ARR - Annual Report

We have examined the effects of powder cohesion and particle size distribution on mixing and segregation processes. As agreed in our Year 1 work plan, we have completed the development of experimental and computational procedures for this study and we have conducted an extensive characterization of non-segregating systems in order to provide a baseline for segregating mixtures (to be examined in years 2 and 3 of this project). We have met all of the stipulated milestones, summarized as follows.

1. We have built a computer-controlled lab scale double-cone blender.
2. We have developed and debugged discrete element simulation code for the double-cone blender.
3. We have simulated mixing in the double-cone over the course of up to 12 tumbler revolutions, using 15,000 monodisperse and 13,000 bidisperse particle blends.
4. We have measured the mixing rate as a function of fill level and vessel speed.
5. We have evaluated effects of filling level and vessel speed on mixing and segregation, both experimentally and computationally.
6. We have examined the effects on mixing and segregation of three-way interactions between particle size, fill level and vessel speed.

At this stage of the project, we have found the following.

1. Mixing in the double-cone occurs by a combination of radial-azimuthal convection and axial diffusion. The chief bottleneck to mixing in the double-cone is diffusion acrass the symmetry plane; we have demonstrated that judicious baffle placement can significantly improve mixing.
2. As particle sizes are reduced below about 200mu, steady and regular flow in tumbling blenders gives way to intermittent and chaotic mixing. This results in a dramatic improvement in mixing rates, overwhelmingly exceeding what would be possible by traditional mixing mechanisms. Moreover, this work demonstrates that traditional analysis cannot be applied, even qualitatively, to the study of flow and mixing of fine grains.
3. Several new segregation modes have been identified. We have charted the phase space of these modes, and we have begun an analysis of a preliminary model of segregational mechanisms which seems to show promise for developing a predictive understanding of flow and transport of polydisperse granular mixtures.

Executive Summary

This is the first annual report for the IFPRI project 37 Powder-Binder Agglomeration. The aims of this project are to:

• Define the controlling groups for each of the following classes of granulation process (1) binder dispersion, wetting and nucleation (2) consolidation and growth (3) attrition and breakage;
• Use appropriate models to link these groups to key product attributes eg. size, size distribution, density (porosity);
• In each case, qualify the models for the effects of complicating powder-binder interactions eg. dissolution, reaction, drying.

The report summarises progress in both wetting and nucleation, and consolidation and growth. Drop penetration time and dimensionless spray flux are proposed as two controlling groups for wetting and nucleation. The drop penetration time, which depends mainly depends on formulation properties, is a promising tool for studying nucleation processes. Preliminary studies show that penetration time varies widely, particularly with binder viscosity. Combining the existing models for drop penetration (Denesuk, 1993) and nuclei growth (Schaafsma, 1998b) will create a more complete picture of nuclei formation and morphology, and the effect of material properties. Ex-granulator experiments demonstrated different nucleation regimes from drop-controlled nucleation to caking. In the drop controlled regime, each drop forms a single nucleus and the nuclei distribution can be controlled by controlling the drop size distribution form the spray. A single dimensionless group, the dimensionless spray flux, characterises the main process parameters with respect to spraying. An experimental plan and methodology for studying wetting and nucleation is presented.

For granule growth and consolidation, a regime map of granule growth behavior is proposed based on granule deformation during collision and the granule liquid content measured as the maximum pore saturation. The granule deformability on collision is represented by a deformation number, which is a ratio of granule impact energy to the plastic energy absorbed per unit strain. Granule growth regimes such as steady growth, induction, nucleation, crumb, and slurry are defined. This regime map qualitatively explains the variations in granulation behavior. Laboratory drum granulation experiments were used to test the regime map. Increasing granule yield stress by decreasing particle size and increasing binder viscosity caused the system to move from steady growth to induction behavior as predicted by the regime map. Preliminary validation with literature data was also encouraging. More work, however, is required to better quantify the boundaries between different growth regimes and to investigate the effect of process agitation intensity. This regime map has great potential to help design and control granulation systems, because it is based on properties of the powder/binder system that can be measured or estimated without performing any granulation tests.

Following completion of the review, experimental work and design stages of the first year of our work, significant developments have since occurred in moving towards providing new sensing strategies to enable on line measurements in concentrated and, using different approaches, for dilute flowing mixtures. Over the last year we have been concerned with the two main tasks of designing and testing instrumentation for these purposes.

