Search Documents

By K. Awuah-Offei

Long-term production scheduling optimization has been a challenging issue for the mining industry because of the size and complexity of the problem. The current planning algorithms have limitations in terms of addressing the stochastic variables underlying the mine planning problem. In this paper, an intelligent agent-based mine planning framework based on reinforcement learning is introduced. The long-term mine planning is modelled as a dynamic decision network. The intelligent agent interacts with the block model by means of a stochastic simulation and employs a Q-learning algorithm to learn the sequence of pushbacks that maximizes the net present value of the mining operation. The intelligent open pit simulator (IOPS) was implemented with an object-oriented design in Java®. A comparative application case study was carried out to verify and validate the models. The proposed method was used in planning an iron ore deposit and the results were compared to the Milawa scheduler used in Whittle® software. The outcome of the study demonstrates that the intelligent agent framework provides a powerful basis for addressing real-size open pit mine planning problems.

Jan 1, 2008

By P. G. Roden, G. A. Maki, P. J. Fullmann

"In October 1984, Rocanville Division began using a stacking system to handle potash mill tailings. Previous to this date, mill tailings were slurried to the top of the tailings pile and allowed to debrine while flowing mobile stackers. The stackers consist of cyclone packs to debrine the tailings and drag conveyors with screw discharges to deposit the tailings onto the existing pile. The paper will deal with the various aspects of the project from conception to its present state. Key Words: stacking, mobile stackers, Linatex Separators, potash mill tailings, reclaim brine ponds, settling ponds, tailings pile. IntroductionPrevious to 1984, when the stacking project was first implemented, potash mill tailings were handled as follows. Mill tailings were first debrined to approximately 65% solids using a cyclone cluster. The cyclone underflows were then introduced onto a vacuum scroll filter table. Here the tailings were further debrined and discharged into the disposal tank where the tailings were reslurried with reclaim brine and pumped to the tailings disposal site. At the disposal site the slurry was discharged onto the tailings pile and by a natural settling process the tailings were debrined with the brine ultimately collected in the reclaim brine ponds. The shape and profile of the tailings pile was controlled by building salt dikes to control the flow of the discharged tailings slurry. As the tailings pile grew in height, booster pumps were added to the disposal system."

Jan 1, 1986

By E. Shum, V. G. Papangelakis

"Attempts to reduce the energy requirements of water recovery in the process industries have resulted in the development of techniques to freeze and purify aqueous solutions that take advantage of local climatic conditions. Natural freezing is a technique by which aqueous solution is exposed to freezing air temperatures, resulting in the production of a layer of ice on the surface. As the ice layer tends to exclude solute, a concentrated solution is formed. However, due to dendritic ice growth along the advancing growth front, small quantities of solution are trapped within the ice, decreasing the purity of the water recovered. The ice produced can then be allowed to thaw in order to recover the water using low grade plant heat. As the more concentrated parts of the ice melt first, a purer ice remains. In this study, 0.5 molal magnesium sulfate solution was frozen under simulated natural freezing conditions at temperatures between –1 and –6°C. The ice produced was thawed at either room temperature or 50°C. Up to 40% of the water was recovered with a separation efficiency of above 95% when thawing at room temperature.INTRODUCTION Advances in refrigeration technology have made low-temperature separation processes increasingly appealing to an energy-conscious society. The latent heat of freezing of water is about seven times smaller than the latent heat of evaporation providing freezing with a strong thermodynamic advantage over the traditional evaporation-based processes. The result is the development of a wide variety of processes that fall under the term Freeze Concentration (FC), which describes a variety of techniques in which an aqueous solution is concentrated by the creation and subsequent removal of ice (Williams et al., 2015).FC processes exploit the freezing behaviour of aqueous solutions as dictated by their respective water-salt phase diagram. When, for example, a solution of magnesium sulfate is cooled to a temperature below its freezing point but above its eutectic temperature of –4.1?C (Figure 1) then, if the concentration of MgSO4 is less than the eutectic concentration (about 1.7 m), ice will form until the solution concentration reaches its equilibrium concentration at the given temperature on its phase diagram. If the concentration is greater than the eutectic concentration, MgSO4 solid will precipitate until the concentration reaches equilibrium. It is for this reason that ice can be selectively produced in FC processes (Randall & Nathoo, 2015)."

