Utilization of operation data for parameter estimation of simulated moving bed chromatography

Abstract

Many model‐based optimization strategies for simulated moving bed (SMB) processes have been proposed to determine operating conditions efficiently; however, there has not been a technique to utilize operation data to enhance the reliability of a mathematical model. This study focuses on developing a parameter estimation method for SMBs to obtain a reliable model utilizing the operation data collected in SMB plants through daily runs, which may contain measurement errors. To use such data, Tikhonov regularization was employed where the regularization parameter is determined by the framework of the discrepancy principle. The potential of our estimation method has been demonstrated by sugar separation described by a nonlinear isotherm. The parameter estimation was conducted with 20 operation data sets from an SMB pilot plant. The prediction of the resulting model was validated against test data sets, and the confidence regions of the parameters were evaluated. These tests confirmed that the model has improved compared with the model initially obtained.

Strategies for Simulated Moving Bed Model Parameter Estimation Based on Minimal System Minimal Knowledge: Adsorption Isotherm Equation Screening and Estimability Analysis

“…Traditionally SMB parameter estimations are carried out by the “Step by Step” approach that might involve tracer tests, breakthroughs, and pulse experiments, followed by validation in an SMB unit . This last step is usually carried out without much statistical rigor, except for a few works in the literature. − If the model does not provide a good representation, then a new mathematical representation must be chosen by redefining its AIE.…”