For concentrated systems - use of conventional electrical tomographic measurement method coupled with analysis of raw and reconstructed data using new statistical methods have been employed. This has been used with some success to quantify mixture properties (structural homogeneity, concentration fluctuations etc). Results are reported for solid-liquid pastes, liquid- liquid and gas-liquid mixtures. This has demonstrated that some interesting and, we believe, unique applications exist using such an approach to enhance quality control procedures in manufacturing plants and processes. It is proposed that direct feedback of the outputs from the measurement and interpretation software could be implemented as part of a control scheme.

For dilute systems - a new multi-sensor approach The Particle Gymnasium has been proposed and recently a test system has been built. Ultimately this is intended to provide a means of particle shape and size measurement in a rapidly flowing stream. Preliminary results show that the method appears to be viable and work can now continue to consider particulate systems of specific interest to IFPRI members working in crystallisation and controlled formation.

During the period under review, discussions and visits have taken place between several industrial members to consider and identify measurement needs. A forward project plan is in place. In future this will include an opportunity to assess the benefits to be gained by incorporating additional sensors (including ultrasound and magnetic resonance imaging). Further fabrication and testing of the two types of sensor systems for dilute and concentrated suspensions is planned in conjunction with industrial members in the year ahead. This will begin to define the practical performance of the sensing methods.

For fundamental discussions or understanding on contact/impact charging of particulate materials, it is essential to address measuring the charge generated due to a single impact or contact between a single particle and target- this is our basic concept at the study. Actually for that, we performed so-called “impact charging experiments,” in which a spherical plastic particle was launched one by one and made impact onto a metal target to measure the impact charge due to a single collision. To realize this, in our previous version of the experiments, the size of sample particles was limited to be big, the diameter of the particle used mainly was 3 mm. In this project, therefore, our major subject is to extend the concept into an actual powder size region. To realize the concept, we proposed two approaches in the direction of the enhancement:

The importance of explicitly considering interparticle interactions in Ihe rheology of dense suspensions and particle slurries is well established, although the exact relationships between particle-level quantities and macroscopic rheology and stability are at best qualitative. Most of the understanding has been developed follow shear rheological (linear viscoelastic) properties and/or for dilute dispersions. This project has the goal of providing experimental evidence for the influence of interparticle surface forces and hydrodynamic forces (due to the presence of the solvent) on the moderate to high shear rheological properties and shear stability of dispersions that span the colloidal to particulate range (colloidal dispersions to slurries). Of particular interest is the shear thickening transition and dilatancy, and how that, explicitly depends on the strategy used to stabilize the dispersion or slurry (i.e. steric, electrostatic, polymeric stabilization), as well as the hydrodynamic forces irnport,ant at higher shear rates. The research to date has the following components:

• Systematically explore the influence of the basic methods of parrticle stabilization on the shear thickening and dilatancy of well-characterized colloidal dispersions and slurries of non-colloidal particles. In this report, in particular, the effect of added adsorbed polymer on the rheology is studied for explicit comparison to sirriulation results.
• Use the experimental data to develop simple force-balance based models for predicting the onset of shear thickening, as well as for use by formulators to prevent shear thickening.

Introduction

Agglomerates of fine particles are pervasive in industry. In many particle processing applications, the agglomerates are carried within a suspending fluid, and hydrodynamic shear is applied to break the agglomerate into fragments and to distribute the fragments throughout the suspending media. The underlying purpose of this research it to obtain a fundamental understanding of the various factors that influence this dispersion process. Such information can lead to the development of interfacial engineering strategies aimed at improving the outcome of dispersion processes, or to better design of dispersion equipment.

Our general approach is to study the dispersion behavior of well-characterized single agglomerates in controlled flow fields. This allows us to establish the links between the fundamental properties of an agglomerate and dispersion characteristics such as critical shear stress for dispersion, mode and kinetics of fragmentation, and the evolution of the fragment size distribution.

The specific focus of the work supported under the IFPRI grant involves investigation of how certain time-dependent (dynamic) behaviors can influence the outcome of dispersion processes. Dynamic effects can arise in several facets of the dispersion process. For instance, in practical processing equipment, complex shear histories are inherent. Also, the wetting and spreading of fluids associated with contacting particles and agglomerates with processing fluids are dynamic effects. Finally, for some materials, dissolution of the solids plays a significant role.

