An integral component in understanding the fundamental mechanisms of granulation is the powder flow behaviour within mixer granulators. All three classes of granulation rate processes: wetting and nucleation, consolidation and coalescence, and attrition and breakage require detailed knowledge of flow fields in mixing systems to enable rational approaches to the prediction of granule attributes. The three objectives of this thesis are to 1) develop a powder flow regime map to characterise the granular flow field, 2) investigate the effect of impeller speed, powder cohesion and blade-rake angle on powder flow and mixing behaviour and, 3) outline a framework for compartment models of vertical-axis high shear mixer (vHSM) granulators.
In the development of the powder flow regime map, two distinct regimes were identified. Ideal roping regime occurs at high impeller speeds when the rotating blades pushes the powder out and up the mixer wall, and then back down the free surface forming the roping flow pattern. This stable flow behaviour promotes high vertical bed turnover which is essential for controlled granulation. In this regime, the powder in the blade region has sufficient rotational inertia to overcome gravitational forces (i.e. powder Froude number greater than unity). If this criterion is not met, the mixing system remains in the undesirable bumping regime. This regime is characterised by poor bed turnover and non-uniform shear distribution within the powder bed. In the bumping regime, multiple transitional states are observed with increasing impeller speeds for this particular system, which has not been fully characterised in the literature before. An explanation for the observed apparent roping behaviour based on bed resonance is given. These intermediary transitional states can be differentiated by blade-powder interactions.
From these observations, a powder flow regime map for vertical-axis mixer granulators is proposed based on two dimensionless groups: powder Froude number, the ratio of the powder rotational inertial to gravitational forces and the Bed Resonance number, which characterises the interaction of the bulk motion of the powder bed with the passage of impeller blades beneath it. The regime map was validated by experimental results utilising the well-established, non-invasive tracking technique of Positron Emission Particle Tracking (PEPT) and torque measurements.
Investigations into the effect of impeller speed, powder cohesion and blade-rake angle on powder flow and mixing behaviour were conducted using PEPT. Results showed that powder velocity, bed porosity and shear rates varied spatially within the mixer granulator. Studies of the stable roping regime revealed that the average powder velocity in the impeller zone was 1.66 m/s (~ 25% of the impeller tip speed). In comparison, the average powder velocity in the circulation and surface zone was approximately half of the impeller zone at 0.9 and 0.74 m/s, respectively (12% and 11% of the impeller tip speed). With the addition of moisture (cohesion) to the dry powder, the average velocity of the powder decreases by 10%. Increasing blade-rake angle from 10 to 90 degrees increased the average powder velocity by approximately 38%. Vertical mixing improved with higher blade-rake angle due to an increased thickness of the shear transmission layer however no significant improvements were observed with changes in bed cohesion.
Compartment models are developed to reduce the highly complex flow field into spatial similar phenomenological compartments. A general framework is outlined to describe the methodology used to identify compartments and powder exchange rates between compartments. A simple two-compartment model was proposed for a system in which growth and breakage occurred simultaneously. The mixing vessel was partitioned into two compartments, the impeller zone and the circulation zone. Expanding on this, a more complex multi-compartment model was developed to incorporate wetting and nucleation processes into a third compartment defined as the spray zone. Both models represented reasonable approximations to the experimental data. These fitted models, in combination with population balance modelling should provide a useful tool to predict wet granulation processes as well as a rational approach to controlling final granule attributes.