Application of Seeding as a Process Actuator in a Model Predictive Control Framework for Fed‐Batch Crystallization of Ammonium Sulphate

Abstract

Seeding is discussed from the viewpoint of reproducible process actuation. Seeded experiments in a 75‐l Draft‐Tube (DT) crystallizer were performed in triplicate to verify reproducibility of the product characteristics. The seeded experiments displayed a significantly improved reproducibility of the results in comparison with the corresponding unseeded batches.Narrow sieve fractions of ground seeds were utilized as a seeding material. The seeds were suspended in a saturated solution before being fed into the crystallizer. It was found that the Crystal Size Distribution (CSD) of crystals suspended in a saturated solution changed during the time of their residence in the seeding vessel. Images taken by the Scanning Electron Microscope (SEM) confirmed that the initial change in the median size was caused by the dissolution of fines. The dissolution was subsequently accompanied by the crystal growth and attrition. An optimal residence time was found.Finally, the influence of the initial supersaturation, the seed load, and the seed size on the product quality was investigated. The experimental studies indicated that the initial supersaturation should be as high as possible yet not high enough to trigger primary nucleation. The lowest admissible sieve fraction and the corresponding minimal seed load were found. The conservatism of the used seeding model with respect to the minimal seed load was addressed.

Optimal seed recipe design for crystal size distribution control for batch cooling crystallisation processes

“…Experimentally, seed can be obtained by mixing different amounts of seeds with different size distributions (Aamir et al, 2009a;Doki et al, 2001;Kalbasenka et al, 2007;Kubota et al, 2001). According to the proposed method here, the seed recipe is represented by a sum of Gaussian distributions, with each Gaussian corresponding to the seed distribution in a particular sieve size range, and the generic optimisation problem with the objective of shaping the distribution at the end of batch is formulated as follows:…”

“…Seeding seems to be treated as an art rather than science (Adi et al, 2007;Jagadesh et al, 1999;Kalbasenka et al, 2007;Kubota et al, 2001;Ludwick and Henderson, 1968;Lung-Somarriba et al, 2004) and generally there is a lack of systematic methodologies related to the amount and size of seeds that should be added into a crystallizer to obtain a product with a desired size distribution. Although it is recognized that the most important manipulated variables for the optimisation of crystallisation processes are supersaturation trajectories as well as the seed characteristics (Bohlin and Rasmuson, 1996;Heffels and Kind, 1999;Kalbasenka et al, 2007;Ruf et al, 2000, Yannick et al, 2009, the number of approaches focusing on temperature or anti-solvent addition trajectory optimisations is disproportionally higher than contributions considering seed recipe optimisations. The approaches proposed so far, mainly consider the optimisation of the width and amount of a particular monomodal seed distribution (Chung et al, 1999;Kalbasenka et al, 2007).…”

“…Although it is recognized that the most important manipulated variables for the optimisation of crystallisation processes are supersaturation trajectories as well as the seed characteristics (Bohlin and Rasmuson, 1996;Heffels and Kind, 1999;Kalbasenka et al, 2007;Ruf et al, 2000, Yannick et al, 2009, the number of approaches focusing on temperature or anti-solvent addition trajectory optimisations is disproportionally higher than contributions considering seed recipe optimisations. The approaches proposed so far, mainly consider the optimisation of the width and amount of a particular monomodal seed distribution (Chung et al, 1999;Kalbasenka et al, 2007).…”

Measurement, modelling, and closed-loop control of crystal shape distribution: Literature review and future perspectives

“…They also developed real-time model-based optimal control of anti-solvent semi-batch crystallisation processes and seeded batch crystallisation processes using real-time single and multi-objective optimization, rigid logic and fuzzy logic control methods [171][172][173][174][175]. Kramer and co-workers [158,176] applied dynamic optimization for throughput maximization of an industrial semi-batch crystallisation process with a control strategy based on a non-linear moment model and the resulting problem being solved by a non-linear programming algorithm. Shi et al [91,133] developed a hybrid predictive control strategy using logic-based switching between model predictive control and a fall-back bounded controller with a well-defined stability region.…”

Estudo dos fundamentos de cristalização assistida por destilação com membranas em aplicação de dessalinização de água.

“…Os cristais/sementes devem possuir tamanhos e quantidades adequados e serem adicionados em condição de supersaturação, dentro da zona metaestável do respectivo sistema (Kalbasenka et al, 2007 Neste processo, a transferência de massa é promovida por um gradiente de pressão de vapor estabelecido através dos poros da membrana originados por um gradiente de temperatura (Bodell, 1968).…”