Modelling of trickle bed reactors with small packings: hydrodynamics, residence time distribution and reactor design

Abstract

This thesis is based on an experimental and theoretical study of three important aspects of trickle bed reactors, namely its hydrodynamic and dispersion effects and their contributions to a theoretically based reactor design procedure.The study formed part of a research program directed at the application of the trickle bed reactor to an immobilised glucose oxidase system. As it will be necessary to minimise diffusional limitations, the enzyme must be immobilised on small packings; hence packings ranging in size from 0.5 to 1.8 mm have been considered in the current work.Experimental measurements of holdup and pressure drop were carried out for cocurrent two-phase downflows through the selected packings using an air-water system. Existing correlations for larger gas absorption packings were tested and generally found to hold, provided modifications were made to the relevant constants. Satisfactory correlations were obtained using the method of Turpin and Huntington and Specchia and Baldi.A multiplicity of hydrodynamic states was observed, a phenomenon not previously reported in the literature. These were characterised by different holdups and pressure drops which depended on the maximum gas flowrate experienced by the bed. The higher the level of the maximum gas flowrate, the lower was the pressure drop and the greater was the total liquid holdup. However, it was noteworthy that the same relationship between the total holdup and pressure drop existed regardless of the hydrodynamic state the bed was in. A Channelling Flow model was proposed to explain the results in terms of varying gas flowpath tortuosity.A residence time model for the liquid phase was developed in which the liquid holdup was divided into a dynamic and a stagnant region. Dispersion was attributed mainly to mass transfer between the dynamic and the stagnant regions, and partly to backmixing in the bulk flowing liquid in the dynamic region. A simpler Ideal Plug Flow Stagnancy model neglects the dispersion due to backmixing in the dynamic region, leaving only three parameters which are the total holdup, the dynamic fraction of it and a dimensionless group termed the Stagnancy number. The relationships between the dynamic fraction of the total holdup and the Stagnancy number to the fluid Reynolds numbers have been interpreted as confirming the theory of rivulet flow of the liquid phase and the mechanism of turbulent backmixing as the mode of mass transfer between the dynamic and the stagnant regions.The Stagnancy model was extended to a system with chemical reactions of simple kinetics. The model predicts that the reactor conversion should increase with holdup; from observations on holdups in various hydrodynamic states, it was concluded that for a given gas flowrate, a higher conversion can always be attained by operating the trickle bed reactor in a condition where the gas flow has been reduced from a higher rate.