CONTINUOUS CENTRIFUGAL CONCENTRATOR OPERATION AND CONTROL by Michael J. McLeavy B.A.SC. T H E UNIVERSITY OF BRITISH C O L U M B I A , MINING A N D M I N E R A L PROCESS ENGINEERING, 1999 A THESIS SUBMITTED IN PARTIAL F U L F I L M E N T OF T H E REQUIREMENTS FOR T H E D E G R E E OF Master of Applied Science in THE FACULTY OF GRADUATE STUDIES MINING AND MINERAL PROCESS ENGINEERING The University of British Columbia March 2005 © Michael J. McLeavy, 2005 ABSTRACT A study was conducted to evaluate the operation and control of a continuous centrifugal gravity concentrator. The unit that was tested was a pilot scale Knelson Continuous Variable Discharge (CVD6) Concentrator. The primary objective was to determine the effects of the machine operating variables on separation performance; concentrate grade and recovery. The secondary objective was to use the results to develop an operating strategy for the concentrator. Thirdly, in testing the machine the practical limits of the equipment were determined in order to develop a basic understanding of where this technology could be applied commercially. To achieve these objectives four sets of pilot scale experiments were conducted including: 1. Quartz/magnetite statistical experimental design 2. Quartz/magnetite incremental testing 3. Synthetic sphalerite/quartz incremental testing 4. Pb/Zn flotation tailings plant testing at Hudson's Bay Mining and Smelting Statistical experimental design was used to rank the effects of operating variables on concentrator grade and recovery. The following table lists the operating variables in order, from most significant to least significant: Rank Variable rankings Variable rankings based on Grade based on Recovery 1 Solid Feed Rate Heavy Particle Size 2 Bowl Speed Pinch Closed 3 Pinch Closed Bowl Speed 4 % Solids %Solids ii 5 Fluidization Fluidization Water Flow 6 Feed Grade Pinch Open 7 Pinch Open Feed Grade 8 Heavies Particle Size Solid Feed Rate The first stage of research in evaluating the operating variables of the CVD was to evaluate the feed and machine variables using a statistical experimental design approach. This methodology was selected because of the large number of variables and was the only practical way to evaluate the individual variables and their interaction effects. A synthetic feed of quartz and magnetite was tested in a pilot plant that included a Knelson CVD6 concentrator. The result of the factorial design was a ranking of effects on both grade and recovery for each of the feed and machine variables. Testing of both synthetic quartz/magnetite and synthetic quartz/sphalerite resulted in a better understanding of the trends that result from incremental changes in operating variables. The plant testing at HBM&S created a large database of information that confirmed the information from the laboratory scale tests. Based on the results an operating strategy was developed that includes the following general strategy: • Identification of performance expectations • Testing to produce an array of results that span the spectrum of concentrator abilities • Identification of a performance boundary layer iii • Qualitative tuning of operating variables to approach boundary layer conditions • Fine tuning of operating variables using knowledge of the impact of individual operating variables Combining information from all of the test programs, the effect of particle specific gravity differential was investigated. The result was that the CVD was shown to be able to separate particles down to a specific gravity differential of 0.2. However, the practical limitation where a reasonable recovery was obtained was a differential of 2. iv T A B L E O F C O N T E N T S ABSTRACT I TABLE OF CONTENTS V LIST OF FIGURES XI LIST OF TABLES IX ACKNOWLEDGEMENTS XIII 1 INTRODUCTION 1 2 LITERATURE REVIEW 2 2.1 HISTORY 2 2.2 BATCH AND USE : 2 2.3 EVOLUTION OF CONTINUOUS CENTRIFUGAL CONCENTRATORS FROM BATCH TECHNOLOGY 4 2.4 FALCON C 5 2.5 KELSEYJ IG 7 2.6 K N E L S O N C V D ". 11 2.6.1 Knelson CVDApplications 14 3 EXPERIMENTAL PROGRAM 14 3.1 TEST PROGRAM 14 3.2 PILOT SCALE CIRCUIT 15 4 QUARTZ MAGNETITE FACTORIAL DESIGN 21 4.1 INTRODUCTION 21 4.2 EXPERIMENTAL 22 4.2.1 Procedure 22 4.2.2 Quartz/Magnetite Assay Procedure 24 4.2.3 Attrition Check 25 4.2.4 Test Program Details 27 V 4.3 RESULTS 31 4.3.1 Introduction 31 4.3.2 Effect of Operating Variables on Grade 33 4.3.3 Effect of Operating Variables on Recovery 35 4.3.4 Conclusion 37 5 QUARTZ/MAGNETITE INCREMENTAL TESTING 40 5.1 INTRODUCTION 40 5.2 EXPERIMENTAL TEST PROGRAM 40 5.3 QUARTZ/MAGNETITE INCREMENTAL TESTING RESULTS 42 5.3.1 Introduction 42 5.3.2 Quartz/Magnetite Trends with Changing Bowl Speed. 43 5.3.3 Quartz/Magnetite Trends with Changing Fluidization Water Flowrate 44 5.3.4 Quartz/Magnetite Trends with Changing Percent Solids 47 5.3.5 Conclusion 48 6 LAB TESTING - SYNTHETIC SPHALERITE/QUARTZ 49 6.1 INTRODUCTION 49 6.2 EXPERIMENTAL - SYNTHETIC SPHALERITE/QUARTZ 50 6.2.1 Feed Characterization 50 6.2.2 Test Program 53 6.2.3 Procedures and Sampling 55 6.2.4 Sample Preparation 55 6.3 RESULTS -SYNTHETIC QUARTZ/SPHALERITE 56 6.4 CONCLUSIONS 59 7 HUDSON'S BAY MINING AND SMELTING PLANT TRIALS 60 7.1 INTRODUCTION 60 7.2 EXPERIMENTAL-PLANT TRIALS AT HUDSON'S BAY MINING AND SMELTING 61 7.2.1 Test Program 61 vi 7.2.2 Procedures and Sampling 63 7.2.3 Sample Preparation 65 13 RESULTS - PLANT TRIALS AT H B M & S 66 7.3.1 Introduction 66 7.3.2 Bowl speed 66 7.3.3 Pinch valve open time 67 7.3.4 Pinch valve closed time 70 7.4 CONCLUSIONS 72 8 S P E C I F I C G R A V I T Y D I F F E R E N T I A L 73 8.1 INTRODUCTION 73 8.2 DISCUSSION 73 8.3 CONCLUSIONS 7 7 9 V A R I A B L E T U N I N G I N A C V D C O N C E N T R A T O R 79 9.1 INTRODUCTION 79 9.2 OPERATING VARIABLE PERFORMANCE RATIO 79 9.3 CONCLUSIONS 84 10 D E V E L O P I N G A N O P E R A T I N G S T R A T E G Y F O R A N E W A P P L I C A T I O N 86 10.1 INTRODUCTION 86 10.2 IDENTIFICATION OF THE OBJECTIVES 86 10.3 C V D BOUNDARY CONDITIONS 87 10.4 QUALITATIVE TUNING 88 10.5 FORMULATION OF A GENERAL TEST PROGRAM 91 10.6 FINE TUNING THE OPERATING VARIABLES 92 10.7 CONCLUSION 93 11 C O N C L U S I O N S A N D R E C O M M E N D A T I O N S 94 11.1 FACTORIAL DESIGN 94 vii 11.2 INCREMENTAL QUARTZ/MAGNETITE 95 11.3 SYNTHETIC SPHALERITE/QUARTZ 96 11.4 H B M & S PLANT TESTING 96 11.5 SPECIFIC GRAVITY DIFFERENTIAL 97 11.6 VARIABLE TUNING IN A C V D CONCENTRATOR 97 11.7 DEVELOPING AND OPERATING STRATEGY FOR A NEW APPLICATION 98 11.8 RECOMMENDATIONS 99 12 REFERENCES 100 13 APPENDIX 1 - QUARTZ/MAGNETITE FRACTIONAL FACTORIAL DESIGN 103 13.1 ALIAS STRUCTURE 104 13.2 FRACTIONAL FACTORIAL DESIGN M A P 105 13.3 FACTORIAL DESIGN TEST RESULTS/ METALLURGICAL B A L A N C E 107 13.4 CALCULATED FACTORIAL DESIGN RESULTS - EFFECT OF OPERATING PARAMETERS 110 13.5 QUARTZ/MAGNETITE ASSAY PROCEDURE I l l 13.6 ATTRITION CHECK 112 14 APPENDIX 2 - HBM&S GOLD ZINC INCREMENTAL TRIALS 115 viii LIST OF TABLES Table 2.3.1 - CVD Commercially available models 5 Table 2.4.1 - Falcon C models and feed capacities 7 Table 2.5.1 - Kelsey Jig models and rated solids feed capacities 9 Table 4.1.1 - Variables impacting product grade and recovery 21 Table 4.2.3.1 - Feed sample standard deviations by size 26 Table 4.2.5.1 - High and low levels for operating variables in factorial design 28 Table 4.2.5.2 - Quartz and magnetite factorial design map 29 Table 4.2.5.3 - Quartz/magnetite factorial design test program 30 Table 4.3.2.1 - Calculated operating variable effects on grade from factorial design results 34 Table 4.3.2.2 - Significant feed variable effects on grade 34 Table 4.3.2.3 - Calculated machine variable effects on grade 35 Table 4.3.3.1 - Calculated operating variable effects on recovery from factorial design results 36 Table 4.3.3.2 - Machine variable effects on recovery 37 Table 4.3.4.1 - Summary of significant variable effects 38 Table 5.2.1 - Synthetic quartz and magnetite incremental test program 41 Table 5.3.2.1 - Results of changing bowl speed with quartz and magnetite 43 Table 5.3.3.1 - Results of changing fluidization water flowrate with quartz and magnetite 46 Table 5.3.4.1 - Results of changing percentage solids with quartz and magnetite 47 Table 6.2.1.1 - Composition of zinc concentrate from HBM&S used in creating the synthetic feed 51 Table 6.2.1.2 - Composition by size of the zinc concentrate used in the synthetic zinc feed 52 Table 6.2.1.3 - Composition of the blended synthetic quartz/sphalerite feed 52 ix Table 6.2.2.1 - Synthetic quartz and sphalerite test program 54 Table 7.2.1.1 - HBM&S incremental testing map of operating variable levels per test 62 Table 8.2.2 - Approximate specific gravity differentials for HBM&S plant trials 76 Table 9.2.1 - General trends for operating variables 82 Table 9.2.2 - Rate of change in grade versus rate of change in recovery 83 Table 11.1 - Summary of significant variable effects 95 x LIST OF FIGURES Figure 2.4.1 Section view of the Falcon C technology, (www.concentrators.net) 6 Figure 2.1 - Kelsey Jig (www.rochemining.com) 8 Figure 2.5.2 View of the Kelsey Jig showing the slurry flow through the concentrator 9 Figure 2.5.3 - Rotating section of the Kelsey Jig showing the mechanism for pulsing the ragging bed in the concentrator 10 Figure 2.6.1 Section view of the Knelson CVD concentrator (Knelson Concentrators) 12 Figure 3.2.1 CVD6 pilot plant 17 Figure 3.2.2 - Main level of the CVD6 pilot plant showing the CVD6, feed tank, product transfer pump and miscellaneous piping 18 Figure 3.2.3 - Sample plant block diagram 20 Figure 4.2.1.1 - CVD6 determination of steady state conditions 23 Figure 4.2.1.2 - Confirmation of steady state conditions for CVD6 24 Figure 5.3.2.1 - Bowl speed relationship with grade, recovery and mass yield with a quartz/magnetite feed 44 Figure 5.3.3.1 - Fluidization water flowrate relationship with grade, recovery and mass yield with a quartz/magnetite feed 46 Figure 5.3.4.1 - Feed percent solids relationship with grade, recovery and mass yield with a quartz/magnetite feed 48 Figure 6.2.1.1 - Size distribution of sphalerite (zinc concentrate from HBM&S) used in synthetic zinc feed 51 Figure 6.2.1.2 - Size distribution of the synthetic quartz and sphalerite feed 53 Figure 6.3.1 - Bowl speed versus grade at varying pinch valve closed times 57 xi Figure 6.3.2 - Upper boundary results for synthetic sphalerite/quartz 58 Figure 7.3.2.1 - Grade and recovery versus bowl speed 67 Figure 7.3.3.1 - Grade versus pinch valve open time 68 Figure 7.3.3.2 - Recovery versus pinch valve open time 69 Figure 7.3.4.1 - Grade versus pinch valve closed time 71 Figure 7.3.4.2 - Recovery versus pinch valve closed time 72 Figure 8.2.1 - Separation performance of fully liberated synthetic feeds 74 Figure 8.2.1 - Separation performance of fully liberated synthetic feeds 75 Figure 8.2.2 - Upper boundary lines for recovery and upgrade ratio from the plant trials at HBM&S 75 Figure 9.2.1 - Recovery versus grade 80 Figure 10.4.1 - Drawing an upper boundary condition on a grade recovery curve 92 xii ACKNOWLEDGEMENTS This study was funded by The Science Council of British Columbia and Knelson Concentrators. The support of these organizations has been greatly appreciated. The work was supervised by Dr. Bern Klein, Department of Mining and Mineral Process Engineering, The University of British Columbia Much of the test work, analysis was performed in collaboration with Michael Lambert. Thank you for your significant contribution to this work. Thank you to my parents for their relentless support in this endeavor. There are many people who have helped along the way, but none more than my loving wife, Candice. Candice, thank you for your never ending patience and support throughout this journey. xiii 1 Introduction There are currently three commercially available continuous centrifugal gravity concentrators: the Kelsey Jig, the Falcon C and the Knelson CVD. This thesis is a study into the understanding of the operation of one continuous centrifugal gravity concentrator, the Knelson Continuous Variable Discharge (CVD) concentrator. In particular the focus of this thesis is to demonstrate the impact of operating variables on the separation performance of this continuous centrifugal gravity concentrator. This systematic study is aimed at gaining a better understanding of how to operate the concentrator for a range of applications and to identify the equipment's capabilities and limitations. Pilot scale testing was conducted in both controlled laboratory conditions and field testing at an operating mine. The laboratory work included: • Testing of a fully liberated synthetic mixture of quartz and magnetite • Evaluation of a synthetic mixture or sphalerite and quartz. With the baseline knowledge that was gained in the laboratory, a second phase of field- testing at an operating Canadian base metal operation was conducted. The combination of laboratory and field-testing was then evaluated in combination to understand the effect of operating variables on grade, recovery. Results were analyzed to determine how the mineral density differential was affected by the same operating variables. 1 2 Literature Review 2.1 History Gravity concentration is one of the oldest and simplest form of separation of valuable minerals from gangue material. The first gravity concentrators to be developed exploited the difference in specific gravity of particles under the force resulting from gravity. Some common devices that fall into this category are: • Sluices • Jigs • Shaking tables • Spirals • Reichert Cones Many of the gravity separation technologies were developed for the separation of gold. In the late 1970's Byron Knelson observed that fine gold was being lost in sluice box tails of a placer operation. Mr. Knelson subsequently set out to develop a machine that could recover this fine gold. This was the first commercial version of a centrifugal gravity concentrator. The machine has similarities to a conceptual machine designed, but not commercialized in 1935 by MacNicol in Australia, (www.concentrators.net). Other batch centrifugal concentrators have been developed since this time including the Falcon SB. 2.2 Batch and Use The original Knelson Concentrator is a batch process machine that utilizes a spinning cone to enhance the force of gravity for improved particle separation. Water is injected into riffles in the 2 cone that creates an enhanced fluidized bed separator in the concentrator to capture gold and/or platinum group metals. The concentrator accepts feed typically for between 0.5 and 5 hours depending on the grade and nature of the feed before it is stopped and flushed. Modern versions of the batch Knelson concentrator are fully automated and flush automatically. In the flush cycle the feed is stopped, the cone stops spinning and the concentrate is flushed through a multi-port hub in the bottom of the machine. A typical flush cycle in an automated Knelson concentrator takes between two and three minutes (www.knelson.com). The mass yield to concentrate of a batch concentrator is in the order of 0.1% of the feed which is a concentration ratio of approximately 1000:1. A typical feed grade for a batch Knelson concentrator can be measured in grams/tonne. There are two traditional applications for a batch gravity concentrator. One is for hard rock gold mines where it typically treats the cyclone underflow within a grinding circuit circulating load. The second application is in alluvial gold applications. Due to the low grade of gold and platinum group metals in relation to gangue material, a shut-down flush cycle is acceptable as it only represents between approximately one and eight percent of operating time. In grinding circuit applications, there is a high probability that free gold in a cyclone feed will report to cyclone underflow, (S. Banisi, A.R. LaPlante, J. Marois, 1999). Taking the recirculation of gold in grinding circuits into consideration, it becomes possible to utilize batch gravity concentration with offline time for flushing. During concentrator flush cycles gold will have a high probability of remaining in the grinding circuit. 