Reducing Grinding Energy and Cost -Magnetite Iron Ore Design Case ...

Reducing Grinding Energy and Cost - Magnetite Iron Ore Design Case Study A. Jankovic and W. Valery Metso Process Technology & Innovation, PO Box 1028, Eagle Farm Qld, 4009, Australia ABSTRACT:Efficiency of the comminution operations is traditionally assessed based on operating cost and energy consumption. Traditionally, the lowest operating cost was achieved by multi stage fully autogenous grinding due to elimination of steel grinding media costs which may represent up to 50% of the grinding cost.Significant reduction of the cost associated with grinding was achieved over the years by increasing the size and improving the design of the crushers and mills, however there was no major break-through in improving the energy efficiency of the comminution process. The principles of particle breakage in crushing and grinding equipment remained mainly unchanged over the years with energy efficiency reducing as the product size decreases. Only in the last 20 years the more energy efficient technologies were successfully implemented at industrial scale including high pressure grinding rolls (HPGR) for fine crushing and stirred milling for fine grinding. The capital and the operating cost play crucial roles in the design of a new processing plant as they govern the project economics. A theoretical design study for a for high capacity processing of a hard, fine grained silica rich magnetite ore, with several circuit options was carried out to assess the energy efficiency, operating cost and the project economics expressed through NPV. The CO2 emission was estimated and the carbon tax added in the operating cost. The results of the study confirm that application of more energy efficient autogenous grinding technologies offers significant benefits over the conventional grinding circuit options. INTRODUCTION Need for the reduction of energy consumption associated with the mining industry is ever greater. Mining and especially minerals processing routes for different ores (base metals, iron, alumina, platinum, etc) varies significantly and the energy requirement and the opportunities for the energy consumption reduction are different. Iron ore have a special place in the global mining industry judging by the volumes of ore processed and the energy usage. The majority of steel production is supported by iron ore sourced from high grade hematite deposits, although a significant fraction comes from the magnetite deposits. Compared to direct ship hematite ores mined from the upper regolith, magnetite deposits require significant beneficiation which typically involves grinding to a particle size where magnetite is liberated from its silicate matrix. Many banded iron formation deposits are very fine grained often requiring a final concentrate grind size P80 of 25-35 µm. The amount of energy required to produce a magnetite product suitable for sale as pellet plant feed from these deposits is order of magnitude higher than an equivalent direct ship lump (< 32 mm > 6 mm) and fines (< 6 mm) hematite project. The cost associated with high capacity processing of a hard, fine grained silica rich magnetite ore are presented in this paper, with the emphasis on comminution circuit options. The objective is to evaluate several options involving different grinding technologies in respect to energy consumption, operating as well as capital cost. Typical conceptual or scoping level assessment methodology used by the engineering companies was therefore applied. MAGNETITE ORE GRINDING Various magnetite ore grinding flowsheets have been implemented in the past, including:

71

• Conventional three (and four) stage crushing followed by primary and secondary

milling, • Primary crushing followed by wet SAG or AG milling and ball or pebble milling, • Air swept AG milling (for coarse grind).

