Heap leaching is a hydrometallurgical process used for the extraction of minerals within complex and typically low-grade ores. An important factor in the mineral dissolution process is the contact efficiency between the irrigation fluid (lixiviant) and the targeted mineral, which is influenced by both the solid and fluid properties of the system. One of the principal challenges related to the contact efficiency is preferential flow, cited to result in low extraction rates and in extreme cases, heap failure. Preferential flow reveals itself on two scales in drip irrigated heaps, referred to as the bed and solution scale. The bed scale takes a macro view of the heap and deals with uneven wetting profiles characterized by the presence of wet and dry sections. Linked to this is capillary suction effects which play an important role in the establishment of fluid flow profiles within the heap. The solution scale focuses on preferential flow behaviour in the wetted sections of the heap characterized by variations in the residence times of fluid elements. Such variations produce fast flowing, slow flowing and stagnant solution pools. Therefore, ideal solution flow behaviour in a heap result in uniform wetting at the bed scale and plug flow behaviour with similar fluid residence times at the solution scale. Though bed scale preferential flow can be visually observed, diagnosing symptoms at the solution scale typically requires the generation, analysis, and modelling of residence time distribution (RTD) curves. The main objectives of this study were to firstly explore the effects that important material and fluid properties have on the steady state fluid flow profiles in drip irrigated beds characteristic of those used in laboratory scale column leaching studies and quantified using step tracer tests. This is based on the underlying principle that the movement of inert tracer molecules within an irrigated bed at steady state is identical to the solution flow path within the bed. The second objective was to test the ability of nine empirical and semi-empirical solute transport models to adequately fit the generated flow profiles or RTD curves. The third was to compare the magnitudes of the quantified model parameters to ascertain the level of solution scale preferential flow in the different beds and determine the adequate level of model complexity needed to describe their flow profiles which facilitates identifying the controlling variables within the system. Properties of the loading material that were identified as potentially most impactful with respect to heap operations were: porosity, wettability, particle shape and size distribution. Therefore, four different materials with unique inherent characteristics were selected for this study: glass beads (GB - spherical and non-porous), glass shards (GS - irregularly shaped and non-porous), greywacke (GW - irregularly shaped, porous, and highly wettable) and malachite ore (MO - irregularly shaped, highly porous, low wettability and non-uniform composition). In terms of fluid properties, current models have already established a correlation between the concentration of dissolved chemical species within a fluid and its viscosity. This was relevant due to the variety of lixiviant compositions used in previous heap hydrology studies and the fact that the composition also varies with time within a reactive heap. To study the effects of this parameter on the establishment of flow profiles, glycerol was used as a viscosity modifier to formulate solutions with viscosities ranging from 0.8 to 2.2 cP, representative of the range experienced in heap leaching systems due to varying SO4 2- concentrations. The packed beds were characterized using their bulk densities, void age, total liquid holdups, total bed saturations, 24-hour drain-down moisture percentages, solution and tracer breakthrough times. Beds containing both narrow and mixed particle size fractions were tested. The nine solute transport models used for RTD modelling included three compartmental model configurations (CM-1, CM-2, CM-3) and the tanks-in-series (TIS) model, all empirical in nature. The five semi-empirical models selected were the advection-dispersion (AD), piston exchange (PE), piston exchange-diffusion variant (PE-D) and piston dispersion and exchange (PDE) models. A novel model formulation called the piston dispersion and exchange-diffusion variant (PDE-D) model was also coded and tested, which incorporated both the longitudinal dispersion coefficient as well as a diffusional flux mass transfer mechanism. The CM-2, AD and TIS were mono porosity models assuming all solution volumes within the beds were actively flowing which limited their ability to account for solution scale preferential flow. The CM-1, CM-3, PE, PE-D, PDE, PDE-D models were dual porous, accounting for the presence of either dead or stagnant solution volumes. The model parameters used to account for preferential flow in the RTD profiles included: the fraction of dead to total solution volume, dynamic to total saturation fraction, number of TIS, ratio of parallel continuously stirred tank volumes, longitudinal dispersion coefficient, overall mass transfer coefficient and maximum diffusional pore length. The cumulative RTD responses for the bed systems composed of narrow size fractions were noticeably impacted by particle size. These systems displayed symptoms of increased solution channelling behaviour at steady state, based on their relatively short tracer breakthrough times, as the average particle size was increased from ∼1 to 15 mm. The incorporation of semi-empirical models which could account for stagnant volumes. The main comparative modelling results across all systems studied showed that the PDE and PDE-D models were the top performers, based on a model fit analysis. This was due to their dual porous nature and relatively higher levels of complexity. The mono porosity models (CM-2, TIS and AD) performed the worst due to their inability to account for isolated and immobile liquid volumes. However, when 10 mm during agglomeration will aid in increasing the fraction of mobile (actively flowing) liquid within the heap due to the increased presence of macro voids. High levels of particle porosities (>2.5 m2/g) will also aid in this aspect. This is proposed to be due to greater void network connectivity with an increase in porosity facilitating better mass transfer. These insights were obtained through the analysis of experimentally generated data and model simulations. They have provided a better understanding of the movement of fluid molecules within drip irrigated beds, which is essential for improved leaching performance. Building on this, the next step is to consider the effects of scale up and reactive systems on both empirical and simulated data.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:uct/oai:localhost:11427/38530 |
| Date | 12 September 2023 |
| Creators | Odidi, Michael Dumisane |
| Contributors | Harrison, Susan, Fagan Marijke |
| Publisher | Faculty of Engineering and the Built Environment, Department of Chemical Engineering |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
| Type | Doctoral Thesis, Doctoral, PhD |
| Format | application/pdf |
Page generated in 0.0029 seconds