Why to use packed bed micro-reactors for catalyst testing? Miniaturized packed bed reactors have a large surface-to-volume ratio at the reactor and particle level that favors the heat- and mass-transfer processes at all scales (intra-particle, inter-phase and inter-particle or reactor level). If the mass-transfer processes are fast in respect to the reaction-rate, then the reaction-rate is under kinetic control over the entire range of conversion and it is possible to measure intrinsic kinetics, from differential to integral conditions, eliminating extrapolation and improving reliability. For reaction mixtures where the catalyst-deactivation phenomena should be included in the reaction-rate expression to predict the reactor performance, these continuous reactors allow the measurement of intrinsic kinetics and, independently, catalyst deactivation over long periods of time. When using several reactors in parallel, as in parallel multi-flow experimentation units, a fast evaluation of the reaction-rate dependency on the process variables can be performed. Why to study the hydrodynamics of multi-phase packed bed micro-reactors? Established correlations to evaluate hydrodynamic parameters are based on large-scale trickle beds (Re>1, dparticle >1mm), where inertia and gravity play a prominent role. For multi-phase packed bed reactors the Reynolds number Re is calculated based on the liquid properties and the particle diameter Re= ?LuLdp/?L where ?L is the liquid density, uL is the liquid velocity, dp is the particle diameter, and ?L is the liquid viscosity. There are important differences in flow behavior upon downscaling. When using small volumes and flow velocities (Re<1, dparticle ~ 100?m) capillary forces become dominant. Therefore, back mixing, hold-up, and residence time can be very different in micro-packed beds as compared with larger systems, and the large-scale trickle bed correlations should not be used for micro-packed beds. From an equipment design perspective it is vital to determine the extent of the reactant dispersion, especially when we aim high conversion in small reactors. Moreover, for gas-liquid-solid reactions, it is instructive to know the difference between the residence time of the gas and that of the liquid, because stripping of reactants and products can significantly impact conversion levels and mask the proper interpretation of kinetic experiments. Residence time distribution experiments, multi-phase flow pattern visualizations and mathematical models were used to gain insight in the intricate interplay of diffusion, gas-liquid interaction, and convection for the surface tension-dominated low Reynolds-number flows in these miniaturized packed bed systems. What we learn trough the Residence Time Distribution (RTD) experiments? In chapter IV we combined liquid phase RTD studies with pressure drop measurements to determine the influence of the flow history of the bed on the hydrodynamic stability. We also studied how much time is required to reach a hydrodynamic steady state at start-up of after a step-change in flow rate. We found that miniaturized multi-phase packed bed reactors have very high liquid hold-up values (always above 0.65, per unit void volume, for our experimental conditions) and, therefore, more stable hydrodynamic conditions when compared with trickle bed reactors. The little hysteresis, that we found, had a higher pressure drop in the branch of increasing flow of both gas and liquid. We observed start-up and switching times of the magnitude of 3-4 liquid-residence times. These long characteristic times can be attributed to a combined effect of compressibility in the gas-feed section and flow resistance in the packed bed itself. During this slow response toward a new steady state, the pressure (i.e., the dissolved gas concentration) and the residence time of liquid-phase components are not constant and kinetic data obtained during this slow response, which can take up to 12 h, cannot reliably be used in the evaluation of kinetic parameters. In chapter V we used liquid phase RTD studies to determine the axial dispersion behavior of the liquid phase in a multi-phase system. Using tracers of different molecular diffusivity we showed that this parameter has no effect on the liquid axial dispersion. Our RTD results indicated limited interaction between gas and liquid at low-Re multi-phase flow. We found that neither hold-up nor dispersion is strongly affected by the gas-flow rate. The dispersion in multi-phase miniature packed beds is very similar to dispersion in equivalent single-phase systems, with only a factor of 2 - 3 difference between multi-phase and single liquid-phase flow. The accurate dispersion data reported allow a good estimation of the bed length needed to achieve plug-flow behavior using Mears’ criteria. What was the insight we gain trough the multi-phase flow pattern visualization? In chapter III we described the co-current flow of