Optimization of simulated moving bed chromatography with fractionation and feedback: Part I. Fractionation of one outlet

A novel modification of simulated moving bed (SMB) technology, referred to as fractionation and feedback SMB (FF-SMB), has been introduced recently. This concept is based on fractionating one or both outlet streams and feeding the off-spec fractions back into the unit alternatingly with the original feed mixture. In this paper, the optimization problem of FF-SMB realizing one outlet fractionation is considered. A mathematical optimization framework based on a detailed process model is presented which allows to evaluate quantitatively the potential of this operating scheme. Detailed optimization studies have been carried out for a difficult separation characterized by small selectivity and low column efficiency. The results reveal that the proposed fractionation and feedback regime can be significantly superior to the classical SMB chromatography, in terms of both feed throughput and desorbent consumption. The effect of the feeding sequence on the performance of FF-SMB is also examined.     

Improving performance of simulated moving bed chromatography by fractionation and feed-back of outlet streams

Simulated moving bed (SMB) chromatography is an important technique for the continuous separation of valuable products. Recently suggested process modifications have shown the potential for further improvement. In this work, a concept is presented that combines non-permanent product withdrawal at one or both outlet ports (leading periodically to a "product" and a "non-product" fraction) with an internal recycle and re-feeding of the "non-product" fraction in alternation to the original feed mixture. Using simulation studies for linear and non-linear isotherms, it was shown that in terms of process performance and product recovery, this fractionation and feed-back approach (FF-SMB) is superior to both the conventional SMB process as well as to a previously reported fractionation and discard strategy.     

Optimizing control of simulated moving beds--linear isotherm

A new optimization based adaptive control strategy for simulated moving beds (SMBs) is proposed. A linearized reduced order model, which accounts for the periodic nature of the SMB process, is used for online optimization and control. The manipulated variables are the four inlet flow rates, the outputs are the raffinate and extract concentrations. Concentration measurements at the raffinate and extract outlets are used as the feedback information. The state estimate from the periodic Kalman filter is used for the prediction of the outlet concentrations over a chosen horizon. Predicted outlet concentrations are the basis for the calculation of the optimal input adjustments, which maximize the productivity and minimize the desorbent consumption subject to constraints on product purities. The realization of this concept is discussed and the implementation on a virtual eight column SMB platform is assessed, in the case of binary linear systems. For a whole series of typical plant disturbances it is shown that the proposed approach is effective in minimizing off-spec products and in achieving optimal SMB operation, also in the case where there are significant model uncertainties.     

Optimization of a hybrid chromatography-crystallization process for the separation of Tröger's base enantiomers

This paper presents an analysis of a hybrid process consisting of simulated moving bed (SMB) chromatography and crystallization and studies its performance for the separation of the Tr?ger's base enantiomers. The SMB is simulated using a detailed model including column efficiency, thus, implying a proper evaluation of the effect of column size on column efficiency and separation performance. The crystallization operations are accounted for through material balances, assuming equilibrium between enantiopure crystals and mother liquor. A genetic algorithm is used to optimize the combined process, using proper definitions of objective functions. Multi-objective optimization of this hybrid process for productivity and evaporation cost in terms of operating parameters, column length, and SMB feed concentration shows an optimum SMB purity value as a trade off between increased SMB performance and recycle of the mother liquor.     

Effect of the homogeneity of the column set on the performance of a simulated moving bed unit. I. Theory

Although it is impossible to manufacture identical columns for use in a simulated moving bed (SMB) process, theoretical studies assume that all the columns in an SMB unit have identical characteristics. In practice, calculations in modeling and optimization studies are made with the average values of each column parameter set. In this report, the effects on SMB process performance caused by column-to-column fluctuations of the parameters are discussed. As a first step, we show how the differences in porosity of the columns may be taken into account with a revised set of separation conditions. Reductions in the purity of the extract and the raffinate streams are quantitatively related to the column-to-column fluctuations of the retention times of the two components arising from these porosity differences. For the sake of simplicity, the discussion first addresses the case of a four-column SMB operating under linear conditions. Then, the scope is extended to the cases of SMB units incorporating several columns in each section and to SMB units operating under nonlinear conditions.     

Comparison of configurations of a four-column simulated moving bed process by multi-objective optimization

Configurations of a four-column simulated moving bed chromatographic process are investigated by multi-objective optimization. Various existing column configurations are compared through a multi-objective optimization problem. Furthermore, an approach based on an SMB superstructure is applied to find novel configurations which have been found to outperform the standard SMB configuration. An efficient numerical optimization technique is applied to the mathematical model of the SMB process. It has been confirmed that although the optimal configuration highly depends on the purity requirement, the superstructure approach is able to find the most efficient configuration without exploring various existing configurations.     