Historically, the lowest operating cost for fine grained ores was achieved by multi stage fully autogenous grinding (Koivistoinen et al, 1989) with integrated magnetic separation steps between the stages. The major benefit of fully autogenous grinding is the elimination of steel grinding media costs and the need to discriminate between steel and magnetite in coarse magnetic separation ahead of pebble crushing. The separation step between grinding stages progressively reduces the amount of material to be ground and in many cases reduces the abrasive properties of the concentrate. Some of the best known magnetite companies using the autogenous milling are the subsidiaries of Cleveland-Cliffs Inc. in Northen America. The original autogenous milling circuit, consisting of an AG mill followed by cobber magnetic separation of pebbles, pebble milling of the magnetic concentrate, a finisher magnetic separation stage and silica flotation, was installed at Empire Mines in 1963 (Weiss, 1985). There have been three expansions since and, in the 1990s, Empire Mines had a total of 24 individual concentrating lines and a total plant capacity of 8 Mtpa of pellets. The target grind size of the circuit varies between the 90-95 per cent minus 500 mesh (32 µm) depending on the ore and operating conditions (Rajala et al., 2007). New Technologies Significant reduction of the cost associated with grinding was achieved over the years by increasing the size and improving the design of the crushers and mills, however there was no major break-through in improving the energy efficiency of the comminution process. The principles of particle breakage in crushing and grinding equipment remained mainly unchanged over the years with energy efficiency of the comminution process reducing as the product size decreases. Only in the last 20 years the more energy efficient technologies were successfully implemented at industrial scale including high pressure grinding rolls (HPGR) for fine crushing (Dunne, 2006) and stirred milling for fine grinding (Gao et al, 2003). Application of more efficient grinding technologies has provided opportunities to further reduce the operating costs associated with grinding. At Empire Mines a HPGR is installed for processing crushed pebbles and its introduction has resulted in a primary AG mill throughput increase in the order of 20 per cent (Dowling et al., 2001). Application of Vertimill® fine grinding technology at Hibbing Taconite Company enabled processing of lower grade ores and increased the concentrate production (Pforr, 2001). A sharp increase in application of HPGR and stirred mill technologies is recorded in the last decade driven by the benefits from energy efficiency and supported by improvements in equipment reliability. Potentials for the reduction of energy consumption in order of 30-45 per cent was suggested to be possible (Valery and Jankovic, 2002), although significantly lover reductions, 9-13 per cent, were reported after detailed engineering studies for two large copper projects (Seidel et al. 2006). This clearly indicate that benefits from new energy efficient technologies are case specific and intention of this paper is to show the potential for the magnetite ore processing. CASE STUDY An option study for a 10 Mtpa ore processing plant for a hard, fine grained silica rich magnetite ore was carried out, with the emphasis on comminution circuit options. The concentrator was assumed to be located within 100 km of a port suitable for facilitating equipment delivery. It was assumed that there were no restrictions on spatial layout and that the process facility would be built on ground of a sound geotechnical character. Any

72

subsequent differences in tailings disposal, water recovery, operation and cost were not considered. A set of ore comminution properties used as the basis for this hypotetical study is provided in Table 1. The magnetite concentrate weight recovery, SG, abrasion index Ai, iron and silica content were based on the following relationships: Concentrate weight recovery % = 10.737 ln(P80) - 3.0945 Concentrate weight Fe % = -8.4667 ln(P80) + 98.455 Concentrate SG = 0.84[(% Fe·5.18)/724) + (1 – (% Fe·3.0)/724)] Concentrate Ai = 0.05(% SiO20.4332) Concentrate silica content % SiO2 = 9.6966 ln(P80) – 29.571 Table 1. Ore Design Parameters Ore Grade

% FeT

32.2

Drop Weight Index (DWi) Ore SG

kWh/m3

11.1 3.40

Concentrate SG Bulk Density

t/m3

4.30 2.01

Bond ball mill work index (BBWi) Bond abrasion Index (BAi)

kWh/t

17.2 0.30

Bond rod mill work index (BRWi) Bond crushing work index (BCWi)

kWh/t kWh/t

17.7 20.6

MPa MPa

14.8 355

Point Load Index (PLI) Unconfined Compressive Strength (UCS) Fibrous Mineral Content

Nil

The fine grained nature of this hypothetical ore results in a relatively late release or liberation curve. This fundamental property of a magnetite ore is generally one of the major drivers of flowsheet design and therefore flowsheet option generation. Four circuit options were selected for comparison (McNab et al, 2009) with the following acronyms used to identify the primary unit process within each: COS – coarse ore stockpile; SC – secondary crush; HPGR – high pressure grinding roll; AGC – autogenous mill in closed circuit with cyclones and pebble crusher; RMS – rougher magnetic separation; CMS – cleaner magnetic separation; CMS2 – second cleaner magnetic separation; PM – pebble mill; PC – primary crusher; SM – stirred mill; TSF – tailings storage facility; Option 1. PC/AGC/RMS/PM/CMS Primary crushing – AG milling in closed circuit with hydrocyclones and pebble crushing – rougher magnetic separation – pebble milling – cleaner magnetic separation. Option 1 resembles the well known fully autogenous LKAB and Cleveland Cliffs style, low operating cost operations. Absence of steel grinding media is major basis for the low operating cost. Pebble mill control and pebble transport and handling requirements add complexity to the design and operation

