MINERAL PROCESSING (SEPARATION, FLOTATION, CRUSHING, GRINDING, ENRICHMENT ETC.) : PRINCIPLES, TECHNOLOGIES AND INDUSTRIAL PRACTICE

In this paper a dynamic model of ore flotation is presented. The model has been conceived in such a way that its tuning is easily carried out by the use of commonly available information: Flotation kinematics experiments results (used in plant design process), plant dimensions and design throughput, valves and nozzles specifications, and some minor assumptions. The basic modeled module is a flotation cell with two phases: Slurry and froth. The phenomena involving the transportation of rich particles from the slurry to the froth by air bubbles is addressed. In spite of its simplicity, the model has proved to be highly effective for operator training, control system verification and commissioning, and plant design verification. The model is presented in detail along with an operator training case study.

The wide plant control integration of mineral processes poses many challenges. Flotation plants consist of complex interacting circuits where minerals are processed in different stages with recycling. Today, a common arrange are the RCS circuits, where a rougher circuit is combined with a cleaning circuit and a scavenger circuit. The global objectives of a flotation plant are to maximize the value metal recovery while the grade of the final concentrate is kept inside a narrow band. To achieve this, for any time variant feed attributes (flow rate, solid percent, grade, particle size distribution, pH, and chemicals), a capability of modifying the operation of each circuit, in order to achieve some local objectives, is demanded. For example, in a rougher circuit one can maximize the circuit Cu recovery while the mass and grade of rougher concentrate are constrained to some values. To achieve this one can find out which pulp level profile, along the flotation bank of cells, should be used. In case of forced air cells one can also find out which air flow rate profile should be used. These local targets are chosen in order to assure the produced concentrate stream can be adequately processed by the cleaning and scavenger circuits, to deliver a final concentrate meeting the product specifications. Furthermore, that will depend on how the flotation columns in the cleaning circuit and the cells of the scavenger circuits are operated. As one can see, the problem is far from trivial and the solution in terms of how the available resources in each circuit are set, will change according to the feed characteristics.Recently, Bergh and Yianatos have proposed an algorithm to change the set of a froth level profile in a rougher circuit aided by a rougher circuit simulator based on phenomenological models with parameters estimated by using industrial data, collected in experiments designed for that purpose. In this work, a simulator for the cleaning and scavenger circuits, including a regrinding stage is presented. This simulator is based on phenomenological models, with parameters estimated from industrial data. Details on the simulators and their use to provide insight on how to use the available resources in each circuit, to obtain some local objectives which are harmonized in a global control strategy, are discussed.

The circulation load of closed circuits calculation is often a problem when performing the mass balance of mineral processing plants. Iterative methods are a family of possible methods to be applied in the resolution of this calculation and consist in a finite loop where at each iteration the calculated solution is refined to produce a solution closer to the real one. The present work presents a simple iterative algorithm to calculate the closed circuit circulation load which allows the construction of reliable balances of mass, grade and water. The equations proposed in the algorithm were obtained through analyses of many circuits taking into account each process parameter and had been validated with industrial data acquired from a phosphate ore processing plant. Three different closed circuits, with different complexity levels, are presented to explain how the proposed algorithm works. The obtained results show the algorithm has converged for any industrial situation and it is able to solve the circulation load calculation with a few iterations and a small computational time.

The mining market is growing in such a way that getting enough specialized workforce for mining operations will be extremely difficult in the short term. As a result, technologies based on dynamic simulation are becoming a key pillar to educate and train inexperienced personnel at the pace required by this industry. Operator Training Simulator (OTS) systems have been widely used in the petrochemical industry, however the knowledge transmitted to the operators has not been "capitalized" in general. OTS systems are usually delivered as a hardware and software platform and not as a long term integral service.In this paper, an integrated solution for operator effectiveness, based on the conjunction of dynamic simulation, process knowledge, virtual reality and knowledge capitalization, through the creation of simulation scenarios and eLearning, is proposed.

Crushing plants are widely used around the world as a pre-processing step in the mineral and mining industries or as standalone processing plants for final products in the aggregates industry. Despite automation and different types of advanced model predictive control, many the processes are still managed by operators. The skill of the operators influences the process performance and thus production yield. Therefore, it is important to train the operators so they know how to behave in different situations and to make them able to operate the process in the best possible way. Different types of models for crushers and other production units have been developed during the years and the latest improvement is the addition of dynamic behavior which gives the crushing plants a time dependent behavior and performance. This can be used as a simulator for operators training. By connecting an Internet Human Machine Interface (WebHMI) to a dynamic simulator with the models incorporated, an on-line training environment for operators can be achieved. In this paper, a dynamic crushing plant simulator implemented in MATLAB/SIMULINK has been connected to a WebHMI. The WebHMI is accessible via the Internet, thus creating a realistic control room for operators’ training. In the created training environment, the operators can be trained under realistic conditions. Simple training scenarios and how they could be simulated are discussed. Apart from the increased level of knowledge and experience among the operators, the time aspect is an important factor. While a real crushing plant is still being built, the operators to be can already be trained, saving a lot of the commissioning and ramp up time.