Jan 1, 2017

By John Russell

"The South American mining community leads the world in recognition and adoption of the latest grinding mill relining technologies. Clearly this region has recognised the benefits of improved relining performance.Six of the seven RUSSELL TWIN 8 Mill Relining Systems, ordered to date, are either operating or being manufactured for South American mine sites.This paradigm shift, towards very short relining times, aims at displacing mill relining activities from the critical path of concentrator maintenance shutdown schedules.The first uses of the RUSSELL TWIN 8 Mill Relining System have seen some of the highest SAG mill new liner placement rates ever observed by RME, as would be expected with two, independent Machines working simultaneously in the one mill.The liner-replacement-rate, averaged over the whole reline, is good (27% improvement over single Machine System in the same mill) but short of full potential. This modest result can be attributed to reline crews unfamiliarity with RUSSELL TWIN 8 System operation.This paper compares a sample of conventional (single machine) RME Mill Relining System performance data with one ‘double head’ mill liner handler by others. Early performance of Escondida’s new RUSSELL TWIN 8 System performance is then compared with recent single Machine data from Escondida, immediately prior to RUSSELL TWIN 8 commissioning. RME and Escondida look to constantly improving RUSSELL TWIN 8 System performance as full operational familiarity is achieved.Also, the preparation of this paper revealed a need to develop a global industry standard for mill relining data collection. A web based platform, to develop and then distribute standardised data collection templates, is discussed in Annexure 1."

Jan 1, 2010

By Ryan W. Ulansky

This paper presents the conceptual design of a magnetically-levitated skip for use in an underground mine. A skip is used to move rock from an underground mine to its surface processing facility. Conventionally the skip is raised and lowered by cables, similar to an elevator. In a magnetically-levitated skip, the cable is replaced with a linear induction or synchronous motor to provide propulsion and levitation. Removal of the cables in a magnetically-levitated skip provides many advantages over a conventional system including the ability: to negotiate corners, to travel vertically in the shaft as well as horizontally to the mining face, and to move a very large volume of material using multiple skips in a small shaft. A magnetically-levitated skip is a safer, more environmentally-friendly, productive alternative to conventional skipping. It has the further advantage to decrease haulage distances for underground vehicles.

May 1, 2002

By Robert Cluff

The Campo Morado precious-metal-bearing, volcanogenic massive sulphide district occurs in a Lower Cretaceous felsic volcanic sequence in a northerly trending belt in the Guerrero Terrane in northeastern Guerrero, Mexico. During upper Cretaceous to Early Tertiary greenschist-facies regional metamorphism, the rocks were deformed strongly into a northeast-verging fold-and-thrust belt. The deposits belong to a low-sulfidation, volcanogenic massive sulphide system formed in a subaqueous environment, and are of the siliciclastic type. Most are on the flanks of felsic domes or in adjacent basins and many are associated with heterolithic volcaniclastic rocks. The Reforma and possibly the El Rey massive sulphide deposits are interpreted to be overturned, and the Naranjo, El Largo, and Estrella de Oro deposits and the El Profundo occurrence to the south are upright. The G9 deposit and the La Lucha and San Rafael occurrences are in an over-thrust plate further to the southwest. In several of the deposits, zinc, lead, silver, and gold are concentrated near the stratigraphic top, and copper is concentrated near the stratigraphic base; copper also is concentrated in replacement bodies near the top of some domes. Major minerals are pyrite, quartz, ankerite, sphalerite, chalcopyrite, and galena. Gold and silver occur in argentian gold, and silver also occurs in tennantite/freibergite and galena. Underlying stockwork zones contain disseminated and vein quartz, pyrite, chlorite, and ankerite, with locally abundant chalcopyrite and/or sphalerite. Exploration since 2004 has enlarged greatly the El Largo and the high-grade G-9 deposits, and slightly enlarged the El Rey deposit. The cumulative in situ massive sulphides for the Reforma, Naranjo, El Rey, El Largo, and G9 deposits exceed 45 Mt.

