A NOVEL AG-CRUSHER-HPGR CIRCUIT FOR HARD, WEATHERED ORES CONTAINING CLAYS

Transcription

1 A NOVEL AG-CRUSHER-HPGR CIRCUIT FOR HARD, WEATHERED ORES CONTAINING CLAYS *P.P. Rosario 1,2, R.A. Hall 2, M. Grundy 3, and B. Klein 2 1 Aura Minerals Inc Dunsmuir Street Vancouver, Canada V7Y 1K4 (*Corresponding author: prosario@auraminerals.com) 2 University of British Columbia Norman B. Keevil Inst. of Mining Eng Stores Road Vancouver, Canada V6T 1Z4 3 AMEC Americas Ltd. Mining & Metals Dunsmuir Street Vancouver, Canada V6B 5W3 1

2 A NOVEL AG-CRUSHER-HPGR CIRCUIT FOR HARD, WEATHERED ORES CONTAINING CLAYS ABSTRACT A research project was conducted at the University of British Columbia (UBC) resulting in the development and evaluation of a novel comminution circuit incorporating autogenous grinding and parallel trains of cone crushers and high-pressure grinding rolls (HPGR). A primary driver for the circuit was to treat ore from deposits with clay and variable hardness. This paper summarizes the unique pilot-plant test program that was developed as a basis for experimental simulation and lists the main findings of the research, including the circuit s potential operational cost savings compared to a conventional SAG circuit. KEYWORDS Comminution, autogenous grinding, crushing, SAG milling, HPGR, non-ferrous metallic ores INTRODUCTION Even though SAG-based comminution circuits are dominant in the industry, they do present some challenges for the treatment of several types of large orebodies. If the orebody contains significant hard ore, the SAG mill becomes extremely energy inefficient as its capacity is highly reduced (Morley and Staples, 2010). High hardness variability throughout the orebody produces significant SAG capacity variation and wide throughput fluctuations. Similar fluctuations occur when the SAG feed size distribution cannot be maintained relatively uniform through time (Morrell and Valery, 2001). Usually, the larger the low-grade deposit, the larger the variance in rock properties. For instance, large porphyry copper deposits (currently the largest source of copper) can present highly variable hardness; examples of such orebodies are Freeport-McMoRan s Chino Mine in New Mexico (Amelunxen et al, 2001) and Newmont s Batu Hijau operation in Indonesia (Burger et al, 2006). A research project was conducted at UBC using rock samples from a large copper-gold mining project. The project is a low-grade copper-gold porphyry deposit, requiring a high-tonnage operation to take advantage of economies of scale. Early testwork and economic pre-assessments had already recognized that a significant portion of this orebody contains high proportions of sericite (clays). Based on this fact, all the work conducted prior to this investigation has been based on a comminution circuit consisting of SAG mills, ball mills and pebble crushing (SABC). Conventional three-stage crushing circuits, with or without HPGR for tertiary crushing, were ruled out because of the high clay content of the orebody. As a result, no HPGR testwork had been conducted. A comprehensive body of grindability testwork has been conducted on ore from this property over the past few years. The data indicates that the two mineral zones contain a mixture of very hard rocks, softer material, and clays, in proportions that will vary throughout the life of the mine. The high variability of hardness implies that an SABC circuit will result in substantial fluctuations in tonnage and product size, unless a very effective ore blending program is in place. However, a considerable amount of material in the moderately soft to very soft classes, along with the presence of high portion of clays, suggest that a conventional crushing or crushing-hpgr circuit will present operational challenges. This combination of facts prompted the authors to develop a research project to investigate the applicability of an innovative HPGR-based flowsheet. 2

