Investigation into the effect of simulating a 3D cylindrical fluidized bed reactor on a 2D plane

Abstract

Abstract2D planar simulations of 3D cylindrical fluidized bed reactors are routinely carried out in order to reduce computational costs. The error involved in this simplification is largely unknown, however, and this study was therefore conducted to quantify this error over a wide range of reactor operating conditions. 2D and 3D simulations were carried out over a wide range of flow conditions in the bubbling fluidization regime by changing the fluidization velocity, bed mass, reaction temperature and particle size. Detailed comparisons revealed that 2D simulations qualitatively behaved similarly to 3D simulations, but overpredicted reactor performance (measured by the degree of conversion achieved) by about 45% on average. Large systematic variations of this error were also observed with changes in all four independent variables investigated. These large errors were due to two primary factors; incorrect predictions of the gas residence time by misrepresentations of the bed expansion and incorrect predictions of the mass transfer by misrepresentations of bubble formation and the splash zone at the top of the expanded bed. The mass transfer error was found to be most influential and was also confirmed as the most important factor to be correctly predicted by CFD simulations. 3D predictions of the mass transfer resistance were further analysed to identify the particle size as a very influential variable through which the mass transfer characteristics in fluidized bed reactors can be influenced. IntroductionFluidized bed reactors find application in a wide range of process industries dealing with gas-solid or solid catalysed reactions. The fluid-like behaviour of these reactors results in excellent heat and mass transfer performance which is highly favourable for any reactor process. Fluidized bed reactors can be very challenging to design and scale up, however, and commercialisation of processes utilising such reactors therefore typically consists of a large number of incremental scale-up and demonstration steps. For this reason, there is a great need for modelling tools that can be used to reduce the risk of taking larger scale-up steps.The fundamental flow modelling framework of computational fluid dynamics (CFD) is considered in this work to be a viable methodology for achieving this purpose. CFD inherently accounts for the complex and tightly interconnected non-linear interactions between reactor hydrodynamics, heat transfer, species transfer and reaction kinetics. Because of this fundamental basis, it is reasoned that CFD can provide the generality demanded of a modelling tool used for fluidized bed reactor design, optimization and scale-up.CFD simulations face their own challenges, however. The primary challenge is presented by the mesoscale structure formation characteristic of any fluidized bed reactor process. Small clusters of particles are formed within these beds due to the non-linear drag interaction between the gas and solids and have a very large influence on all physics occurring within the reactor...

Grid independence behaviour of fluidized bed reactor simulations using the Two Fluid Model: Effect of particle size

“…This approach has been confirmed to give adequate representations of the hydrodynamics of fluidized bed units [6,14,15], although it should be stated that the 2D approximations of 3D cylindrical beds will lead to systematic deviations within the parameter space of interest [16] and that the simulation becomes very sensitive to exact cluster resolution when fast reactions are simulated [17]. The complete equation system can be viewed in [1].…”

“…The degree of bed expansion will influence the gas residence time inside the bed and therefore also the reactor performance, but a previous study [16] showed that the mass transfer effect was much more influential than this residence time effect. For this reason, the reactor performance grid independence behaviour was even more stringent than that for phase segregation.…”

The effect of frictional pressure, geometry and wall friction on the modelling of a pseudo-2D bubbling fluidised bed reactor

“…The increase in reactor performance with particle size is due to improved bubble-toemulsion mass transfer. As discussed in more detail in a previous study [22], this is primarily due to the fact that an emulsion of coarser particles is much more penetrable to gaseous reactants than an emulsion of fine particles.…”

Use of α -shapes for the measurement of 3D bubbles in fluidized beds from two-fluid model simulations

“…A first order Backward Euler method was used for the discretization of the transient terms. The second order time discretization is usually preferred for accurate solution of fast-moving riser flows with the TFM but this is not the case for dense bubbling beds at low relative gas velocities, in which the majority of the bed moves relatively slowly [38,46]. The numerical discretization of the system variables was done with an upwind scheme with convergence criteria based on a RMS residual target of 10 -3 , as recommended from previous works [24].…”

“…The model closures and parameter values were adapted to the pseudo-2D system and might not give physical realistic predictions of a 3D TS-TZFBR bed, as already observed by some authors [38,39]. The qualitative comparison of the model predictions on the bubbling characteristics performed in this work for both bed configurations aims to bring some light on the extrapolation of pseudo-2D bubble data to the real 3D column bubbling.…”