Valérie Grosfils

Parametric uncertainties and influence of the dead volume representation in modelling simulated moving bed separation processes

In this study, a systematic numerical procedure for identifying the model parameters of simulated moving bed (SMB) separation processes is developed. The parameters are first estimated by minimizing a weighted least-squares criterion using experimental data from batch experiments, e.g. the time evolution of the concentration of elution peaks. Then, a cross-validation is achieved using data from experiments in SMB operation. At this stage, the importance of a careful modelling of the dead volumes within the SMB process is highlighted. In addition, confidence intervals on the estimated parameters and on the predicted concentration profiles are evaluated.

On simplified modelling approaches to SMB processes

The simulated moving bed (SMB) technology is important in various fields, from sugar to enantiomer separation, and operating conditions must be carefully selected and regulated, based on an appropriate process model. Basically, two modelling approaches exist, i.e., true moving bed (TMB) and SMB models. Both approaches show advantages and drawbacks. In this work, an attempt is made to develop a model which retains the simplicity of the original TMB model and its modest computational load, while capturing the essential features of the cyclic steady state reproduced by the SMB model.

Modelling and identification from batch experiments of a SMB process

Abstract The Simulated Moving Bed (SMB) technology is a continuous chromatographic process which is important in various fields, from sugar to enantiomer separation. In this paper, a systematic identification procedure for determining parameters of SMB models from batch experiments is validated with experimental SMB data. Parameters are first estimated from elution peaks. Then a cross-validation with SMB experiments is performed so as to assess whether the parameters identified from batch experiments may be used in a SMB model. This part of the work requires a careful modelling of the dead volumes within the SMB process.

A SYSTEMATIC APPROACH TO SMB PROCESS MODEL IDENTIFICATION FROM SMB PROCESS DATA

Abstract The simulated moving bed process is a continuous chromatographic separation process, which is important in various fields, from sugar to enantiomer separation. In this paper, a systematic approach to parameter identification of a kinetic SMB model is presented. In contrast to previous studies, identifiability and the influence of local minima are investigated from fictitious data obtained by simulation of different SMB processes at various operating conditions.

TUNING ISSUES IN LOOP MONITORING

The aim of this study is to develop an automatic control loop monitoring system. The first step includes the detection of changes and of oscillations in the control error and in the control signal. To determine the parameters of the statistical change detection test (a CUSUM test), the average run length (ARL) function is exploited. To detect oscillations, a modified version of Hagglunds approach is presented. Methods are proposed in order to help the operator in the choice of the parameters for both detection tests. The second step should consist in determining the origin of the oscillations or of the changes. Only the first step is presented in this work.

Fouling resistance modelling, identification and monitoring in a thermosiphon reboiler

Abstract The aim of this paper is to build a mathematical model of the fouling resistance that is able to give the evolution of the fouling, which causes economic and energetic losses by degrading the heat transfer coefficient First, a mathematical model of the fouling resistance based on the overall heat transfer coefficient and on the inside film heat transfer coefficient is developed. Then, the overall heat transfer coefficient is obtained from the mass and heat balances of the reboiler. The inside ftlm heat transfer coefficient is determined from correlations with some process variables. The resulting model is a function of measured quantities and is validated with industrial data. To estimate the fouling resistance, only two parameters have to be identified.