May 1, 2006

By Hiroaki Sakurai

Using batch reactors equipped with CO and CO2 gas sensor, CO oxidation over Au catalysts in room air has been investigated and compared with Pt catalysts. For both Au and Pt catalysts, the logarithm plot of CO concentration data during initial 10 min are well fitted by straight line, which mean the first order dependence on the CO concentration between 10 and 10,000ppm. The rate constant value (k) over Au and Pt catalyst generally decrease with increasing initial CO concentration above 100ppm. The value decrease especially rapidly between 100ppm and 1,000ppm over Pt catalysts, which should be due to the strong adsorption of CO on Pt. The k value over Au catalyst has become ten times or more as large as the Pt catalysts at 1,000ppm. Already for the initial concentration of 100ppm CO, the strongly adsorbed CO on Pt metal has been clearly observed by FT-IR during the reaction over Pt/Al2O3. A part of the strongly adsorbed CO has been remained on Pt after the disappearance of gas phase CO, and has gradually changed to gas phase CO2. Such strongly adsorbed species on Au has never been observed on Au/TiO2 under the same reaction conditions. These features of Au catalysts are favorable for indoor air purification.

Oct 1, 2003

By Claude Dufresne

The technical development of the niobium-bearing products is to a great extent influenced by the striving of the respective industries to maintain or strengthen their positions in all fields of application against competing materials. South East Asia, China, Eastern Europe and CIS are slowly becoming significant consumers of ferroniobium. As a result, the world niobium market has been stable despite the slowdown in steel production in 2001. This paper will describe present producers, capacities and the current market conditions.

May 1, 2003

By T. Belem

The main objective of this study is to better understand the effect of underground CPB placement conditions on its microstructural, hydro-mechanical and geochemical properties. To reach this objective, three PVC columns, each 3 m high, were built, and were filled with CPB material. The columns allow simulating undrained (UD), half-drained (HD) and fully-drained (FD) conditions along with the measurement of resulting self-weight consolidation settlement of the CPB. Bleed and drainage waters were also collected from each column for geochemical analyses (pH, Eh, EC, SO42- content). After 90-days of curing time, a total of sixty CPB samples were cored from the backfill columns and subjected to uniaxial compression tests followed by physical characterization (Gs, void ratio, water content, degree of saturation, specific surface area). The results showed that the total drainage water and the maximum observed settlement occur mainly within the first 72 hours after the columns are filled. The corresponding volumetric strains suggest that in situ backfilled stopes simulated behave in a similar way to the fully-drained (FD) or the half-drained (HD) conditions. The MIP test results show that the pore structure varies between the top and the bottom of each column.

May 1, 2007

By P. Tittes, M. Deal, M. Spicher, S. McDaniel, J. Wickens

"Haile Gold Mine is located near Kershaw, South Carolina, approximately 50 miles south of Charlotte, North Carolina. Gold was first discovered at the site in 1827, and the property has been through several operating periods since. The relatively low grade and refractory nature of the deposit has been a challenge for efficient processing since the start of mining on the property. Early processing techniques included gravity separation and barrel chlorination. The last operation employed heap leaching in 1985 through 1992. After being closed and reclaimed for over 20 years, the property is poised to re-open and produce once again. In this chapter of the Mine’s legacy, flotation and ultra-fine grinding will be employed to achieve economic extraction of gold and silver from the pyrite. This paper briefly reviews some of the mine’s history, the metallurgical testing, the flowsheet and discusses design considerations.INTRODUCTIONHaile Gold Mine is located near Kershaw, South Carolina, approximately 50 miles south of Charlotte, North Carolina. Gold was first discovered at the site in 1827, and the property has been through several operating periods since. The last operation employed heap leaching in 1985 through 1992. Since then, the property has been closed and fully reclaimed. Romarco Minerals acquired the property in 2007, and began an exploration program that has delineated over 62 tonnes (+2M ozs.) of gold in reserve at an overall grade of 2 ppm (0.06 opt). This paper provides a brief history of the property, a summary of metallurgical testing, description of the process flowsheet and discusses design considerations.HISTORY1827 – 1837Gold was discovered in creek beds on the Haile property in 1827. The property was owned by Benjamin Haile, and the first few years of production came from placer mining. After gold was discovered in schist beds to the sides of the streams, Haile expanded the operation by leasing 18 meter-square (50 ft.) plots to local planters. These planters used slaves to work the small excavations in the agricultural offseason. The gold ore was taken to arrastras where the free gold was liberated and recovered. The excavations were above the water table, generally less than 18 meters (60 ft.) down, and it was noted then that gold could be found embedded in sulfides and difficult to extract."

