One of the major product specifications of a crystalline material is the crystal size distribution (CSD). In order to monitor and control the CSD in an industrial crystallization process, on‐line sensors are required. Over the years, a number of techniques to measure the CSD have been established. In this paper, three instruments operated in an on‐line fashion and an off‐line method are compared. The instruments were the OPUS, a HELOS/VARIO (both manufactured by Sympatec) and a Malvern 2600c (manufactured by Malvern). They were implemented on an 1100‐l evaporative draft tube baffle (DTB) crystallizer producing ammonium sulfate crystals from aqueous solution. Samples from this crystallizer were also analyzed offline by wet sieving. The results show reasonably good agreement between the different on‐line techniques and the wet sieving technique concerning the shape of the distribution. However, there is a discrepancy regarding the absolute values, which can be explained by the fact that the techniques used are based on different measuring principles.

No abstract

No abstract

No abstract

Experiments on a pilot-scale 150-litre forced circulation crystallizer with external circulation flow rates of 2 and 3 l/s revealed a high sensitivity of the product crystal size distribution for this operational variable. It is believed that a change in the circulation flow rate influences (1) the nucleation rate, which is mainly caused by the mechanical action of the pump, (2) the degree of internal dissolution in the heat exchanger and (3) the settling of the larger crystals in the crystallizer body. This paper presents a modelling framework which allows for a mechanistic description of growth and dissolution, attrition and secondary nucleation and settling. Such a description is seen as a prerequisite for the development of crystallization models with a predictive value for different operational conditions or for another scale of operation. Growth, attrition and nucleation kinetics are modelled using relationships from both Eek [1] and Ó Meadhra [2], whereas a diffusioncontrolled model is applied to describe size-dependent dissolution kinetics. The settling behaviour of crystals is described using Stokes law combined with Richardson-Zaki's equation for hindered settling. From the simulation results it can be concluded that the modelling framework presented here is capable of qualitatively predicting the effects of a change in the circulation flow rate. The quantitative predictions of the median crystal size are within 15% of the experimental values, whereas the oscillations predicted during start-up show a far larger deviation from the experimental results.