3 AG-CRUSHER-HPGR CIRCUIT A novel HPGR flowsheet was proposed to take advantage of the potential operating cost savings in processing hard rocks, to stabilize production, and to mitigate issues caused by the soft and clayish materials. The proposed flowsheet is shown in Figure 1; evaluation of its capabilities was the object of the research. Crusher Feed Bin Cone Crusher HPGR Trommel Screen Washing Screen Diverter Autogenous Mill/Scrubber Coarse Ore To Ball Mills Figure 1 - Proposed HPGR Flowsheet for Clayish Ore In this circuit, the primary-crushed ore is slurried in a low-power autogenous mill, where weaknesses in the fresh rock are immediately exploited and hard material is scrubbed. The autogenous mill is an unconventional design in that it has an overflow discharge. The mill product is screened in two stages. The first stage is a trommel screen, which removes the bulk of the fine product. The trommel screen oversize is then washed on a vibrating screen, and now free from fines, is conveyed to a cone crushing stage followed by an HPGR. An automatic bypass arrangement protects the HPGR when the cone crushers release coarse material to relieve jams. The HPGR product is recycled to the autogenous mill, where it is scrubbed, slurried and screened with the fresh feed. are: The potential advantages of this circuit over conventional SAG-based or crusher-based circuits the production rate is less sensitive to variations in ore hardness than a SAG-based circuit the autogenous mill deals with any clays and obviates possible issues with high-capacity, wet, fine screening competent HPGR cakes are de-agglomerated in the autogenous mill before screening dust generation is less than in conventional crushing circuits there is no SAG steel media consumption there is no potential for steel ball scats to report to the crusher (as in SAG-based circuits). The circuit was evaluated by simulations using pilot plant data obtained from tests run at UBC. A sample of ore from the deposit was sent to the mineral processing laboratory at UBC, and a detailed test program was developed to represent a full-scale circuit. These tests (including HPGR pilot-scale) provided inputs to a modelling program in which the circuit was compared to conventional SABC technology. In 3

4 parallel, a single HPGR test was conducted on the same ore for a preliminary evaluation of HPGR performance in a conventional tertiary-crushing application. Sample TESTWORK An ore sample for the test program was obtained from the mining company. The sample was a blend of material remaining from an extensive comminution characterization testwork program conducted on several HQ and PQ drill cores from the main mineral zone of the property. The mining company extracted two sub-samples from the composite, and had their contracted laboratory perform one SAG Mill Comminution (SMC) test and one MinnovEX SPI test. Testwork Design The testwork was designed to represent the full scale process as much as possible. As part of this, effort was made to generate a feed for the HPGR that approximated the feed that it would receive in the actual processing plant. The test procedure is summarized as follows: 1. Conduct a preliminary circuit simulation, using JKSimMet, to obtain a target screen analysis of the cone crusher product. 2. Tumble and screen the sample to scrub and remove the clays and very soft material, to simulate the effect of the autogenous mill in the full circuit. 3. Crush a portion of the tumbled sample in the laboratory crusher. 4. Run the synthesised sample through the HPGR, at three different pressing forces. 5. Repeat the procedure twice; at each repetition, before the tumbling stage, mix the product from the previous stage with a calculated quantity of the fresh ore sample, to simulate the recycle in the plant (approximately 85%). Following the HPGR tests, a series of tests was done to determine the effect, in terms of screening efficiency, of passing the HPGR product through the autogenous mill/scrubber. First, product cake from the closed-cycle tests was scrubbed in the tumbling mill, then washed on a screen with high-pressure water for one minute, and the screen efficiency determined. Second, a standard HPGR test on the unscrubbed sample, full feed, was conducted, following which the product, still unscrubbed, was washed on a screen with high-pressure water for one minute, and the screen efficiency determined. Third, a standard HPGR test on the full feed sample was conducted, then the product was scrubbed in the tumbling mill, followed by washing on a screen with high-pressure water for one minute, and the screen efficiency determined. Test Equipment and Circuit The HPGR at UBC was manufactured specifically for pilot plant work. The unit is fitted with 0.75 m diameter by 0.22 m width rollers and with Koeppern s patented Hexadur wear lining with profiled surface (hexagonal tiles with different heights). The machine is capable of crushing at specific pressing forces (F SP ) of up to 8.5 N/mm 2. A tumbling mill (60mm diameter x 1m long at 22 rpm), a vibrating screen (Sweco Vibro- Energy ZS40), laboratory-scale gyratory and cone crushers, a standard Bond ball mill, and dry and wet screen shaking apparatuses were also applied in the testwork program. Figure 2 shows the schematic of the lab-scale circuit for the tests. 4

