constraints may reduce significantly the conversion of xylene isomers over ZSM-5; therefore, zeolites with larger pores must be studied. Fixed-bed catalytic experiments were conducted over zeolites ZSM-5, Beta, and Mordenite with different acidity also at above 180 °C. Acid catalysts in pellet form are limited to few zeolites with specific silica-to-alumina ratio, which defines the acidity of the solid. To overcome that, several zeolites in powder form were shaped to obtain appropriate pellets for fixed-bed studies through an extrusion-spheronization process. Large-pore zeolites Beta and Mordenite exhibited higher conversion and longer catalyst life than ZSM-5. Furthermore, side reactions were analyzed under the conditions studied in the presence of toluene. Contrary to what is found in the literature, transalkylation of toluene with ethylbenzene were observed in addition to disproportionation of ethylbenzene. Disproportionation of toluene was not detected confirming that that reaction goes through a different mechanism.

The literature reports several SMBR optimization studies. However, the majority of these investigations focus on the SMBR exclusively, without considering its integration with the subsequent desorbent recovery units. Thus, a short economical evaluation of the SMBR with the associated distillation columns was conducted. Since all units are connected, they shall be optimized all together. The objective function to be maximized was the price of the product minus the operating costs. The distillation columns were modeled as simple as possible (i.e., shortcut methods) but validated against commercial software to determine the heat duty. Since the SMBR consists of 24 columns (the same as the existing SMB), the unit was modeled with the true moving bed reactor approach. The variables to be manipulated during the optimization involved the arrangement of columns and flow rates in each zone of the SMBR, switching time, temperature, desorbent consumption, catalyst-to-adsorbent ratio, among others. The constraints included the minimum purity in the extract and maximum pressure drop per bed. The SMBR offering the best performance consisted of 4, 3, 15, and 2 columns in zones 1, 2, 3, and 4 respectively, operating at about 280 °C, and resulting in a reduction of the cycle loop in about 30 %.

(1) DB Broughton and C Gerhold, Continuous sorption process employing fixed bed of sorbent and moving inlets and outlets, US Patent 2985589 (1961)

(2) DB Carson and FV Purse, Rotary valve US Patent 3040777 (1962)

(3) RM Nicoud, The separation of optical isomers by simulated moving bed chromatography, Pharm. Technol. Eur 11, 36-44 (1999)

(4) P Sá Gomes, M Minceva and AE RodriguesSimulated moving bed technology:old and new, Adsorption, 12, 375-392(2006)

(5) P Adam et al, Process and device for separation with variable length, US Patent 6136198 (2000)

(6) Storti et al, Optimal design of multicomponent countercurrent adsorption separation processesinvolving nonlinear equilibria,, Chem Eng Sci, 44, 1329-1345 (1989)

(7) D Azevedo and AE Rodrigues, Design of simulated moving bed in presence of mass transfer resistances, AIChEJ, 45, 956-966 (1999)

(8) Z Ma and NHL Wang, Standing wave analysis of SMB chromatography: linear systems, AIChEJ, 43, 2488-2508 (1997)

(9) MSP Silva et al, Modeling and simulation of an industrial scale Parex process, AIChEJ, 61, 1345-1363 (2015)

(10) R. Arafahet al, Separation of Nadolol Stereoisomers Using Chiralpak IA Chiral Stationary Phase, Chirality 28 (2016) 399-408.

(11) Vanessa F. D. Martins, et al, “ Development of gas-phase SMB technology for light olefin/paraffin separations”, AIChEJ 62(7) 2490-2500 (2016).

(12) Carla Pereira, Viviana Silva and A.E. Rodrigues, “Coupled PermSMBR- Process design and development for 1,1 dibutoxyethane production, ChERD, 92, 2017-2026 (2014)

(13) Dânia S Constantino, et al “Simulated Moving Bed Reactor for Butyl Acrylate Synthesis: from Pilot to Industrial Scale”, Chem Eng and Processing 97, 153-168 (2015)