Viatcheslav Kulikov, Heiko Briesen, Wolfgang Marquardt:
Scale integration for the coupled simulation of crystallization and fluid dynamics
Chemical Engineering Research and Design, 2005, 83(A6), 706-717
The behaviour of real particulate processes, among them crystallization, is determined by the interaction of multiple process phenomena, which have to be modelled to fully describe the process. A traditional way of modelling only accounts for the process kinetics under the assumption of an ideal mixing. However, often this is not sufficient to obtain a rigorous process description. In order to consider an inhomogeneous flow field in the crystallizer, the fluid dynamics has to be modelled. Since the considered process phenomena act on different time and length scales, the development of scale integration methods is necessary. Scale integration methods enable the interaction between the phenomena on different scales and allow them to be incorporated into an overall process model. The aim of this contribution is the practical application of multi-scale modelling concepts to the coupled dynamic simulation of a crystallization process. During the coupled simulation the models of the process phenomena are implemented in appropriately selected specialized software tools. The full process description is then obtained by a software-technological integration. In this study, FLUENT and Parsival are used for the modelling of fluid dynamics and crystallization, respectively. A fine CFD grid is used for the fluid dynamics model. The population balance model of the crystallization is represented by a set of interconnected ideally mixed compartments. During the coupled simulation this set is represented by a flowsheet and is solved by means of an iterative modular simulation algorithm. In the present study, the algorithm is extended with aggregation and disaggregation methods in order to enable the scale integration. Aggregation methods provide the averaging of the local properties (temperature, etc.) to use them for the computation of crystallization kinetics in the compartments. Disaggregation methods based here on Gaussian smoothing or on spline interpolation are needed to update the properties on the CFD grid. The presented method has been evaluated for an exemplary crystallization process. The implemented framework for the multi-scale coupled simulation enabled a rigorous description of the crystallization process accounting for an inhomogeneous flow field. The software-technological coupling approach allows the distribution of the computational load on specialized software tools.