Application of X-ray microtomography to numerical simulations of agglomerate breakage by distinct element method

Computer-aided tomography has been used for many years to provide significant information about the internal properties of an object, particularly in the medical fraternity. By reconstructing one-dimensional (ID) X-ray images, 2D cross-sections and 3D renders can provide a wealth of information about an object's internal structure. An extension of the methodology is reported here to enable the characterization of a model agglomerate structure. It is demonstrated that methods based on X-ray microtomography offer considerable potential in the validation and utilization of distinct element method simulations also examined.

The effect of the binder size and viscosity on agglomerate growth in fluidised hot melt granulation

Fluidised hot melt granulation (FHMG) is a novel granulation technique for processing pharmaceutical powders. Several process and formulation parameters have been shown to significantly influence granulation characteristics within FHMG. In this study we have investigated the effect of the binder properties (binder particle size and binder viscosity) on agglomerate growth mechanisms within FHMG. Low-melting point co-polymers of polyoxyethylene–polyoxypropylene (Lutrol® F68 Poloxamer 188 and Lutrol® F127 Poloxamer 407) were used as meltable binders for FHMG, while standard ballotini beads were used as model fillers for this process. Standard sieve analysis was used to determine the size distribution of granules whereas we utilised fluorescence microscopy to investigate the distribution of binder within granules. This provided further insight into the growth mechanisms during FHMG. Binder particle size and viscosity were found to affect the onset time of granulation. Agglomerate growth achieved equilibrium within short time-scales and was shown to proceed by two competing processes, breakage of formed granules and re-agglomeration of fractured granules. Breakage was affected by the initial material properties (binder size and viscosity). When using binder with a small particle size (<250 µm), agglomerate growth via a distribution mechanism dominated. Increasing the binder particle size shifted the granulation mechanism such that agglomerates were formed predominantly via immersion. A critical ratio between binder diameter and filler has been calculated and this value may be useful for predicting or controlling granulation growth processes.

Effect of granule morphology on breakage behaviour during compression

In the area of dry particle breakage, Discrete Element Method (DEM) simulations have been widely used to analyse the sensitivity of various physical parameters to the behaviour of agglomerates during breakage. This paper looks at the effect of agglomerate shape and structure on the mechanisms and extent of breakage of dry agglomerates under compressive load using DEM simulations. In the simulations, a spherical-shaped agglomerate produced within the DEM code is compared with an irregularly shaped agglomerate, whose structure is that of an actual granule that was characterised with X-ray microtomography (muCT). Both agglomerates have identical particle size distribution, coordination number and surface energy values, with only the agglomerate shape and structure differing between the two. The work here details the breakage behaviour with a number of traditional DEM output parameters (i.e., contact/cluster distributions) with showing vastly different behaviour between the two agglomerates. (C) 2004 Elsevier B.V. All rights reserved.

Fibre agglomerate transport in a horizontal flow

An experimental and theoretical study of the transport of mineral wool fibre agglomerates in nuclear power plant containment sumps is being performed. A racetrack channel was devised to provide data for the validation of numerical models, which are intended to model the transport of fibre agglomerates. The racetrack channel provides near uniform and steady conditions that lead to either the sedimentation or suspension of the agglomerates. Various experimental techniques were used to determine the velocity conditions and the distribution of the fibre agglomerates in the channel. The fibre agglomerates are modelled as fluid particles in the Eulerian reference frame. Simulations of pure sedimentation of a known mass and volume of agglomerations show that the transport of the fibre agglomerates can be replicated. The suspension of the fibres is also replicated in the simulations; however, the definition of the fibre agglomerate phase is strongly dependent on the selected density and diameter. Detailed information on the morphology of the fibre agglomerates is lacking for the suspension conditions, as the fibre agglomerates may undergo breakage and erosion. Therefore, ongoing work, which is described here, is being pursued to improve the experimental characterisation of the suspended transport of the fibre agglomerates.

Micromechanics of agglomerate damage processes

This thesis reports a detailed investigation of the micromechanics of agglomerate behaviour under free-fall impact, double (punch) impact and diametrical compression tests using the simulation software TRUBAL. The software is based on the discrete element method (DEM) which incorporates the Newtonian equations of motion and contact mechanics theory to model the interparticle interactions. Four agglomerates have been used: three dense (differing in interface energy and contact density) and one loose. Although the simulated agglomerates are relatively coarse-grained, the results obtained are in good agreement with laboratory test results reported in the literature. The computer simulation results show that, in all three types of test, the loose agglomerate cannot fracture as it is unable to store sufficient elastic energy. Instead, it becomes flattened for low loading-rates and shattered or crushed at higher loading-rates. In impact tests, the dense agglomerates experience only local damage at low impact velocities. Semi-brittle fracture and fragmentation are produced over a range of higher impact velocities and at very high impact velocities shattering occurs. The dense agglomerates fracture in two or three large fragments in the diametrical compression tests. Local damage at the agglomerate-platen interface always occurs prior to fracture and consists of local bond breakage (microcrack formation) and local dislocations (compaction). The fracture process is dynamic and much more complex than that suggested by continuum fracture mechanics theory. Cracks are always initiated from the contact zones and propagate towards the agglomerate centre. Fracture occurs a short time after the start of unloading when a fracture crack "selection" process takes place. The detailed investigation of the agglomerate damage processes includes an examination of the evolution of the fracture surface. Detailed comparisons of the behaviour of the same agglomerate in all three types of test are presented. The particle size distribution curves of the debris are also examined, for both free-fall and double impact tests.