3 2.3 Evolution of Continuous Centrifugal Concentrators from Batch Technology In applications where the grade of the desired mineral is at percentage levels, the use of a batch concentrator becomes unfeasible. A batch concentrator would have to be stopped and flushed so frequently that it would effectively have zero feed availability. In order to address this type of application with high mass yield requirements, centrifugal gravity concentrators that could remove concentrate quickly and efficiently were required. This led to the development of the general class of gravity concentrators commonly known as "continuous centrifugal concentrators." There are three commercially available machines that qualify as continuous centrifugal concentrators: Knelson Continuous Variable Discharge (CVD), Falcon C, and Kelsey Jig. The Knelson CVD and the Falcon C have the capability of producing a wide range of mass yields with a practical range from 0 to 65%. The Kelsey Jig is more effective at sharp separations at relatively low mass yields to concentrate (typically less than 10%> mass yield to concentrate). In the next sections each of these machines is reviewed in more detail. 4 A list of the commercially available continuous concentrators is shown in Table 2.1. Table 2.1 The CVD has either one or two concentrating rings as indicated by the model number (eg. CVD 32-1 has one concentrating ring and CVD32-2 has 2 concentrating rings. The bowl dimensions range from 6 inches to 42 inches and is indicated by the model number. Table 2.3.1 CVD Commercially available models Concentrator Model Feed Tonnage (tonnes/hour) Knelson CVD6-1 0.5-2 Knelson CVD20-1 15-35 Knelson CVD32-1 40-70 Knelson CVD32-2 40-70 Knelson CVD42-1 70-100 2.4 Falcon C The Falcon Model C is the simplest machine as it has only two operating variables; bowl speed and valve aperture size. It is important to note that the Falcon C has no fluidization water added. Any water that is required for the operation of the concentrator is added to the Falcon C feed. The Falcon utilizes a large feed preparation distance in the lower portion of the Falcon C bowl. The smooth feed preparation section of the concentrator's bowl leads into capture rings in the top section of the Falcon bowl. At the back of each of the rings is a pneumatically operated annular discharge valve. The physical features of the Falcon C can be seen in Figure 2.4.1. The valve 5 technology sets the valve to a fixed open aperture size and remains open at all times. The valve aperture size is adjustable through the automated control on the Falcon C. Figure 2.4.1 Section view of the Falcon C technology, (www.concentrators.net) { TC "Figure 2.4.1 Section view of the Falcon C technology, (www.concentrators.net)" \f F \1 "1"} The operating parameters of the Falcon C include: • Valve aperture size • G-force Feed enters the Falcon C through a feed tube in the top of the concentrator and is directed to a simple feed accelerator that accelerates the slurry to the lower bowl wall. The bottom of the Falcon C bowl is smooth and acts as a feed preparation area for the capture rings at the top of the Falcon bowl. The particles travel up the smooth bowl section where heavy particles displace light particles along the bowl wall. At the top of the bowl is a concentrate collection ring with 6 controllable and remains open to expel concentrate from the capture rings into a dedicated concentrate launder. The tailing material forms the innermost layer on the bowl wall and overflows the bowl to a tailings launder. Table 2.4.1 Falcon C models and feed capacities Concentrator Model Rated Feed Capacity (tones solids/hour) Falcon C400 1-4.5 Falcon CI000 5.5-30 Falcon C2000 22-66 Falcon C4000 45 - 100 The applications for the Falcon C as listed on the Falcon Concentrators Website are Fine coal cleaning (-1mm coal) and Scavenging of metal values from tailings (www.concentrators.net). 2.5 Kelsey Jig The Kelsey Jig differs from the Knelson CVD and the Falcon C in that it utilizes a rotating jig bed. The combination of enhanced g-force and the use of a ragging bed enable the Kelsey Jig to be used in applications where a narrow specific gravity differential is present. Figure 2.5.1 is a photograph of a Kelsey Jig and Figure 2.5.2 is a section schematic showing the slurry flow paths through the concentrator. 7 Figure 2.1 - Kelsey Jig (wvvav.rochemining.com) The ragging size and density can be altered to affect the crispness of separation in the Kelsey Jig. Generally the density and size of the ragging in a Kelsey jig is intermediate to the target mineral and the gangue sg. A ragging recovery circuit is utilized with the Kelsey Jig. 8 Figure 2.5.2 Section view of the Kelsey Jig showing the slurry flow through the concentrator, (www.rochemining. com) Table 2.5.1 Kelsey Jig models and rated solids feed capacities. Concentrator Model Rated Capacity (tonnes solids/hour) Kelsey KCJ - Laboratory 15-100 kg/hr Kelsey J1300 Mk II KCJ 2-30 Kelsey J1800 KCJ 5-60 In the Kelsey Jig, particles are fed into the top and enter the vertical jig bed. The water pulsation cycles in the hutch behind the bed facilitates particle separation. The centrifugal acceleration 9 imparted on the heavy particles causes them to travel radially outward through ragging and into a hutch. Low density particles travel upwards across the bed and overflow to a launder. The three controllable variables in the Kelsey Jig are bowl speed, ragging size/density and jigging pulsation. Figure 2.5.3 shows a photo of the rotating mechanism in the Kelsey Jig with the pulsating rubber baffles that provide the movement in the jigging bed. The picture of the mechanism shows how this concentrator is more mechanically complex than the Knelson CVD and the Falcon C. Figure 2.5.3 - Rotating section of the Kelsey Jig showing the mechanism for pulsing the ragging bed in the concentrator The Kelsey Jig applications are not well publicized. However the flagship commercial installation for the Kelsey Jig is in the Australia at the Greenbushes plant, owned by Sons of Gwalia. In this application the Kelsey jig is being used for the recovery of fine Tantalum. 10 2.6 Knelson CVD A section view of the Knelson CVD is shown in Figure 2.6.1. Feed is introduced into the top of the machine through a feed tube into the center of the bowl section. The feed hits a plate at the bottom of the bowl section and is dispersed radially to the bowl wall. The particles are accelerated to a g-force defined by the bowl speed and travel up a short section of smooth wall towards the ring. The short smooth lower section of the CVD concentrating cone is designed to smooth the turbulence of the incoming slurry prior to entering the separation ring(s) where fluidization water, supplied through holes in the ring wall, is added to fluidize the bed of packed particles. Concentrate is extracted through pinch valves at the back of the ring. The pinch valve timing (open/closed) can be adjusted. Light particles overflow the bowl into a tailings launder. 11 Feed Tube Figure 2.6.1 Section view of the Knelson CVD concentrator (Knelson Concentrators) The Knelson CVD has four main operating variables: • Bowl Speed (g-force) • Pinch Valve Open Time • Pinch Valve Closed Time • Fluidization Water Flowrate 12 The combination of variables and their associated effects on separation performance produces a complicated problem of selecting appropriate levels for an effective separation of a mineral system. All models of the CVD are fully automated. The machine operating variables listed above can all be changed via a touch screen user interface connected to PLC control. The rotating section of the CVD concentrator is driven by a motor on a variable frequency drive. By entering bowl speed values into the PLC interface the frequency on the drive is altered to the entered set point. The bowl speed is continually monitored with proximity sensors on the driven sheave to insure that the set point g-force is maintained. The feature in the CVD that makes it a continuous concentrator is the ability to remove concentrate from the separation ring without stopping the feed to the machine. Pneumatically operated rubber pinch valves that can be opened and closed for specific intervals enable tight control over the removal of material from the concentrating ring(s). The pinch valve intervals -open and closed time, are both adjusted by entering set-points into the touch screen interface. There is an automated piping assembly on the CVD that includes a segmented ball valve in a feed forward control loop with a digital flowmeter. The water flow to the concentrating ring is delivered through small straight holes in the back of the concentrating ring. A setpoint for the water flowrate to the concentrator is set and can be adjusted from the PLC touchscreen interface. 13 2.6.1 Knelson CVD Applications The CVD has been commercially applied in the following applications (www.knelson.com): • Separation of iron oxides from talc (Byron, R.O., K. Roberts, 2004) • Cassiterite • Chromite • Gold sulphides (Simpson P., 2003) Other applications have been or are currently being tested by Knelson Concentrators. These applications may be in laboratory pilot testing or plant pilot/commercial testing but have not been accepted as an economic process yet by commercial producers. Some of these applications include: • Separation of Tantalite • Separation of ash and sulphur from coal • Gold removal from bio-oxidation residues • Separation of precious metals from used circuit boards The applications for the CVD are diverse. 3 Experimental Program 3.1 Test Program The evaluation of the CVD6 concentrator was performed at the laboratory pilot scale and at an operating mill. 14 The machine was first tested in the laboratory setting with the objective being to develop a framework understanding of the influence of operating variables on separation performance. The laboratory testwork was performed with synthetic mixtures in a controlled environment. The results of the laboratory testing with synthetic mixtures was then verified and expanded on with field pilot testing at an operating mill. Field-testing enabled the results from the laboratory testing to be verified and expanded. The field pilot testing enabled large numbers of tests to be performed in a relatively short period of time. Field testing was performed on a sulphide stream at Hudson's Bay Mining and Smelting (HBM&S). The CVD6 was installed with a bleed stream of cleaner flotation feed. The feed, concentrate and tailings could all be sampled directly at the concentrator. 3.2 Pilot Scale Circuit A Knelson CVD6 unit was used for pilot scale testing in the UBC Centre for Coal and mineral Processing (CCMP). Some of the factors that were considered when designing and constructing the pilot plant were as follows: • The feed had to remain consistent for the duration of the testing. The products (concentrate and tailings) from the CVD could not be recombined with the feed until the sampling was complete. • The feed tank had to be agitated in such a way as to produce a homogeneous slurry. 15 • The feed tank needed to be sufficiently sized to accommodate enough feed slurry for a 15 minute feed duration at 1 tonne/hour feed rate to the CVD6. • The products from the CVD needed to be stored in a separate tank that would accommodate the slurry plus any water that was added in the CVD6. • Decant systems were necessary to adjust the feed slurry density prior to each test. • The feed rate to the CVD6 needed to be consistent, adjustable and reproducible. • Sample points were required to collect representative samples of feed, concentrate and tailings. A pilot scale circuit was designed and constructed. Primary goals of circuit design were to enable full control of feed rate and enable quality sampling of concentrate, tailings, and feed for trial runs. Feed is agitated in a 145 x 145cm (57" x 57") flat bottom polyethylene baffled tank by a 2 V2 hp motor and 38cm (15") axial flow impeller assembly. The feed is pumped through a side discharge in the feed tank to the CVD by a fixed speed slurry pump. Concentrate and tailings from the CVD is pumped to a second 145 x 145cm (57" x 57") polyethylene tank where it is stored until the run is complete. After each run is complete, the feed slurry is transferred from the holding tank to the feed tank in preparation for the next trial. Figure 3.2.1 shows a picture of the completed pilot plant 16 17 Figure 3.2.2 - Main level of the CVD6 pilot plant showing the CVD6, feed tank, product transfer pump and miscellaneous piping. Much of the equipment used in the construction of the pilot plant was salvaged and assembled for this duty. Figure 3.2.2 shows the CVD6 in the foreground with water, compressed air and electricity attached to the equipment. Leading off to the right of the CVD6 are the tails in the top pipe and concentrate in the lower pipe. Both product pipes are routed to a Sala tank pump that is used to transfer the combined products to a holding tank on the second level. Sampling was performed by collecting timed flows of concentrate and multiple cuts of the tails stream. Figure 3.2.3 shows a conceptual block diagram of the pilot plant. A valved recycle line was used after the feed pump to adjust the feed rate to the CVD6. An oversized fixed speed pump was 18 used to keep a high line velocity. Tests were performed to ensure that a representative feed consistently delivered to the unit. 19 Holding Tank Feed Recycle and Feed Sample Point Slurry Transfer (between batches) Agitated Feed Excess Water Fluidization Compressed Feed Pump Tails Pump and Sample Point Tails Concentrate Concentrate Pump and Sample Point Figure 3.2.3 - Sample plant block diagram 20 4 Quartz Magnetite Factorial Design 4.1 Introduction A factorial design test program was utilized to determine the effects of the CVD operating variables on product grade and recovery. Originally eight variables were identified as having a possible effect on the separation performance of the CVD. These operating variables can be classified as feed and machine variables. The variables are outlined in Table 4.1.1. Table 4.1.1 - Variables impacting product grade and recovery Machine Variables Feed Variables Bowl speed Feed rate Pinch valve open time % solids Pinch valve closed time Particle size Fluidization water flowrate Grade In order to test one variable at a time at two levels would have required 256 tests and for three levels would have required 6561 tests. Therefore, a statistical experimental design program was designed and implemented. A 28"4 fractional factorial design was used that required 16 main tests and 3 center point runs. Using a factorial design provided sufficient information to rank the effect of variables on product grade and recovery. The results of this test program provided an empirical understanding of the influence of the variables on process performance. The results were used to assist in the development and implementation of subsequent test programs. 