5 FULL FEED CLOSED CIRCUIT (CYCLES 2 & 3) Open Circuit (Cycle 1) Sample Tumbling Mill Wet Vibratory Screen Pilot HPGR -6mm (~30%) ~70% ~14% Laboratory Pilot Gyratory HPGR Crusher Split ~56% Figure 2 - Lab-Scale Circuits Used for the Tests Testwork Results and Discussion Tumbling Test The tumbling tests were conducted wet at 70% solids by weight and provided mild autogenous grinding action. It was observed that very little size reduction occurs on material larger than 10 mm but there is a significant increase in the fine portion. Even though the full-scale circuit is designed with the application of a non-conventional AG mill with a deliberately low applied power, the mild breakage action achieved by this tumbling test is probably lower than what would be seen in the full-scale mill. This difference is acceptable for two reasons. First, the assessment of the full-scale AG mill performance is conducted through modelling and simulation based on material hardness characteristics (assessed by other tests) and a very distinct feed size distribution. Secondly, the main purpose of the laboratory tumbling test is to simply scrub off the clays as a first phase in the preparation of an adequate feed for the HPGR test. HPGR Feed Size In the full-scale circuit, the oversize material from the screen reports to the cone crushers and their product in turn feeds the HPGR. For the laboratory testwork, there was a concern regarding the crushing action provided by the small scale crushers, as the available laboratory crushers would produce too fine a product which would not represent what would happen at a full-scale operation. However, if no crushing was applied, the HPGR feed could also be misleading, as no fines would be present and the particle size distribution (PSD) would be notably truncated. To address this issue only 20% of the screen oversize material was crushed and blended with the uncrushed material prior to feeding the HPGR. Such a blend provided a PSD with the approximate shape of the full scale circuit (based on preliminary simulations) and with minimum size reduction. Figure 3 shows the final blend PSD as well as the original products and the full-scale simulated cone crusher product. 5

6 Particle Size Distributions Cumulative Passing [%] 100% 90% 80% 70% 60% 50% 40% 30% Pre-simulated HPGR feed 100% crushed material Non-crushed (100% screen O/S) Optimum blend (~1/5 crushed) 20% 10% 0% Particle Size [mm] Figure 3 - PSDs for the Optimum Blend, Original Products and Simulated Product HPGR Feed Moisture Content The moisture content in the HPGR feed is an important operational parameter as it affects the machine throughput as well as the energy consumption. In addition, feed moisture has an effect on the characteristics of the cakes produced. In conventional crusher-hpgr circuits, such as Cerro Verde and Boddington, the HPGR feed moisture content is driven directly by the fresh feed moisture and the moisture content of the recirculation stream (wet fine screen oversize). In the proposed AG-Crusher-HPGR circuit, the fresh feed moisture has no effect as all material reporting to the HPGR is first slurried in the AG mill and then screened through a combination of trommel and wet vibrating screens for high efficiency. The approach taken for this research was to estimate the full-scale HPGR feed moisture based on two main parameters. First, the tested material surface moisture, i.e. the amount of water that remains retained on the rock surface subsequent to tumbling and screening. Second, the full-scale vibrating screen oversize PSD based on preliminary simulations. Table 1 lists the main observed values for the test parameters along with the assumed and calculated values for the full-scale operation. Based on the 3.5% moisture content observed on the tests, it was calculated that the moisture content for full-scale operation would be approximately 1.7%. 6

7 Table 1: Summary of Parameters and Calculated Results for Moisture Content Test Full-scale Screen aperture (mm) Oversize P80 (mm) Oversize P50 (mm) Total surface area (dm 2 ) 164.4** 78.3** Solids weight (g) * Water weight (g) ** Liquid content (%) ** Surface moisture (g/dm 2 ) 1.92** 1.92* * assumed, ** calculated HPGR Tests In total, six pilot-hpgr tests were conducted during the period of this research. Table 2 summarizes the test results, and the schematic of the procedures is shown in Figure 2. Detailed test results and operational parameters were reported in a previous paper (Rosario et al, 2011). Table 2: Summary of main parameters and results for all HPGR pilot tests Test Nbr Circ. type Open Open Open Close Close Open Unit Pressure Tests - Cycle 1 Cycle 2 Cycle 3 Full Feed Spec. Pressing Force N/mm Speed m/s Roller gap mm Hydraulic Pressure bar Pressing Force kn Net Spec. Energy Con. kwh/t Spec. Thr. Cst. (m-dot) ts/hm Flake Thickness Average mm Feed Moisture % 1.60% 1.60% 1.60% 1.60% 1.20% 1.70% Particle Size Distribution 80% Passing Size mm % Passing Size mm Reduction Ratio F80/P Reduction Ratio F50/P To investigate the effects of closed-circuit operation on the HPGR s performance, pilot closedcircuit tests were performed by repeating the complete test until a good convergence was achieved. At each repetition, before the tumbling stage, the product from the previous stage was mixed with a calculated amount of fresh sample, to simulate the recirculation expected in the plant. After two iterations the results showed an acceptable level of convergence. The specific throughput constant showed very little variation 7