gas and liquid through micro-fabricated beds with regular or distorted hexagonal pillar-array at Ca numbers (Ca ? ?LuL/?, ?L is the liquid viscosity, uL is the liquid velocity, and ? is the surface tension) in the order of 10-5 – 10-3 to get a better physical picture of the flow pattern in miniaturized packed beds of cylindrical geometry (2 mm inner diameter) packed with spherical non-porous particles (dparticle = 100 ?m). We observed that the surface tension-driven multi-phase flow patterns tend to show segregation of the phases with high liquid hold-up values (> 0.5), the gas-liquid interaction can be boosted by increasing flow-rates, the size of void-channels (inter-pillar or inter-particle space) has a major impact not only on the bed porosity but also on the distribution of the phases inside the bed, and changing from a full wetting condition to a partial wetting condition produces a significant change in flow pattern. It is still difficult to predict accurately the multi-phase flow pattern for micro-fabricated and miniaturized packed beds because many variables (Ca number, contact angle, pore-connectivity, and feed-section design) seem to be important. It is worthwhile to perform visualization or residence time distribution experiments in connection with hot flow experiments to establish the relation between bed packing characteristics and the gas-liquid-solid reaction performance as well as analyzing trends with the parameters. What taught us the mathematical modelling? In chapter VI we described the effect of volatility on residence time distribution and conversion in multiphase reactors. This is relevant for the many processes where substantial vaporization of the liquid feed occurs. The typical situation is that the evaporated molecules not only lower the concentration in the liquid phase but also travel faster through the reactor. This has direct importance in understanding multiphase reactors with light reactants and products (such as HDS), offering insight into practical issues such as bypassing and stripping of relevant compounds. Our complete model uses two mobile zones, one for the liquid phase and one for the gas phase, with dispersion in each zone and mutual mass transfer. We explored the entire parameter space for our model numerically. We learned that the relevant parameters are the velocities and volumetric fluxes of the phases, the thermodynamic equilibrium between the phases, the dispersion behavior of the individual phases and the mass transfer rate between the phases. We described quantitatively how the mean residence time of a component decreases when it significantly evaporates to a faster-moving gas phase. We explored how slow mass transfer contributes to the broadening of the residence time distribution. Experimentally, we validated the model in a more limited parameter space in a gas-liquid micro-packed bed with volatile and non-volatile compounds in different solvents. Additionally, using reactor simulations we developed a simple model to predict the conversion for systems with a volatile reactant at mass-transfer rate values relevant for micro-reactors and trickle-bed reactors. We used a liquid-only reaction to represent the case of a fully wetted heterogeneous catalyst or a homogeneous non-volatile catalyst in the liquid phase. Our results indicate that: for multi-phase continuous micro-reactors with high mass-transfer rates (St ? 100, St=(ka)glØL/uL, where (ka)gl is the gas-liquid mass-transfer coefficient, Ø is the liquid hold-up, L is the reactor length, and uL is the liquid velocity), the volatile compounds have a lower concentration in the reactive phase and the overall residence time for such compounds is lower. These two effects, lower concentration and shorter residence time, are related manifestations of the same phenomenon. Accounting for only the change of concentration reliably predicts the conversion for all reaction orders higher than zero. Liquid by-passing via the gas phase is not important since thermodynamic equilibrium is already reached in the top of the bed. For large scale trickle-beds the typical range of St is 5-50, at these intermediate mass-transfer rates the conversion is further reduced due to a significant evaporation of the reactant to the gas phase without re-condensation before leaving the reactor. At even lower mass-transfer rates (St ? 1), too few molecules evaporate before leaving the reactor and the effect of the volatility vanishes. Wherever possible, we extracted from the detailed numerical model practical engineering correlations for average residence time and conversion. The results presented in this work teach whether reactant volatility should be considered in a reactor design. Wrap up
The whole thesis work is summarized in more detail in chapter VII, where also the impact of the micro-reactor hydrodynamic characteristics on catalyst kinetic testing is discussed and recommendations for future work are given.