Crushing plays an important role in the aggregates and mining industries by reducing the particle size of granular solids, such as rocks and ores. Nevertheless, the control of crushers has received a fairly little attention over the past decades. As a result, the main control objective today is still to regulate the crusher power or the closed side setting, not the actual size reduction or the product size. This paper presents two approximate methods for regulation of crusher product size distribution: Specific energy consumption (SEC)-based control and ratio control of selected product fractions. The basic idea behind ratio control is to regulate the crusher product size distribution by using a ratio of two measured product mass flows (e. G. Recirculated oversize fraction and total throughput) as a controlled variable. In SEC-control, on the other hand, the product size is not measured at all, but instead the disturbance rejection is achieved as a regulatory feature of a selected controlled variable. For both methods, the objective is still the same; Minimize the variation in the crusher output. The performance of the proposed methods is evaluated against currently used control methods in a simulation study. Moreover, the implementation of the proposed methods and their suitability for different process layouts will be discussed in detail. The paper will finally summarize the strengths and the weaknesses of each available control method and categorize the methods according to application and layout-specific suitability.

To achieve a balanced and optimal production at a mine, three tasks at least must be considered. The first one is to have an optimal production plan, including a set of optimal production targets for all process units. And the second task is to make sure that all those optimal targets in the mentioned optimal production plan must be achieved at all process units by using all resources available. Furthermore, to achieve the mentioned two tasks, all measurements, equipment and systems at the mine must be running and available in real time, particularly those critical measurements such as weightometers, densitometers, ore types; Stockpile levels. With the advancement and applications of technologies in mineral processing industry, an automated mine optimization system is developed to include the following three functions: (1) mine production optimizer; (2) various dynamic controllers; And (3) a set of soft sensors. A large amount of work has been done to develop the system, started from forming the concept to testing and implementation at various mines. The system can potentially help improve the production throughput up to 30% for a mineral processing operation. The system can be applied to various mining operations, such as coal mines, chromite mines, manganese ore mines, iron ore mines, and diamond mines.

This paper describes a practical implementation of a control scheme for two impact crushers operating in parallel as secondary stage crushers in a large-scale aggregate production plant in the United Kingdom. The control is based on measuring crusher power draw, mass flows and the size distribution of produced rock. The size distribution is estimated using image analysis. The settings, i.e. Rotor speed and impact plate gaps, of the impact crushers are then automatically adjusted to produce the desired end products. The system compensates for crusher blow bar wear and changes in feed material properties.This is the first time that such a control scheme has been implemented in a production plant. This paper describes the methods and controls used as well as future developments including an optimizing control based on a second, already installed camera. Unfortunately the plant in question has lately experienced equipment breakages so an analysis of the observed benefits is not available in time for paper preparation deadlines. Such an analysis can be presented at the conference, however.

Stable operation of grinding plants is of great importance as it ensures improved efficiency and mineral recovery. The majority of current control solutions in mineral grinding plants are based largely on expert control systems which aim to maximize throughput while keeping operational variables within predefined safe limits and a stable process operation. Nevertheless, these systems are not without disadvantages: They tend to systematize bad operational practices; There are no clear procedures to tune them and they exhibit poor response to unmeasured disturbances.Strategies based on predictive control, on the other hand, allow the handling of operational constrains, unmeasured disturbances and coupling of operational variables. Additionally, they are comparably easier to tune and are less sensitive to modeling errors.This paper presents a comparative analysis of four control strategies applied to a mineral grinding plant. The tested controllers are: (i) single centralized MPC, (ii) decentralized MPC for SAG mills and ball mills, (iii) multi-level control with a higher optimization layer and a lower decentralized MPC regulatory layer and (iv) multi-level control with a higher optimization layer including a coordinating expert module and a lower decentralized MPC regulatory layer. These four control strategies are implemented using various Honeywell's Profit software applications. In order to analyze the strategies, several performance indices are defined and tests are performed using a MATLAB/Simulink-based dynamic simulator.

New hydro-electrochemical technology is designed for processing concentrates of copper sulfides of Kajaran and Kapan (Armenia) deposits in a special mode (know-how). New technology provides a high degree of extraction of copper, iron, sulfur and precious metals, an economic efficiency and satisfies modern ecological requirements. The essence of technology lies in the fact that the copper concentrate, without drying, is immediately subjected to electrochemical dissolution. As a result the copper in the form of powder is deposited on the cathode, and then gradually deposited on the bottom of the electrolytic bath, and the iron goes into solution and subsequently can be removed by crystallization. Part of the sulfur is separated in the form of elemental sulfur from the decomposition of chalcopyrite and remains in the undissolved mass. The process is environmentally friendly because no gaseous or liquid waste is produced: Electrolyte is circulated in closed cycle. Thermodynamic and kinetic parameters of the process are presented.