8 in the first repetition, Cycle 2, and a slight decrease in Cycle 3. A good convergence in feed and product size distributions was also observed. To assess the differences on HPGR performance between the proposed circuit and a conventional HPGR circuit, an additional HPGR test was performed (Test 7 full feed). This test was the type of test that is commonly used for conventional HPGR circuit assessment, i.e., the entire (full) feed is processed through the pilot-hpgr without any tumbling or scalping of fines or clays (a sketch of the test procedure is shown in Figure 2). The first observation during the test execution was the difference in the cakes produced. Visually, the product cakes from the full-feed test were large, approximately double the size of the cakes produced in the previous tests. Subsequently, when handling this product through the rotary splitter, the chute ahead of the vibrating feeder plugged a number of times, which gives an indication of the kind of problems that may occur in full scale production. The results of the full-feed test (Test 7) are compared to those from Test 3, which was also an open-circuit test performed at the F SP of 3.5 N/mm², but with the prepared feed. The full-feed test (Test 7) product PSD revealed a marked similarity to the product obtained from Test 3 (tumbled and screened) as shown in Figure 4 (along with feed PSDs). Interestingly, although the HPGR feed in Test 7 feed was considerably finer feed than in Test 3 (31.4% and 9.4% mm, respectively), the HPGR product size distribution is almost the same. The authors suspect that the observed higher performance, i.e. increased size reduction, for the Test 3 (tumbled and screened) occurred because its feed had most of the softer and very fine material scalped off and thus was comprised of almost entirely hard material. Similar observations have been recently reported by researchers that indicate a higher HPGR performance when treating a homogenous feed of hard material if compared to a blended feed containing the same hard material plus some other softer component(s) (Abouzeid and Fuerstenau, 2009; Benzer et al, 2010). 100% PSDs HPGR Open Circ. Tests: Standard & Tumbled-Screened Cumulative Passing [%] 90% 80% 70% 60% 50% 40% 30% 20% 10% Tumbled-Screened Feed Tumbled-Screened Product Standard Feed Standard Product 0% Particle Size [m m] Measured 68-32% Center-Edge - Calculated 90-10% C-E Figure 4 - Feed and Product PSDs for Full Feed and Tumbled-Screened Open-Circuit HPGR Tests 8