21 4.2 Experimental 4.2.1 Procedure A mixture of quartz and magnetite was used for the test program. Sized silica sand (quartz) was obtained and the size distributions were confirmed by sieve analysis. The magnetite was prepared in two size fractions with P80's of 125 and 425 pm. For testing, the quartz and the appropriate size fraction of magnetite were added to the feed tank. The percent solids were adjusted to target level and the slurry was agitated. The preparation procedure for the magnetite involved crushing, grinding and magnetic separation. The crushing was performed in three stages by jaw crushing and 2 stages of cone crushing. The magnetite was then cleaned using a lab scale drum magnetic separator. The purity of the magnetite was checked with a Davis tube and was determined to be 99.99% pure. Sixteen pilot tests plus four center point tests were conducted. In all tests the same testing and sampling procedure was employed. Parameter levels including: pinch valve open time, pinch valve close time, bowl speed, and fluidization water flow rate were set before the feed was introduced to the concentrator. The feed tank was agitated thoroughly to produce a homogeneous slurry and then the slurry was pumped through a flow control valve to the concentrator. A test was conducted to determine the time required for the CVD6 to reach steady state operation. Samples of concentrate were taken every 30 seconds and assayed. During the first several minutes the magnetite grades were erratic but stabilized within 6 minutes. Two 22 additional tests were performed to confirm that grades stabilized with time. Figure 4.2.1.1 shows the results of the first test. Figure 4.2.1.2 shows the results of the two additional tests that confirm that steady state is achieved within 8 minutes. Based on the results of this test, the concentrator was run for 8 minutes prior to sampling in order to ensure steady-state conditions were achieved. Sampling lasted for 4 minutes during which the concentrate was collected and composite samples of feed and tailings were collected. Each of the samples was then pressure filtered, dried and riffled to obtain two representative sub samples. One of the sub-samples was subjected to assay to determine magnetite grade and the other retained sample was used for size assay analysis. Steady State Testing 25% 0% -I , , 1 1 0 100 200 300 400 Time (seconds) Figure 4.2.1.1 - CVD6 determination of steady state conditions 23 Steady State Determination 50.0% -i , 1 1 : , , , , , , , , r -Figure 4.2.1.2 - Confirmation of steady state conditions for CVD6 The magnetite grade was determined using a Davis tube. For size assay analysis each sample was screened into three fractions: +50, -50 +100, and -100 mesh. Each fraction was assayed with the Davis tube for % magnetics. Sub-samples were retained for assay checks as required to ensure reconciliation of head assays. 4.2.2 Quartz/Magnetite Assay Procedure A Davis tube was used to determine magnetite grades by recovering ferromagnetic magnetite from representative sub-samples. The samples were prepared for assay using the following procedure: • Pressure filter • Dry in oven at 100 degrees Celsius • Riffle to manageable sample size as follows: o 100 grams for feed 24 o 100 grams for tails o 50 grams for concentrate The riffled samples were weighed and transferred into 500ml glass beakers. Water was run through the Davis tube and the electro-magnet was started. The water level in the Davis Tube was kept constant by setting the height of the water discharge tube. The agitation of the Davis Tube was turned on. A feed funnel assembly was attached to the Davis tube to prevent spillage of sample. At this point the sample was carefully slurried with a small water jet in the beaker and washed slowly into the Davis Tube. Care was taken to avoid overfeeding the Davis tube. Once the entire sample was run through the Davis tube it was allowed to continue agitation for another 30 seconds. The water and the agitation were then turned off and a beaker placed under the tube discharge. The magnet was then turned off and the sample was thoroughly washed out of the tube and into the beaker. After settling for at least 5 minutes the excess water was decanted from the beaker. The beaker with the magnetite was then dried completely and weighed. Once weighed, all sample was returned to the feed tank to prevent depletion. 4.2.3 Attrition Check Many tests were run with the same synthetic quartz/magnetite feed. A test was conducted to determine if magnetite was being ground during pumping and mixing. Feed samples from all tests were subjected to size assay analyses using 50 and 100 mesh screens. Table 4.2.3.1 summarizes the standard deviations in feed assays by size over 15 tests. It was determined from the test that attrition grinding of magnetite was not a factor. 25 Table 4.2.3.1 - Feed sample standard deviations by size Size Fraction (mesh) Standard Deviation in Feed Grade from 15 Tests +50 3% 50 x 100 12% -100 6% As an additional precaution, the tests were carried out in a sequence that minimized the number of feed changes required to complete the program and maintain consistent size distributions and grades of magnetite. 26 4.2.4 Test Program Details The 2s"4 fractional factorial experimental design required 16 main tests plus 3 center point tests. Variables tested in this design included: • Feed Rate (t/h) • Pulp density of feed • Magnetite particle size • Pinch Valve Open Timing (seconds) • Pinch Valve Closed Timing (seconds) • Fluidization Flow Rate (gal/min) • Bowl Speed (RPM) • Feed Grade (% Magnetite) For the experimental design, two levels were selected for each variable, a high and a low level. These levels were selected based on results from preliminary scoping tests. Table 4.2.5.1 shows the selected high and low levels for each operating and feed variable. 27 Table 4.2.5.1 - High and low levels for operating variables in factorial design Variable High Low Heavies (%) 4 1 Fluidization (gpm) 15 5 Pinch Valve Open (s) 0.08 0.03 Pinch Valve Closed (s) 8 2 Bowl Speed (G's) 75 45 Solids Feed Rate (tph) 2 1 % Solids 45 30 Heavies Particle Size (p80) (microns) 425 125 A fractional factorial map was utilized to determine what the combination of high and low values for each variable should be used for each test. The map, shown in Table 4.2.5.2, was then converted into a test program by adding the variable levels into the map. The test program is shown in Table 4.2.5.3. 28 Table 4.2.5.2 - Quartz and magnetite factorial design Test Fluidization Water % Solids Feed Grade Heavies Particle Size Bowl Speed Pinch Open Pinch Closed Solid Feed Rate 1 High High High High High High High High 2 Low High High High High Low Low Low 3 High Low High High Low High Low Low 4 Low Low High High Low Low High High 5 High High Low High Low Low Low High 6 Low High Low High Low High High Low 7 High Low Low High High Low High Low 8 Low Low Low High High High Low High 29 Table 4.2.5.3 - Quartz/magnetite factorial design test program Heavies Solid Fluidization Feed Particle Bowl Pinch Pinch Feed Water % Grade Size (p80 Speed Open Closed Rate Test (gal/min) Solids (%) microns) (RPM) (s) (s) (t/h) 1 14 45 4 425 925 0.05 8 2 2 5 45 4 425 925 0.03 2 1 3 14 30 4 425 725 0.05 2 1 4 5 30 4 425 725 0.03 8 2 5 14 45 1 425 725 0.03 2 2 6 5 45 1 425 725 0.05 8 1 7 14 30 1 425 925 0.03 8 1 8 5 30 1 425 925 0.05 2 2 9 14 45 4 125 725 0.03 8 1 10 5 45 4 125 725 0.05 2 2 11 14 30 4 125 925 0.03 2 2 12 5 30 4 125 925 0.05 8 1 13 14 45 1 125 925 0.05 2 1 14 5 45 1 125 925 0.03 8 2 15 14 30 1 125 725 0.05 8 2 16 5 30 1 125 725 0.03 2 1 30 4.3 Results 4.3.1 Introduction The main measured responses from the program were product grade, mass yield and recovery. Mass balances were generated for each test to obtain mass yields and recoveries. The mass balance data and operating data was combined into one spreadsheet. Variable effects were calculated to rank the relative importance of the variables for each of the responses. The effects of operating variable can be interpreted as the percentage change in the response (grade or recovery) resulting from the change in that operating variable level. 4.3.1.1 Method of Calculating Operating Variable Effects from Factorial Design Results For each variable tested in the factorial design an effect was calculated. Two effects were calculated, one for grade and one for recovery. The effect is a way of quantifying the impact that individual or interacting variables have on an outcome (grade or recovery). The design and associated grades and recoveries for each test are presented in Appendix A. The following methodology outlines the procedure used for calculating the effects for grade and recovery. For the effect on grade, the following process was used: • Concentrate grade was recorded for each set of conditions. • For each operating variable the grades from all of the tests run at high levels for that specific variable were averaged. • For each operating variable the grades from all of the tests run at low levels for that specific variable were averaged. 31 The average grade of the high levels was then subtracted from the average of the low levels. The resulting number (in percent) is the effect each variable had on grade when it was increased from the low to the high level. 32 This procedure can also be summarized by the following equation for calculating effects in a factorial design: main effect = y+- y_ The equation above (Box et All., 1978) is used for the calculation of main effects in a factorial design. As described previously the main effect is the average of the high parameter level results minus the low parameter level results. 4.3.2 Effect of Operating Variables on Grade The effects of operating variables on concentrate grade are shown in Table 4.3.2.1. The error associated with these results is ± 3.4%. The error was calculated by running a series of repeated tests. The repeat tests were performed at center point levels of variables. The standard deviations of the duplicate tests are used as the error. Parameters with effects larger than the error are considered significant. As shown in Table 4.3.2.1, the parameter that most influenced the grade of the concentrate was found to be the solid feed rate. As the solid feed rate was increased from 1 tph to 2 tph, the impact was an 8.2% increase in concentrate grade. Similarly, the effects of other variables are shown in Table 4.3.2.1. High and low variable levels are summarized in Table 4.2.5.1. Pinch valve open time, and heavy particle P80 size did not qualify as significant effects because their effect was less than the standard deviation of +/-3.4%. 33 Table 4.3.2.1 - Calculated operating variable effects on magnetite grade Rank Variable Effect 1 Solid Feed Rate (t/h) 8.2% 2 Bowl Speed (G's) -7.3% 3 Pinch Closed (s) 6.3% 4 % Solids -6.2% 5 Fluidization (g/min) -4.6% 6 Feed Grade (%) 4.2% 7 Pinch Open (s) -3.1% 8 Heavies Particle Size (p80 microns) 2.0% In a plant environment, the feed parameters are not easily manipulated. The results have been split into feed variables and operating variables in Tables 4.3.2.2 and 4.3.2.3. The effect of feed variables on grade can be used to evaluate the installation location within a plant. However, the machine operating variables can be used to directly influence the performance of the concentrator. Table 4.3.2.2 - Significant feed variable effects on grade Rank Feed Variable Effect 1 Solid Feed Rate (t/h) 8.2% 2 % Solids -6.2% 3 Feed Grade (%) 4.2% 34 Table 4.3.2.3 - Calculated machine variable effects on grade Rank Operating Variable Effect 1 Bowl Speed (G's) -7.3% 2 Pinch Valve Closed (s) 6.3% 3 Fluidization (gal/min) -4.6% 4.3.3 Effect of Operating Variables on Recovery The effects that single operating variables had on recovery of magnetite to the concentrate are shown from most significant to least significant in Table 4.3.3.1. The error associated with these results is ± 9.0%. The error was calculated using a standard deviation from repeat tests. 35 Table 4.3.3.1 - Calculated operating variable effects on recovery from factorial design results Rank Variable Recovery Effect 1 Heavy Particle Size (pgo microns) -21.9% 2 Pinch Closed (seconds) -12.5% 3 Bowl Speed (RPM) -12% 4 %Solids -5.9% 5 Fluidization Water Flow (gpm) 4.8% 6 Pinch Open (seconds) 4.0% 7 Feed Grade (%) -3.2% 8 Solid Feed Rate (tons/hour) -1.6% Considering the results in Table 4.3.3.1, the most influential parameter was found to be the Pgo size of the heavy particle. It was found that increasing the heavy particle Pgo (magnetite particle size) in the feed from 125 microns to 425 microns caused the recovery to decrease by 21.9 percent. The decrease in recovery is due to fewer large particles being capable of passing through the pinch valves in a discrete time. The effects of all parameters tested are summarized in Table 4.3.3.1. Table 4.3.3.2 classifies the effects of machine variables on recovery performance. Only variables that had an effect in excess of +1-9% were considered significant in this test program. The only feed variable with a significant effect was heavy particle size. This effect was -21.9%. 36 Table 4.3.3.2 - Machine variable effects on recovery Rank Machine Variable Effect 1 Pinch Closed (seconds) -12.5% 2 Bowl Speed (rpm) -12% 4.3.4 Conclusion The factorial design testwork with a synthetic feed of quartz and magnetite allowed the effects of the variables to be ranked. The results can be used to prioritize the impact that variations in the feed or the machine operating variable levels have on the separation performance. The results were classified as machine variables and feed variables and rated in order of importance with respect to both grade and recovery. Table 4.3.4.1 presents the significant variables in order of significance from the test program. The table classifies the variables as either feed variables or machine variables. The effects are measured as either an effect on grade or recovery. Of the 8 variables tested, the only variable that did not have a significant effect was pinch valve open time. It is possible that the high and low levels for this variable were set too close together to have a significant effect. Further incremental testing will determine whether this hypothesis is correct. 