Johan Jager

Control of industrial crystallizers

The particle size distribution produced in particulate processes generally has a large impact on process economics, and the crystal size distribution produced in industrial crystallizers is no exception. Crystal Size Distribution (CSD) control is therefore desirable. Its establishment is the main objective of a joint research program between three research groups within the Delft University of Technology and several industrial participants. This paper presents a review of this project and intends (1) to illustrate the development of an on-line crystal size measuring technique, (2) to illustrate how information on CSD can be used to derive a dynamic process model and (3) to emphasize the need for effective process inputs and to make suitable suggestions in that direction.

On-line particle size measurement in dense slurries

Abstract An automatic dilution unit which enables fast and representative dilution of a process stream, down to the solids concentration required for Fraunhofer diffraction size measurements, has been developed. This unit has been built and tested on a 1-m3 continuous crystallizer producing ammonium sulphate. A Malvern 2600c particle size analyser is used as the actual measuring device. The combined system of dilution unit and sizer is integrated into a supervisory control system. From experimental data, the reproducibility of the average size (m4/m3), the second, third and fourth moments of the Particle Size Distribution (PSD) is estimated to be 2.2%, 2.7%, 2.3%, 3.3%, respectively. The observed trends of the PSD are in accordance with results obtained by sieve analysis. The measurement frequency attained is 30 times per hour.

Estimation of nucleation kinetics from crystal size distribution transients of a continuous crystallizer

Abstract Four techniques which can be used for the estimation of nucleation kinetics under unsteadystate conditions are applied to transients of a 20-1 evaporative ammonium sulfate crystallizer. A moment and a Laplace transformation technique appear not to be feasible because they rely heavily on the measurement of the number of small crystals. The present crystal size distribution (CSD) measurement technique is inaccurate in this size range and therefore both methods suffer severely from measurement noise. The third technique, using the method of characteristics, indicates the appearance of so-called secondary nucleation kinetics. The technique is unsuitable for estimating the kinetic parameters quantitatively, again as a result of the measurement noise. Therefore a fourth technique, which uses a non-linear parameter estimation routine, has been developed. This technique is very successful in estimating the parameters in a secondary nucleation model. With the estimated parameters the second and third moments of the CSD can be predicted. If additional moments are to be predicted it is shown that other phenomena which affect the crystal size distribution, such as crystal attrition, must be included in the dynamic model of the process.

Derivation of State-Space Models of Continuous Crystallizers

Abstract For effective control of crystallizers, multivariable controllers are required. To design such controllers, a model in state-space representation is required. In deriving such a model, the population balance has to be transformed into a set of ordinary differential equations. Three transformation methods were reported earlier in the literature. Two of these methods are limited to MSMPR crystallizers. The third method has no restrictions in its applicability, however it results in very high model orders. Therefore another technique is proposed which can also be applied directly on experimental data without the intermediate step of calculating the kinetic parameters.