9 As shown in Table 2, the Net Specific Energy Consumption (E SP net ) result for Test 7 was 1.68kWh/t, an approximate 6% reduction when compared to Test 3 E SP net (1.79 kwh/t). In addition, there is a gain of approximately 15% in the value for Specific Throughput Constant from ts/hm 3 in Test 3 to ts/hm 3 in Test 7. These results appear to concur with the findings of other researchers (van der Meer and Gruendken, 2010; Morley, 2006) which advise against the use of HPGR for the treatment of truncated feeds. However, the full-feed case has 15% higher specific throughput, but the tumbled-and-screened case only includes 70% of the ore, because 30% (fine material) bypasses the circuit. Therefore the total plant throughput of the tumbled-and-screened case is actually 24% higher. Similarly, the 6% specific power reduction refers only to the feed to the HPGR; the specific power applied to the total fresh feed is actually less, even when taking into account the minimal extra power required for screening and partial crushing. The difference between these findings and those of the earlier research is that the feed to the HPGR is only partially truncated, because a portion of the feed to the HPGR was crushed using a laboratory crusher (as previously described in the HPGR Feed section). In a full scale operation, the entire feed to HPGR will be crushed upstream. This analysis indicates that the removal of a portion of fines from the feed proved to be beneficial. Therefore, in this case it seems that a partially truncated feed can be effectively processed by the HPGR. During the development of the AG-HPGR circuit, it was anticipated that, even with the tumbling and screening phases ahead of the HPGR, its product could still contain competent cakes due to the clayish nature of the ore, which prompted the concept of having the HPGR product re-circulated through the AG mill. To assess the properties of the cakes produced during the testwork, a wet-screening based test was developed and applied. Through these tests, it was possible to observe the lower cake-competence (screening efficiency equivalent) between HPGR products from the test with prepared feed and the ones with the full feed. In addition, the tests confirmed the benefit of a wet scrubbing phase in the HPGR circuit. The scrubbing increases screening efficiency and reduces moisture in the recycle stream, even following a wet processing step (Rosario, 2010). The HPGR effect in reducing the Bond ball mill work index was also assessed in this research and the methodology was based on the standard Bond ball mill test procedure. On one occasion, following the routine Bond test on the feed and product of one HPGR test (Test 6 with 19.3 and 17.4 kwh/t, respectively), the test on the product was repeated with a PSD artificially adjusted to that of the HPGR feed. This was done to investigate if the high amount of fines in such a product was affecting the work-index result. The new WI BM result for the product was virtually the same (17.5 kwh/t). It is suspected that because of the nature of the Bond test, (several cycles and makeup of the feed based on the production of fines in the product until test convergence), it is only marginally affected by the fines, and may turn out to be the proper test to investigate the particle weakening effect alone. As a summary, both the full and prepared feed HPGR products showed a sizeable reduction in the index (12.7% and 9.3% respectively). A higher WI BM for the screen oversize portion than for the feed was observed (19.3 and 16.6 kwh/t, respectively) which may indicate that the softer material does constitute the finer fractions of the sample, as one would expect. 9

10 ASSESSMENT OF THE AG-CRUSHER-HPGR CIRCUIT The authors had been involved in the assessment of a SAG-based comminution circuit originally proposed for the deposit and based on the testwork results and using a structured methodology (summarized in Figure 5; detailed in Rosario and Hall, 2010), the original SABC circuit and the proposed novel-hpgr circuit were evaluated and their benefits and disadvantages discussed. Figure 5 Summary of the Circuit Evaluation Methodology A JKSimMet model was developed for the proposed AG-Crusher-HPGR circuit. The main inputs were: data derived from results obtained through the SMC test (A=62.9, b=0.55, and ta=0.33 and specific comminution energy) Bond ball-mill work index (16.6 kwh/t) specific gravity tests (S.G.=2.70) HPGR pilot machine physical parameters(diameter, length, rpm of rolls) HPGR pilot machine operational parameters (e.g. speed, power) The results from the HPGR Test 7, Closed-Circuit Cycle 3 (m-dot, PSD of feed and product). The primary crusher product PSD (F80 of 125 mm) was estimated from survey data from different operations with similar hardness and also on the correlation with the t a parameter recommended by JKTech (Bailey et al, 2009). 10

11 During the prefeasibility assessment, the original project equipment (SABC circuit) was selected to provide a life-of-mine average throughput in the order of 180,000 t/d. Consequently, the AG-Crusher- HPGR circuit was designed and its model developed to provide a throughput of approximately 180,000 t/d for the sample provided by the mining company. The SABC circuit was simulated using the previously developed JKSimMet model, the same feed PSD, and the grindability characteristics of the sample utilized in the research (based on the SMC test and Bond ball-mill work index). This simulation estimated that the SABC circuit would deliver 139,000 t/d when fed with this ore. Several simulation iterations were performed for the refinement of the AG-Crusher-HPGR circuit. During this process, different values for full-scale machine/operational parameters, such as HPGR roll dimensions and speed, were tried. In addition, this fine-tuning process took into consideration equipment vendor information, and the provision for parallel systems to increase the availability of the circuit. Energy Requirements Ball Mill Energy The Ball Mill energy requirement for both circuits were based on Bond s third theory of comminution, the application of the phantom cyclone method (Napier-Munn et al, 1996), and the value of 16.6 kwh/t for the WI BM as per the test results conducted on the sample. Although a reduction on the WI BM was observed for the HPGR product, no discount was applied for the Ball Mill power requirement for the HPGR circuit. This decision may have added some conservatism to the design but seems appropriated due to the difficulty in precisely estimate the WI BM for the effective Ball Mill feed. As previously discussed, although a reduction in the HPGR product WI BM was observed, the HPGR feed is expected to have a higher WI BM than the fresh feed as per the screen oversize test results. For the required final product P80 of 200 microns, the calculation for the Ball Mill specific energy resulted in 7.78 and 8.21 KWh/t for the AG-HPGR-Crusher and the SABC circuits respectively. This difference is mainly due to a higher amount of fines in the Ball Mill feed for the HPGR circuit. As the SABC circuit was designed to cope with the hard and extremely hard ores that are part of the orebody, the SAG mill has to operate at a high ball load and speed which implies a coarser product. In addition, the SAG mill discharge screen was designed with a larger aperture (15.9 mm) than the one for AG circuit (12.7 mm). Pure Comminution Energy Based on the simulations results and the ball mill energy requirement calculations, the total energy applied by the comminution equipment was assessed. Table 3 shows the description of the selected machine sizes and energy simulation results for the comminution equipment for both circuits. 11