37 Table 4.3.4.1 - Summary of significant variable effects Feed Variables Machine Variables Grade Recovery Grade Recovery Solids Feed Rate Heavy Particle Size Bowl Speed Pinch Valve Closed Time % Solids Pinch Valve Closed Time Bowl Speed Feed Grade Fluidization Water Testing of the impact of feed characteristics is more important for understanding the implications of different feed types. This section did not reveal a great deal of useful information on the impact of feed characteristics. The most interesting result was that heavy particle size had a very significant impact on recovery. This indicates that separation by size is a key parameter to study when evaluating CVD performance. It is believed that the particle size impacts the rate at which particles will travel through the pinch valves. Perhaps longer open times would counteract the recovery drop effect as particle size gets larger. This result demonstrates that it is necessary to study a more detailed range of parameter levels in order to fully understand the effect of the operating variables. The next section of this report studies the effect of operating variables over a broader spectrum than the 3 parameter level factorial design approach. This test program ranked the operating variables of the CVD based on selected high, low and midpoint levels for each parameter. During calculation of the effects, interaction effects were ignored and confounded results were eliminated. In order to better understand the impact of the variables over a wide range of levels, further incremental testing is recommended. This program 38 was an effective tool in identifying that all variables (other than pinch valve open time) significant. No variables can be eliminated from further testing. 39 5 Quartz/Magnetite Incremental Testing 5.7 Introduction Incremental testing is a necessary compan ion to a statistical experimental design. In a statistical experimental design a range o f each operating variable is selected. The h igh , l o w and center point o f each variable is tested i n the design i n order to establish each ind iv idua l parameter 's effect o n the results. T h i s type o f testing is very useful i n establishing w h i c h variables are significant i n a system o f variables that is very complex . Howeve r , a factorial design does not show what the grades and recoveries are over the who le range o f levels for each operating variable. In fact, because on ly three points f rom each operating var iab le ' s range are tested i n the design, an inf lec t ion point i n the results may be missed. I f a range for the variables is selected that is too smal l then it may not encompass the cr i t ica l m a x i m u m or m i n i m u m point . I f the range is too large then the m a x i m u m or m i n i m u m point may be br idged and not seen. A n incremental test program is required to show detail over the who le operable range o f the machine. It w o u l d not be feasible to test a l l operating variables i n an incremental des ign so the factorial des ign is essential i n nar rowing the focus. 5.2 Experimental Test Program The incremental test program for quartz and magnetite tested the effects o f f lu id iza t ion water flowrate, feed percent sol ids, and b o w l speed. P i n c h va lve open and c losed t ime was not tested i n this program as it was planned that they w o u l d be tested i n plant trials. T h i s test program was designed to determine operating parameter trends. The test program is s h o w n i n Table 5 .2 .1. O n l y one variable at a t ime was changed i n each subset o f tests i n this incremental test program. 40 Table 5.2.1 - Synthetic quartz and magnetite incremental test program Heavies Lights Solid Fluidization Feed Particle Particle Bowl Pinch Feed Water Grade Size (p80 Size (p80 Speed Pinch Closed Rate Test (g/min) % Solids (%) microns) microns) (G's) Open (s) (s) (t/h) 1 6 25 2 275 425 45 0.03 5 1.5 2 6 30 2 275 425 25 0.05 5 1.5 3 6 30 2 275 425 35 0.05 5 1.5 4 6 30 2 275 425 45 0.05 5 1.5 5 6 30 2 275 425 55 0.05 5 1.5 6 6 30 2 275 425 65 0.05 5 1.5 7 6 30 2 275 425 75 0.05 5 1.5 8 2 30 2 425 425 55 0.05 5 1.5 9 4 30 2 425 425 55 0.05 5 1.5 10 6 30 2 425 425 55' 0.05 5 1.5 11 8 30 2 425 425 55 0.05 5 1.5 12 10 30 2 425 425 55 0.05 5 1.5 13 6 25 2 425 425 55 0.05 5 1.5 14 6 30 2 425 425 55 0.05 5 1.5 15 6 35 2 425 425 55 0.05 5 1.5 16 6 40 2 425 425 55 0.05 5 1.5 41 5.3 Quartz/Magnetite Incremental Testing Results 5 . 3 . 1 Introduction The incremental testing with quartz and magnetite focused on bowl speed, fluidization water flowrate and feed percent solids. Not all of the operating and feed variables were tested in this program. The reason for this is that it takes considerable time and effort to run pilot plant synthetic testing. Therefore the variables that were tested in this program were selected based on the criteria as described below. Variables that were not tested in this incremental program were left for testing in subsequent plant testing or eliminated as not significant from the factorial design. Bowl speed was investigated because it was identified as having a significant effect on grade and recovery in the factorial design results. Therefore, analysis of the general trends in grade and recovery with changing bowl speed was desired. Fluidization water flowrate was of particular interest as the preliminary testing and the factorial design did not clearly indicate an appropriate operating range for fluidization water. The incremental testing allowed this variable to be tested without any other factors impacting the results. Feed percent solids was identified as having a significant effect on grade. This variable would not be easily tested in a plant scenario, as feed parameters cannot easily be manipulated in an operating mill. Therefore, a controlled lab scale pilot study was considered to be ideal for testing this parameter. 42 5.3.2 Quartz/Magnetite Trends with Changing Bowl Speed Incrementally changing the CVD bowl speed between 25 (550rpm) and 75g's (900rpm) resulted in linear trends in recovery, grade and mass yield as shown in Figure 5.3.2.1. The bowl speed had the most significant effect on recovery. Increasing bowl speed increased recovery. The mass yield contributed in part to the increase in recovery with increasing bowl speed. As the bowl speed increased, the mass yield increased, but at a lesser slope than the recovery increase. Increasing the bowl speed decreased the grade of the concentrate. Table 5.3.2.1 - Results of changing bowl speed with quartz and magnetite Mass Bowl Con Grade Tails Grade Recovery Yield Speed (%) (%) (%) (%) 25 29.5 1.35 42.2 3.2 35 22.4 1.16 51.4 5.2 45 20.6 1.11 53.9 5.9 55 19.0 0.86 64.9 7.7 65 18.9 0.77 68.7 8.2 75 16.0 0.67 73.4 10.4 43 Bowl Speed Relationship 20 30 40 50 60 70 80 Bowl Velocity (G's) • Recovery • Con Grade • Mass Yield Figure 5.3.2.1 - Bowl speed relationship with grade, recovery and mass yield with a quartz/magnetite feed 5.3.3 Quartz/Magnetite Trends with Changing Fluidization Water Flowrate Fluidization water was incrementally changed between 2 and 10 gpm with all other operating and feed variables constant. The effect on grade, recovery and mass yield are shown in Figure 5.3.3.1. A small increase in mass yield was observed with increasing mass yield. This increase can be attributed to lower percentage solids in the concentrate flowing through the pinch valves. The 44 fluidization water acts as a lubricant for the particles and prevents plugging or sticking in the valves. The grade of the concentrate was not significantly affected by changing the fluidization water flowrate. However, the grade decreased slightly over the tested range. Fluidization water flowrate had a very interesting result on recovery. Between 2 and 6 gpm there was a sharp increase in recovery. At 6gpm there is a transition point in the recovery trend. Beyond 6 gpm there was no change in recovery. This result indicates that a minimum amount of fluidization water is necessary to achieve efficient recovery. Beyond a threshold, 6 gpm, additional fluidization water only affects grade slightly. In operating mills water addition is minimized. The result of this test is that fluidization water should be maintained as close as possible to the threshold of 6gpm. Beyond 6gpm the grade continues to decrease and needless water is consumed. However, this result also shows that it is more conservative to run the CVD with more water than the threshold limit as the slope of the recovery line is very steep before the threshold. The grade and mass yield lines have very small slopes beyond the threshold. It is conservative to operate the CVD with 7 or 8 gpm. The result of this test with fluidization water flowrate was informally observed in preliminary and plant testing. The existence of the transition point was justification to eliminate fluidization water flowrate from subsequent test programs. 45 Table 5.3.3.1 - Results of changing fluidization water flowrate with quartz and magnetite Mass Con Grade Tails Grade Recovery Yield Fluidization (%) (%) (%) (%) 2 16.6 1.46 30.6 3.7 4 22.0 1.34 35.8 3.2 6 21.6 1.12 47.2 4.4 8 19.3 1.13 47.0 4.9 10 17.6 1.12 47.7 5.5 60% 50% 40% S9 30% 20% 10% 0% Fluidization Relationship 4 6 8 Fluidization (gpm) 10 • Con Grade A Recovery • Mass Pull Poly. (Con Grade) 12 Figure 5.3.3.1 - Fluidization water flowrate relationship with grade, recovery and mass yield with a quartz/magnetite feed 46 5.3.4 Quartz/Magnetite Trends with Changing Percent Solids The percent solids in the feed was changed to determine the effect on separation performance. As shown in Figure 5.3.4.1, the percentage solids in the feed only affected the recovery. This is similar to the relationship seen when the fluidization water was changed. The higher the solids content in the feed, the lower the recovery. This test does not have enough data points to clearly define an inflection point; it does appear that the optimum feed percent solids was approximately 30%. Mass yield and concentrate grade were not affected by percent solids within experimental error. Table 5.3.4.1 - Results of changing percentage solids with quartz and magnetite Mass Con Grade Tails Grade Recovery Yield % Solids (%) (%) (%) (%) 25 22.3 1.2 42.2 3.8 30 20.6 1.1 45.3 4.5 35 20.9 1.2 40.0 3.9 40 22.3 1.5 22.9 2.1 47 % Solids Relationship 10% • •- • . 0% -I , 1 T 20 25 30 35 40 45 % Solids • Con Grade A Recovery BMass Pull Figure 5.3.4.1 - Feed percent solids relationship with grade, recovery and mass yield with a quartz/magnetite feed 5.3.5 Conclusion Incremental testing of quartz and magnetite feed tested the effects of bowl speed, fluidization water flowrate and feed percent solids. Increasing bowl speed was found to increase recovery and mass yield while decreasing concentrate grade. Fluidization water flowrate was determined to have the greatest effect on recovery. Threshold minimum fluidization water was found at 6 gpm. Above 6 gpm additional fluidization water did not affect recovery and only slightly reduced grade. It was determined from this result that fluidization water could be held constant for all subsequent testing. The C V D was found to perform best with feed percent solids of 30%. 48 6 Lab Testing - Synthetic Sphalerite/Quartz 6.1 Introduction A synthetic feed of quartz and sphalerite was created to test in the lab. This feed was used because it models a typical sulfide and silicate ore at a fine size. A test program where one variable was changed incrementally was used to generate data that could be used for plotting trends. Specifically this test program was targeted at testing the capabilities of the CVD concentrator at separating minerals with a smaller SG differential than was present with the quartz/magnetite system. This system was fully liberated and particle size distributions were similar for both the denser sphalerite and the less dense quartz. The information gathered in this component of the test work can be used to determine the following: • Whether the CVD concentrator can make a separation at a specific gravity differential of 1.35 with a fully liberated system. The sphalerite (SG 4.0) and quartz (SG 2.65) system had a specific gravity differential of 1.35. This is a smaller differential than that of quartz/magnetite at 2.55. This testing would identify whether the CVD could separate minerals at a close mineral density. • If the CVD can make a separation with this material then plant testing at Hudson's Bay Mining and Smelting (HBM&S) can be pursued to investigate how the CVD will behave on a non-synthetic feed. 49 6.2 Experimental - Synthetic Sphalerite/Quartz 6.2.1 Feed Characterization The sphalerite component was a final zinc concentrate obtained from HBM&S. The chemical composition of the zinc concentrate is summarized in Table 6.2.1.1. The zinc concentrate used in making the synthetic feed was primarily sphalerite with some pyrite and small amounts of galena and chalcopyrite. The Pgo of the zinc concentrate was 121pm with a size distribution as shown in Figure 6.2.1.1. Table 6.2.1.2 is a size-assay of the zinc concentrate that shows the component grades are evenly distributed over the size distribution. 50 Table 6.2.1.1 - C o m p o s i t i o n o f z inc concentrate f rom H B M & S used i n creating the synthetic feed Assayed Components % Composition Z n 53.5 S 31.4 Fe 10.3 P b 0.8 C u 0.7 Figure 6.2.1.1 - S i ze dis t r ibut ion o f sphalerite (z inc concentrate f rom H B M & S ) used i n synthetic z inc feed 100% 90% 80% D) 70% C "35 (A ra 60% a. 9) > 50% to 3 C 40% C 3 o 30% 20% 10% 0% Size Distribution of Zinc Concentrate from HBM&S for Use in Synthetic Sphalerite Feed y = 0.0091x-0.3003 R 2 = 0.9806 25 50 75 100 Size (um) 125 150 175 51 Table 6.2.1.2 - Composition by size of the zinc concentrate used in the synthetic zinc feed Size (um) S (%) Cu Pb (%) Zn Fe (%) (%) (%) +150 31.9 0.68 0.84 51.2 9.9 -150+106 30.8 0.84 0.79 52.3 10.0 -106+75 31.4 0.65 0.71 53.7 11.0 -75+53 32.0 0.64 0.87 53.6 10.5 -53+38 29.2 0.68 1.00 52.6 9.6 -38 32.4 0.76 0.76 54.1 8.9 The zinc concentrate was blended with fine quartz with a Pso of approximately 135pm. The synthetic feed had a final Pgo of 135um and a size distribution that is shown in Figure 6.2.1.2. Effectively the size distribution of the quartz and the sphalerite concentrate were equal. The composition of the synthetic feed is shown in Table 6.2.1.3. The zinc grade of the synthetic feed averaged 8.76%. Table 6.2.1.3 - Composition of the blended synthetic quartz/sphalerite feed Assayed Components % Composition Zn 8.76 S 5.21 Fe 1.75 Pb 0.13 Cu 0.10 52 Synthetic Quartz/Sphalerite Feed Size Distribution y = 0.0081x-0.2915 175 Size (microns) Figure 6.2.1.2 - Size distribution of the synthetic quartz and sphalerite feed 6.2.2 Test Program The synthetic zinc test program investigated the effects of bowl speed, pinch valve open time and pinch valve closed time. The experimental program consisted of 31 tests. The feed was changed completely twice to prevent depletion of sphalerite in discrete size fractions. It was unknown whether the CVD would preferentially separate certain particle sizes and shapes. If this were to have happened it could have altered the performance results in tests that were conducted later in the program. As well, between tests all the sample that was not required for assay purposes was returned to the feed tank. An outline of the experimental program and the order in which the tests were run is shown in Table 6.2.2.1. Note that the test order was randomized so minimize systematic experimental error. 