Masters Thesis: Identification of Models of Industrial Batch Cooling Crystallization system

Crystallization is the formulation of a material in the solid state, it is a core separation and purification technology in major sectors of chemical process industry because the crystalline material is formed from a fluid phase. The vast majority of crystallization operations are performed in batch mode, which is most suitable when the production is low-volumed and high value-added. The significance of batch crystallization processes can be described by the global market trends of semiconductor industries which would surpass US$ 340 billion in revenues in 2012. In batch crystallization, it is difficult to obtain the system model because of scarce fundamental understanding of nucleation and growth phenomena. According to this problem which can be divided into two subproblems such as: 1. Model structure discrimination 2. Parameter estimation, this Msc thesis project depicts an application of model-based experimental design to provide a strategy for model identification: First an optimal cooling temperature profile is designed for mode structure discrimination, which can be solved by weighted least square approach. Several outputs such as concentration, second moment are used to find out which measurements can be used to distinguish the true model structure from the wrong one. Subsequently the second subproblem parameter estimation is explored, including measurements, cooling temperature profile as well as initial condition. In addition, accuracy analysis is also explored both in Monte Carlo study and Sensitivity analysis. Finally cooling temperature profile for parameter estimation is improved by solving an optimization problem. Master of Science Thesis Xinwei Sun