Jan 1, 2013

By D. Wolfe, R. Cleminson

"The Aurora North mine has been operating for more than ten years, and is the source of over half of Syncrude’s bitumen production. The slurry systems at Aurora experience high rates of wear, and mission lengths on current slurry pumps are nominally 1500-2000 hours. Recent work has enabled a better understanding of the significant adverse effect of large particle impingement damage to impeller vane leading edges - which is the known life determining issue for Aurora hydrotransport pumps. A new pump design has been developed with a larger diameter impeller and increased vane lengths, and is expected to triple Aurora’s mission length for hydrotransport pumps.INTRODUCTIONSlurry Pumps are Critical Equipment at SyncrudeSlurry pumps are a critical component in Syncrude’s operations. In a typical Aurora application, each pump stage helps move over 8000 dry tonnes per hour of oil sand, plus an equal volume of process water that has been added to create the slurry. The average impeller life in oil sand hydrotransport service at Aurora has been nominally 1500-2000 hours, or 12-16 million tonnes of oil sand pumped. Generally, the slurry pump application at Aurora has proven very difficult to improve due to the unique combination of high sand abrasivity, 125 mm (5”) large particle top size (lumps), and high flow rate relative to basic pump size; a factor that results in higher than optimal impeller vane speeds for the service conditions.PROJECT BASISSlurry Pumps are the Reliability Bottleneck at AuroraSlurry pumps are currently the shortest-lived components that determine the overall system run length at Aurora, in Hydrotransport, Extraction and Tailings. Vibrating screens and piping had been previous bottlenecks, but their reliability has been improved through upgrades in equipment design and wear materials. Both vibrating screens and piping can now achieve 4000 hour mission lengths."

Jan 1, 2013

By C. Doucet

Since 2001, CANMET Mining and Mineral Sciences Laboratories (MMSL) have worked with Falconbridge Ltd. (now Xstrata Nickel) to evaluate trends in the strength development and water retention properties of a novel backfill material for underground mining, called gelfill. Between 2004 and 2006, extensive laboratory tests were undertaken under the leadership of the Deep Mining Research Consortium (DMRC). This paper compares results obtained from samples circulated through a mini-flow loop and a simple mixer. Although many parameters were studied over the last five (5) years, this paper focuses on the investigation of the effect of the mixing method, sequence and time, and the curing environment on strength development and water retention of gelfill. It was found that binder should be added as a powder to the mix in the mini-flow loop, and that mixing slurry in the mini-flow loop as opposed to the mixer generally yielded higher strengths. The practical implications of these observations are the subject of an on-going underground trial that will compare laboratory and in situ properties of regular slurry fill with gelfill.

May 1, 2007

By M. Helinski

In the authors? experience inappropriate water management in hydraulic fill masses has been the root cause of most barricade failures. Inappropriate water management can lead to high water pressures, reduced effective stresses and increased barricade stresses (Kuganathan 2002). In addition the high hydraulic conductivity and non-plastic nature of most hydraulic fills makes susceptibility to piping failure (Cowling et al. 1987) a high possibility in many hydraulic backfills where a static water head is available. In order to avert both of these failure mechanisms it is essential to control the elevation of the phreatic surface within the fill mass. This is usually achieved through campaigned filling and resting periods. The duration of filling and resting periods are often estimated using some form of stope drainage simulator. Due to the random nature of many characteristics that influence the fill mass hydraulic conductivity, simulation of hydraulic fill drainage can be difficult. The high consequences of inaccurate phreatic surface estimates combined with the high variability in drainage properties often results in operators taking a conservative approach to this modelling. This conservative approach can lead to significant delays in the filling cycle which can have a major impact on the productivity of many hydraulic fill operations and can often results in an unnecessary movement towards capital intensive paste fill systems. This paper presents a water management technique which combines a simplified drainage model with in-situ monitoring to provide a rational approach to water management. With an accurate drainage model filling operations can be carried out at optimum efficiency without compromising safety. The implementation of this strategy has lead to a significant increase in filling rates at a number of Australian filling operations. A brief case study from one of these sites has been presented.

