A nested loop for simultaneous model topology screening, parameters estimation, and identification of the optimal number of experiments: Application to a Simulated Moving Bed unit

TL;DR

This paper introduces a nested loop optimization method for efficient model topology screening, parameter estimation, and determining the minimum number of experiments needed in SMB chromatography, validated through noisy data and cross-validation.

Contribution

It presents a novel nested loop approach that simultaneously addresses topology screening, parameter estimation, and experiment minimization in SMB modeling.

Findings

The methodology accurately estimates parameters from noisy data.

It effectively screens model topologies based on isotherm definitions.

The approach reduces the number of experiments needed for reliable modeling.

Abstract

Simulated Moving Bed (SMB) chromatography is a well-known technique for the resolution of several high-value-added compounds. Parameters identification and model topology definition are arduous when one is dealing with complex systems such as a Simulated Moving Bed unit. Moreover, the large number of experiments necessary might be an expansive-long process. Hence, this work proposes a novel methodology for parameter estimation, screening the most suitable topology of the models sink-source (defined by the adsorption isotherm equation) and defining the minimum number of experiments necessary to identify the model. Therefore, a nested loop optimization problem is proposed with three levels considering the three main goals of the work: parameters estimation; topology screening by isotherm definition; minimum number of experiments necessary to yield a precise model. The proposed methodology emulated a real scenario by introducing noise in the data and using a Software-in-the-Loop (SIL) approach. Data reconciliation and uncertainty evaluation add robustness to the parameter estimation adding precision and reliability to the model. The methodology is validated considering experimental data from literature apart from the samples applied for parameter estimation, following a cross-validation. The results corroborate that it is possible to carry out trustworthy parameter estimation directly from an SMB unit with minimal system knowledge.