On the Influence of Mixing and Scaling-Up in Semi-Batch Reaction Crystallization

Abstract

Semi-batch crystallization experiments have been performed both in a loop reactor and in stirred tank reactors. Hydrochloric acid was fed to a stirred solution of sodium benzoate, and benzoic acid immediately formed. Benzoic acid is formed in excess of the solubility making the solution supersaturated. The loop reactor is U-shaped. In one leg a propeller stirrer was placed to circulate the solution and in the other a turbine stirrer was placed in front of the feed point to vary the local mixing intensity. The objective was to analyse the relative importance of different levels of mixing on the product size distribution. The importance of mixing as well as scaling-up effects on the product size distribution were studied in three stirred tank reactors of volumes 2.5 L, 10 L, and 200 L. The stirred tank reactors had different geometry and were equipped with either a marine propeller or a pitched blade turbine. The weight mean size generally increases with increasing total feeding time and increasing mixing intensity. The weight mean size increases by locating an extra turbine impeller at the feed point in the 10 L stirred tank reactor. The turbine impeller provides the desired feed point mixing intensity without raising the mixing intensity of the whole tank. The weight mean size increases with decreasing feed pipe diameter in the loop reactor and for low feed rates in the 10 L stirred tank reactor. The weight mean size increases significantly by changing the feed pipe opening from circular to rectangular with a constant cross-sectional area at equal feed rates. Backmixing is visually observed in the largest feed pipe diameter in the loop reactor, thus, reducing the weight mean size. English However, backmixing is not considered to be a dominant phenomenon in the present work. Mesomixing time constants have been calculated according to the turbulent dispersion mechanism and the inertial-convective mechanism. The time constants for mesomixing are generally longer than the time constant for micromixing. Thus, the ratio of the mesomixing and the micromixing time constants shows an influence of mesomixing as is shown by the experimental results. The experimental results are best described by the inertial-convective disintegration mechanism showing that the feed plume mixing increases with decreasing feed pipe diameter and increased feed point mixing. The weight mean size is not strongly affected by the reactor volume. However, the mixing conditions in the reactors have a strong influence on the weight mean size. No suggested scaling-up rule can satisfactorily predict the weight mean size in the different volumes. No single physical parameter, such as the local energy dissipation rate, the mean energy dissipation rate or the circulation time, can satisfactorily explain the experimental results. A new dimensionless mixing parameter, TR, has been defined as the ratio of the total feeding time and the mesomixing time constant. The mesomixing time constant is defined as the shortest dimension of the feed pipe divided by the resultant bulk velocity passing the feed pipe entrance. The experimental results from both the loop reactor and the stirred tank reactors of different volumes can be correlated with TR. The weight mean size increases with increasing TR.