12 Description Table 3: Simulation Results Pure Comminution Energy Requirements Average Unit Power Consumption AG-Crusher-HPGR Option (2 lines) 4,000 t/h per line (simulated instantaneous) 176,640 92% Availability Comminution equipment Specific Energy Qt. Inst. (kw) Sim. (kw) Total (kw) kwh/t AG Mills m D x 6.1m EGL (34 x 20 ft) 2 8,000 7,411 13, SecondaryCone Crushers XL-1100 (8-6 oper. 2 stdby) , Tertiary HPGR - 2.4m D x 1.7m W 4 5,000 3,665 13, Ball Mill - 7.9m D x 12.8m L ( 26 x 42 ft ) 4 18,000 15,563 57, Totals 112,920 88, SABC Option (2 lines) 3,080 t/h per line (simulated instantaneous) 138,970 94% Availability Comminution equipment SAG Mills m D x 6.7m EGL (40 x 22 ft) 2 25,000 23,279 43, Pebble Crusher XL , Ball Mill - 7.9m D x 12.8m L ( 26 x 42 ft ) 4 18,000 12,636 47, Totals 125,279 92, Difference in Specific Energy 24.50% Even though the ore provided by the mining company for the research was harder than previously estimated for the life-of-mine average, the two circuits were simulated using the same ore parameters. Therefore any energy requirement comparison should be focused on specific energy values (last column of the table), as the installed equipment may be larger than necessary for the AG-Crusher-HPGR circuit. As shown in the table, the proposed HPGR circuit provides savings in pure comminution energy in the order of 24.5%. However, most HPGR circuits add a level of complexity, as more auxiliary equipment is usually required. Although, the proposed circuit targets simpler operation than conventional HPGR circuits, the determination of the real difference in the energy savings for the complete circuit involves a higher level of circuit detail. Complete Circuit Comminution Energy For the complete circuit energy requirement comparison, auxiliary equipment was sized and added to the HPGR flowsheet. This refined flowsheet was arranged in parallel to the development of a preliminary plant layout. Once the circuit was detailed, the estimation of the energy usage of the complete circuit was also performed. And Table 4 summarizes the findings and shows that the complete circuit energy savings are in the order of 22.7%. 12