May 1, 2007

By Daniel Hodouin, Claude Bazin, René del Villar

"A traditional linkage has long been established at Laval University's Department of Mining & Metallurgy between the Mineral Process (MP) and the Mining Exploitation (Min) on one hand, and between the Extractive Metallurgy (EM) and the Materials Engineering (Mat) on the other. This academic separation of Mineral Processing from Extractive Metallurgy at the educational level has become increasingly at odds with today's industrial practice in which both disciplines are not only complementary but also often intertwined. For instance, in gold ore processing, comminution, flotation, and gravity separation units are interfaced with atmospheric or pressure leaching modules and with carbon adsorption and electrowinning units. The new copper plants in South America combine comminution and flotation circuits with heap leaching, solvent extraction and electrowinning processes.In order to reflect this new industrial reality and to harmonise the options offered in both its engineering programs, the Department of Mining and Metallurgy of Laval University has introduced major changes in its curriculum in 1996. These changes reflect also the recognition that the somewhat secondary status given so far to Mineral Processing and to Extractive Metallurgy within their ""mother"" programs, Mining/Mineral Engineering and Materials metallurgical Engineering respectively, was in sharp contrast with their actual significance for the Canadian economy. Leading industrial voices have raised similar concerns and the Canadian Engineering Accreditation Board has also addressed this problem in its latest program evaluation.The changes implemented at Laval aimed at a) better representing MP and EM within the two undergraduate programs; b) training top-level versatile and multidisciplinary engineers, fluent in all aspects of MP and EM, irrespectively of which of the two ((mother)) programs they had chosen and capable of meeting today's industrial requirements; c) training students in process analysis, optimisation and automatic control using state-of-the-art software, and d) taking into account the co-operative nature of both undergraduate programs.Results are very encouraging in terms of student enrolment and industrial response. However, some concern is expressed on the sustainability of the renewed curriculum considering the drastic budget restrictions in effect at Laval as in most other Canadian universities."

Jan 1, 1999

By Q. Zhang

The necessary capability of the gravity transportation pipeline system for cemented coal gangue backfilling at the Suncun Coal Mine, China, is about 74.4 m3?h-1 according to mining production and working times in the mine. It is estimated that the working velocity and maximum pipe length-backfill depth ratio of the backfill system are 3.82m?s-1 and 5.4, respectively and that the probable backfill capability of the system is 110.18m3?h-1. Results of hydraulic calculations for the backfill transportation system show that the designed backfill system is reliable in flowability and sufficient for backfill capability needs.

May 1, 2007

By K. D. Mills, R. G. Robins, C. G. Anderson, L. G. Twidwell

Arsenic is a metalloid which has attracted great environmental scrutiny in recent decades. Moreover, its control and deportment is becoming increasingly important in the metallurgical processing of modern complex ores and concentrates. As such the authors of this publication have a combined professional experience of well over 100 years in the study and hydrometallurgical treatment of arsenic. Hence, this paper will cover arsenic fundamentals along with a review of technology for removal, fixation and stabilization. Finally, an application of industrial Alkaline Sulfide Leaching (ASL) of a copper enargite concentrate will be elucidated.

Jan 1, 2014

The #3 Shaft at Brunswick Mine, Bathurst, New Brunswick, Canada, intersects a soft graphite zone just below the 3400 Level. Ground movement within this zone causes the concrete liner of the shaft to be deflected towards the shaft?s interior, affecting the steel framework of the conveyance. In some areas, the clearance between the cage and the liner has been reduced to less than 2 cm, which is a concern. An innovative surface-mounted extensometer was designed in 2008 to monitor potential movements of the shaft?s liner. Four such novel extensometers were installed in the shaft for the surveillance of a particularly critical 3.0m by 5.5m area. This article describes the approach taken to instrument the shaft, the principle of the new extensometer, as well as the performance of the instrumentation following its installation.

