Joachim Werther

For better understanding the process of particle coating by chemical vapor deposition (CVD) in the fluidized bed, the simulation of the deposition process was combined with a discrete particle model (DPM). Based on the experimental results of the thermal decomposition of tri-isobutyl-aluminum (TIBA) to produce aluminum onto glass beads, mechanisms on the micro-scale were investigated by single particle tracking. Zones of excessive growth as well as zones of insufficient mixing were identified. In particular, the take-up of aluminum was traced for selected particles that exhibited a large mass of deposited aluminum what in turn provides insight into the homogeneity and quantity of the coating throughout the bed material.

The mechanism of liquid injection into fluidized bed reactors was investigated by injecting water and ethanol, respectively, into a pilot‐scale bubbling fluidized bed (cross section: 1 × 0.5 m) that was kept at temperatures between 120 and 180°C. Quartz sand and FCC catalyst were used as bed materials. The injected liquid was found to form agglomerates with the bed particles at the nozzle exit and become transported into the bed interior by mixing of the large‐scale solids. There the liquids evaporate from the surface of the particles. © 2005 American Institute of Chemical Engineers AIChE J, 51: 766–775, 2005

A three-dimensional model has been developed to describe the injection of liquid reactants into fluidized bed reactors operating in the bubbling fluidized bed regime. The model considers the processes of liquid transport and evaporation in the vicinity of the point of injection. The underlying idea, which is supported by previous measurements, is that the particles in the dense suspension phase are wetted by the liquid or gas-liquid spray. The wetted particles are subsequently dried while they are following the gross solids circulation within the bed. The model considers the flow structure of the bubbling fluidized bed and the solids mixing with the aid of a hybrid model which combines semi-empirical models for bubble growth by coalescence and for bubble splitting with a CFD approach for the continuous emulsion phase surrounding the bubbles. Submodels for heat and mass transfer are used to describe the temperature and concentration fields in the vicinity of the injection nozzle and the drying process of the wetted particles with the resulting release of the vaporized injection liquid. The model was validated separately against flow structure measurements, solids tracer measurements and experiments with the injection of water and ethanol, respectively, into beds of FCC particles.

In a circulating fluidized bed (CFB) combustor the reacting solids are locally fed into the combustion chamber. These reactants have to be dispersed across the reactor's cross-sectional area. Since the rate of mixing is limited this leads to a mal-distribution of the reactants and to locally varying reaction conditions. In order to describe the influence of mixing a three-dimensional model of the combustion chamber is suggested. The model is divided into three sub-topics. First, the flow structure in terms of local gas and solids velocities and solids volume concentrations is described. Second, mixing of the solids and the gas phase is quantified by defining dispersion coefficients, and finally the combustion process itself, i.e. the reaction kinetics, is modelled. The model was validated against data from measurements in the large-scale combustor of Chalmers University of Technology in Göteborg/Sweden. Insufficient fuel mixing generated mal-distributions of locally released volatiles, which were the basis for the uneven reactants distribution at steady-state. In the case of two-stage operation, the injected secondary air did not reach immediately the reactor's center but was slowly mixed with the main gas flow. The concentration gradients hardly vanish before the exit of the combustion chamber.

The fluidized bed reactor has a lot of advantages: excellent gas-solid contacting, no hot spots even with highly exothermal reactions, good gas-to-particle and bed-to-wall heat transfer and the ease of solids handling which is particularly important if the catalyst is quickly ageing. However, the list of disadvantages is as long: broad residence time distribution of the gas due to dispersion and gas-bypass in the form of bubbles, broad residence time distribution of solids due to intense mixing, erosion of bed internals and the attrition of the catalyst particles. A particular disadvantage of the fluidized bed reactor is its difficult scale-up. The historical experience with the FCC process is that in the early 40's of the last century this process was successfully scaled up from a 5 cm dia. pilot-scale unit to a 4.5 m dia. bed in the production unit. On the other hand, around 1950 the scale up of the Fischer-Tropsch synthesis in the fluidized bed failed completely. Modern process design should be able to avoid such disasters by making use of modeling and simulation tools. However, a modeling tool which is really helpful in planning and designing of an industrial fluidized bed reactor has to fulfill a lot of requirements. It should be able to describe the influence of the several changes which are typical for the scale-up process, for example enlargement of bed diameter, bed height and fluidizing velocity, changes of gas distributor design, introduction of in-bed heat exchanger tubes and baffles. In the present work a modelling approach is presented which is able to handle the most important aspects of industrial fluidized bed reactors. A particular focus is to describe the relationship between catalyst attrition, solids recovery in the reactor system and chemical performance of the fluidized bed reactor. The competing influences of attrition of the catalyst particles and efficiency of the solids recovery lead to the establishment of a catalyst particle size distribution (PSD) in the bed inventory which in turn influences via the hydrodynamic characteristics of the fluidized bed the performance of the chemical reactor. The usefulness of this approach is illustrated with model calculations for a fictituous first order reaction where the fluidized bed is equipped with different solids recovery systems including one single stage cyclone, several cyclones in parallel, two- and three-stage cyclone systems, respectively. Model calculations illustrate the significance of a high efficiency of the solids recovery in order to keep the fines in the system which is decisive for a high performance of the reactor. The calculations reveal that it may take months until a quasi steady state of the bed particle size distribution is obtained.