Copper recovery and acid consumption are two key variables in heap leaching operations. The main idea of this work is to estimate these variables based on historic data and the previous knowledge available in the plant. In order to tackle the problem three key tools are required; I.e. Data exploration analysis, fuzzy logic modelling and optimization techniques. The data exploration methodology is used to process the data and it considers classification, variable selection and data correlation. On the other hand, the empirical knowledge used by the operators, and described in terms of fuzzy rules, is included in the model in order to provide consistent predictions even though there is not enough data available to support such predictions. Finally, an optimization technique is used to solve the problem of adjusting the model’s parameters by considering both the data and previous knowledge. As a result, it is possible to obtain a model that can be used to estimate the recovery and the acid consumption with a low uncertainty compared with other methods such as the one normally used in practice; I.e., simple linear in the parameters correlation models. This modelling methodology is applied to model the operation of a real industrial heap leaching process. The results clearly show that the time evolution of the recovery mainly depends on the mineralogy, and acid consumption on the temperature and the acid feeding. In addition, they also illustrate that model can predict these variables over a wide range of conditions and time span.

The paper will present a case where fuzzy logic was the logical choice to improve performances of a semi-autogenous grinding (SAG) mill. The SAG mill stability had to be improved and throughput increased. The process is multivariable, strongly non-linear, and before implementing this system, the operators were actively manipulating many variables with varying success depending on operator experience and occurring disturbances. The paper is divided into three main sections. The first section gives an introduction in which the characteristics of model-based versus expert systems and fuzzy logic control are examined: How to determine the best approach? How to decide between ruled-based approaches and model-based approaches? How to balance advantages and disadvantages, complexity and simplicity, investment and results? A decision tree is presented to select the right approach. The second section describes the approach selection and the SAG mill process. For the SAG mill process, fuzzy logic was the logical choice. The solution is robust, simple and is implemented in the actual control system, a programmable logic controller with fuzzy logic function blocks.Finally, the third section is devoted to the implementation, commissioning and optimization of the controller. This section will also touch on training, maintenance, improvements and results achieved. Specific energy reduction, improved stability and throughput increase are among the main recommendations and conclusions.

Several dynamic models are able to represent the phenomenological behavior of flotation. On the other side the interest in developing control strategies for large-scale processes has led to formulate novel methodologies which allow to consider the global behavior of a plant, facilitating the design, validation and evaluation of more complex optimizing control strategies.In this work, we first develop a dynamic hybrid model for a flotation rougher circuit with two production lines composed by four cell banks in series, connected by controlled valves. Each line is actuated by discrete valves, whose interaction with different bank pulp levels define several operation modes. The model was validated with data from an industrial plant.Subsequently, a hybrid model predictive control (HMPC) strategy is proposed, valid for the different operation modes established. This optimizing controller considers a hybrid prediction model obtained by applying identification techniques to the different scenario. Our simulation results show that the proposed methodology is suitable for modeling the global behavior of a rougher flotation circuit, representing its dynamic and providing a robust control for several operation modes of the circuit.

Hydrocyclones are devices worldwide used in mineral processing and used for desliming, classification, selective classification, thickening and pre-concentration. Versatile in application, the hydrocyclone is the standard classifier used in closed circuit milling in mineral processing plants. A hydrocyclone is composed by a cylindrical and a conical section joint together, without any moving parts and it is capable of perform granular material separation in pulp. The pulp is feed under pressure tangentially to the cylindrical section. The granular separation mechanism is complex and it´s mathematical modelling is empirical. The most used model for hydrocyclone dimensioning was proposed by Plitt (1976). Combining the first industrial database on cyclones generated at JKMRC (Rao, 1966) with his own laboratory data, Plitt developed an alternative general- purpose cyclone model. Over the years many revisions and corrections in Plitt´s model were proposed. In the present paper the influence of two operational parameters (percent solids in feed by volume and the pulp feed flow) in seven models based in Plitt´s models were studied in reference to the corrected classification size (d50c). To do so a hydrocyclone using Ritema´s geometry (10 cm of diameter) and iron ore pulp was used. The founded results allow the proposal a different value for Plitt´s model constant, resulting in another revision for the model.

Hydrocyclones are devices worldwide used in mineral processing and used for desliming, classification, selective classification, thickening and pre-concentration. Versatile in application, the hydrocyclone is the standard classifier used in closed circuit milling in mineral processing plants. A hydrocyclone is composed by a cylindrical and a conical section joint together, without any moving parts and it is capable of performing granular material separation in pulp. The pulp is feed under pressure tangentially to the cylindrical section. The granular separation mechanism is complex and its mathematical modelling is empirical. The most used model for hydrocyclone dimensioning was proposed by Plitt (1976). Combining the first industrial database on cyclones generated at JKMRC (Rao, 1966) with his own laboratory data, Plitt developed an alternative general- purpose cyclone model. Over the years many revisions and corrections in Plitt´s model were proposed. In the present paper the influence of two operational parameters (percent solids in feed by volume and the pulp feed flow) in seven models based in Plitt´s models were studied in reference to the corrected classification size (d50c). To do so, a hydrocyclone using Ritema´s geometry (10 cm of diameter) and iron ore pulp were used. The founded results allow the proposal a different value for Plitt´s model constant, resulting in another revision for the model.