(148d) Optimal Control of Crystal Properties for Nonisothermal Antisolvent Crystallization

Optimal Control of Crystal Properties for Nonisothermal Antisolvent Crystallization

   Crystallization is one of the main processes in industry that occurs as a result of changes in solubility equilibrium and is widely used for solid-liquid separation. Today various sections of the chemical industry, at some stage, utilize crystallization as a method of production, purification or recovery of material. Mean crystal size and crystal size distribution are significant properties for a crystalline product that influence factors such as filtration rate, de-watering rate, dissolution rate and bioavailability. Therefore, it is of paramount importance to have control over particle size.

   Several ways to change the equilibrium solubility include: cooling, evaporation, and addition of anti-solvent. In this paper we consider the non-isothermal anti-solvent crystallization. It is hypothesized that, for systems with solubility weakly dependent on temperature, it is possible to impart significantly improved control over both the distribution mean size and coefficient of variation by manipulating temperature together with anti-solvent feed rate. The one dimensional Fokker-Planck equation (FPE) with linear stochastic term represents the dynamic characteristic of the crystal growth and used as process model. Analytical solution of the FPE gives the relationship between controlled variables and manipulated variables to obtain stability of the global model in a process control framework. Taking advantage of image analysis as an online method of data sampling during experiments, the crystals’ characteristics can be measured in real time and a feedback control loop is achievable. A three mode PID controller is tuned and its transient performance is investigated for set-point tracking to ensure that the process operates at its design specifications. Additionally, the model of the process is directly incorporated to implement a feedforward control and improve the performance of the closed loop. Knowing the exact transfer function of both mean size of crystals and CSD variance as a function of the anti-solvent feed-rate and temperature, it is possible to develop an Internal Model-based Controller (IMC) and use it as feedback controller. Results are illustrated for the crystallization of sodium chloride in water using ethanol as antisolvent and performed in an experimental bench-scale semi-batch crystallizer.