Numerical modelling of agglomerate degradation

In the processing industries particulate materials are often in the form of powders which themselves are agglomerations of much smaller sized particles. During powder processing operations agglomerate degradation occurs primarily as a result of collisions between agglomerates and between agglomerates and the process equipment. Due to the small size of the agglomerates and the very short duration of the collisions it is currently not possible to obtain sufficiently detailed quantitative information from real experiments to provide a sound theoretically based strategy for designing particles to prevent or guarantee breakage. However, with the aid of computer simulated experiments, the micro-examination of these short duration dynamic events is made possible. This thesis presents the results of computer simulated experiments on a 2D monodisperse agglomerate in which the algorithms used to model the particle-particle interactions have been derived from contact mechanics theories and, necessarily, incorporate contact adhesion. A detailed description of the theoretical background is included in the thesis. The results of the agglomerate impact simulations show three types of behaviour depending on whether the initial impact velocity is high, moderate or low. It is demonstrated that high velocity impacts produce extensive plastic deformation which leads to subsequent shattering of the agglomerate. At moderate impact velocities semi-brittle fracture is observed and there is a threshold velocity below which the agglomerate bounces off the wall with little or no visible damage. The micromechanical processes controlling these different types of behaviour are discussed and illustrated by computer graphics. Further work is reported to demonstrate the effect of impact velocity and bond strength on the damage produced. Empirical relationships between impact velocity, bond strength and damage are presented and their relevance to attrition and comminution is discussed. The particle size distribution curves resulting from the agglomerate impacts are also provided. Computer simulated diametrical compression tests on the same agglomerate have also been carried out. Simulations were performed for different platen velocities and different bond strengths. The results show that high platen velocities produce extensive plastic deformation and crushing. Low platen velocities produce semi-brittle failure in which cracks propagate from the platens inwards towards the centre of the agglomerate. The results are compared with the results of the agglomerate impact tests in terms of work input, applied velocity and damage produced.

Breakage of viscoelastic drops in turbulent stirred dispersions

The existing models of drop breakage in stirred turbulent dispersions are applicable only to purely viscous dispersed phases. In their present form, they are found to underpredict the diameters of the largest stable drops formed when a viscoelastic fluid is dispersed into a Newtonian liquid. In purely viscous fluids, the turbulent stresses are opposed both by the stresses due to interfacial tension and the viscous stresses generated as the drop deforms. In viscoelastic fluids, drop deformation produces additional retractive elastic stresses which also oppose turbulent stresses. As the deformation rates are large, the retractive stresses can be large in magnitude. Assuming that these additional stresses decay with time, a model of viscoelastic drop breakage in turbulent stirred dispersions has been developed. The new model quantitatively predicts the dmax of viscoelastic fluids. The model, however, does not predict the observation that when the time constant of the fluid becomes large (λ > 0.5 s), the fluid can not be dispersed into droplets up to agitator speeds of about 10 rps in our equipment.

Breakage of viscous and non-Newtonian drops in stirred dispersions

A model of breakage of drops in a stirred vessel has been proposed to account for the effect of rheology of the dispersed phase. The deformation of the drop is represented by a Voigt element. A realistic description of the role of interfacial tension is incorporated by treating it as a restoring force which passes through a maximum as the drop deforms and eventually reaching a zero value at the break point. It is considered that the drop will break when the strain of the drop has reached a value equal to its diameter. An expression for maximum stable drop diameter, dmax, is derived from the model and found to be applicable over a wide range of variables, as well as to data already existing in literature. The model could be naturally extended to predict observed values of dmax when the dispersed phase is a power law fluid or a Bingham plastic.

Mechanics of Deformation under Traction and Friction of a Micrometric Monolithic MoS2 Particle in Comparison with those of an Agglomerate of Nanometric MoS2 Particles

Tribology of small inorganic nanoparticles in suspension in a liquid lubricant is often impaired because these particles agglomerate even when organic dispersants are used. In this paper we use lateral force microscopy to study the deformation mechanism and dissipation under traction of two extreme configurations (1) a large MoS2 particle (similar to 20 mu m width) of about 1 mu m height and (2) an agglomerate (similar to 20 mu m width), constituting 50 nm MoS2 crystallites, of about 1 mu m height. The agglomerate records a friction coefficient which is about 5-7 times that of monolithic particle. The paper examines the mechanisms of material removal for both the particles using continuum modeling and microscopy and infers that while the agglomerate response to traction can be accounted for by the bulk mechanical properties of the material, intralayer and interlayer basal planar slips determine the friction and wear of monolithic particles. The results provide a rationale for selection of layered particles, for suspension in liquid lubricants.

