Multiphase Reactors: Models and Experimental Verification

Abstract
Identification and quantification of flow regimes, phase holdup distributions, flow patterns and backmixing is essential for proper design and scale-up of multiphase reactors. Existing models often suffer from inadequate experimental confirmation. Here, we describe recent progress made in our laboratory in characterizing liquid circulation and backmixing in bubble columns and in quantifying, via a phenomenological model, the behavior of trickle-beds in the low interaction regime. The need for imaging techniques such as fan-beam tomography and PET is illustrated. Liquid mixing in bubble columns caused by global liquid convection and by turbulent eddies is usually described by the axial dispersion model (ADM) which, at least in the churn turbulent regime, lacks a theoretical basis. In the ADM the two mixing mechanisms are lumped into a single effective dispersion coefficient. An alternative backmixing description assumes multiple liquid circulation cells, with height equal to column diameter, to occur in the column. We have successfully demonstrated that instantaneous and time averaged liquid velocity profiles in the column can be determined by monitoring the motion of a neutrally bouyant tracer particle via a Computer Aided Radioactive Particle Tracking (CARPT) technique. It can now be shown that a single strong liquid circulation cell exists in columns of a variety of diameters and at different operating conditions but that smaller circulation cells can form in the distributor region. Model predicted one dimensional axial time averaged liquid velocity profile agrees well with the data and applies to the middle section of the column. The CARPT technique allows the computation of Lagrangian autocorrelation coefficients, the rms fluctuating velocities, Lagrangian integral time scales and the turbulent dispersion coefficients. Thus CARPT provides the necessary information i. e. velocity profiles and turbulent dispersion coefficients, for implementation of a convection dispersion model. The planned implementation of fan-beam tomography would quantify the two-dimensional holdup distribution and together with CARPT provide all the data necessary for verification of fundamental two phase flow models. Pressure drop, holdup, flow regime transition and phase distribution in trickle-beds has received considerable attention but no generally accepted theory for prediction of these quantities exists. Here we describe a phenomenological model for the low interaction regime which views the bed as an array of slits with liquid film flow. Universal velocity profile is used to describe the flow in both liquid films and the gas core and a large data bank for pressure drop and holdup was used to confirm that in the low interaction regime each of the two phases ignores the presence of the other one. The final model contains only two parameters, which are determined from single phase flow experiments, and contains no constants fitted to two phase flow data. The model predicts pressure drop and holdup in the uniform low interaction regime better than any of the existing models. Furthermore, introducing the Kapitza's criterion for laminar liquid film instability into the model the flow regime transition to pulsing is predicted well for all the data that satisfy the conditions for the theory to apply. Other pore level transition mechanisms are suspected for other data. The above phenomenological model is also used as a basis for predicting liquid distribution in a cell model of a trickle-bed. Experimental confirmation awaits the application of PET technology. L'identification et la quantification des régimes, des distributions des rétentions de phase, des schémas de flux et du mélange en retour sont d'une importance primordiale pour extrapoler à l'échelle et concevoir correctement les réacteurs polyphasés. Les modèles existants sont, malheureusement, bien souvent étayés par des résultats expérimentaux inadéquats. Nous décrivons ici les progrès récents accomplis dans notre laboratoire dans la caractérisation de la circulation et du mélange en retour de liquide dans les colonnes à bulles et dans la quantification, à l'aide d'un modèle phénoménologique, du comportement des ruissellements cocourants du régime de faible interaction. On trouvera une illustration de la nécessité de faire appel à des techniques d'imagerie telles que la tomographie et la technique PET. Le mélange des liquides dans les colonnes à bulles provoqué par la convection liquide globale et par des tourbillons est habituellement décrit par le modèle de dispersion axiale (ADM) qui, du moins pour le régime de turbulence, est dépourvu de base théorique. Dans ce modèle, les deux mécanismes de mélange sont associés par un coefficient de dispersion unique. Dans une autre description de mélange en retour, on suppose que des cellules de circulation de liquide, de hauteur égale au diamètre de la colonne, se produisent dans la colonne. Nous avons réussi à démontrer que les courbes de vitesse instantanée et moyenne de propagation de liquide peuvent être déterminées en pilotant le déplacement d'une particule de traceur par la technique du marquage radioactif des particules assisté par ordinateur (en anglais, technique CARPT). On peut maintenant démontrer qu'une cellule unique de circulation forte de liquide existe dans des colonnes de diamètres variés et à différentes conditions d'exploitation, mais que des cellules de circulation plus petites peuvent se former dans la région de distribution. La courbe de vitesse moyenne de propagation de liquide est en...