53 Table 6.2.2.1 - Synthetic quartz and sphalerite test program Bowl Pinch Pinch Speed Open Closed 5 600 0.07 3 7 800 0.07 3 3 700 0.08 3 9 600 0.07 3 4 700 0.07 3 6 800 0.07 3 8 900 0.07 3 10 1000 0.07 3 1 600 0.05 3 2 700 0.05 3 11 800 0.05 3 12 900 0.05 3 13 1000 0.05 3 14 600 0.04 3 15 700 0.04 3 16 800 0.04 3 17 900 0.04 3 18 1000 0.04 3 19 600 0.05 4 20 700 0.05 4 21 800 0.05 4 22 900 0.05 4 23 1000 0.05 4 24 600 0.05 5 25 700 0.05 5 26 800 0.05 5 27 900 0.05 5 28 1000 0.05 5 29 800 0.05 4 30 800 0.05 4 31 800 0.05 4 54 6.2.3 Procedures and Sampling Samples of feed, concentrate and tailings were taken for each test. A slightly different procedure was used in sampling the synthetic quartz/sphalerite feed than was employed for the quartz/magnetite ore. The reason for this is that the samples for the quartz/sphalerite had to be chemically assayed which took longer than an assay performed by determining percent magnetics. As well, the sphalerite had to be dried at a lower temperature to prevent oxidation that took 24 hours. Any sample that was removed from the circuit was depleting the synthetic feed for 24 hours. A minimum amount of sample for each test had to be taken while still getting a representative sample. The machine was run with feed at the set operating parameters for eight minutes before sampling. Sampling then took place over the next four minutes. 25 liters of feed were taken for each run. There was no concern in depletion from feed samples because it was a homogeneous sample. The tailings were sampled in the same manner as in the quartz/magnetite testing. Four full pinch valve cycles of tailings were sampled over the four minutes of sampling. The concentrate was carefully taken to avoid feed depletion effects. Cuts of concentrate were taken over the four minutes of sampling. Each cut of concentrate included three full pinch valve cycles. An attempt was made to keep the concentrate samples close to 200 grams each. 6.2.4 Sample Preparation Slurry samples of feed, tailings and concentrate were all prepared for assay with the same procedure. The samples were pressure filtered, oven dried at 40°C, cooled, rolled, 10 grams split 55 for assay, and 80-100 grams split for size analysis or duplicate assaying. All remaining sample was added back to the feed tank. 6.3 Results -Synthetic Quartz/Sphalerite A wide scatter of results was obtained from the incremental testing of the synthetic quartz/sphalerite. Some basic trends with mass yield, recovery and grade could be seen when the results were graphed. However, the results were more qualitative than quantitative due to the small number of tests in each data set. An example of some trending results is shown in Figure 6.3.1 where the grade is shown to be affected by altering the pinch valve closed times and bowl speeds. Sulphur grade has been used to track sphalerite grade in these results and a dip is evident in 4 and 5 second pinch valve closed timings and is not present in the 3 second pinch valve closed timing at 700 RPM. A more thorough test program is required to determine whether a dip does in fact exist or whether there was an experimental error at this point. What the graph does show is that the CVD can clearly upgrade the synthetic feed and that increasing bowl speed and increasing pinch valve closed time reduces the concentrate grade. More detailed analysis of operating variable trends will have to be determined in plant testing. 56 Figure 6.3.1 - Bowl speed versus grade at varying pinch valve closed times Bowl Speed versus Grade (PO 0.05, FW 7) Pinch Closed = 4 sec Pinch Closed = 5 sec • A - Pinch Closed = 3 sec 500 700 900 1100 Bowl Speed (RPM) All of the results from the sphalerite/quartz synthetic feed testing were plotted as upgrade ratio versus recovery. The upgrade ratio was defined as the ratio between the CVD concentrate grade and the CVD feed grade. The Sphalerite grade was calculated based on sulphur and Zn assays. All the data points that formed an upper boundary condition were plotted independently and a trend line was drawn using these points. Figure 6.3.2 shows the plotted upper boundary condition that was derived from this test program. Since a broad combination of parameters was tested this plot represents the typical separation performance of a CVD6 at 135 micron P80 and a 1.35 SG differential. From the graph it can be seen that upgrade ratios range up to 4.25 and recoveries range from 10 - 70%. 57 Figure 6.3.2 - Upper boundary results for synthetic sphalerite/quartz Recovery versus Upgrade Ratio Synthetic Sphalerite - Quartz Background 80% Upgrade Ratio 58 6.4 Conclusions The synthetic sphalerite/quartz testing showed that the CVD is capable of making a separation at an SG differential of 1.35 with similar particle size distributions of the sphalerite and the quartz fractions. There was not sufficient data in this program to determine any distinct individual machine operating parameter trends. A larger data set is required to determine distinct trends. What was clear was that similar grade and recovery results could be obtained with a variety of machine operating variable combinations. The positive results from the lab testing of this synthetic ore indicate that it would be advantageous to test the CVD in the HBM&S mill. The plant testing will be performed in an effort to identify CVD machine operating variable trends. 59 7 Hudson's Bay Mining and Smelting Plant Trials 7.1 Introduction A CVD 6 was installed at HBM&S in Flin Flon Manitoba on a zinc flotation cleaner tails stream. Incremental plant testing was aimed at thoroughly evaluating the CVD operating variable ranges using a real feed. The following factors of processing real feed as opposed to synthetic feeds were present: • Particles were not always fully liberated • Size distribution was not as uniform • Ore consisted of a range of mineral densities One of the major advantages of plant testing is the speed at which data can be collected. In the lab each test takes hours to perform due to the amount of preparation required to maintain a consistent feed for each test. In the plant, a test turnaround is generally half an hour. The ability to run between 10 and 15 tests per day with relatively consistent feed enables large amounts of data to be collected for analysis. This data was generated to provide a large database of grade and recovery information to be used in the creation of an operating strategy. The incremental nature of the testing provides the ability to plot trends. These trends can show how three variables at a time interact to produce an overall impact on separation performance. 60 7.2 Experimental - Plant Trials at Hudson's Bay Mining and Smelting 7.2.1 Test Program The objective of this test program was to investigate how grade and recovery of gold, copper, and zinc changed with variations in operating parameter levels. This testing took place in an operational mill where ore conditions, general mill operating conditions, and maintenance needed to be factored into the design. Tests were organized into manageable modules that could be carried out in one day. Ore conditions were assumed to be constant throughout a six-hour time frame. Feed samples were taken for most tests, or at least one feed sample per hour. Each module changed only one variable and left the remaining variables fixed. An incremental test program was performed to determine the individual effects that bowl speed, pinch valve open time and pinch valve closed time had on the grade (upgrade ration), recovery, and mass yield. All other operating variables were held fixed. Fluidization water flowrate was held constant throughout the tests. It has been demonstrated in the laboratory quartz/magnetite testing that fluidization water has little effect on the grade and recovery if set within the range of 6-12 g/min. The solids feed rate to the machine was maintained throughout each module; however it varied from module to module with changing percent solids. The test program was first laid out with approximate ranges of operating variable levels set. Preliminary testing was performed with the CVD on site to fine-tune the tested ranges. A test program map is shown in table 7.2.1.1. 61 Table 7.2.1.1 - HBM&S incremental testing map of operating variable levels per test Bowl Feed Test Speed Pinch Pinch Rate Date ID RPM Open (s) Closed (s) % solids (tph) Oct 04/00 62 500 0.18 4 11.55 0.491 Oct 04/00 63 600 0.18 4 11.55 0.491 Oct 04/00 61 700 0.18 4 11.55 0.491 Oct 04/00 64 800 0.18 4 11.55 0.491 Oct 04/00 65 900 0.18 4 11.55 0.491 Oct 04/00 66 1000 0.18 4 11.55 0.491 Oct 05/00 67 500 0.18 2 21.15 1.01 Oct 05/00 68 625 0.18 2 21.15 1.01 Oct 05/00 82 625 0.18 3 21.15 1.01 Oct 05/00 69 625 0.18 4 21.15 1.01 Oct 05/00 70 625 0.18 6 21.15 1.01 Oct 05/00 71 625 0.18 8 21.15 1.01 Oct 05/00 72 700 0.18 2 21.15 1.01 Oct 05/00 73 700 0.18 3 21.15 1.01 Oct 05/00 74 700 0.18 4 21.15 1.01 Oct 05/00 75 700 0.18 6 21.15 1.01 Oct 05/00 76 700 0.18 8 21.15 1.01 Oct 05/00 77 850 0.18 2 21.15 1.01 Oct 05/00 78 850 0.18 3 21.15 1.01 Oct 05/00 79 850 0.18 4 21.15 1.01 Oct 05/00 80 850 0.18 6 21.15 1.01 Oct 05/00 81 850 0.18 8 21.15 1.01 Oct 06/00 83 700 0.16 4 31.15 1.72 Oct 06/00 84 700 0.17 4 31.15 1.72 Oct 06/00 85 700 0.18 4 31.15 1.72 Oct 06/00 86 700 0.19 4 31.15 1.72 Oct 06/00 87 700 0.2 4 31.15 1.72 Oct 06/00 88 850 0.14 4 " 31.15 1.72 Oct 06/00 89 850 0.15 4 31.15 1.72 Oct 06/00 90 850 0.16 4 31.15 1.72 Oct 06/00 91 850 0.17 4 31.15 1.72 Oct 06/00 92 850 0.18 4 31.15 1.72 Oct 10/00 93 625 0.18 4 38.9 2.04 Oct 10/00 94 625 0.19 4 38.9 2.04 Oct 10/00 95 625 0.2 4 38.9 2.04 Oct 10/00 96 625 0.21 4 38.9 2.04 Oct 10/00 97 625 0.22 4 38.9 2.04 Oct 10/00 98 700 0.18 4 38.9 2.04 62 The tests for this incremental design were run over two weeks. The test program was set-up in a modular format where each module tested a different variable with all of the other feed and operating variables held constant. Modules were never interrupted and were always run on the same day. It was advantageous to be running the concentrator on a cleaner tailings stream because the effect of changing ore conditions and operational difficulties in the mill were dampened at that point in the circuit. 7.2.2 Procedures and Sampling Each day before the CVD was run; a thorough visual inspection was performed. The valves were set to 5 seconds open time and 2 seconds closed time. When the valves were in the open position a nylon tie-wrap was inserted in the valve to ensure there was no blockage. The strip was left in the valve until the valve went into the closed cycle at which time the tie-wrap was gently pulled to make sure that the valve was actuating properly. This procedure was repeated for all eight valves. Next Fluidization water was run and a visual inspection was made of the flow through the fluidization holes. The water was left on for about five minutes to ensure the machine was flushed. At this point the machine was ready for start-up. With the water running the machine was started at the testing conditions. The feed rate was adjusted by using a stopwatch and a bucket. Once the desired feed rate was established the feed tube was inserted in the top of the machine and the test timer started. In order to achieve accurate and reproducible results, careful sampling procedures were established. The two major considerations for obtaining good samples from the CVD were 63 establishing a steady state within the machine and getting a representative cut of feed, concentrate, and tailings. When any variable is adjusted in the CVD a certain amount of time is required for the machine to achieve steady state. In the laboratory with synthetic feeds the steady state time was eight minutes. In the mill longer stabilization time was allowed. Each test was given a minimum of 15 minutes to stabilize. In tests where the mass yields were exceptionally low the machine was sometimes left for up to two hours to stabilize. There is no data to justify leaving the machine to stabilize for that long. The basis for leaving the machine longer to stabilize when very little mass was being extracted was so that all of the material that may have been resident in the machine or the sample pipes would be flushed between tests. Samples were required for feed, concentrate and tailings from each test. The tailings line was always the highest flowrate and was too high to take the entire stream for any length of time. Due to the semi-continuous extraction of concentrate, the tailings stream needed to be sampled over entire pinch valve cycles. If the sample did not accurately represent an entire cycle, erroneous results may have been created. Therefore, several cuts of the entire tailings stream over at least two whole pinch valve cycles were taken. Due to the low flowrate of concentrate, the entire concentrate was collected over varying times between 30 seconds and 6 minutes. At the end of a test where a feed sample was desired the entire feed stream was collected in a bucket for ten seconds. 64 7.2.3 Sample Preparation Slurry samples of feed, concentrate and tailings were all prepared for assay in the same way. Slurry samples were pressure filtered and transferred to pans and dried overnight in a 60 degree C oven. Once dry, the samples were screened at 8-mesh to break the dry cake into a powder. The powder was then mixed on waxed paper and a portion was split out for assay and a second was retained. 65 7.3 Results - Plant Trials at HBM&S 7.3.1 Introduction Incremental testing of the CVD 6 in Flin Flon, Manitoba at Hudson's Bay Mining and Smelting was performed to determine the trends of operating variables on separation performance and to evaluate the effect of specific gravity differential on particle separation. The results are based on gold assays as they were the most consistent results out of all the assays. 7.3.2 Bowl speed Tests were conducted on the zinc cleaner tailings at bowl speeds ranging from 500 to 1000 rpm. The relationship between gold grade and recovery versus bowl speed is shown in Figure 7.3.2.1. The slope of the grade versus bowl speed curve shows two distinct zones. Between 500 rpm and 700 rpm, the grade decreased significantly from 35 g/t to 15 g/t gold. However, beyond 700 rpm the slope flattens; increasing the speed from 700 rpm to 1000 rpm lowers the grade by only 5%. There is a transition zone for grade between 650 rpm and 750 rpm. 66 Grade and Recovery versus Bowl Speed 500 600 700 800 900 1000 Bowl Speed (rpm) Figure 7.3.2.1 - Grade and recovery versus bowl speed The recovery mirrors grade, having two distinct slopes at low and high bowl speeds. Between 500 rpm and 700 rpm there is only a 3% increase in the recovery, but between 700 rpm and 1000 rpm there is a 55% increase. There is clearly a change in separation mechanism that occurs at the transition speed of about 700 rpm that affects the performance. 7.3.3 Pinch valve open time Pinch valve open times were varied from 0.14 to 0.22 seconds, while other operating variables were set (Table 1). The effect of open time was evaluated at three different bowl speeds (625 rpm, 700 rpm and 850 rpm). These test conditions were selected with consideration of the gold recovery objectives for the mine. 67 Figure 7.3.3.1 shows that the concentrate grade decreased with increasing pinch valve open time. The trend was the same at all three bowl speeds except that the curves were shifted downwards with increasing bowl speed. Grade versus Pinch Valve Open Time 40 35 30 S 25 m 20 o < 10 5 850 rpm \ 700 rpm \ 625 rpm \ v • 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.2 0.21 Pinch Valve Open Time (seconds) 0.22 0.23 Figure 7.3.3.1 - Grade versus pinch valve open time Figure 7.3.3.2 shows that the recovery increases with increasing pinch valve open time. Increasing bowl speed shifts the recovery upwards. Recovery increases more rapidly at higher rpm's per incremental change in pinch valve open time. This trend is more visible beyond the transition point in the recovery versus pinch valve open time curves. The rate of change in recovery versus pinch valve open time decreases with decreasing bowl speed. This effect can be seen in Figure 7.3.3.2 where the linear trend line slopes become flatter when comparing the 850 rpm, 700 rpm, and 625 rpm lines. 