Bi-level optimizing control of a simulated moving bed process with nonlinear adsorption isotherms

A bi-level optimizing control scheme originally proposed for a simulated moving bed (SMB) with linear isotherms has been extended to an SMB with nonlinear isotherms. Cyclic steady state optimization is performed in the upper level to determine the optimum switching period and time-varying feed/desorbent flow rates, and repetitive model predictive control is run in the lower level for purity regulation, taking the decision variables from the upper level as feed-forward information. Experimental as well as numerical study for an SMB process separating a high-concentration mixture of aqueous L-ribose and L-arabinose solutions showed that the proposed scheme performs satisfactorily against various disturbances. In contrast, an alternative scheme based on an SMB model with linear isotherms showed a limitation in the control performance; this scheme was apt to fail in purity regulation.     

Optimization of simulated moving bed and Varicol processes

A new continuous chromatographic process (Varicol) has been presented recently. Its basic principle consists, in contrast to the traditional simulated moving bed (SMB) technology, of an asynchronous shift of the inlet/outlet lines in a multi-column system with a recycle loop. Due to the stronger influence of the discrete dynamics on the plant behavior, the design of a Varicol process requires the use of model-based optimization to take advantage of the very high flexibility of this process. The equilibrium theory which has been successfully applied to SMB by many practitioners fails to predict the region of complete separation accurately. In this paper, we present a rigorous model-based optimization framework, which can handle the SMB and the novel Varicol process in a systematic manner. The feasibility of the approach is demonstrated by the separation of a mixture of propranolol isomers which exhibits a highly non-linear multi-component adsorption behavior. Experimental results are presented and discussed.     

Simulated moving bed chromatography for the separation of enantiomers

Simulated moving bed (SMB) chromatography, a continuous multi-column chromatographic process, has become one of the preferred techniques for the separation of the enantiomers of a chiral compound. Several active pharmaceutical ingredients, including blockbuster drugs, are manufactured using the SMB technology. Compared to single column preparative chromatography, SMB separations achieve higher productivity and purity, while reducing the solvent consumption. The SMB technology has found applications both at small and large scales. Design methods have been developed for robust operation and scale-up, using data obtained from analytical experiments. In the last few years, rapid developments have been made in the areas of design, improved process schemes, optimization and robust control. This review addresses these developments, as well as both the fundamentals of the SMB science and technology and some practical issues concerning the operation of SMB units. Particular emphasis is placed on the consolidation of the “triangle theory”, a design tool that is used both in the academia and industry for the design of SMB processes.     

Impact of adsorption isotherm parameters on the performance of enantioseparation using simulated moving bed chromatography

Often there are several chromatographic systems, i.e., combinations of mobile and stationary phases, available to solve a certain separation problem. Essential differences of these chromatographic systems are the separation factors and the efficiencies. For preparative applications in addition also the column saturation capacities and solubility limits are of importance. The impact of all these parameters appears to be rather well understood for conventional overloaded elution chromatography using a single column. In the last years the continuous simulated moving bed (SMB) process was increasingly used as a powerful alternative to batch elution since increased productivities and reduced solvent consumptions could be realised. However, the selection of suitable chromatographic systems is more sophisticated for this process. In this paper five different chromatographic systems capable of separating the enantiomers of mandelic acid are compared based on the achievable productivities using SMB chromatography. For these five systems the adsorption isotherms have been determined experimentally. Subsequently, an analysis of the SMB process was performed numerically using a well-established model.     

Simulated moving bed technology with a simplified approach for protein purification. Separation of lactoperoxidase and lactoferrin from whey protein concentrate

Simulated moving bed (SMB) technology is a continuous chromatographic technique proven to have many advantages compared to conventional batch chromatography, such as: raised productivity and product concentration, reduced buffer consumption as well as more efficient use of raw material. In this study a 20 column SMB process for the separation of lactoperoxidase and lactoferrin from whey protein concentrate (WPC) was developed. A simplified approach with data from a single column experiment was used when designing the process. The SMB process data were compared to a theoretical scale-up of the breakthrough experiment reflecting the same 20 column set-up run in non-moving bed mode. The outcome of the comparison is a 48% raise in productivity, a 4.3 times decrease in buffer consumption, 6.5 times raise in target protein concentration with a raw material utilization which is slightly better for the SMB process.     

