A Predictive Model for Wet High Intensity Magnetic Separator (WHIMS) using Artificial Neural Networks

Abstract

Abstract—Materials can be classified into three major categories based on the magnetic susceptibility thereof; the property governing the response of the material subjected to a magnetic field. Chromium has many uses in the industry, with stainless steel production being the most important. Chromite ore in South Africa is mainly mined in the Bushveld complex situated in the central western region of the Highveld. Magnetic separation is the physical separation of discrete particles. Wet high-intensity magnetic separation (WHIMS) is commonly used in the gold, uranium, iron and chromite recovery industries. The WHIMS system is based on the imbalance of forces on particles. These forces are magnetic, gravitational, centrifugal, frictional or inertial, and attractive or repulsive forces in favour of the magnetic forces, all due to the production of a magnetic field. During the experimental procedure, single stage separation was used for the aim of this project. Operational parameters such as magnetic intensity (flux), wash water flow rate, feed flow rate and feed density along with particle size are varied. The primary objectives in this study were to obtain experimental data from a laboratory scale WHIMS and to use this data to construct an artificial neural network (ANN) able accurately predict grade, yield and recovery. Sampling and analysis were used to determine the recoveries, grades and yields for the varied operating conditions. The material used during this study is chromite ore. The ANN’s predicted the grade, recovery and yield with high accuracy. The data from experimentation suggest that the WHIMS system recovers best at smaller particle sizes.