68 Recovery versus Pinch Valve Open Time Pinch Valve Open Time (seconds) Figure 7.3.3.2 - Recovery versus pinch valve open time The data trends were drawn as curves in Figures 7.3.3.1 and 7.3.3.2 with straight lines superimposed. The straight lines indicate two distinct sections for each rpm tested with a transition zone located around the intersection of the lines (transition point). The trend lines extend beyond the transition zone to show that the transition is smooth. This suggests that the mechanism responsible for the change in slope does not occur at a specific point. However, the transition point is useful to distinguish between zones dominated by different separation mechanisms. For pinch valve open times up to the transition point, the steep decrease in grade can be explained by the dilution of the high-grade material near the pinch valve resulting from the recovery of lower grade material. Increasing the pinch valve open time further, beyond the transition point, further increases the proportion of the low-grade material in the ring that is 69 recovered. However, the incrementally increased recovery of low-grade material has a diminishing effect on dilution. Therefore the grade levels off and eventually approaches the feed grade. 7.3.4 Pinch valve closed time Pinch valve closed times were varied from two to eight seconds while hold other operating variable levels constant (Table 7.2.1.1). As shown in Figure 7.3.4.1, the concentrate grade increased with increasing pinch valve closed time. A longer pinch valve closed time allows for more upgrading in the concentrating ring resulting from heavy particles displacing light ones. Figure 7.3.4.1 also shows that while upgrading took place at a bowl speed of 625 rpm, at higher bowl speeds (700rpm and 850 rpm) the concentrate grades were not affected significantly. These results support those described above, which suggested that high bowl speeds inhibit upgrading in the ring. At high bowl speeds, packing in the ring likely prevents particle displacement. 18 16 14 5" 12 " © 1 0 •a re O Grade versus Pinch Valve Closed Time 625 rpm 850 rpm 3 4 5 6 7 Pinch Valve Closed Time (seconds) 70 Figure 7.3.4.1 - Grade versus pinch valve closed time At 625 rpm, the absence of a clear transition point indicates that the grade will continue to increase with closed time at a uniform rate. Eventually the grade would approach a constant value that depends on particle specific gravity and size. As described above, visual observations reveal that cones of coarse high density material radiates from the pinch valves. The particle size gradation within the cone ranges from coarse near the pinch valve to fine further away from the valve. As shown in Figure 7.3.4.2, the recovery decreases significantly for pinch valve closed times up to about 3 seconds and then decreases at a much slower rate. This transition point represents an operating limit for pinch valve closed time. Based on operating experience with the CVD, it is not practical to use pinch valve closed times of less than 3 seconds because the upgrade ratios are low. 71 Recovery versus Pinch Valve Closed Time Pinch Valve Closed Time (seconds) Figure 7.3.4.2 - Recovery versus pinch valve closed time At 625 rpm, there was almost no recovery over the whole range of pinch valve closed times. This result was explained by the small pinch valve open time used for these tests, which did not allow significant mass pull. The recovery curves shift upward with increased bowl speed. 7.4 Conclusions Plant testing at the HBM&S mill generated significant data that showed the trends formed by changing operating variables. The one parameter at a time testing methodology showed how the response curves for grade and recovery formed consistently shaped curves that were shifted with varying parameter levels. The consistency of the shapes will enable a control strategy to be developed. 72 8 Specific Gravity Differential 8.1 Introduction Particle separation in the CVD is based on a combination of specific gravity separation and particle size classification. The studied mineral systems all have different specific gravity differentials. The effect of specific gravity differential on separation performance was investigated in the lab and then verified in the field. This chapter investigates the upper boundary condition (recovery versus upgrade ratio) results for quartz/magnetite, quartz/sphalerite, and zinc flotation cleaner tails. Knowing the particle specific gravity differentials between the target and background minerals the separation limitations of the CVD are investigated. It is the objective of this chapter to determine the impact of particle specific gravity on separation in a CVD concentrator. 8.2 Discussion Figure 8.2.1 shows the upper boundary results for recovery and upgrade ratio for the two studied synthetic ores; quartz/magnetite and quartz/sphalerite. When plotted on a normal - log plot, two straight lines are produced. The specific gravity (sg) differential between magnetite and quartz is 2.55 and sphalerite and quartz is 1.35. As expected, Figure 8.2.1 shows that in a fully liberated system that the higher the specific gravity differential the better the separation performance. Also note that although the sphalerite system produced lower upgrade ratios, the CVD was still able to make a separation at a sg differential of 1.35. 73 Recovery versus Upgrade Ratio Synthetic Ores - Quartz Background 100% 1 10 100 Upgrade Ratio ure 8.2.1 - Separation performance of fully liberated synthetic feeds 74 Recovery - Upgrade Ratio HBM&S Plant Trials 100% T - T -> 0% 1.0 Upgrade Ratio 10.0 • Au • Chalcopyrite * Sphalerite • Galena Figure 8.2.2 - Upper boundary lines for recovery and upgrade ratio from the plant trials at HBM&S Results from the plant trials are analysed separately from the synthetic ores tested in the lab. It is not clear exactly what the sg differential is as there are 6 components in the natural ore from HBM&S that are competing. As well, liberation is not 100% as it was in the synthetic systems. The main composition of the feed to the CVD was pyrite and sphalerite. The feed material in the HBM&S trials was a flotation cleaner tailings and has a relatively narrow particle size distribution with a P80 of approximately 100 microns. For this analysis the impact of particle size has been ignored. 75 Table 8.2.1 - Approximate specific gravity differentials for HBM&S plant trials Gold Chalcopyrite Sphalerite Galena Pyrite silicates % Composition 5.2 X 10"' 5.5 21.9 1.7 60.2 10.7 SG 19.3 4.2 4 7.5 5 2.7 Background SG 4.53 4.55 4.68 4.48 3.83 4.75 Theoretical SG Differential 14.77 -0.35 -0.68 3.02 1.17 -2.05 The gold particles were associated with pyrite. The specific gravity of a gold particle is approximately 19.3. When gold is locked in pyrite the sg becomes a blend between pyrite and gold. The gold in this ore was not free gold. If approximately 3 0 % of a particle was gold and 70%o were pyrite, then the particle sg would be about 8.5. This is the estimate used for this analysis as the sg differential becomes 3.94 for a gold/pyrite particle against a background of sg 4.53. Figure 8.2.2 shows that the Au line is just above the galena line. The galena has an sg differential of 3.02 and is typically well liberated in this concentrate. Table 8.2.1 shows a theoretical calculation of sg differential. The composition of the components is based on chemical assay. The background sg differs for each component because the component in consideration is removed from the calculation in each case. It can be seen that chalcopyrite, sphalerite and silicates are lower than the background sg in this material. Therefore it is expected that if these components are present in the CVD concentrate it is only due to mass yield (feed grade material that reports to concentrate). This can be seen to be true in Figure 8.2.2 where these component lines are all very close to an upgrade ratio of 1. Figure 8.2.2 shows that Chalcopyrite is slightly upgraded in higher recovery conditions. Sphalerite is very slightly upgraded when the machine is pulling large mass yields and 76 correspondingly high recoveries. Chalcopyrite is always upgraded more than sphalerite even though the sg differential between the two minerals is only 0.2. Under no set of operating conditions did a lower density particle get upgraded more than a higher density particle. There are clearly four boundary conditions that do not intersect over the entire set of tests that are defined by the lines drawn for each mineral in Figure 8.2.2. This testing demonstrates that separation is possible down to an sg differential of 0.2. However, selectivity would be impossible unless extremely small mass yields were obtained. For small mass yield applications a batch centrifugal gravity concentrator or other similar technology is better suited. In order to get upgrade ratios in excess of 3:1 with recoveries over 50%, a differential between the target and gangue mineral needs to be 2 or greater. This statement is based on the fully liberated synthetic systems shown in figure 8.2.1. Also, as seen in the plant trial example, with multi component feeds the target mineral would need to have a sg differential of 2 or more between the next highest gangue mineral to achieve selectivity. The CVD cannot be selective at high mass yields unless there is a sg differential of 2 or more. 8.3 Conclusions This chapter has shown that the CVD6 concentrator can make separations at a specific gravity (sg) differential of 0.2 between the target and the background sg. However in order to achieve good selectivity a practicle sg differential of 2 or more is required. 77 It is recommended that further work be performed on narrower size classes to further understand the interaction between particle density separation and particle size separation. 78 9 Variable Tuning in a CVD Concentrator 9.1 Introduction This chapter demonstrates the effects of operating variables on separation performance. The data was obtained from laboratory and plant testing of a pilot scale Knelson CVD6. By comparing the rates of change of grade and recovery with incremental changes in the operating variables, a set of basic guidelines for developing a control strategy is proposed. 9.2 Operating variable performance ratio As demonstrated by the results of this study, the CVD has the ability to achieve a wide range of metallurgical results. Figure 9.2.1 shows a plot of recovery versus grade for gold. Each point on the plot represents a different set of operating conditions. The upper bound on the plot identifies the maximum performance of a CVD6 for this ore. It is important to know how to manipulate the operating variables in order to operate a CVD as close to the upper bound as possible. 79 Recovery versus Grade Au Grade (g/t) Figure 9.2.1 - Recovery versus grade Knowing the effect of each variable on both grade and recovery aids in developing an operating strategy. For all variables tested, changes in the level to increase the concentrate grade resulted in a decrease in recovery and vice versa. To assist with the development of an operating strategy, a methodology for comparing the change in grade and recovery at a specified operating variable range was established. Each of the operating variables has different ranges and increments (units) of change. Bowl speed is measured in rpm, pinch valve closed times in seconds, and pinch valve open time in fractions of a second. The difference in units makes direct comparison of slopes impossible. Instead, broad operating ranges were selected for the three operating variables. Each operating range was broken down into equal increments called levels. A midpoint in each operating range 80 was identified for each operating variable. The midpoints were all selected from previous operating experience as conservative operating levels. Tests were selected where one variable at a time was changed while the other three variables were held at midpoint levels. For each operating variable, grade versus parameter level and recovery versus parameter level plots were generated. Linear trend sections on these plots were identified and trend lines drawn. The slopes for grade versus parameter level and recovery versus parameter level were divided to produce a ratio of the change in grade versus the change in recovery. This ratio of slopes was used to compare the overall impact of a variable on the separation performance and is referred to as the operating variable performance ratio (ovp ratio). Comparing the ovp ratio for each operating variable quantifies the combined grade and recovery effect of changing each operating variable. Since grade and recovery always have opposing slopes a net negative ratio results in all cases. For comparison, the absolute value of the ovp ratio was used. Table 9.2.1 shows the direction of the trends. The ovp ratio is a semi-quantitative tool for selecting an operating variable to change when tuning the operation of a CVD. As discussed before, the CVD will have a maximum performance boundary condition for grade and recovery as demonstrated in figure 9.2.1. The ovp ratio gives the operator a map of which operating variable to change in order to achieve boundary condition performance or to move to another point on the boundary line. If the 81 objective is to increase predominantly grade (shift horizontally on Figure 9.2.1), then a high ovp ratio is desired. However if recovery is the objective a low ovp ratio is desired. The ovp ratio is a guideline for roughly quantifying an operating variable's overall effect on separation performance. Table 9.2.2 summarizes the ovp ratios that were generated at each variable level. Due to transition points in the trends, the ratios change for bowl speed and pinch valve closed time. In Table 9.2.2, a high ovp ratio at low bowl speed (403), with all other variables at midpoints indicates that an increase in bowl speed will influence grade the greatest. Pinch valve open time had a constant ovp ratio throughout the tested range meaning that it would have the same result on grade with any incremental change in open time. Other than at low levels, pinch valve closed time also has a uniform effect on grade. In general, these results indicate that pinch valve open time generates the greatest change in grade per incremental change in operating variable level. Table 9.2.1 - General trends for operating variables Grade Recovery Increasing Fluidization Decreases Increases (to a limit) Water Flowrate Increasing Bowl Speed Decreases Increases Increasing Pinch Valve Decreases Increases Open Time Increasing Pinch Valve Increases Decreases Closed Time For the majority of the operating range (levels 3 - 7), pinch valve open time should be used in making coarse adjustments to concentrate grade. Fine-tuning of the grade should be performed with bowl speed. If intermediate tuning is needed, the pinch valve closed time can be effective. 