Optimization of simulated moving bed and column chromatography for a plasmid DNA purification step and for a chiral separation

This work analyzes the performance of the SMB and the column chromatography processes for two different case studies: the first stage of the plasmid DNA (pDNA) polishing, and the Tr?ger's base enantiomer separation, in which the adsorption isotherms are linear and non-linear, respectively. Simulation tools are used together with an optimization routine (Non-Sorting Genetic Algorithm (NSGA)) in order to find the optimum operating conditions leading to maximum productivity and minimum solvent consumption; the optimum solution for each of the processes is a curve on the productivity-solvent consumption plane, the so-called Pareto set. The comparison between the column and the SMB processes is based on the relative position of the two Pareto sets calculated at equal conditions and for the same final purity and recovery of the target species. The results show that SMB is superior to column chromatography in the two case studies investigated, i.e. in the case of the linear isotherm (pDNA), the productivity gain is up to a factor two for a given value of the solvent consumption. Furthermore, the flexibility of the SMB operation is larger, since the Pareto sets are flatter and they prolong into regions of the productivity-solvent consumption plane that are not accessible with the column chromatography process.     

A low thermal mass fast gas chromatograph and its implementation in fast gas chromatography mass spectrometry with supersonic molecular beams

A new type of low thermal mass (LTM) fast gas chromatograph (GC) was designed and operated in combination with gas chromatography mass spectrometry (GC-MS) with supersonic molecular beams (SMB), including GC-MS-MS with SMB, thereby providing a novel combination with unique capabilities. The LTM fast GC is based on a short capillary column inserted inside a stainless steel tube that is resistively heated. It is located and mounted outside the standard GC oven on its available top detector port, while the capillary column is connected as usual to the standard GC injector and supersonic molecular beam interface transfer line. This new type of fast GC-MS with SMB enables less than 1 min full range temperature programming and cooling down analysis cycle time. The operation of the fast GC-MS with SMB was explored and 1 min full analysis cycle time of a mixture of 16 hydrocarbons in the C(10)H(22) up to C(44)H(90) range was achieved. The use of 35 mL/min high column flow rate enabled the elution of C(44)H(90) in less than 45 s while the SMB interface enabled splitless acceptance of this high flow rate and the provision of dominant molecular ions. A novel compound 9-benzylazidanthracene was analyzed for its purity and a synthetic chemistry process was monitored for the optimization of the chemical reaction yield. Biodiesel was analyzed in jet fuel (by both GC-MS and GC-MS-MS) in under 1 min as 5 ppm fatty acid methyl esters. Authentic iprodion and cypermethrin pesticides were analyzed in grapes extract in both full scan mode and fast GC-MS-MS mode in under 1 min cycle time and explosive mixture including TATP, TNT and RDX was analyzed in under 1 min combined with exhibiting dominant molecular ion for TATP. Fast GC-MS with SMB is based on trading GC separation for speed of analysis while enhancing the separation power of the MS via the enhancement of the molecular ion in the electron ionization of cold molecules in the SMB. This paper further discusses several features of fast GC and fast GC-MS and the various trade-offs involved in having powerful and practical fast GC-MS.     

Experimental implementation of identification-based optimizing control of a simulated moving bed process

Experimental implementation of an optimizing controller based on identified model for the separation of nucleosides in a laboratory scale simulated moving bed (SMB) unit is reported in this study. The manipulative variables are the three external and one internal flow rates while the outputs are productivity, solvent consumption, and purities of extract and raffinate streams averaged over a switching period. The feedback information is the concentration profile of extract and raffinate measured online using two ultraviolet (UV) detectors. Experimental results show that the designed controller is able to operate the SMB units under optimal condition fulfilling the purity requirements. Besides, the controller demonstrated excellent performance in terms of rejecting disturbances that may occur during SMB operations.     

Optimal operation of simulated moving bed chromatographic processes by means of simple feedback control

In this contribution, simple methods are presented for controlling a simulated moving bed (SMB) chromatographic process with standard PI (proportional integral) controllers. The first method represents a simple and model-free inferential control scheme which was motivated from common distillation column control. The SMB unit is equipped with UV detectors. The UV signals in the four separation zones of the unit are fixed by four corresponding PI controllers calculating the ratio of liquid and solid flow in the respective separation zone. In order to be able to adjust the product purity a second, model-based control scheme is proposed. It makes use of the nonlinear wave propagation phenomena in the apparatus. The controlled chromatographic unit is automatically working with minimum solvent consumption and maximum feed throughput--without any numerical optimization calculations. This control algorithm can therefore also be applied for fast optimization of SMB processes.     

Optimizing control of simulated moving bed separations of mixtures subject to the generalized Langmuir isotherm

Simulated moving bed (SMB) is a cost-efficient separation technique that offers high productivity and low solvent consumption. SMB has gained importance in the pharmaceutical and fine chemical industry to perform complex separation tasks. However, an open and challenging problem is the optimal, robust operation of the SMB process. We have developed a control scheme that integrates the optimization and control of the SMB unit. A significant feature of the controller is that only minimal information of the system has to be provided, i.e. the linear adsorption behavior of the mixture to be separated and the average void fraction of the columns. Therefore, a full characterization of the adsorption behavior of the mixture and the columns is no longer required. In this ‘cycle to cycle’ control scheme, the measurements, optimization and control actions are performed once in every cycle. This paper presents simulation results of the control scheme applied to the separation of binary mixtures characterized by generalized Langmuir isotherms. The results are presented and analyzed in the frame of the triangle theory that has been recently extended to encompass these types of isotherms. Besides, online optimum performance of the SMB unit is compared with off-line optimization carried out using genetic algorithm. The results show that the controller fulfills the product and process specifications while operating the SMB unit optimally, regardless of the different types of Langmuir isotherms that the systems exhibit.     

