INTRODUCTION OF TURBO-LIFTER SYSTEM AT KAROWE DIAMOND MINE P. Morgan 1, L. van Niekerk 1 and M. Duddy 1, *G. Underwood 2, *A. Paz 3 1 DRA Global No.3 Inyanga Close Sunninghill, 2157, South Africa 2 Minopex 7 Pinewood Office Park 33 Riley Road, Woodmead, 2196 (*Corresponding author: gavin.underwood@minopex.co.za) 3 Outotec Australia 40 Kings Park Road 6005 WA West Perth, Australia (*Corresponding author: andres.paz@outotec.com) 1
INTRODUCTION OF TURBO-LIFTER SYSTEM AT KAROWE DIAMOND MINE ABSTRACT Boteti are due to begin treating the significantly more competent (Unit13) kimberlites at their Karowe Diamond plant, Botswana and wish to continue with AG milling as the heart of their process flow sheet. This synopsis looks at the current performance of the AG mill, and reports on the assessment of the impact of treating harder kimberlites in the near future. The introduction of a Turbo-Pulp Lifter (TPL TM ) system is anticipated to improve upon the difficulties experienced in the original mill grate/pan lifter operation, and reports upon the estimated impact of increased throughput and hard kimberlite reflected in the newly proposed and modified flow-sheet. KEYWORDS AG Milling, Diamonds, Grate Design INTRODUCTION Karowe Mine is the mine developed from Lucara Diamond Corp s AK6 Project and is 100% owned by Lucara s subsidiary Boteti Mining (Pty) Ltd. The Karowe Mine is located in north-central Botswana and is part of the Orapa/Letlhakane Kimberlite district, one of the world s most prolific diamond producing areas. The kimberlite at the Karowe Mine (the AK6 kimberlite ) comprises a single, tri-lobate kimberlite pipe, which is pinched at the surface and its sub-outcrop consists of a core of kimberlite, covering an area of 4.2ha, surrounded by an area where the kimberlite is capped by basalt or basalt breccia. Drilling has shown that the kimberlite bulges to a maximum area of 7ha at a depth of 120m. The majority of the ore body consists of a competent ore whose main component, identified as Unit 13, is unusually hard for a kimberlite. It is also very abrasive and displays other abnormal properties including high DMS yields, high crushing strength and low amenability to scrubbing. The autogenous high aspect ratio variable speed mill (Outotec 8.5mØ x 3.96m, 4MW) was required to treat various materials and kimberlite from the Karowe ore body, at a rate of 350mtph (in closed-circuit with a cone crusher) to provide a minus 30mm product with a waste fraction below 1.5mm (van Niekerk et al., 2013). The cut-off size was changed to 35mm due to the recovery of large, high quality diamonds. The various minerals encountered during treatment ranged from soft sandstones and weathered kimberlites, to more competent kimberlites, mudstones and basalts. A large degree of variability was therefore expected, due to the fact that there is a fivefold variation in overall power requirements. As a result of the high variability of the expected ore, a mill was selected as it has the flexibility to operate over the range of different types of material as well as combining crushing and scrubbing steps in one unit. It was anticipated that the mill would treat 300mtph of the hard Unit 13 (Phase 2); but Boteti wished to maintain the 350mtph name-plate throughout the mine life which would require some modifications to the circuit to accommodate the hardness and yield variations. It was also envisaged that during the initial stages of operation the mill will receive the bulk of weathered and softer materials (referred to as Unit 1) and will later treat more competent materials like Unit 13, together with mudstones and sandstones making up the major diluents. 2