13 Description Table 4: Energy Requirements for the Complete Circuits Connected Power Sim. Average Consumption Specific Energy kw kw kwh/t AG-Crusher-HPGR Option (2 lines) 4,000 t/h per line (simulated instantaneous) 176,640 92% Availability Comminution equipment 112,920 88, Conveyors and feeders 5,108 3, Pumps 12,080 8, Screens Comm. Eq. Lube and Cooling systems 1,537 1, Dust and metal collection requirements Heat & Ventilation Systems Totals 132, , SABC Option (2 lines) 3,080 t/h per line (simulated instantaneous) 138,970 94% Availability Comminution equipment 125,279 92, Conveyors and feeders 2,554 1, Pumps 11,968 8, Screens Comm. Eq. Lube and Cooling systems 1,476 1, Dust and metal collection requirements Heat & Ventilation Systems Totals 141, , Difference in Specific Energy 22.71% Operating and Capital Costs Operating Costs For the comparison of operating costs, it was assumed that the HPGR circuits and the SAG mill circuits would have similar steel consumption for liners (including HPGR roll surfaces and SAG mill liners, grates and pulp lifter bars) and ball mill media. In other words, it is assumed that the main difference in steel consumption lies in the SAG mill balls. For the SAG steel media consumption the same methodology as described in Rosario and Hall, 2010 was used and resulted in 480 grams of steel per processed ton. Assuming a cost of Can$ 1,000 per ton of SAG steel ball and 180,000 t/d, the operating cost savings in steel grinding media for the HPGR-circuit would be Can$ 86,400 per day. Furthermore, by assuming the unit power cost of Can$ 0.08 kwh, and having a specific energy difference of 4.1 kwh/t, it is estimated that the energy savings with the HPGR circuit are Can$ 59,000 per day. Thus the total estimated savings including steel grinding media is Can$ 145,400 per day. 13

14 Capital Costs For the purposes of this research, a rigorous materials take-off was not undertaken. To arrive at an order-of-magnitude capital cost for the AG-Crusher-HPGR circuit, it was assumed that the area (civil) cost is proportional to the installed equipment cost. The SABC circuit costs were generated during the original pre-feasibility study and are summarized in Table 5. This table also shows the summary of the calculated costs for AG-Crusher-HPGR circuit. Table 5: Capital Cost Summary Item Cost (kusd) Ratio SABC Circuit SAG Area Total Cost from Pre-feasibility Study 225,940 Pebble Crushing Area Total Cost from Pre-feasibility Study 54,480 Total SABC Area Cost 280,430 Installed Major Process Equipment (SABC) 123,840 Factor, SABC Area Cost: Process Equipment Cost 2.26 AG-Crusher-HPGR Circuit Installed Major Process Equipment 152,640 Factor (Area Cost: Process Equipment Cost) 2.26 Total AG-Crusher-HPGR Area Direct Cost 345,660 Difference (AG-Crusher-HPGR vs. SABC) 65,230 Indirects 43% 28,050 Contingency at 20% of Direct plus Indirect 18,660 Total Cost Difference AG-Crusher- HPGR > SABC 111,930 23% Based on the difference in the operating costs, the payout period for the additional capital cost for the HPGR-circuit is estimated in approximately 2.1 years. DISCUSSION Copper porphyry deposits are a major source of global copper production and, in several cases, present quite heterogeneous orebodies (in terms of rock hardness) which present challenges to SAG mill operation. Based on the promising results from this application of the novel circuit for heterogeneous, hard ore with high clay content, it is possible that this circuit, or one derived from it, may become an important alternative for SAG-based circuits for processing of copper porphyry deposits. This work has demonstrated that the change in energy savings using the novel circuit when additional equipment is added is much lower (from 24.5% to 22.7%) than those calculated when comparing SABC circuits against conventional crushing-hpgr circuits (Rosario and Hall, 2010). It is believed this is due to some unique characteristics of the proposed circuit, such as the circuit simplicity and the scalping of the softer components of the ore prior to the crushers-hpgr portion of the circuit. The circuit emphasises simplicity; compared to current HPGR-based circuit designs, the amount and complexity of ancillary equipment is reduced by the elimination of the nested closed circuits. The use of high-angle conveyors ( sandwich conveyor ) reduces overall circuit footprint and thus building requirements. In addition, the gain in efficiency is augmented by the fact that the crusher-hpgr portion of the circuit deals with the harder component of the feed only. In the proposed AG-Crusher-HPGR circuit design, a single AG mill per line is used. This mill has some non-conventional features. The AG is designed with unusual limited power to perform the 14