Effect of surfactants on drop breakage in turbulent liquid dispersions

The existingm odels of drop breakage in stirred dispersions grossly overpredict the maximum drop size when surface active agents are present inspite of using the lowered value of interfacial tension. It is shown that the difference in the values of dynamic and static interfacial tension, aids the turbulent stresses in drop breakage. When the difference is zero, e.g. for pure liquids and for high concentration of surfactants, the influence of the addition of surfactant is merely to reduce the interfacial tension and can be accounted for by existingm odels. A modified model has been developed, where the drop breakage is assumed to be represented by a Voigt element. The deforming stresses are due to turbulence and the difference between dynamic and static interfacial tensions. The resisting stresses arise due to interfacial tension and the viscous flow inside the drop. The model yields the existing expressions for dmax as special cases. The model has been found to be satisfactory when tested against experimental results using the styrene-water-teepol system.

A multi-stage model for drop breakage in stirred vessels

Existing models for dmax predict that, in the limit of μd → ∞, dmax increases with 3/4 power of μd. Further, at low values of interfacial tension, dmax becomes independent of σ even at moderate values of μd. However, experiments contradict both the predictions show that dmax dependence on μd is much weaker, and that, even at very low values of σ,dmax does not become independent of it. A model is proposed to explain these results. The model assumes that a drop circulates in a stirred vessel along with the bulk fluid and repeatedly passes through a deformation zone followed by a relaxation zone. In the deformation zone, the turbulent inertial stress tends to deform the drop, while the viscous stress generated in the drop and the interfacial stress resist deformation. The relaxation zone is characterized by absence of turbulent stress and hence the drop tends to relax back to undeformed state. It is shown that a circulating drop, starting with some initial deformation, either reaches a steady state or breaks in one or several cycles. dmax is defined as the maximum size of a drop which, starting with an undeformed initial state for the first cycle, passes through deformation zone infinite number of times without breaking. The model predictions reduce to that of Lagisetty. (1986) for moderate values of μd and σ. The model successfully predicts the reduced dependence of dmax on μd at high values of μd as well as the dependence of dmax on σ at low values of σ. The data available in literature on dmax could be predicted to a greater accuracy by the model in comparison with existing models and correlations.

A multi-stage model for drop breakage in stirred vessels

Existing models for dmax predict that, in the limit of μd → ∞, dmax increases with 3/4 power of μd. Further, at low values of interfacial tension, dmax becomes independent of σ even at moderate values of μd. However, experiments contradict both the predictions show that dmax dependence on μd is much weaker, and that, even at very low values of σ,dmax does not become independent of it. A model is proposed to explain these results. The model assumes that a drop circulates in a stirred vessel along with the bulk fluid and repeatedly passes through a deformation zone followed by a relaxation zone. In the deformation zone, the turbulent inertial stress tends to deform the drop, while the viscous stress generated in the drop and the interfacial stress resist deformation. The relaxation zone is characterized by absence of turbulent stress and hence the drop tends to relax back to undeformed state. It is shown that a circulating drop, starting with some initial deformation, either reaches a steady state or breaks in one or several cycles. dmax is defined as the maximum size of a drop which, starting with an undeformed initial state for the first cycle, passes through deformation zone infinite number of times without breaking. The model predictions reduce to that of Lagisetty. (1986) for moderate values of μd and σ. The model successfully predicts the reduced dependence of dmax on μd at high values of μd as well as the dependence of dmax on σ at low values of σ. The data available in literature on dmax could be predicted to a greater accuracy by the model in comparison with existing models and correlations.

Breakage and coalescence of drops in turbulent stirred dispersions

The various existing models for predicting the maximum stable drop diameterd max in turbulent stirred dispersions have been reviewed. Variations in the basic framework dictated by additional complexities such as the presence of drag reducing agents in the continuous phase, or viscoelasticity of the dispersed phase have been outlined. Drop breakage in the presence of surfactants in the continuous phase has also been analysed. Finally, the various approaches to obtaining expressions for the breakage and coalescence frequencies, needed to solve the population balance equation for the number density function of the dispersed phase droplets, have been discussed.

A new model for the breakage frequency of drops in turbulent stirred dispersions

A model of drop breakage in turbulent stirred dispersions based on interaction of a drop with eddies of a length scale smaller than the drop diameter has been developed. It predicts that, unlike the equal breakage assumed by earlier models, a large drop reduces in size due to stripping of smaller segments off it through unequal breakage. It is only when the drop nears the value of the maximum stable drop diameter that it breaks into equal parts. This new model of drop breakage, coupled with the pattern of interaction of drops with eddies of different sizes existing in the vessel, has been used to evaluate not only the breakage frequency, but also the size distribution of the daughter droplets(which was hitherto assumed). The model has been incorporated in the population balance equation and the resulting cumulative size distributions compared with those availble in the literature.