Initially, rubber lined pebble ports of 70mm and 30mm grate slots, together inter-spaced with blank panels were proposed, allowing for a 5.5% to 10% open area range. As the mine development progressed into more competent areas, it was envisaged that the grate design would need further optimisation; and 90mm port designs were made up and later installed. As more competent material was being treated earlier than planned, principally fragmented kimberlites, the mill production rates reduced and the causes for the shortfall were specifically investigated during two milling plant sampling audits, referred to as S1 and S2. APPARENT HARDNESS AND RELATIVE GRINDING RATES The variability of the diamondiferous horizons to be encountered were originally tested and compared using Drop-Weight criteria, and later minor scrubbing and grind milling tests were used by DRA Global to compare relative grinding rates and for population balance modelling. The table below describes the observed peak specific rates of breakage (energy based) relative to the bulk sample (South Lobe) for the laboratory tested materials and breakage function φ values. Table 1 Observed relative grinding rates Type Relative Grinding Rates φ Unit 1(U1) 1.45 0.40 Unit 13(13) 0.25 0.25 Sandstone 1.12 1.00 Unit2456(FK) 0.37 0.40 Bulk (South Lobe) 1.00 0.80 SMCC and JK Tech DWT tests were conducted regularly during the mine ramp-up period and the total sample population history is shown in the graph below, with the published drop weight index (DWi, kw/m 3 ) database distribution superimposed (red thin line). It became apparent that the Unit2456 facies originally tested was akin to fragmented kimberlite (FK). The deeper material is apparently increasingly more competent. Figure 1 SMCC Test sample population data plot to date or Karowe 3
The graphic (Figure 1) clearly shows a significant degree of hardness variation that prescribes virtually the complete spectrum of DWi values. The abrasion factor, t a, is remarkably similar to the observed breakage function φ-values. Figure 2 Karowe sample t a population MILL GRATE PERFORMANCE Mill circuit sampling surveys were specifically undertaken whilst treating progressively harder horizon material, principally fragmented kimberlite, and the internal condition was also monitored by crash stopping the mill. The first survey (S1) was undertaken when the mill grate was severely worn and partly breached. The integrity of several rubber grates was severely compromised. A second survey (S2) was undertaken shortly after the damaged mill grates and worn liners were replaced following scheduled relining maintenance. The two surveys indicated that the grate ports were clogging up and more specifically, that the pan lifters were retaining material that should escape the mill. Simulating the circuit according to direct relative rate mixtures of known material facies enabled benchmarking of the milling-pebble crusher circuit model. Transfer functions were used to describe the pebble crusher operation, according size and to close-side setting. Previous data on Excel crushers had been used to develop a crusher transfer function model, and check survey performance (Hinde et al., 2001). Population balance modelling techniques made use of a grate function (modified Lynch-Rao partition coefficient equation) to describe hold-up load in the mill and thus influence the mill power development. Pc(x i ) = β + (1 β). exp α.(x i d50c ) exp α.(x i d50c ) +exp α 2 (1) Where d 50c - Port cut size, mm Pc(x i ) - Fraction of size retained by grate α - Sharpness coefficient β - Backflow fraction, % 4
Plant data from the surveys was fitted in terms of mill power, mill load, measured tonnage and product stream particle size distributions to ascertain the grate performance and apparent relative grinding rates (See Table 2). Table 2 Survey grate functions and perceived relative grinding rate Survey β (%) d 50c (mm) Rel. Rate φ DWi (kw/m 3 ) 1 12.3 64 0.565 0.52 3.99~4.68 2 14.0 87 0.420 0.43 5.71 1 1 Not direct measurement but best least square error fit. The data was simply fitted using combinations of Fragmented Kimberlite and South Lobe grinding parameters. Survey 2 mill feed actual drop weight index (taken at the end of the sampling period) was lower than the data fitted; but contamination of sample with softer kimberlite was suspected. The results clearly show that a significant fraction of finished product is retained by the mill grate-pan lifter system, and that partial clogging and bridging was apparent with the old grates and was borne-out by physical observation. Following the benchmarking exercise, it was apparent that the mill would in the future receive material significantly harder than seen in the surveys. The benchmarking revealed the following proposed relationship