Effect of the homogeneity of the column set on the performance of a simulated moving bed unit. II. Experimental study

Previous studies of the simulated moving bed (SMB) process assume identical characteristics of all the columns incorporated in a given unit. Due to the practical impossibility to manufacture identical columns, numerical applications of the theory to modeling and optimization use for each of the needed column parameter the average value for the entire column set. In this study, the effects of these simplifications on the actual productivity of the SMB process are evaluated by making exact calculations, i.e., by taking the differences in the porosity values into account. We apply a revised set of separation conditions previously introduced and derived from the equilibrium theory. Earlier theoretical results are compared to experimental results obtained in the study of the enantiomeric separation of Tr?ger's base on Chiralpak AD. Due to the nonLangmuirian character of the adsorption isotherms of these two compounds on the packing material used, the separation area cannot be determined analytically. As an alternative, a reliable numerical algorithm was used to scan a wider region and to define the separation area. The form of this area depends on the applied porosity values. A UV detector and a laser polarimeter located at one node of the SMB monitor on-line the internal concentration profiles. Excellent agreement between the calculated and the experimentally determined concentration profiles was obtained under nonlinear conditions. The influence of column-to-column variations on the performance of the SMB process was found to be more significant under nonlinear than under linear conditions.     

Model-based design of a pilot-scale simulated moving bed for purification of citric acid from fermentation broth

One of the conventional processes used for the recovery of citric acid from its fermentation broth is environmentally harmful and cost intensive. In this work an innovative benign process, which comprises simulated moving bed (SMB) technology and use of a tailor-made tertiary poly(4-vinylpyridine) (PVP) resin as a stationary phase is proposed. This paper focuses on a model-based design of the operation conditions for an existing pilot-scale SMB plant. The SMB unit is modeled on the basis of experimentally determined hydrodynamics, thermodynamics and mass transfer characteristics in a single chromatographic column. Three mathematical models are applied and validated for the prediction of the experimentally attained breakthrough and elution profiles of citric acid and the main impurity component (glucose). The transport dispersive model was selected for the SMB simulation and design studies, since it gives a satisfactory prediction of the elution profiles within acceptable computational time. The equivalent true moving bed (TMB) and SMB models give a good prediction of the experimentally attained SMB separation performances, obtained with a real clarified and concentrated fermentation broth as a feed mixture. The SMB separation requirements are set to at least 99.8% citric acid purity and 90% citric acid recovery in the extract stream. The complete regeneration in sections 1 and 4 is unnecessary. Therefore the net flow rates in all four SMB sections have been considered in the unit design. The influences of the operating conditions (the flow rate in each section, switching time and unit configuration) on the SMB performances were investigated systematically. The resulting SMB design provides 99.8% citric acid purity and 97.2% citric acid recovery in the extract. In addition the citric acid concentration in the extract is a half of its concentration in the pretreated fermentation broth (feed).     

PowerFeed operation of simulated moving bed units: changing flow-rates during the switching interval

A possible way to improve the separation performance of simulated moving bed (SMB) units is to change the internal and external liquid flow-rates during the switching period. This operation mode, referred to as PowerFeed, is examined in this work through a model analysis. Similar to the Varicol process, which allows for the asynchronous movement of the ports, the PowerFeed process exhibits more degrees of freedom than the classical SMB process and therefore allows more room for optimization. Using an optimization technique based on a genetic algorithm, all three processes have been optimized for a few case studies in order to determine their relative potentials. It is found that PowerFeed and Varicol provide substantially equivalent performances, which are however significantly superior to those of the classical SMB process.     

Standing wave optimization of SMB using a hybrid simulated annealing and genetic algorithm (SAGA)

In this paper we draw on two stochastic optimization techniques, Simulated Annealing and Genetic Algorithm (SAGA), to create a hybrid to determine the optimal design of nonlinear Simulated Moving Bed (SMB) systems. A mathematical programming model based on the Standing Wave Design (SWD) offers a significant advantage in optimizing SMB systems. SAGA builds upon the strength of SA and GA to optimize the 16 variables of the mixed-integer nonlinear programming model for single- and multi-objective optimizations. The SAGA procedure is shown to be robust with computational time in minutes for single-objective optimization and in a few hours for a multi-objective optimization, which is comprised of more than one hundred points.