15 scrubbing action of conventional scrubbers and to provide limited breakage on the soft portions of the ore. This will be achieved with a higher applied power than conventional scrubbers provided by the higher diameter-length aspect ratio and the proportionally high internal load inherent to the AG. However, one potential risk of this design is the required high volumetric flow of slurry (flux) through the mill. To deal with this, an overflow-discharge type of mill is proposed. Another option would be a discharge mechanism with modern pulp lift design such as the kind proposed by Latchireddi in 2009, in combination with larger than usual grate openings (slots) positioned at the far circular periphery region of the mill. The sample utilized in this research was a carefully prepared composite for proper correlation with the average properties of the orebody. However, as a single sample was used for the evaluations, the results are quite specific for the corresponding grindability parameters of this sample. CONCLUSION This research has unveiled the potential for innovative HPGR-based circuits, and may serve as incentive for the future use of HPGR in other applications currently labelled as ideal for different technologies. ACKNOWLEDGEMENTS The authors would like to thank the (anonymous) mining company that supplied the sample, Aura Minerals Inc., AMEC Mining & Metals, the University of British Columbia, and BC Mining Research for supporting this presentation, cooperating with this research and sponsoring the investigation of the feasibility of the proposed HPGR circuit. REFERENCES Abouzeid A.-Z.M. and Fuerstenau D.W. (2009). Grinding of Mixtures in High-Pressure Grinding Rolls. Int. J. Miner. Process, vol. 93, pp Amelunxen P., Bennett C., Garretson P., and Mertig H. (2001). Use of Geostatistics to Generate an Orebody Hardness Dataset and to Quantify the Relationship between Sample Spacing and the Precision of the Throughput Predictions. SAG2001-SAG Mill Circuit. Int. Conf. on Autogenous and Semiautogenous Grinding Technology, vol. 4, pp Bailey C., Lane G., Morrell S., and Staples P. (2009). What Can Go Wrong in Comminution Circuit Design? Proceedings of the Tenth Mill Operators Conference, Adelaide, SA, pp Benzer H., Aydogan N.A., and Dundar H. (2010). Investigation of the Breakage of Hard and Soft Components under High Compression: HPGR Application. Minerals Engineering, Special Issue: Comminution 10, vol. 24, issues 3-4, Feb-Mar 2011, pp Burger, B., McCaffery K., McGaffin I., Jankovic A., Valery W., and La Rosa D. (2006) Batu Hijau Model for Throughput Forecast, Mining and Milling Optimization, and Expansion Studies, Advances in Comminution, eds. Kawatra, K.S., Soc. for Mining, Metallurgy, and Exploration, Inc. (SME), pp Latchireddi S. (2009). Improving Autogenous/Semi-Autogenous Grinding Performance and Energy Efficiency. Proceedings of the 10th Mill Operators' Conference, Adelaide, SA, pp Morley C. (2006). Flowsheets for HPGR. SAG2006-SAG Mill Circuit. Int. Conf. on Autogenous and Semiautogenous Grinding Technology, vol. 4, pp

16 Morley C. and Staples P. (2010). SAG or HPGR? The Current Dilemma. Proceedings of the 42 nd Annual Meeting of the Canadian Mineral Processors, January 2010, Ottawa, Canada Morrell, S. and Valery Jr. W. (2001). Influence of feed size on AG/SAG mill performance. SAG2001-SAG Mill Circuit. Int. Conf. on Autogenous and Semiautogenous Grinding Technology, vol. 1, pp Napier-Munn, T.J., Morrell, S., Morrison, R.D., Kojovic, T. (1996). Mineral Comminution Circuits Their Operation and Optimisation, Julius Kruttschnitt Mineral Research Centre, Monograph, vol. 2, The University of Queensland, Brisbane, Australia, pp Rosario, P.P. and Hall, R.A. (2010). A Structured Approach to the Evaluation of the Energy Requirements of HPGR and SAG Mill Circuits in Hard Ore Applications. The Journal of the Southern African Institute of Mining, vol. 110, pp Rosario, P.P. (2010). Comminution Circuit Design and Simulation for the Development of a Novel High Pressure Grinding Roll Circuit. PhD thesis, University of British Columbia, Vancouver, BC. Rosario, P.P., and Hall, R.A., Grundy M., Klein B. (2011). A Preliminary Investigation into the Feasibility of a Novel HPGR-based Circuit for Hard, Weathered Ores Containing Clayish Material. Minerals Engineering, Special Issue: Comminution 10, vol. 24, issues 3-4, Feb-Mar 2011, pp van der Meer, F.P. and Gruendken, A (2010). Flowsheet considerations for optimal use of high pressure grinding rolls. Minerals Engineering, vol. 23, issues 9, August 2010, pp