between DWi (kw/m 3 ) and observed relative grinding rates. Figure 3 Observed relationship between relative grinding rates and DWi Using this relationship it was possible to simulate the mill-crusher circuit whilst treating the harder drop-weight (DWi) referenced ores HARD KIMBERLITE SIMULATIONS It became apparent from initial design trial simulations that partial pre-crushing, modified grate design and bypass screening was necessary to control the mill load and alleviate any overloading mill condition. It was also apparent that less fines were going to be produced, potentially overloading the existing DMS circuit. High DMS yields are also anticipated on the harder ores, thus potentially 5
exacerbating the problem. XRT technology was incorporated into the circuit design for the Large Diamond Recovery (LDR) and DMS replacement as a consequence, whilst the existing DMS will process the fines. TURBO PULP LIFTER SYSTEM Outotec were approached with the intention of providing their innovative Turbo Pulp Lifter (TPL TM ) system to improve the discharge, and grate efficiency. The TPL TM has a curved pulp lifter and a side wall to minimise transport issues such as pebble carry over and flow back into the mill (Latchireddi, 2007). Outotec have used JK SimMet modelling tool to assess the proposed circuit taking into consideration changes to pebble port area, grate and pebble apertures, grate open area, pan depth and discharge angle (A. Paz, personal communication, April 2014). In consultation with DRA they have produced two TPL designs with 90mm and 115mm pebble ports, the first of which is for fragmented kimberlites and the latter due to be installed for the treatment of hard Unit13 mid-year (2015) and based upon feedback from the mine. Table 3 Post TPL grate function performance Survey β (%) d 50c (mm) Rel. Rate φ DWi (kw/m 3 ) 3 0~9 110 0.622 0.56 3.86 4 0~4 95.8 0.391 0.38 6.13 Survey 3 and much later 4 were conducted after the installation of the first combination set of TPL ports, and showed that after an initial period of stabilisation (Survey 3) of about a week, the cut size of the lifters settled down to 90~95mm. The sharpness coefficient, also improved from 4.75~5.3, to 8.0~12.5 indicating a much sharper separation. PROPOSED NEW CIRCUIT DESIGN Figure 4 illustrates the flow sheet for the proposed new circuit. 6
ROM FEED 1 JAW CRUSHER 8 2 5 4 MILL FEED PRE-CRUSHER 6 MILL FEED STOCKPILE PEBBLE BLEED SPLITTER 9 10 7-32mm PEBBLE CRUSHER LARGE DIAMOND RECOVERY (LDR) 14 ROM CONCENTRATE FEED +60mm +20x32mm 17 TERTIARY CRUSHER and DEWATERING EXISTING MILL +1.5x60mm 19 THICKENER ROM FEED FEED 15 +14x32mm +32x60mm 13 +8x60mm +1.5x8mm 11-1.5mm GRITS ROM FEED LARGE MIDDLES DIAMOND RECOVERY BULK SORTER (LDR) 16 +8x14mm LARGE FINES DIAMOND RECOVERY BULK SORTER (LDR) 12 FINES DMS -20mm ROM CONCENTRATE FEED 18 RECOVERY ROM FEED FEED 20 TAILS ROM FEED Figure 4 Proposed new flow sheet A 1300mmØ secondary gyratory crusher (KG 4513) will partly crush a proportion of jaw crusher product depending upon the coarseness, to prepare a suitably consistent feed to the AG mill. The mill discharge screen recycles plus 60mm directly and combined with the -60mm+32mm LDR tails to the existing pebble crusher. Pebble crusher product can be partially split to bypass mill recycle. XRT bulk sorting will be used to recover diamonds on -32mm+14mm, and14mm+8mm fractions, with a recycle tertiary crush on plus 20mm tail using a 1200mmØ Cybas Fine crusher. Minus 8mm plus 1.25mm fraction will be treated in the DMS. REFERENCES Van Niekerk, L.M., Ndlovu, G.N., & Sikwa, N.A. (2013, March). Commissioning and Operating an Autogenous Mill at Karowe Diamond Mine. Diamonds. Paper presented at 2013 SAIMM Conference, Muldersdrift, South Africa. Hinde, A., Wiseman, D., Hand, P., & Morgan, P. (2001, February). Using the World s Most Popular Software for Design and Optimisation of Crushing and Grinding Circuits. Paper presented at the 3 rd IRR Conference, Crushing and Grinding in Mining Conference, Johannesburg, South Africa. Latchireddi, S. (2007, January). Turbo Pulp Lifter (TPL TM ) An Efficient Discharger to Improve SAG Mill Performance. Paper presented at 39 th Annual Meeting of the Canadian Mineral Processors, Ottawa, Canada. 7