Open Collections

Abstract

This thesis focuses on investigating compression breakage in High-Pressure Grinding Rolls (HPGR) and developing a throughput prediction model employing the Piston Press Test (PPT). The research aims to reduce the amount of material required for sizing and selecting an HPGR, hence lowering the testing costs and time, which is particularly beneficial in the early development stages of new projects with limited sample availability and for geometallurgy studies. Statistical analysis and design of experiments (DOE) have been leveraged to determine the significant factors influencing compression breakage in PPT and pilot-scale HPGR tests. The study reveals the importance of parameters such as pressing force, sample volume, and moisture content. Findings highlight the similar behaviour of different ore types under compression and illustrate the viability of applying developed models to other geometallurgical units. Furthermore, the thesis presents an innovative PPT procedure designed to predict the HPGR operational gap using pilot-scale data, advancing the ability to estimate HPGR throughput at the laboratory scale without the need for extensive pilot-scale calibration. The validation of this method revealed low average error rates, demonstrating a high degree of accuracy and reliability in its predictions. Moreover, the study developed a specific throughput model for pilot-scale HPGR that accurately predicts throughput based on available pilot-scale test data. The model demonstrates a high degree of precision, offering strong confidence in its predictive performance. Lastly, an in-depth analysis of how feed characteristics, such as the sample moisture content and particle size distribution, affect breakage by compression was conducted. The outcomes provide insights for improving HPGR operation in mining and enhancing efficiency while reducing operating costs. The research contributes valuable knowledge to HPGR modelling, which has the potential to be considered for industrial applications and software.