The behaviour of liquid fluidised beds following stepwise changes in flowrate

The behaviour of liquid fluidised beds following a stepwise increase in fluidising velocities has been studied. It is shown that the interface formed between two regions of different voidage fraction is unstable. Theory developed by Layzer[1] for Taylor instabilities which lead to slugging adequately describes the behaviour and development of these instabilities, provided allowance is made for the effects of density difference and geometry.     

Resonant and non-linear behaviour in vibrationally fluidised beds

This paper presents an experimental examination of the behaviour of vibrationally fluidised granular materials. This work was principally focused on examining the resonant behaviour of these systems. The harmonic response of a column of various granular materials (Geldart type A and B) was measured using a small apparatus. These materials displayed several non-linear effects such as; a normalised response that was dependent on excitation level, multiple harmonic components in the response to a single excitation frequency, and odd behaviour in terms of the motion of the emulsion.Two resonant peaks were found for each material. The relative magnitudes of these two peaks were heavily dependent on excitation level, so much so that at high levels of excitation only the higher resonant frequency was present. In each case, some correlation was found between this resonant peak and a prediction based on a “granular-gas” estimate of the speed of sound in the fluidised emulsion. This correlation was further examined by studying the resonant of the system under various levels of partial vacuum. The prediction followed the trend correctly.     

Heat transfer from immersed surfaces in liquid fluidised beds

Heat transfer data were obtained using a small electrically-heated simulated plate immersed in a liquid—fluidised bed. For steel balls and for spherical and cylindrical particles of glass fluidised by dimethyl phthalate, two correlations valid over the voidage range 0.40–0.85 were obtained; one was based on the Reynolds number (Re₀), involving the particle diameter and its terminal falling velocity, and the other on a modified Reynolds number (Re_m) incorporating the interstitial velocity and the volume: surface ratio of the voids.

and

In the above eqns, St is the Stanton number, J_h the Colburn heat transfer factor and e the voidage. K is a constant depending on the properties of the solid particles.     

A comprehensive interpretation of solid layer inversion in liquid fluidised beds

Solid mixing and segregation in liquid fluidised beds containing binary mixtures of spherical particles of different density and size has been studied for a range of liquid velocities, bulk bed compositions and particle properties. It was shown that a bed of denser particles expands with liquid velocity independently of the presence of the lighter particles. When the bulk volume fraction of the lighter particles is high and the liquid velocity is relatively low, the bed forms two layers, i.e. the upper layer consisting almost entirely of the lighter and the lower mixed layer consisting of both components in which the volume of the lighter increases with liquid velocity. A completely mixed bed is obtained at a certain velocity and then a further increase of the velocity causes “layer inversion”. The liquid velocity at which complete mixing occurs depends on the bulk bed composition, and at that velocity the volume fraction of the lighter in the lower mixed layer is constant regardless of the bulk bed composition. It is shown that layer inversion occurs for a given particle mixture when the liquid velocity passes through a value at which the volume fraction of the lighter in the lower layer becomes equal to the bulk bed composition; or for a given velocity, when the bulk bed composition becomes equal to the fraction of the lighter component which exists in the lower layer. The dependency of the fraction on the liquid velocity and the particle properties is examined to some extent.     

Gas distributor in fluidised beds

Satisfactory performance of fluidised bed reactors in terms of bed characteristics, chemical conversion and power requirement for pumping the fluidising medium is influenced by the design features of gas distributors. In industrial practice multi-orifice types of distributors are frequently used.This paper presents the results of experiments carried out in a 100-mm-diam. glass column using multi-orifice plate distributors. The effect of bed height and bed materials on the number of operating orifices is reported. The number of operating orifices is found to be a function of gas velocity and the ratio of pressure drops across the distributor and the bed.     

Mechanism of elutriation from fluidised beds

Elutriation can be used as a convenient method for classification of particles. However, it has been observed that complete elutriation is not possible. Elutriation stops when the concentration of the elutriating component in the bed reaches an equilibrium value. The rate equation for elutriation has been found to be analogous to that of a first order reversible chemical reaction. In this paper two correlations—one for evaluating the equilibrium bed concentration and the other for the elutriation rate constant have been proposed by relating these quantities with the relevant system parameters.     

The transition from the fixed to the fluidised state of binary-solid liquid beds

A peculiar phenomenon is observed when a binary-solid bed, in which the particle species possessing the larger minimum fluidization velocity is placed initially on top of the other, is brought from the fixed to the fluidised state. Contrary to the well known behaviour of mono-component beds, a single minimum fluidization velocity is not observed in this case and, more importantly, the fluid pressure drop can turn out to be considerably greater than the total effective weight of the particles. Particle–wall interactions are shown to be responsible for this latter effect. The problem is approached by considering differential horizontal slices of the bed with the simplification suggested by Janssen in 1895 that the ratio of vertical to horizontal solid pressure in a slice is constant—the Janssen constant K. Predictions of the model, which contains no adjustable parameters, are shown to be in excellent agreement with experimental measurements.     

