Estudo experimental e numérico da dinâmica de partículas em um disco rotatório

Abstract

The granulation process directly depends on how the particles or grains move inside the dish granulator, this behavior being quite complex and still poorly understood. Knowledge about particle dynamics is essential for optimizing and scaling projects involving the dish granulator and improving granulation process efficiency. In this context, the present work aimed to study, experimentally and numerically, the influence of the operating conditions of the dish granulator on the dynamics of soy and glass granular materials. It was found that higher rotation speeds and lower inclination angles increased the solids holdup inside the dish granulator. It was also noticed that, as it has a more rigid and compacted granular bed, the soybean particles presented a greater solids holdup when compared to the glass spheres. The transition of the flow regimes was evaluated by the image analysis method and it was possible to observe that the operational conditions analyzed (filling degree: 2, 5, and 8% and inclination angle: 40, 50, and 60°) and the physical properties of the materials influenced the rotation speed necessary for the transition of the flow regimes. The images obtained by this method were also used for the training and validity of a convolutional neural network. The proposed network presented an accuracy of 98%, correctly classifying 353 of the 360 images provided to it. Because image analysis is a qualitative method, the acoustic analysis methodology was proposed as a new method for identifying the transition of flow regimes. The acoustic analysis method proved to be an excellent alternative in the study of particle dynamics, as it was able to identify the cascading-cataracting transition for soybean particles, in addition to determining the exact moment in which the particles centrifugation took place. A modification was also carried out in the Papadakis and Bombled (1961) model to calculate the critical centrifugation speed. The filling degree addition resulted in an average deviation of 14.27%. As a complementary tool to experimental studies, the Lagrangian numerical approach was used, through the Discrete Element Method (DEM), as an aid to understanding the flow of granular materials. A sensitivity analysis of both the DEM numerical parameters and the operational conditions on the contact number between particles was performed. Regarding the DEM parameters, it was found that the particle-wall static friction coefficient was the parameter that had the greatest effect on the response, whereas, for the operational conditions, the particle-particle contact number was more affected by changes in the filling degree and the speed rotation. DEM was also used to assess runoff regimes. It was noted that the experimentally obtained values of the DEM parameters, when implemented in the model, could not adequately represent the experimental observations, requiring a calibration process. For soybean particles, it was found that the calibration of the particle-wall static friction coefficient was not sufficient for the numerical response to reproduce the experimental one, requiring more study of the influence of other parameters and the particle sphericity. As for the glass spheres, after calibration, it was found that the Discrete Element Method was able to represent the granular behavior inside the dish granulator and that the values of the parameters obtained by calibration were within the standard deviation of the experimental measurement, maintaining the physical properties.