Coal Preparation » Dewatering
Centrifugal dewatering is one of the main processes used for fine and ultrafine coal dewatering, but its efficiency is heavily impacted by variations in the feed stream, including flow rate, particle size distribution and solids content. It is essential to continuously monitor the feed and effluent streams to understand the performance of the centrifuges and to adjust the operational variables accordingly to ensure optimal solids recovery with consistent product moisture level.
This report describes the development of an online tool that utilises image analysis and machine learning techniques to simultaneously measure solids content and particle size distribution (PSD) of the feed and effluent streams of centrifuges.
Two classic image analysis approaches and two machine learning algorithms were tested to achieve accurate and speedy analysis of the PSD and solids concentration in the slurry. One of the trained machine learning models showed a promising result - the model could be used to analyse the slurry accurately in real-time.
The monitoring system was also tested with the feed and effluent samples collected from a pilot-scale screen bowl centrifuge at University of Queensland under various operation conditions. It demonstrated a great potential for simultaneous analysis of solids concentration and PSD. There is scope for further improving the tool's performance for the effluent sample as the majority of the particles in the effluent stream (i.e., 50 - 90%) are below the detection limit of the monitoring tool (i.e., 8 μm with a 5x magnifying lens). It is recommended that a 40x magnifying lens be mounted to improve the detection limit of the particle to 1 μm for accurate analysis of the effluent stream.
The centrifuge model previously developed in project C25018 was redefined and then combined it with the design of experiments (DoE) method to systematically examine the impact of the various operation variables on the solids loss in the effluent. The DoE helped find the optimal combination of the operation parameters that can maximise the solids recovery at a given PSD of the feed. The analysis revealed that feed flow rate would play a crucial factor in minimising the loss of the solids in the effluent stream when the solids concentration in the feed exceeded a certain threshold. It is important to tightly control these variables to prevent unwanted loss of valuable coal, especially when the centrifuge is operating at a low G-force and high feed solids throughput. At a given feed solids throughput, an optimal combination of feed flow rate and feed solids concentration could significantly reduce the loss of the solids compared to non-optimal conditions.
An effective online tool and protocols for monitoring feed and effluent streams of the centrifuges were developed. The project also demonstrated that the implementation of the centrifuge model into the slurry monitoring tool had great potential for improving the operation and control of the centrifuges for ultrafine and fine coal products. The developed tool and associated software, including the SBC model, are expected to assist engineers and operators in achieving efficient operation of centrifuges and provide important guidelines for process improvement, which will ultimately lead to improvement in productivity and environmental sustainability by minimising solids loss to the tailings.