An algorithm for predicting the hydrodynamic and mass transfer parameters in bubble column and slurry bubble column reactors

Abstract

A large number of experimental data points obtained in our laboratory as well as from the literature, covering wide ranges of reactor geometry (column diameter, gas distributor type/open area), physicochemical properties (liquid and gas densities and molecular weights, liquid viscosity and surface tension, gas diffusivity, solid particles size/density), and operating variables (superficial gas velocity, temperature and pressure, solid loading, impurities concentration, mixtures) were used to develop empirical as well as Back-Propagation Neural Network (BPNN) correlations in order to predict the hydrodynamic and mass transfer parameters in bubble column reactors (BCRs) and slurry bubble column reactors (SBCRs). The empirical and BPNN correlations developed were incorporated in an algorithm for predicting gas holdups (εG, εG-Small, εG-Large); volumetric liquid-side mass transfer coefficients (kLa, kLa-Small, kLa-Large); Sauter mean bubble diameters (dS, dS-Small, dS-Large); gas–liquid interfacial areas (a, aSmall, aLarge); and liquid-side mass transfer coefficients (kL, kL-Large, kL-Small) for total, small and large gas bubbles in BCRs and SBCRs.The developed algorithm was used to predict the effects of reactor diameter and solid (alumina) loading on the hydrodynamic and mass transfer parameters in the Fisher–Tropsch (F–T) synthesis for the hydrogenation of carbon monoxide in a SBCR, and to predict the effects of presence of organic impurities (which decrease the liquid surface tension) and air superficial mass velocity in the Loprox process for the wet air oxidation of organic pollutants in a BCR. In the F–T process, the predictions showed that increasing the reactor diameter from 0.1 to 7.0 m and/or increasing the alumina loading from 25 to 50 wt.% significantly decreased εG, kLaH2 and kLaCO and increased dS. The decrease of the total gas holdup was found to be controlled by the holdup of small gas bubbles. The increase of the Sauter mean bubble diameter increased both kLH2 and kLCO, however, the decrease of the total gas holdup coupled with the increase of dS resulted in a dramatic decrease of the gas–liquid interfacial area, a, and subsequently kLaH2 and kLaCO. Thus, in the churn-turbulent flow regime, the hydrodynamic and mass transfer behaviors of the F–T SBCR were controlled by the holdup and the gas–liquid interfacial area of small bubbles. In the Loprox process, the predictions showed that increasing the liquid surface tension (removal of organic impurities from water) significantly increased dS and decreased both εG and kLaO2. The decrease of the total gas holdup with increasing liquid-phase surface tension was due mainly to the decrease of the liquid-phase foamability which led to the decrease of the holdup of small gas bubbles. The increase of the Sauter mean bubble diameter and the decrease of the total gas holdup resulted in a strong decrease of the gas–liquid interfacial area, and subsequently kLaO2. Increasing the air superficial mass velocity increased εG, dS, a, kL-O2 and kLaO2. Within the conditions used in the Loprox BCR, the hydrodynamics and mass transfer parameter behaviors of the process appeared also to be controlled by the gas holdup of small gas bubbles; and the gas–liquid interfacial area.