Prediction of bubble behaviour in fluidised beds based on solid motion and flow structure

It has been demonstrated that the non-intrusive positron emission particle tracking (PEPT) could be a potential technique for observing bubble flow pattern, measuring bubble size and rise velocity in bubbling fluidised beds according to the solid motion in bubble and its wake. The results indicate that the behaviour of air bubbles varies greatly with the bed materials and superficial gas velocity. Three types of bubbling patterns (namely A, B and C) have been reported in this study, in which the pattern C is observed when the polyethylene fluidised bed is operated at the superficial gas velocity (U − U_mf) of 0.25–0.5 m/s and the ratio of bed height to bed diameter is unity. After the comparison of the results measured by the PEPT technique with the values calculated by using a number of empirical correlations, two modified correlations are recommended to calculate the bubble size based on the PEPT data.     

Particle-to-liquid mass transfer in fluidised beds

Nelson and Galloway's theory of particle to fluid mass transfer in dense systems of fine particles is re-examined and modified slightly to make it applicable to liquid fluidised beds. The resulting expression agrees with published data although these are inadequate to test it critically.     

Imposing dynamic structures on fluidised beds

Structuring fluidised beds can increase the conversion and selectivity, and facilitate control and scale-up. Two methods for introducing a dynamic structure into gas–solid fluidised beds are compared based on their overall hydrodynamics: electric field enhanced fluidisation and distributed secondary gas injection by a fractal injector. It is shown that, under various conditions, these systems lead to significant decreases in bubble size and bubble hold-up and to an increase in the number of bubbles. It was found that the electric field enhancement can lead to homogeneous fluidisation at lower flow rates, and the distributed secondary flow leads to smaller bubbles at higher flow rates.     

Measurement and simulation of bubbling fluidised beds

The effective control of systems requires the formulation of suitably robust models of their behaviour. The work described in this paper describes the simulation and modelling of the behaviour of a bubbling fluidised bed. A simple system is investigated consisting of a vertical planar bed. The performance of the bed is characterised by measuring the proportion of the bed occupied by the voids associated with bubbles. From these measurements it is possible to evaluate the response of the bed to changes in the gas flow rate into it in the time domain and through transformation into the frequency domain. These techniques allow a simulation of the bed based on the work of Clift and Grace [R. Clift, J. Grace, Coalescence of bubbles in fluidised beds, A.I.Ch.E. Symp. Ser. 67 (116) (1970) 23–33.] to be validated. The simulation can then be used to evaluate a simple but effective physical model of a bubbling fluidised bed which treats it as being primarily a temporary store of gas. The model represents the dynamics of the bed well and in the form of a transfer function which can be used successfully as a basis for controlling the bed.     

Particle-fluid heat transfer and dispersion in fluidised beds

The dynamic response of a gas fluidised bed has been measured for a range of particle sizes of lead glass ballotini and a range of particle Reynolds numbers. A dispersion model has been formulated that includes the effects of gas and particle mixing, fluid-to-particle heat transfer and intraparticle thermal conductivity, and the dynamic thermal response in theory has been found by solving the partial differential equations in the Laplace transform domain. The coefficient of thermal dispersion, the particle-to-fluid heat transfer coefficient and the intraparticle thermal conductivity have been found for the experimental response by non-linear regression. The coefficient of axial dispersion was found to be large and the particle to fluid heat transfer coefficients agreed with an established correlation for fixed and fluidised beds. The intraparticle thermal conductivity agreed with literature values for lead glass, the estimates showed no trend with flowrate, and the standard deviation of the estimate was three times smaller than the deviation found from similar experiments in fixed beds.     

Particle residence time distributions in circulating fluidised beds

This paper gives experimental measurements of the particle residence time distribution (RTD) made in the riser of a square cross section, cold model, circulating fluidised bed, using the fast response particle RTD technique developed by Harris et al. (Chem. Eng. J. 89 (2002a) 127). This technique depends upon all particles having phosphorescent properties. A small proportion of the particles become tracers when activated by a flash of light at the riser entry; the concentration of these phosphorescent particles can subsequently be detected by a photomultiplier. The influence of the solids circulation rate and superficial gas velocity on the RTD were investigated. The results presented are novel because (i) the experiments were performed in a system with closed boundaries and hence give the true residence time distribution in the riser and (ii) the measurement of the tracer concentration is exceedingly fast. The majority of previous studies have measured the RTD in risers with open boundaries, giving an erroneous measure of the RTD.Analysis of the results suggests that using pressure measurements in a riser to infer the solids inventory leads to erroneous estimates of the mean residence time. In particular, the results cast doubt on the assumption that friction and acceleration effects can be neglected when inferring the axial solids concentration profile from riser pressure measurements.An assessment of particle RTD models is also given. A stochastic particle RTD model was coupled to a riser hydrodynamic model incorporating the four main hydrodynamic regions observed in a fast-fluidised bed riser namely (i) the entrance region, (ii) a transition region, (iii) a core-annulus region and (iv) an exit region. This model successfully predicts the experimental residence time distributions.