Solution Concentration Measurement and Operation Optimization of Batch Cooling Crystallization

Author:Zhang Fang Kun

Supervisor:liu tao


Degree Year:2019





Crystallization technology has been widely used for product separation and purification in fine chemical and pharmaceutical industries.For monitoring process dynamics along with crystal growth during crystallization,on-line methods for measuring the solution concentration based on infrared spectroscopy have been increasingly used in engineering applications in the recent years.Although some methods based on in-situ Attenuation total reflection-Fourier transform infrared(ATR-FTIR)spectroscopy had been proposed in the literature for real-time measurement of solution concentration,their real applications are limited to some extent,especially for low-concentration or multi-component crystallization processes.Besides,there are more attentions and discussions deveted to control and optimization of crystal product quality,based on the dynamic information of crystallization process acquired by real-time measurement.This dissertation focuses on spectral calibration model building for real-time measurement of solution concentration via in-situ ATR-FTIR spectroscopy,operating variable selection and process optimization,with case study on batch cooling crystallization process of L-glutamic acid(LGA).The main research contents include:To overcome the influence from overlap with the solvent absorption peak and temperature variation to ATR-FTIR spectra real-time measured during cooling crystallization,a differential spectra measurement method is proposed based on real-time temperature correction.The operating temperature is referenced to eliminate the influence from the solvent absorption peak,according to the Lambert-Beer law and the additivity of the Fourier absorbance spectra.Subsequently,the partial least-squares(PLS)algorithm is used to establish a calibration model of the solution concentration for the preprocessed ATR-FTIR spectra.Experiments on measuring the solution concentration during LGA cooling crystallization are performed to demonstrate the effectiveness and accuracy of the proposed method.The measurement accuracy is evidently improved compared to the existing methods based on unpreprocessed spectra or constant temperature correction.To overcome the nonlinearity problem of in-situ ATR-FTIR spectra due to the influence from operating conditions and dynamics properties of crystallization processes,a spectral calibration modeling method is proposed based on the operational zone of a crystallization process.The infrared spectral data collected are classified in terms of the unsaturated zone(USZ)and metastable zone(MSZ),according to the concentration-temperature phase diagram of the cooling crystallization process.Four different calibration models for solution concentration prediction are established based on different uses of MSZ and USZ data.By comparison,it is found that using the MSZ data for model calibration could give the best prediction accuracy for monitoring MSZ that is the operating window of cooling crystallization,while maintaining accuracy when the solution concentration and operating temperature vary during crystallization.In contrast,the calibration model based on the USZ spectra could give the best prediction accuracy for monitoring USZ,but with a larger prediction error for MSZ.Therefore,the zone-partition based calibration model building method for applying ATR-FTIR spectroscopy can ensure good measurement accuracy in MSZ and USZ,.respectively.Concerning different crystal shapes produced by different operation conditions in batch cooling crystallization processes,the effects of the main operating variables including cooling rate,seeding temperature and initial supersaturation on crystal size distribution(CSD)and crystal shape distribution(CShD)are investigated based on the β-LGA cooling crystallization process.Through experimental design of these operating variables,some guidelines for β-LGA crystal growth are obtained by comparing the crystal shape,CSD and CShD measured by an in-situ imaging system.Correspondingly,some strategies are given for control and optimization of β-LGA crystallization process,in order to obtain the desired crystal shape and CSD.For batch cooling crystallization,a seed recipe design is proposed for obtaining the desired product yield,mean size and size variance.The effects of seed recipe and batch time on product quality are investigated based on the population balance model(PBM)simulations with respect to the size-dependent growth of crystals.By introducing a constrained objective function for seed recipe optimization,the desired product yield and size distribution can be predicted with regard to a specified manipulating seed recipe.The effectiveness and practicability of the proposed method are well demonstrated by the simulation tests and experiments of β-LGA cooling crystallization.