Coal Preparation » Process Control
Teeter Bed Separators (TBS) are widely used in Australian coal preparation plants to process material that is too fine for dense medium separation and too coarse for flotation. While TBS units are effective density-based separators, their performance is sensitive to feed characteristics and operating conditions, and opportunities exist to improve stability, product quality, and operational consistency through improved understanding and control.
The primary objective of this project was the development and validation of a computational fluid dynamics (CFD) based digital twin capable of reproducing key TBS separation behaviours observed under industrial operating conditions. A comprehensive experimental survey program was undertaken to provide full-scale data for model validation and to review what insight could be gained from plant measurements.
In total, 80 full-scale TBS surveys were conducted across two operating coal preparation plants, involving different unit configurations, feed size distributions, and coal types. The experimental results were internally consistent and confirmed several fundamental aspects of TBS operation.
Regression analysis of the survey data showed that bed density set point, teeter water flowrate, and feed ash content are key variables influencing yield, ash recovery, and product ash. The analysis also confirmed that key performance metrics are bounded and non-linear, indicating that simple linear models are insufficient to describe TBS behaviour across a wide operating range. CFD simulations performed further demonstrated that feed properties, which could not be systematically varied in the experimental program, including feed rate, feed percent solids, and feed size distribution, can also exert a significant influence on TBS performance.
The central outcome is the successful development and validation of CFD digital twins for two full-scale TBS units and a laboratory fluidisation device. Using an Eulerian multiphase modelling framework with embedded bed density control, the CFD simulations were able to reproduce density-driven separation behaviour and key performance trends observed experimentally. Validation against site survey data demonstrated good agreement for mass pull, partition behaviour, ash recovery, and product ash grade for the +0.25mm particle size fraction. This confirms that the digital twin captures the dominant separation physics governing coarser material. Further work is required to confirm its capability to accurately represent fine particle behaviour, which was shown experimentally to be largely unselective.
It is concluded that the project successfully achieved its primary objective of developing and validating a CFD-based digital twin capable of reproducing key TBS separation behaviour under industrial conditions. The model demonstrated good agreement with plant data and provides a reliable platform for predicting and understanding the drivers of TBS performance.