Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Chemical and Biochemical Engineering


Ajay K. Ray


Simulated Moving Bed (SMB) systems are used for separations that are difficult using traditional separation techniques. Due to the advantage of adsorption-based chromatographic separation, SMB has shown promising application in petrochemical and sugar industries, and of late, for chiral drug separations. In recent years, the concept of integration of reaction and in-situ separation in a single unit has achieved considerable attention. The simulated moving bed reactor (SMBR) couples both these unit operations bringing down the operation costs while improving the process performance, particularly for products that require mild operating conditions. However, its application has been limited due to complexity of the SMBR process. Hence, to successfully implement a reaction in SMB, a detailed understanding of the design and operating conditions of the SMBR corresponding to that particular reaction process is necessary.

Biodiesel has emerged has a viable alternative to petroleum-based diesel as a renewable energy source in recent years. Biodiesel can be produced by esterification of free fatty acids (present in large amounts in waste oil) with alcohol. The reaction is equilibrium-limited, and hence, to achieve high purity, additional purification steps increases the production cost. Therefore, combining reaction and separation in SMBR to produce high purity biodiesel is quite promising in terms of bringing down the production cost.

In this work, the reversible esterification reaction of oleic acid with methanol catalyzed by Amberlyst 15 resin to form methyl oleate (biodiesel) in SMBR has been investigated both theoretically and experimentally. First, the adsorption and kinetic constants were determined for the biodiesel synthesis reaction by performing experiments in a single column packed with Amberlyst 15, which acts as both adsorbent and catalyst. Thereafter, a rigorous model was used to describe the dynamic behaviour of multi-column SMBR followed by experimental verification of the mathematical model. Sensitivity analysis is done to determine robustness of the model. Finally, a few simple multi-objective optimization problems were solved that included both existing and design-stage SMBRs using non-dominated sorting genetic algorithm (NSGA). Pareto-optimal solutions were obtained in both cases, and moreover, it was found that the performance of the SMBR could be improved significantly under optimal operating conditions.