May 1, 2009

By Norman O. Lotter

Drill core is preferred as the sample material for metallurgical and mineralogical characterisation and flowsheeting of an orebody. The question arises as to how to do this at the economic optimum, since geologists have several other strategic uses for their drill core. The limiting factors are the cost of extra drilling for the metallurgical testwork and the need to preserve the drill core record. In this study, the representative sampling of drill core at an economic optimum for metallurgical testing is addressed. This phase of the drill core work is preceded by earlier drill core characterisation - for which other appropriate sampling protocols are used - that has provisionally categorised the orebody into geometallurgical units. In stratified sampling of drill core, use is made of a version of the stratified approach to compound distributions proposed by Cochran in 1946. Use is made of Gy’s 50-piece minimum sample mass model to estimate minimum requisite sample mass. In the case where sampling and testwork are required on a Lifeof- Mine blend of a given ore deposit, whereby defined ore types or geometallurgical units are present, the sample mass requirement is significantly reduced by working from drill-core at the geometallurgical unit level. A case study using Raglan drill core is discussed. Using nickel as an indicator paymetal, and sampling at a fundamental variance level of 8%, the minimum sample masses for disseminated sulphides, net-textured sulphides, and massive sulphides are 3.80, 7.37 and 4.39 kg respectively. When the same drill-core is sampled as a Life-of-Mine mixture without knowledge of the geometallurgical units, the minimum sample mass increases to 84.28 kg. It is thus possible to assemble a representative Life-of-Mine sample of drill-core more economically, at the same time validating its representativity. Equivalent calculations for copper, platinum and palladium are discussed, with outcomes consistent with this finding. When all four paymetals are considered, one selects the largest minimum mass out of all four cases. This results in minimum sample masses of 11.67, 74.90 and 175.16 kg for disseminated sulphides, net-textured sulphides and massive sulphides respectively. The simulated Life-of- Mine blend would require a minimum sample mass of 377.33 kg to cover all four paymetals.

Jan 1, 2010

By C. Patel, George H. Hope

"INTRODUCTIONThe Hemlo area is located in Northern Ontario, some 350 km east of Thunder Bay along highway 17. Hemlo is Canada's largest, recent discovery of gold ore. Three companies, Lac, Noranda, and Teck are currently developing mines and processing facilities in the area. A combined milling capacity of approximately 10,000 tonnes per day will ultimately be available in the area. The gold yield will be between 750,000 to 1,000,000 ozs. per year. This will slightly more than double Ontario's gold production but only represents 2.5% of the total world gold production.Lac Minerals Ltd. is substantially committed to the gold mining business. Lac currently operates four (4) gold mines and mills and produces 300,000 ozs. per year. Lac's Hemlo development will put the company in the position of number one gold producer in Canada.The Lac Hemlo ore body has an estimated reserve of 37 million tonnes at 6.2 g/t. The mining will take place at an initial rate of 3,000 t/d commencing in the first quarter of 1986. The mine and plant will be expanded to 6,000 t/d in 1988. At this capacity, Lac will be the largest producer in the Hemlo area.1. GEOLOGYThe orebody is divided into two (2) zones for mining purposes. The 'A' Zone is to the east and dips to the Northeast from surface to 300 m below surface. This zone has a strike length of 250 m to 300 m at an average width of 20 m. The upperpart of the zone will be mined by open pit and the remainder from underground. The 'B' Zone, to the north west of the 'A' Zone, although lined up is not necessarily an extension of 'A' Zone. 'B' Zone has a known strike length of 600 m and is open to depth and along the strike. It is over 70 m wide in the core. This zone holds most of the property's known reserves. 'C' Zone with narrow gold zones is under exploration in the south west portion of the property."

Jan 1, 1985

By G. A. Kosick, C. A. Harris

"Metallurgy at the Polaris Mine is approaching an impasse, with lead and zinc concentrate grades and recoveries approaching their theoretical maximums. In an effort to increase productivity, plant throughput and operating material costs became a major focus of milling personnel.one method of increasing plant throughput was to minimize flotation circuit upsets by; standardizing operating strategy, decreasing control action response time of operators, increasing training of operators, and providing a more available decision support system for operators.To achieve these goals, Polaris personnel decided to utilize expert system technology to develop a real-time process controller for the flotation circuit which would be capable of justifying its actions to operating personnel.This paper describes the procedures followed in constructing the first process controller in the mineral industry which is based on the principles of Artificial Intelligence.IntroductionThe Polaris Concentrator was constructed in 1980 on a barge in Trois Rivieres, Quebec and towed to its present location on Little Cornwallis Island, North West Territories, about 1450 kilometers from the North Pole. The plant presently treats about 2,800 MTPD of new feed with average feed grades of 3% Pb, 13% Zn and 5% Fe. Current production averages 36,000 tonnes of lead concentrate and 220,000 tonnes of zinc concentrate annually.In an effort to increase the efficiency of the concentrator, the decision was made to implement an innovative control system on the flotation circuit. This system would embody and utilize heuristics ""rules-of-thumb"" gained from operators to make on-line process decisions. It would also justify these decisions on a demand basis."

Jan 1, 1988