crystallization; monitoring; particle size distribution; control
(2015), Agglomeration of Needle-like Crystals in Suspension: I. Measurements, in Crystal Growth and Design
, 15(4), 1923-1933.
(2015), Modeling the Facet Growth Rate Dispersion of β L-Glutamic Acid - Combining Single Crystal Experiments with nD Particle Size Distribution Data, in Chemical Engineering Science
, 133, 30-43.
(2014), Growth Rate Estimation of β l-Glutamic Acid from Online Measurements of Multidimensional Particle Size Distributions and Concentration, in Industrial & Engineering Chemistry Research
(2013), High accuracy online measurement of multidimensional particle size distributions during crystallization, in Chemical Engineering Science
, 105, 155-168.
(2013), Parametric Polytope Reconstruction, an Application to Crystal Shape Estimation
, Measuring multidimensional particle size distributions during crystallization, in Chemical Engineering Science
, 77, 130-142.
, Parametric Polytope Reconstruction, an Application to Crystal Shape Estimation, in IEEE Transactions on Image Processing
, 23(10), 4474-4485.
The CrystOCAM project aims at developing a new generation of tools to model, to measure and monitor online, to optimize and control the distribution of sizes and shapes of crystals of a suspension of particles and crystals during the crystallization from solution of organic compounds of interest for pharmaceutical, food and fine chemical applications.Crystallization is very often the key purification and pre-formulation step of products in the above mentioned industrial sectors. By tuning operating conditions mechanical properties such as tabletability and flowability, as well as features such as purity, chemical stability and bioavailability can be controlled. Such properties are critically influenced by the distribution of sizes and shapes of crystals, which must be characterized through the so called multi-dimensional particle size distribution (nD-PSD), i.e. the distribution of probabilities of the occurrence of certain combinations of two,three or more characteristic dimensions in the particle population. Describing, measuring and controlling in an optimal manner during process development, scale-up and operation such nD-PSD are at the same time major scientific and technical challenges and great opportunities to make an important change in product quality and process robustness in thecrystallization of organic compounds.The proposed three-year project will rely on the expertise of three laboratories, in each of which one Ph.D. student will be working: crystallization know-how at the Separation Processes Laboratory (SPL) of Prof. Mazzotti at ETH Zurich; control knowledge at the Automatic Control Laboratory (IfA) of Prof. Morari at ETH Zurich; computer vision and optimization-based control expertise at the new Laboratoire d'Automatique LA) of Prof. Jones at EPF Lausanne.The project partners will tackle the challenges above by exploiting important and promising background knowledge, such as a microscopic flow-through cell recently developed at SPL that images individual crystals in suspension and provides sufficient information to reconstruct their three-dimensional structure, or the image analysis and shape reconstruction algorithms developed by Prof. Jones, or the model predictive control androbust control techniques mastered at IfA and already applied for complex optimal control tasks in chemical processes such as azeotropic distillation or multicolumn continuous chromatography.The following deliverables are expected:- Advanced vision algorithms capable of estimating the three-dimensional shape of individual crystals and therefore the nD-PSD of the crystal ensemble online during the crystallization process, so as to enable process monitoring- Dynamic models of the evolution of the nD-PSD during crystallization and protocols for the characterization of the rates of the key mechanisms involved (nucleation, growth, agglomeration), which will enhance the understanding of the evolution of size and shape during crystallization- An optimization-in-the-loop controller that will allow for the first time optimization and control of the final nD-PSD- A collection of experimental case studies that will assess and demonstrate the effectiveness of the tools developed within the project