82 When considering ovp ratios from a recovery perspective the results mirror those for grade. Levels 3 through 6 in Table 9.2.2 represent a reasonable operating range. Low ovp ratios in this range for bowl speed indicate that bowl speed has the greatest effect on recovery per change in grade. If a significant change in recovery is required with a minimum effect on grade, bowl speed should be adjusted. A mid-range ovp ratio of 32 for pinch valve closed time classifies this variable as a mid-range tuning variable for recovery. The ovp ratio of 182 for pinch valve open time indicates that it is not an effective variable for tuning recovery. Table 9.2.2 - Rate of change in grade versus rate of change in recovery Level Actual Bowl Speed Setting (rpm) Actual Pinch Open Setting (seconds) Actual Pinch Closed Setting (seconds) Bowl Speed OVP Ratio Pinch Open Time OVP Ratio Pinch Closed Time OVP Ratio 1 500 0.17 2 403 182 1 2 600 0.18 3 403 182 32 3 700 0.19 4 10 182 32 4 800 0.20 5 10 182 32 5 900 6 10 32 6 1000 7 10 32 7 8 32 For fine-tuning of the machine, the operating variables can be adjusted in smaller increments. The resolution of control on the operating levels will influence how much fine control a variable can have. The bowl speed resolution is lrpm, pinch valve open time 0.01 seconds, and the pinch valve closed time 0.1 seconds. The pinch valve open time is the only operating variable that was tested in the smallest increment possible. When testing a new feed in a CVD, a systematic approach for finding the appropriate operating variable levels should be employed. • First run the machine at the center-point conditions: 700 rpm bowl speed, mid-range (0.18 seconds) pinch valve open, 4 seconds pinch valve closed. Pinch valve open time is 83 described as mid-range because all CVD's have unique valve histeresis that changes the pinch valve open ranges. • Determine the grade and recovery for the center-point test. The relationships for each operating variable's effect on grade and recovery can be used to tune the performance. It is recommended to only change one variable at a time. 9.3 Conclusions The CVD has four operating variables that affect the metallurgical performance and therefore levels must be selected to achieve the desired metallurgical results. For all operating variables, changing the levels to increase grade cause a reduction in recovery and vice versa. A good understanding of the effects of each variable is important to developing an operating strategy for optimum performance. The CVD's operating variables are fluidization water flowrate, bowl speed, pinch valve open time, and pinch valve closed time. Table 9.2.1 summarizes the general trends observed when increasing each of the operating variables. After setting the fluidization water flow rate and selecting an appropriate bowl speed, product grade and recovery can be optimized by adjust the pinch valve timing. The pinch valve open and closed times have an interrelated effect for achieving separation performance. Increasing the closed times allows for more upgrading in the ring. The required open time is the timed need to drain the upgraded product from each cycle of operation. 8 4 A ratio of the change in grade versus the change in recovery was defined as the operating variable performance ratio (ovp ratio). The ovp ratio was used to quantify the operating variable's overall effect on separation performance. The ovp ratio is considered a useful parameter to help decide which variable should be changed to cause an improvement in grade recovery. 85 10 Developing an Operating Strategy for a New Application 10.1 Introduction The research performed for this project has shown that the CVD can produce a wide range of mass yield, grade and recovery results. Each CVD application will have a unique set of objectives depending on the nature of the material, the complimentary processing equipment and the required product specifications. As shown in the preceding sections of this report, manipulation of the operating variables and their interaction will affect the final concentrate outcome. This section is dedicated to outlining a procedure for tuning the CVD6 concentrator for a new application. It is assumed that the concentrator is on site and will be tested under constant feed conditions. This is a critical outcome and is the result of combining the knowledge of operating parameter ranges and interactions. 10.2 Identification of the Objectives The CVD concentrator by nature is capable of mass yields to concentrate of 0 - 100% (typically 1 - 60%) of the feed to the unit. By adjustment of all the operating variables in conjunction, one can produce a random array of results. The ever present trade-off between grade and recovery is present in the results of a CVD concentrator as they are with any piece of mineral processing equipment. Therefore it is paramount that the first task in the tuning of a CVD be in the clear identification of the expectations of the equipment. 86 The first step in making the determination of the objectives is to understand and characterize the nature of the feed to the concentrator. The target mineral grade should be known. The association of the target mineral with other species along with the degree of liberation is useful information. This information can then be used to make a determination of the approximate mass yield range that will be required to attain the desired recovery. For example, if an iron oxide material were to be removed from talc and the content of iron oxide in the material was 0.5% by weight. Then it would be reasonable to assume that in order to remove the iron oxide from the talc the target mass yield would have to be greater than 0.5% and likely less than 5% of the feed reporting to concentrate. In this case, the product specification will require that any content of iron oxide in the talc will render the talc unsaleable. Therefore the mass yield range will more likely be 3 - 10%. The more information that is available to narrow the optimization window the faster and more accurately a solution will be attained. Once a realistic expectation of the result has been formulated then the "goal posts" have been set and the test program can be organized. 10.3 CVD Boundary Conditions In order to effectively set out realistic expectations for the CVD6 it is necessary to understand what its general capabilities are. These capabilities are only semi-quantitative and are limited to the experience that has been gained in the test programs performed in the production of this report. Below is a list of basic rule of thumb expectations for the use of a CVD6 concentrator: 87 • The CVD has been shown to be a very effective "rougher" machine and can generally produce concentrate grades that are 3 to 6 times as high as the feed grade. • Recoveries are typically in the range of 70 - 95% 10.4 Qualitative Tuning When the CVD6 is first started it can be run under any set of conditions. Once running it will be clear to see approximately how much concentrate is coming out of the concentrator. Since a basic range of what mass yield is to be targeted has already been established, the operating parameters can be changed to get within the testing range. This is where knowledge of the operating variable effects and interactions can be applied. In the fractional factorial design testing using a feed of quartz and magnetite the effect of both operating variables and feed variables were investigated to determine the relative effects of the variables. When a concentrator is tested in the field there is little opportunity to affect the percentage solids in the feed and the particle size of the target mineral. These results will not be considered in the field testing and tuning of a CVD as they are not variables that can be practically changed. For this machine tuning procedure there will be a focus on only the physical CVD machine variables that include the bowl speed, inch valve open time, pinch valve closed time, and fluidization water flowrate. The testing that was performed with the quartz and magnetite material was a useful starting point for the understanding of the relative importance of the operating variables and their interaction effects. Unfortunately, the nature of such a test program requires the selection of a low mid and high value for each of the variables. Because this was the first test program that was used to 88 evaluate the CVD some of the parameter ranges were not selected ideally. The results of the testing were not incorrect; however in combining the knowledge gained from the incremental testing, further field testing, and experience the machine variables can be ranked by importance follows: pinch valve closed time, bowl speed, pinch valve open time, and fluidization water flowrate. The rankings will help in tuning the approximate range that the operating variables should be adjusted within. However, what will be demonstrated in the following section is that although the operating variable adjustments can be ranked the interaction of different combinations of operating variable levels can produce the identical result. It is because of this complex and overlapping interaction that necessitates a tuning procedure such as the one that is outlined in this section. The most valuable outcome of this thesis is the understanding that it is not the impact of any singular variable on grade and recovery but the achievement of a desired objective by understanding the capabilities and limitations of the CVD6 and how to tune the machine to achieve the desired objective. It is the aim that this procedure can be used to tune all sizes of CVD concentrator and not just the CVD6. There is a lack of information on the effect of performance scale-up from a CVD6 to larger CVD models, however it is the opinion of the author that the rankings of operating variable importance and testing methodology would not change as the size of the machine increases. It is possible that the performance characteristics would change due to changes in the geometry and size within the machine, however the fundamentals should remain relatively constant. 89 It was shown that fluidization water flowrate had no measurable effect on recovery or grade as long as the CVD6 was operated between 7-10 gpm water addition. Other CVD models would be similar but the range of fluidization water addition would be different. For this reason the fluidization water flowrate should be set in the middle of the range at approximately 8.5gpm and this variable can be eliminated as a variable for the duration of the testing. Once the final operating variables have been established the fluidization water can be altered slightly in an attempt to fine tune the results, however it is expected that the impact will be minimal. The three remaining variables, namely bowl speed, pinch valve open time and pinch valve closed time, will greatly affect the performance of the concentrator. Bowl speed should be set initially in the middle of the range at 750 rpm as a starting point. The pinch valves can then be adjusted to affect the amount of material that is reporting to the concentrate. The exact valve timings cannot be reported here as each installation will vary depending on the air inlet pressure, length of air delivery line and size of the concentrator. This is why it is recommended to qualitatively adjust the valves and take note of the mass yield produced to concentrate. If the nature of the material is such that any grade determination can be determined by hand panning or observation of a change in colour, then this is recommended in order to identify a good starting range for the parameters. Within an hour, the CVD operator should be able to determine an effective range of pinch valve timings that will cover the range. If in doubt simply test a very wide range and the results will drop out. The qualitative portion of this program simply reduces the time and effort that will be spent on the next stage of the test program. 90 10.5 Formulation of a General Test Program A wide range of results can be obtained from the manipulation of the CVD operating variables and the consequential interaction between those variables. In fact, in a test program that covers the major ranges of all of the operating variables the result will plotted on a grade recovery curve will yield a very broad spectrum of results. What is possible to determine from any of these sets of data is that all of the data points will lie beneath a boundary condition line. If sufficient data points can be generated that suitably span the operating range for the key operating variables (bowl speed, pinch valve open time, and pinch valve closed time) then an upper boundary line can be drawn on the plot that defines the optimum operating conditions for the CVD. By moving along the upper boundary condition line the optimum operating conditions for the desired objective can be selected. In the case where the optimum condition lies between two or more points a more focused optimization test program can be run to fine tune the operating variables and get closer to an optimum condition. 91 Recovery • • • • Grade Figure 10.4.1 - Drawing an upper boundary condition on a grade recovery curve The results from a very comprehensive test program will yield points on the grade recovery curve that are equal within error and yet have been derived from different operating variable combinations. It is this phenomenon that necessitates the use of the boundary condition evaluation of the results. As long as the operating variable combination that is selected results in a result that lies on or in close proximity to the upper boundary condition then an acceptable set of conditions has been selected. 10.6 Fine Tuning the Operating Variables As demonstrated in the section of this report on the trends in Au recovery and the discussion on the Operating Variable Performance ratio (OVP), the impact and trends of individual operating variables can be carefully changed in order to get closer to the desired response. The OVP methodology should be utilized for the initial fine tuning of the concentrator and also for 92 example when changes in the ore conditions are presented that require a small change in approach. 10.7 Conclusion The complexity and interaction effects of the operating variables that exist with a CVD lead to the possibility of producing the same result with different combinations of operating variables. As well, if a wide enough spectrum of each operating variable is not tested then a non optimized result can be attained. It is therefore necessary to have a desired objective and then test the CVD capabilities for a specific feed material to determine the upper boundary condition that exists for that application. With the knowledge of what effect individual operating variables have on grade and recovery as outlined in the Operating Variable Performance Ratio section the results can be fine tuned to move along the upper boundary condition line. 93 11 Conclusions and Recommendations This study included fundamental laboratory pilot testing coupled with field pilot work to evaluate the operation and control of a continuous centrifugal concentrator. Specifically the Knelson Continuous Variable Discharge (CVD) concentrator was tested. The program included the following main sections: • Literature Review • Experimental Program • Quartz/Magnetite Factorial Design • Quartz/Magnetite Incremental Testing • Synthetic Sphalerite/Quartz • Hudson's Bay Mining and Smelting Plant Trials • Specific Gravity Differential • Variable Tuning in a CVD Concentrator • Developing an Operating Strategy for a New Application 11.1 Factorial Design Four feed variables; heavy particle size, feed percent solids, feed rate, and magnetite grade were identified. Four CVD machine operating variable were identified; bowl speed, pinch valve open time and pinch valve closed time. With a total of eight variables to investigate a factorial design was used to rank the relative importance of each of the variables, quantified as the effect on grade and recovery. The results are summarized in Table 11.1. 