Cooling Crystallization in an Oscillatory Flow Baffled Crystallizer (OFBC): Influence of Fluid Dynamics on Crystal Product

The Oscillatory Flow Baffled Crystallizer (OFBC) design has been proven to offer improved temperature control and near plug plow residence time distributions, while at the same time providing large residence times. These OFBC properties hold the promise of resulting in a narrow crystal size distribution, reduced residence time requirements and improved process control. However to obtain such results the operation of the OFBC along with the crystallization process should be optimized. One of the most important operational aspects to optimize is the local turbulence that results in setting the dispersion in the crystallizer to a minimum, without compromising on the mass and heat transfer processes. In this study the residence time distribution of a tracer resulting from various possible operating configurations (amplitude and frequency) in the OFBC is analyzed, for water flows with a net flow Reynolds number of 140 (flowrate 100 ml/min) using dye tracing with an in situ transmission dip probe (absorption spectrophotometry). The best combinations of frequency and amplitude at which narrowest distribution (least dispersion) of the tracer concentration profile is achieved are found to be 1 Hz and 1 mm, 2 Hz and 3.5 mm and 4 Hz and 1 mm. The experimental results show that the ratio of the oscillatory Reynolds number and the net flow Reynolds number should be between 0.7 and 5. This is a broader range than the reported 2 to 4 range in literature. The measured residence time distributions for the best oscillatory setting have been successfully fitted to a tanks-in-series model with a 3 % maximum error of the coefficient of determination. It can therefore be concluded that the number of tanks in series is an accurate characteristic parameter of the fluid dynamics. The assumption of ideal plug flow, which is shown to not resemble the real RTD well, can therefore be avoided when developing the model for crystallization in an OFBC. The results of a parameter sensitivity study, based on the developed model, shows that the final seed CSD is weakly related to secondary nucleation for the used kinetic parameters. This can be explained as the small mass of the nucleated crystals does not have a large effect on the supersaturation. Furthermore, the final seed CSD has a dependence on all varied parameters but mostly to the initial seed loading. This is logical as double the initial seed loading will consume double or more amount of solute and therefore slow down the growth rate significantly. Secondary nucleation is very sensitive to the supersaturation profile (with the used kinetics) and so to the imposed temperature profile and the seed loading. The degree of plug flow has very little effect on secondary nucleation because the supersaturation profile is hardly influenced upon variation. A temperature profile optimization showed that in order to minimize secondary nucleation using a five zone temperature control, both a constant zone temperature and a linear zone temperature approach could lower the secondary nucleation by a factor of 2 compared to a single linear zone temperature approach. This underlines the importance of optimization in the OFBC.

Subject

oscillatory flow
baffles
residence time distribution
cooling crystallization
modeling
tanks-in-series
high resolution finite volume scheme
nucleation
growth

To reference this document use:

http://resolver.tudelft.nl/uuid:3786e0d9-9b04-412a-881b-0ea4d0eb05e7

Embargo date

2016-03-15

Part of collection

Student theses

Document type

master thesis

Rights

Files

PDF

Cooling_Crystallization_i ... roduct.pdf

3.11 MB

Close viewer