For the first year of this IFPRI grant, the bulk of the research effort was devoted to the development of a new experimental approach for the investigation of the influence of dynamic effects on dispersion behavior. This entailed the design (and redesign) of a dynamic dispersion chamber, and construction of it and the ancillary equipment. Preliminary experiments were done to validate the experimental techniques and to refine the analytical procedures.

The fluidised dense-phase conveying of powders and low-velocity slug-flow of granular bulk solids are the most common and popular modes of dense-phase used in industry. However, the accurate prediction of conveying performance still is not possible from first principles and relies heavily on empiricism. The main aim of this project is to develop the necessary understanding, databases, duidelincs and models for the purpose of predicting accurate optimal operating conditions for the fwo modes of dense-phase. This Annual Progress Report summarises the project objectives, research progress and major achievements for the period September 1998 to November 1999, and includes objectives for the second year of the project.

In the particle technology, it is fundamentally important to know the interaction and adhesive forces between particles and to find the correlation of those forces with the microscopic characteristics of particle surface, because these forces are the origin of many phenomena which particles exhibit in industrial processes. The aim of this project is to clarify in-situ at the molecular level the microstructure of surfaces in solutions of industrial importance and the correlation with interaction and adhesive forces between surfaces, using not only an atomic force microscope (AFM) but also computer simulations.

It was planned to clarify the phenomena and mechanism on the subjects shown in the map of the following figure, which were considered to be fundamental in understanding the phenomena in industrial particle processes.

The objective of this work is systematic understanding of particle-particle nanorheology based on the single particle-particle contact of two atomically-smooth solid surfaces in molecularly-thin proximity. The main relevance is to understand the origins of suspension rheology, especially the origins of rheological anomalies that arise when interfacial films between two solid bodies are so thin that the intuition of what to expect based on bulk rheology no longer applies. Based on this understanding, we are seeking to develop new methods to control and manipulate the properties of their interfacial films.

It is desired to understand the flow with particle clusters, because particle clustering has a definite effect on transport, phenomena in risers of circulating fluidized beds (CFBs). For example it is known that particle clustering largely increases the particle slip velocity against gas, consequently, the pcarticle residence time in the riser is largely changed from that of an isolated particle. In addition, properties of gas turbulence must be significantly modified due to the clusters. Frorn the view point of application to industrial facilities, macroscopic models for predicting flows in the risers should be developed finally. On the other hand, to develop such models, it is important to study the physics or the dynamics of the clusters from both of the experimental and numerical stand points.

According to this context, Tanaka et al., (1998) have carried out the first, year project. They applied their numerical model, in which inviscid gas and a stochastic particle-particle collision model were assumed (Tanaka et al., 1995), in the three-dimensional flows corresponding to the experiments by Louge et al. (1999). They found that their numerical model is capable of predicting the fully developped cluster flows, and examined the effects of pressurized gas condition on the flow structure. Furthermore, they evaluated the quantities for characterizing the cluster structure, such as number density distributions of cluster diameter, probability density functions of solid volume fraction, etc.

In the second year project, quantitative comparison between the sirnulation and the corresponding experiment at Cornell University was intcndcd. The collaboration between Osaka and Cornell Universities began with discussions at the Brighton Annual Meeting in July 1998. There, Tanaka asd Louge proposed to compare the predictions of the numerical simulations at Osaka with solid volume fraction mcasurcments carried out at Cornell. The Cornell group obtained solid volume fraction with an optical fiber bundle in the fully-developed region of the Cornell riser. The Osaka group then carried out the numerical simulation at the conditions of the Cornell experiments. They then compared the probability density functions and power spectra of the data sets and the corresponding simulations.

In usual industrial applications, particles do not have a uniform diameter but a wide distribution. Tarmka et al. (1995) performed Lagrangian/Eulcrian numerical simulations of two-dimensional cluster flows, and found that the spatial scale of cluster structure largely depends on the particle diameter. Therefore, it is expected that the particle size distribution may affect the flows. The Osaka group has ttaken particular care to match the particle size distribution of the experimental powders.

Tanaka then visited Cornell in March 1999 to prcscnt these comparisons and to discuss details of the simulations with Profs. Louge, Jenkins, Koch and their respcctive collaborators. The status of their collaboration was reported at the Spring TC meeting in Newark, NJ, and at, the Annual Meeting in Somerset, NJ.