94 Table 11.1 - Summary of significant variable effects Feed Variables Machine Variables Grade Recovery Grade Recovery Solids Feed Rate Heavy Particle Size Bowl Speed Pinch Valve Closed Time % Solids Pinch Valve Closed Time Bowl Speed Feed Grade Fluidization Water 11.2 Incremental Quartz/Magnetite The nature of a statistical experimental design such as the factorial design that was employed in Chapter 4 sets variable levels at high, low and midpoints. In order to understand the trends that occur in the grade and recovery results over a full operational range an incremental test program was employed. A quartz/magnetite feed was used, but only machine variables were tested. One variable at a time was manipulated and the resulting trends were plotted. Incremental testing resulted in the following conclusions. Increasing bowl speed was found to increase recovery and mass yield while decreasing concentrate grade. Fluidization water flowrate was determined to have the greatest effect on recovery. Threshold minimum fluidization water was found at 6 gpm. Above 6 gpm additional fluidization water did not affect recovery and only slightly reduced grade. It was determined from this result that fluidization water could be held constant for all subsequent testing. Fluidization water is necessary to enhance separation, but is not a variable that should be manipulated to enhance separation 95 performance. The CVD was found to perform best with feed percent solids of approximately 30%. 11.3 Synthetic Sphalerite/Quartz In order to evaluate the concentrator's performance on a different feed with a lower specific gravity differential, but a fully liberated system, a synthetic feed of quartz and sphalerite was tested. The feed was made in the lab by mixing a sphalerite flotation concentrate from Hudson's Bay Mining and Smelting (HBM&S) with quartz sand. The results of the sphalerite/quartz testing satisfied three objectives: contributed separation performance at a narrow sg differential in a fully liberated system, identified more information on machine and feed operating parameter ranges, and the successful particle separation justified further testing of the concentrator at HBM&S on a non-synthetic feed. The synthetic sphalerite/quartz testing showed that the CVD is capable of making a separation at an SG differential of 1.35 with similar particle size distributions of the sphalerite and the quartz fractions. 11.4 HBM&S Plant Testing Plant testing at the HBM&S mill generated significant data that showed the trends formed by changing operating variables. The one parameter at a time testing methodology showed how the response curves for grade and recovery formed consistently shaped curves that were shifted with varying parameter levels. The consistency of the shapes provided data for use in the 96 development of an operating strategy. Plant testing also provided important information on selective particle separation by specific gravity. 77.5 Specific Gravity Differential Information from all of the incremental test programs contributed to this section. It was shown that the CVD6 concentrator can make separations at a specific gravity (sg) differential of 0.2 between the target and the background sg. However in order to achieve good selectivity a practical sg differential of 2 or more is required. It is recommended that further work be performed on narrower size classes to further understand the interaction between particle density separation and particle size separation. 11.6 Variable Tuning in a CVD Concentrator Primarily utilizing the information gained from the field testing of the CVD at HBM&S, a tuning methodology for the machine operating variables; fluidization water, bowl speed, pinch valve open time and pinch valve closed time, was developed. After setting the fluidization water flow rate and selecting an appropriate bowl speed, product grade and recovery can be optimized by adjusting the pinch valve timing. The pinch valve open and closed times have an interrelated effect for achieving separation performance. Increasing the closed times allows for more upgrading in the ring. The required open time is the time needed to drain the upgraded product from each cycle of operation. 97 A ratio of the change in grade versus the change in recovery was defined as the operating variable performance ratio (ovp ratio). The ovp ratio was used to quantify the operating variable's overall effect on separation performance. The ovp ratio is considered a useful parameter to help decide which variable should be changed to cause an improvement in grade or recovery. 11.7 Developing and Operating Strategy for a New Application The CVD concentrator has the opportunity to be applied in a wide spectrum of applications. However, with the complexity and interaction effects of the operating variables a methodology for testing a new application is beneficial. This section takes the results and experience from all of the test programs in this study and introduces a methodology that can be employed in testing any new application for a CVD concentrator. Fundamentally this methodology includes: • Identification of performance objectives • Understanding the CVD's boundary conditions • Qualitative tuning (preliminary testing) • Formulation of a general test program • Fine tuning of the operating variables 98 11.8 Recommendations This study exclusively used the pilot scale Knelson CVD6 concentrator. It would be useful to investigate the scale-up effects from pilot scale to production scale. Finally, the effect of particle size on separation performance was not studied in detail in this study. Further work to investigate the effect of particle size interactions is recommended. 99 12 References 1. Ancia, Ph., J. Frenay, Ph. Dandois, "Comparison of the Knelson and Falcon Centrifugal Separators", Mozley International Conference, Falmouth G.B., June 1997. 2. Burt, R.O., 1984, "Gravity Concentration - from Bench Scale to Plant", Annual Meeting of Canadian Mineral Processors, Ottawa. 3. Byron R., K. Roberts, 2004, "Flotation Improvements in the Luzenac Penhorwood Talc Concentrator", Proceedings 36th Annual Meeting of the Canadian Mineral Processors, Ottawa, Canada, pp 177-188. 4. Falcon Concentrators http://www.concentrators.net 5. Holland - Batt, A.B. Gravity Separation: A Revitalized Technology - Mining Engineering. September 1998. 6. Honaker R. Q., B. C. Paul, D. Wang and M Huang, 1995, "The application of Centrifugal Washing for Fine Coal Cleaning", Minerals and Metallurgical Processing, Vol. 12, pp. 80-84. 7. Knelson Gravity Solutions http://www.knelson.com 100 8. Lambert M., M. McLeavy, B. Klein, I. Grewal, 1999, "Preliminary Studies with a New Continuous Centrifugal Concentrator", Presentation at BC and Yukon CMP Annual Meeting. 9. Banisi S., Laplante A.R.; McGill University, Montreal, Marois J.; Hemlo Gold Mines Ltd., 1991, "The Behaviour of Gold in Hemlo Mines Ltd. Grinding Circuit". 10. McLeavy M., B. Klein and I. Grewal, 2001, "Knelson Continuous Variable Discharge Concentrator, Analysis of Operating Variables", International Heavy Minerals Conference, Fremantle, Australia, pp. 119-125 11. Richards R. G. and M. K. Palmer, 1997, "High Capacity Gravity Separators - A Review of Current Status", Minerals Engineering, Vol. 10, No. 9, pp. 973-982. 12. Roche Mining (MT), 2004, Kelsey Jig http://www.Geologics.com.au 13. Rogan, Chris. Pilot Scale Evaluation of Knelson Variable Discharge Concentrator. University of New South Wales, School of Chemical Engineering. November 1995. 14. Silva, E.C., N. A. Santos, and V. M. Torres, 1999, "Centrifugal Concentrators, a New Era in Gravity Concentration, the Experience of CVRD Research Centre", pp. 1-6. 101 15. Simpson P., 2003, "The Knelson Continuous Variable Discharge Concentrator (KC-CVD) Application on Gold Sulphides", 25th Anniversary of Knelson internal meeting, Vancouver, Canada. 16. Statistics for Experimenters, An Introduction to Design, Data Analysis and Model Building. George E.P. Box, William G. Hunter, J. Stuart Hunter. John Wiley and Sons New York, Chichester, Brisbane, Toronto, Singapore. Copyright 1978. 17. Wyslouzil, H.E., 1990, "Evaluation of the Kelsey Centrifugal Jig at Rio Kemptville Tin' 22nd Annual Meeting of the Canadian Mineral Processors, Ottawa, Paper No 23, pp. 461 472. 102 13 Appendix 1 - Quartz/Magnetite Fractional Factorial Desig 103 13.1 Alias Structure A statistical experimental design was used in the quartz/magnetite test program. Below is the alias structure that was used for this testing. Alias structure for experimental design showing main, and two factor interaction effects. Interactions between three or more factors ignored. h = 1 h= 2 h= 3 1 4 = 4 l s = 5 1 6 = 6 1 7 = 7 1 8 = 8 ll2 = 1 - 2 + 3 - 7 + 4 - 8 + 5 - 6 ll3 = 1 - 3 + 2 - 7 + 4 - 6 + 5 - 8 ll4 = 1 - 4 + 2 - 8 + 3 - 6 + 5 - 7 lis = 1 - 5 + 2 - 6 + 3 - 8 + 4 - 7 ll6 = 1 - 6 + 2 - 5 + 3 - 4 + 7 - 8 ll7 = 1 - 7 + 2 - 3 + 6 - 8 + 4 - 5 ll8 = 1 - 8 + 2 - 4 + 3 - 5 + 6 - 7 104 13.2 Fractional Factorial Design Map Fractional Factorial Map Test Fluidization % Solids Feed Grade Heavies Particle Size Bowl Speed Pinch Open Pinch Closed Solid Feed Rate 1 1 1 1 1 1 1 1 1 2 -1 1 1 1 1 .1 .1 .1 3 1 -1 1 1 -1 1 -1 -1 4 -1 -1 1 1 -1 -1 1 1 5 1 1 -1 1 -1 -1 -1 1 6 -1 1 -1 1 .1 1 1 .1 7 1 -1 -1 1 1 -1 1 -1 8 -1 -1 -1 1 1 1 .1 1 9 1 1 1 .1 -1 .1 1 .1 10 -1 1 1 -1 -1 1 .1 1 11 1 -1 1 -1 1 -1 -1 1 12 -1 -1 1 -1 1 1 1 -1 13 1 1 -1 -1 1 1 -1 -1 14 -1 1 -1 -1 1 -1 1 1 15 1 -1 -1 -1 -1 1 1 1 16 -1 -1 -1 -1 -1 -1 -1 -1 Low variable levels are indicai ed by "-1" in the chart and high values are indicated by "1". 105 Operating Variable Levels for Factorial Design Variable High Low Center Point Heavies (%) 4.0 1.0 2.5 Fluidization (gpm) 14 5 10 Pinch Valve Open (s) 0.05 0.03 0.04 Pinch Valve Closed (s) 8 2 5 Bowl Speed (RPM) 925 725 825 Solids Feed Rate (tph) 2 1 1.5 % Solids 45 30 37 Heavies Particle Size (p80) (microns) 425 125 275 106 I 13.3 Factorial Design Test Results/Metallurgical Balance Factorial Design Test Results Size Fraction - p80 Pinch Valves Results Date Design # RPM Feed Rate (t/h) Percent Solids Mag (microns) Silica (microns) Open (s) Closed (s) Fluidization (g/min) Feed Grade (%) Con Grade (%) Tails Grade (%) Recovery (%) Mass Pull 11/8/99 1 925 2 45 425 425 0.05 8 14 3.5% 30.9% 2.9% 20.1% 2.3% 11/10/99 2 925 1 45 425 425 0.03 2 5 3.9% 55.0% 3.6% 8.7% 0.6% 11/9/99 3 725 1 30 425 425 0.05 2 14 4.3% 18.1% 0.9% 82.5% 19.3% 11/9/99 4 725 2 30 425 425 0.03 8 5 4.2% 69.9% 4.3% -2.0% -0.1% 11/11/99 5 725 2 45 425 425 0.03 2 14 1.0% 26.4% 0.7% 24.6% 0.9% 11/11/99 6 725 1 45 425 425 0.05 8 5 0.9% 11.4% 0.7% 25.5% 2.1% 11/11/99 7 925 1 30 425 425 0.03 8 14 0.7% 24.2% 0.6% 13.9% 0.4% 11/12/99 8 925 2 30 425 425 0.05 2 5 0.6% 3.7% 0.3% 63.3% 10.8% 11/15/99 9 725 1 45 125 425 0.03 8 14 4.0% 28.0% 3.0% 28.1% 4.0% 11/15/99 10 725 2 45 125 425 0.05 2 5 4.4% 20.5% 1.7% 66.8% 14.4% 11/14/99 11 925 2 30 125 425 0.03 2 14 4.0% 69.0% 2.2% 46.3% 2.7% 11/16/99 12 925 1 30 125 425 0.05 8 5 4.4% 31.4% 1.6% 68.2% 9.6% 11/12/99 13 925 1 45 125 425 0.05 2 14 1.1% 3.5% 0.3% 78.2% 24.4% 11/13/99 14 925 2 45 125 425 0.03 8 5 1.1% 30.9% 0.6% 48.8% 1.8% 11/13/99 15 725 2 30 125 425 0.05 8 14 1.0% 12.9% 0.5% 49.8% 3.9% 11/13/99 16 725 1 30 125 425 0.03 2 5 1.3% 27.1% 1.0% 25.5% 1.2% I 107 Grade 30.9% 55.0% 18.1% 69.9% 26.4% 11.4% 24.2% 3.7% 28.0% 20.5% 69.0% 31.4% Pinch Closed Feed Rate _ _ Pinch Open (s) Feed Rate _ _ rt rt Pinch Open (s) Pinch Closed rt rt _ H Bowl Speed (G's) Feed Rate — rt — — Bowl Speed (G's) Pinch Closed _ _ Bowl Speed (G's) Pinch Open _ rt — _ 1 Heavies Particle Size (p80 microns) Feed Rate rt rt Heavies Particle Size (p80 microns) Pinch Closed _ rt rt Heavies Particle Size (p80 microns) Pinch Open rt rt rt Heavies Particle Size (p80 microns) Bowl Speed rt rt Feed Grade (%) Feed Rate _ rt rt — Feed Grade (%) Pinch Closed rt rt — Feed Grade (%) Pinch Open rt — _ rt 1 Feed Grade (%) Bowl Speed >—H _ rt Feed Grade (%) Heavy Particle Size _ rt % Solids Feed Rate rt _ ^_ % Solids Pinch Closed _ _ % Solids Pinch Open rt rt rt % Solids Bowl Speed rt rt % Solids Heavy Particle Size rt _ % Solids Feed Grade rt _ 1 1 00 ~z Fluidization (g/min) Feed Rate rt _ rt rt — — n Fluidization (g/min) Pinch Closed — — • — NO n Fluidization (g/min) Pinch Open rt rt rt rt "x Fluidization (g/min) Bowl Speed rt rt i rt Fluidization (g/min) Heavies Particle Size _ _- rt rt n Fluidization (g/min) Feed Grade — — — n Fluidization (g/min) % Solids rt rt rt — 00 X Solid Feed Rate (t/h) rt rt rt _ X Pinch Closed (s) _ _ NO X Pinch Open (s) rt rt 'f, X Bowl Speed (G's) _ — _ — X Heavies Particle Size (p80 microns) C"l X Feed Grade (%) • • ( N X % Solids _ — rt rt _ H Fluidization (g/min) 1 rt rt rt Factorial Design Test Number i±! 4% a S 3% o §) 2% n S 1% 0% 1 2 3 4 5 6 Run Order • +50 • 50X 1001 A-100 114 14 Appendix 2 - HBM&S Gold, Zinc Incremental Trials 115 I One Variable at a Time Test Conditions Date Test ID Bowl Speed RPM Pinch Open (s) Pinch Closed (s) % solids Feed Rate (tph) Oct 04/00 62 500 0.18 4 11.55 0.491 Oct 04/00 63 600 0.18 4 11.55 0.491 Oct 04/00 61 700 0.18 4 11.55 0.491 Oct 04/00 64 800 0.18 4 11.55 0.491 Oct 04/00 65 900 0.18 4 11.55 0.491 Oct 04/00 66 1000 0.18 4 11.55 0.491 Oct 05/00 67 500 0.18 2 21.15 1.01 Oct 05/00 68 625 0.18 2 21.15 1.01 Oct 05/00 82 625 0.18 3 21.15 1.01 Oct 05/00 69 625 0.18 4 21.15 1.01 Oct 05/00 70 625 0.18 6 21.15 1.01 Oct 05/00 71 625 0.18 8 21.15 1.01 Oct 05/00 72 700 0.18 2 21.15 1.01 Oct 05/00 73 700 0.18 3 21.15 1.01 Oct 05/00 74 700 0.18 4 21.15 1.01 Oct 05/00 75 700 0.18 6 21.15 1.01 Oct 05/00 76 700 0.18 8 21.15 1.01 Oct 05/00 77 850 0.18 2 21.15 1.01 Oct 05/00 78 850 0.18 3 21.15 1.01 Oct 05/00 79 850 0.18 4 21.15 1.01 Oct 05/00 80 850 0.18 6 21.15 1.01 Oct 05/00 81 850 0.18 8 21.15 1.01 Oct 06/00 83 700 0.16 4 31.15 1.72 Oct 06/00 84 700 0.17 4 31.15 1.72 Oct 06/00 85 700 0.18 4 31.15 1.72 Oct 06/00 86 700 0.19 4 31.15 1.72 116 i Date Test ID Bowl Speed RPM Pinch Open (s) Pinch Closed (s) % solids Feed Rate (tph) Oct 06/00 87 700 0.2 4 31.15 1.72 Oct 06/00 88 850 0.14 4 31.15 1.72 Oct 06/00 89 850 0.15 4 31.15 1.72 Oct 06/00 90 850 0.16 4 31.15 1.72 Oct 06/00 91 850 0.17 4 31.15 1.72 Oct 06/00 92 850 0.18 4 31.15 1.72 Oct 10/00 93 625 0.18 4 38.9 2.04 Oct 10/00 94 625 0.19 4 38.9 2.04 Oct 10/00 95 625 0.2 4 38.9 2.04 Oct 10/00 96 625 0.21 4 38.9 2.04 Oct 10/00 97 625 0.22 4 38.9 2.04 Oct 10/00 98 700 0.18 4 38.9 2.04 117