X-ray motion analysis of charge particles in a laboratory mill
Doctoral Thesis
2005
Permanent link to this Item
Authors
Supervisors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
University of Cape Town
Department
License
Series
Abstract
The work reported herein is of an automated X-ray vision system used to track the 3D trajectories of a typical bulk charge particle within an experimental Perspex mill, the intention of which is to simulate the grinding motion of rock found in typical industrial mills The experimental rig is constructed through an optimisation scheme that determines the maximum allowed dimensions of the mill that can be imaged by the X-ray system, however, the optimisation principle is not limited to the current study and can be employed to maximise any volume being investigated with the X-ray system. The raw data for each position of the tracked particle comprises of two X-ray images of the tumbling mill generated orthogonal to each other, with a phase lag between them. The correction of the phase lag between the biplanar images is a new addition to the usual usage of the system resulting in an effective sampling rate of 100 frames per second, thereby ensuring that the resolution is sufficient to conduct detailed kinematic studies. The processing of the raw images are achieved with a dynamic template matching algorithm followed by a modified, and improved, implementation of the Canny edge detector, while the centering of the edge images are based on an adapted conic fitting routine, resulting in an overall subpixel centering accuracy. The processed images are then mapped to object space using the direct linear transformation (DLT), equipped with a physically valid variance model that is shown to improve the standard implementation even when robust solvers are employed. The final reconstruction accuracy of the tracked particle was shown to be 0.15 mm and is achieved without iteration due to the appropriateness of the variance model. The high accuracy data. was initially used to benchmark the discrete element method (DEM), providing the first numerical comparisons that surpassed the usual end-window snapshots employed by other investigators for the purpose of validation. The analysis of the data was not restricted to DEM verification, and in some sense, surpassed the initial objective by yielding trends useful to communication practices. Amongst these analyses was the finding that the circulation rate of the charge is not once per mill revolution but greater, depicting a linear trend with mill speed. The slip between charge layers was shown to follow a linear pattern, with the degree of slip increasing linearly with mill speed. The phenomenon of charge surging was quantified, resulting in a trend for the variation of the surge amplitude with mill speed. A mechanism for the surging phenomenon was also proposed in this study. A particularly useful outcome of the data analyses was the formulation of a power model through heuristic trends of the center of mass (CoM) and center of circulation (CoC) of the charge. The methodology outlined by the model was shown to be robust, providing a correct approach to obtaining a truly fundamental power model based on generally applicable principles.
Description
Includes bibliographical references (p. [266]-273).
Keywords
Reference:
Govender, I. 2005. X-ray motion analysis of charge particles in a laboratory mill. University of Cape Town.