Scale-Up of the agitator speed based on computational fluid dynamics

Abstract

Siegfried in Zofingen is currently working on a hydrogenation reaction. To mimic the production condition for the production, the reaction is carried out on a scale-down reactor laboratory scale (approx. 1 liter). One important process parameter that should be similar during the experiments of the scale-down reactor compared to the later production scale reactor is the mixing behaviour of the agitator. The goal of this work is to find a concept for the appropriate scale-down of the agitator speed based on computational fluid dynamics. This concept shall be compared to conventional scale up methods. To validate this concept different experiment with water and glycerol, as well with and without baffles, were carried out in a 5-litre scale reactor. For the simulation without baffles, the size of the vortex has been compared to the real experiment as well as conventional calculation. In general, the simulation predicted only a vortex 63.6 % of the measured size. In further experiments with suspending of catalyst particles, simulation with baffles were validated with catalyst particle behaviour. The particle behaviour in the experiments could not be measured accurately due to the small size of the particles and the insufficient camera quality. The simulation with particles had a critical error where particles got stuck on the wall which made the results unusable for particle movement prediction. Even with the simulation working correctly, it would not be beneficial for this project, due to the large amount of particles and the simplification done for the simulation. A second method was uses, where the mean velocity of cells in the steady state were compared to each other. Out of 3 different agitator speeds, the agitator speed calculated with keeping P/V constant (571 rpm) was most similar to the agitator speed of the reactor MZA1 (120 rpm). Due to the low sample size this only supports the result but does not confirm it. As a consequence, the results of this thesis give an indication for a possible solution to the stirring behaviour but conventional ways of finding the agitator speed are still necessary. It can be used as steppingstone, but it is still required to refine these simulations. Despite the dissatisfying result with the catalyst particles it could be shown that scale down of agitator speed can still be supported by the mean of computational fluid dynamics