Multiscale modelling of industrial flighted rotary dryers

Rotary dryers are commonly used in the food and mineral processing industries for drying granular or particulate solids due to their simplicity, low cost and versatility compared to other dryers. The co-current industrial rotary dryer (MMG, Karumba) examined in this study is used in drying zinc and lead concentrate. The dryer is 22.2 metres long with a diameter of 3.9 metres. The slope and the typical rotational speed of the dryer are 4 degrees and 3 rpm respectively. The dryer has both unflighted and flighted sections with different flight configurations. Operational issues associated with the dryer that lead to the requirement for a dynamic model of the dryer include issues such as high fuel consumption and the build-up of scale on the internal surfaces. In order to operate an optimum dryer, it is necessary to understand the mechanisms occurring within the dryer. The important transport mechanisms that govern the performance of rotary dryers are: solids transportation, heat, and mass transfer. Studies have shown that the knowledge of the solid transport is important to solve the heat and mass transfer differential equations that describe completely the temperature and moisture content profiles along the dryer for both solid and gas phases. Solid transport within the dryer can be characterised through the solid residence time distribution, which is the distribution of times taken for the solids to travel through the dryer. Solid residence time distribution can be determined experimentally. The most common experimental approach is to introduce tracer at the inlet and monitor tracer concentration at the outlet as a function time. Several modelling approaches have been taken to determine the residence time and the residence time distribution and these approaches have varied from empirical correlations to compartment modelling. In many of these approaches, loading state, residence time and operational feed rates are strongly linked. The loading state also influences the effectiveness of particle to gas heat and mass transfer as well as the residence time distribution of solids through the dryer. There are three potential degrees of loading in a rotary dryer namely under-loaded, design loaded and overloaded. However, most industrial rotary dryers are operated at under-loaded or overloaded, which results into poor efficiency of the dryer and the optimal economics of the dryer will not be achieved. As such, accurate estimation of the design load is critical to the optimal performance of flighted rotary dryers and is an important characteristic of flighted rotary dryer models. To experimentally characterise MMG rotary dryer, industrial and laboratory experiments were undertaken. The industrial experiments included residence time distributions (RTD), shell temperature measurements, spatial sampling of the solids along the length of dryer, moisture content analysis and Process Information (PI) data collection. Residence time distribution experiments were carried out by injecting lithium chloride as tracer at the inlet of the dryer while sampling outlet solids over a period of time. Zinc concentrate properties such as dynamic angle of repose, bulk density and particle size were also determined. A series of different experiments were undertaken to examine the effect of speed and loading. Flight loading experiments were carried out at pilot scale to determine the effect of moisture content and rotational speed on dryer design loadings and to facilitate accurate determination of model parameters. The flight holdup experiments involved taking photographs of the crosssectional area of the dryer. An image analysis technique was developed to estimate the amount of material within the flights and in the airborne phase. The analysis involved developing a combined ImageJ thresholding process and in-house MATLAB code to estimate the cross-sectional area of material within the flight. The suitability of the developed methodology was established. In addition, saturation of both the airborne and upper drum flight-borne solids was observed. To select an appropriate geometrically derived design load model, comparison of existing design load models from the literature was undertaken. The proportion of airborne to flight-borne solids within the drum was characterised through a combination of photographic analysis coupled with Computational Fluid Dynamics (CFD) simulation. In particular, solid volume fractions of the airborne solids were characterised using a CFD technique based on the Eulerian-Eulerian approach. The suitability of using geometric models of flight unloading to predict these proportions in a design loaded dryer were discussed and a modified version of Baker's (1988) design load model was proposed. A multiscale dynamic mass and energy process model was developed and validated for the dryer in order to characterise the performance of MMG rotary dryer. The mass and energy balance equations involved ordinary differential equations for describing the flighted sections and partial differential equations for modelling the unflighted sections. Solids in unflighted sections were modelled as the axially-dispersed plug flow system. In the flighted sections, the solids were modelled using a compartment modelling approach involving well-mixed tanks (Sheehan et al., 2005). The gas phase was modelled as a plug flow system. Simulations were undertaken using gPROMS (process modelling software). As much as possible, model coefficients were determined using geometric modelling based on material properties and dryer operational conditions. The use of this approach is termed a pseudo-physical compartment model. The solid transport model was validated using full scale residence time distribution at different experimental conditions. The model results predicted well the effect of rotational speed, internal diameter and solid feed rate. Estimated parameters included the kilning velocity, axial dispersed coefficient and area correction factors. The validation of the energy balances was based on Process Information (PI), experimental residence time distribution and moisture content data of the studied dryer. Model parameters involving the surface area in contact with the incoming gas data were manipulated to fit experimental moisture content. The gas and solid temperature profiles were also predicted, which provide a firm basis upon which additional studies may be undertaken. Gas inlet temperature was identified as the most suitable manipulated variable for the dryer with clean internal condition. However, to achieve desired product quality within a scaled dryer, the study suggested the solid feed rate should be reduced so as to achieve optimum gas-solid interaction. To address the high fuel consumption associated with the dryer, the study proposed externally lagging of the dryer and reduction in the gas inlet temperature to meet the desired product quality