Wet Low-Intensity Magnetic Separation: Measurement methods and Modelling

Sammanfattning: In the mining industry, ferromagnetic particles (e.g. magnetite) are concentrated using wet low-intensity magnetic separation (LIMS). Mineral particles in suspension with water are pumped into the separator tank, and a magnetic concentrate is extracted by use of magnetic forces. The performance of the process is to a large extent controlled by the internal flow conditions in the separator, governed by process and machine settings. Due to the machine design these settings are not independent, and in some cases it can be difficult to reach optimal process performance. The main purpose of this work has been to find a measurement method capable of monitoring internal material transport in the wet LIMS, and use these data, together with numerical flow modelling, to get an increased understanding about the separation process.Since the mineral slurry entering the separator is essentially opaque, and the solids concentration is rather high, an ultrasound-based method was selected for the internal measurements. It is of interest to monitor both the internal flow patterns, as well as material build-up resulting from the magnetic field. The method development and evaluation proceeded in steps with increasing complexity, with later stages building on experience from the former. Initial measurements were done in model systems with simple geometries, and over a range of flow velocities corresponding to flow velocities in full-scale magnetic separators. Additional measurements were done on model systems under influence of a magnetic field of varying strength. After the measurement methods were verified in controlled laboratory conditions they were evaluated in real world conditions; in situ at the LKAB pilot plant in Malmberget, Sweden.For the pilot scale experiments a setup with two ultrasound transducers, mounted at the bottom of the separator tank, was used. The factors included in the designed experiment were the feed solids concentration, drum rotational speed, position of the concentrate weir, and the magnet assembly angle. Based on this investigation the drum rotational speed was the factor having the strongest influence on the overall flow velocity in the dewatering zone. Also, the presence of a recirculating flow transporting gangue particles away from the concentrate was confirmed. The factor with strongest influence on this flow is also the drum rotational speed, together with the magnet assembly angle. Using this method it is possible make high quality measurements of internal flow velocity profiles. It is also possible to monitor material build-up on the separator drum, and e.g. detect overload of magnetic material.The ultrasound based measurement system resulting from this work measures particle velocity based on a cross-correlation principle. Two consecutive ultrasound pulse-echo signals are cross-correlated piece-wise, to obtain a local velocity estimate. By measuring the suspension flow from two directions, using two transducers, 2D velocity vectors can be estimated. Using the same measured data, but instead studying how the spectral contents of the signal vary with axial distance from the transducer, a qualitative measure of variations in local solids concentration can be obtained.During this work several aspects of wet LIMS have been studied, with focus on the internal material transport processes inside the separators. State-of-the-art measurement methods have been utilized to monitor the material flow inside the separators. Particle capture and entrainment have been studied on the particle level using numerical flow modelling. Measurement results have been linked to operational conditions of the separators. The insights gained, and the methods developed, have generated new possibilities to control, optimise, and develop the wet LIMS process.Keywords: Wet low-intensity magnetic separation, magnetite beneficiation, in-line process monitoring, pulse-echo ultrasound, ultrasonic velocity profiling, solids concentration, signal processing, windowed cross-correlation, power spectral density, numerical flow modelling.