Simulation, Optimization And Scaling Up Of Simulated Moving Bed Systems For Separations Of Citric Acid

Citric acid or 2-hydroxy-1,2,3-propanetricarboxylic acid (molecular formula: C6H8O7), an organic carboxylic acid containing three carboxyl groups, is applied in the field of food, medication, chemistry and etc.As well known, citric acid (CA) industry has been already receiving increasing attentions for its environmental problem, high consumption etc. Many late-model processes have been developed. In all processes, the simulated moving bed (SMB) chromatographic separation process, in that only the hot water is used as the impetus, has been developed by Southern Yangtze University and Wuxi Green Separation Applied Technology Institute Co. Ltd, is very promising.As can be seen form the flowchart of SMB process for separating CA, except hot water no any other chemical medicine has been added into this very short process. These characteristics are come-hither in both environment and economy. It will be a revolution for CA production, if this process comes true into the mass production. Excitingly, we have already set up a pilot scale SMB system (52mmID×1700mmL×10columns) for CA separation in our institute, which performs very well. A 5,000 T/Y CA industrial plant (1200mmID×4500mmL×10 columns) is running in Yixing Xielian Biochemistry Co., Ltd. and a 10,000 T/Y CA industrial plant also will be build soon.Chromatography is a very powerful separation method. The SMB process allows continuous separation operation, which provides many advantages and also has excellent perspectives in the field of CA production.But due to the operational complexity of SMB system, to enlarge the equipment from the laboratorial to the industrial size is quite challenging. The conventional methods of scaling up are not suited. In this case, computer simulation is becoming a powerful and challenging tool.To achieve this scaling up goal, four steps can be designed:1. Firstly, determining the model parameters and investigating influence on sepertation of model parameters, algorithm using analytical chromatographic column.2. Firstly, determining the model parameters and investigating influence on sepertation of model parameters, algorithm with 52mmID×1700mmL column.3. modeling and simulating on the SMB system with 52mmID×1700mmL×10 columns, and checking the concentration profiles of CA and glucose to make sure a model and optimize its separation performances.4. Lastly, optimizing its separation performances on this preparative SMB.Step 1, several experiments are designed to dertermine model parameters on analytical column (4.6mmID×250mmL): external porosity (tracer method):0.34; transfer coefficient (gulcose, Van-Deemter method): 0.5698min^-1; transfer coefficient (citric acid, Van-Deemter method): 0.3576min^-1; adsorption isotherms (citric acid, frontal analysis): 72. 9730·c_CA (1+0.4840·c_CA)+0.3273·c_CA; adsorption isotherms (glucose, frontal analysis): 1. 3066·c_glucose. Then, the simulations of breakthrough curve of glucose and CA on a single analytical column are implemented with several different dynamic models: ideal model, equilibrium dispersed model (ED model) and dispersed plug flow model (DPF model). For the purpose of process simulation, several methods for the solution of the model equations are available: finite difference methods (FDM), finite element methods (FEM), orthogonal collection methods (OCM) and methods of lines (MOL). After much calculation, the PDF model with the MOL is thought to advisable (cpu time=7s, AS=97%). On second thoughts, the influences of various system parameters on the simulation results have been researched.The following conclusions can be drawn:1. Adsorption isotherm is the most hypersensitive in all parameters, so, usually requires an accurate estimation.2. Influences of chromatographic model on CPU time: DPF>EDM>IDM. Influences of chromatographic model on simulated result : DPF> IDM > EDM.3. Influences of simualtion algorithm on CPU time: FEM>FDM>MOL. Influences of simualtion algorithm on simulated result : FEM> MOL > FDM.Step 2, firstly, the parameters of analytical column are enlarged to the preparative column (52mmID×1700mmL). Several experiments are designed: external porosity (retention time method):0.35; transfer coefficient (gulcose, Van-Deemter method): 0.5358min-1; transfer coefficient (citric acid, Van-Deemter method): 0.33736min-1. Adsoption isotherm of citric acid and glucose is very inmportant and is determined with inverse method: Lastly, this chromatograpic model (PDF) is solved by self-adaption method of lines and result shows the deviation of experimental curve and simulated curve is neglectable (0.012).Step 3, Connecting the dynamic models of the single chromatographic columns while considering the cyclic port switching, a dynamic model of the SMB process () can be developed. Comparisons were made between the results obtained from the simulation and experiment and the agreement performs very well (FIT=96.35%). In particular, the chosen model and method which discredited the partial differential equations linking MOL prove to be very efficient and robust.Step 4, two methods are employed to optimize SMB chromatography: triangle theory based on the equilibrium theory and mathematic optimization model with objection of minimum consumption. first, complete seperata triangle of citric acid and glucose is built and the best operate set is S3: desorbent flux (5.551L/h), feed flux (1.026L/h), extract flux (2.738L/h), raffinate flux (3.839L/h), switching time(0.340L/h). Then the mathematic model is: which solved by SQP algorithm and the optimal set is desorbent flux (5.139L/h), feed flux (1.343L/h), extract flux (2.939L/h), raffinate flux (3.543L/h), switching time(0.356L/h).