A framework for understanding and controlling batch cooling crystallization

Batch cooling crystallization;Crystal size control;Process analytical technology;Mass-count plot;Focused beam reflectance measurements;FBRM;Attenuated total reflectance Fourier transform infrared;ATR-FTIRpopulation balance;Population balance;Dynamic programming;Optimal control;Model predictive control;Machine learning;Learning from datadata-driven learning;Data-driven learning;Feedback control

Griffin, Daniel J. ; Rousseau, Ronald W. Grover, Martha A. Kawajiri, Yoshiaki Chemical and Biomolecular Engineering Realff, Matthew J. Petrovic, Bojan ; Rousseau, Ronald W.

University:Georgia Institute of Technology

Department:Chemical and Biomolecular Engineering

关键词: Batch cooling crystallization;    Crystal size control;    Process analytical technology;    Mass-count plot;    Focused beam reflectance measurements;    FBRM;    Attenuated total reflectance Fourier transform infrared;    ATR-FTIRpopulation balance;    Population balance;    Dynamic programming;    Optimal control;    Model predictive control;    Machine learning;    Learning from datadata-driven learning;    Data-driven learning;    Feedback control;   

Others  :  https://smartech.gatech.edu/bitstream/1853/55619/1/GRIFFIN-DISSERTATION-2016.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

In taking a different view of crystallization dynamics, this thesis reveals a new framework for addressing a prevalent process engineering challenge: control over the size of crystals produced by batch cooling crystallization. The thesis divides roughly into halves. In the first half, the crystal size control problem is introduced and the proposed framework for addressing this problem—termed the mass-count (MC) framework—is developed. This new framework is laid out along side the population balance (PB) framework, which is the prevailing framework for modeling crystallization dynamics and addressing the crystal size control problem. In putting the proposed and established frameworks side by side, the intent is not to say that one or the other is correct. Rather, the point is to show that they are different perspectives that facilitate different control approaches. The PB framework is built up from first principles; it is intellectually stimulating and mathematically complete, but it has a drawback for application: it does not directly enable feedback control. The MC framework, on the other hand, takes a less detailed view of crystallization dynamics and does not connect to crystallization theory as directly; it is also more conducive to application. In the second half of the thesis, the utility of the MC framework is put to the test. The framework is first applied to understand and model the crystallization dynamics for two widely different systems: darapskite salt crystallization from water and paracetamol crystallization from ethanol. Once the dynamics have been modeled, the framework is then used to develop feedback control schemes. These schemes are applied to both experimental systems and, in both cases, crystal size control is demonstrated.