73

Option 2. PC/AGC/RMS/BM/CMS/SM/CMS2 Primary crushing – AG milling in closed circuit with hydrocyclones and pebble crushing – rougher magnetic separation – ball milling – cleaner magnetic separation – tertiary milling using stirred mills – second cleaner magnetic separation. Option 2 has an additional grinding and magnetic separation stage compared to Option 1 and is considered to be simple for design and operation. Final milling stage is carried out using energy efficient stirred mills. Steel grinding media usage significantly increases the operating cost. Option 3. PC/C SC/C HPGR/RMS/BM/CMS1/SM/CMS2 Primary crushing – closed circuit secondary crushing – closed circuit HPGR – rougher magnetic separation - ball milling – first cleaner magnetic separation – tertiary milling using stirred mills – second cleaner magnetic separation. In Option 3 secondary crushing and HPGR effectively replace AG milling with pebble crushing. Application of HPGR, stirred milling and additional magnetic separation stage reduces the power requirements compared to Options 1 and 2. Option 4. PC/SC/O HPGR/PM1/RMS/PM2/CMS1/SM/CMS2 Primary crushing – secondary crushing – screening – Open HPGR – coarse pebble milling – rougher magnetic separation – fine pebble milling – first cleaner magnetic separation – tertiary milling using autogenous stirred mills – second cleaner magnetic separation. Option 4 is an attempt to design a circuit with the lowest operating cost through increased grinding energy efficiency using three stages of magnetic separation, traditional autogenous milling, HPGR and stirred milling technology. In this conceptual flowsheet steel grinding media is eliminated. Circuit complexity is partially reduced by open secondary crushing, HPGR grinding and stirred milling operation although recovery, storage and control of three separate sized media streams are introduced. ENERGY CONSUMPTION With the exception of the primary crushing module, which is consistent between options, estimates were developed for the total power drawn in the comminution, classification and magnetic separation areas of each circuit. Energy consumed by material transport machinery related to pumping between areas was not considered at this level of study. A summary of the resultant unit circuit energy for each Option is shown in Fig. 1. A significant circuit energy reduction is predicted with Options 3 and 4 which include HPGR and stirred milling. Some 33 per cent of additional energy separates the most energy efficient option (Option 4) from the least efficient, the two stage AGC Pebble circuit, Option 1. Note that part of the reduction is partially contributed to the additional separation step at coarse grind which reduces the amount of material for milling. According to Seidel et al. (2006), the basic comminution energy requirement for the Boddington HPGR circuit option was 14 per cent lower than the SAG option; however the overall energy requirement including conveying, screening, etc, was reduced to 9 per cent. The Boddington copper gold ore is of similar rock competency to that selected for this study and so provides a good contrast between comminution processes designed to liberate minerals for flotation, in which the whole ore is ground to fine size, and comminution process with the staged rejection of silicates. In the latter case energy consumption difference between flowsheet options can be significantly higher.

74

Figure 1. Energy Comparison. PROCESS OPERATING COST (OPEX) A fairly detailed approach was taken to the development of operating costs for each option. Consumption rates for power, wear and other consumables, labor and maintenance and materials were generated considering each process flowsheet from the COS reclaim feeders to either the final magnetic separator concentrate discharge or the magnetic separator tailings discharge. As such, no concentrate or tailings handling, filtration or storage costs were considered. For simplicity, some minor operating costs such as metallurgical testwork and analysis, which is considered common to all options, have been omitted. Unit costs for power, grinding media, wear consumables and labor were referenced from average values within the GRD Minproc database for similar sized and located projects. A factoring approach from direct capital cost was used to develop cost estimates for maintenance materials. Key assumptions are listed in Table 2. All costs are estimated in Australian dollars and are presented as 1st quarter 2009 costs. Table 2. Key Operating Cost Inputs Power Ball mill steel media Stirred mill steel media Labour on-cost Total HPGR cost

$/MWh $/t delivered $/t delivered % $/t of HPGR feed

120 1501 1814 50 0.35

The estimates as summarised below are judged to have an accuracy of ±35%. Unit cost breakdowns are presented and shown graphically in Fig. 2. Option 1 – 6.17 $/t, Option 2 – 6.42 $/t, Option 3 –6.66 $/t, Option 4 –5.38 $/t The most significant operating cost (OPEX) variables between options are those relating to power, media and liner consumption. The two options including AG mill circuits have

75

between 27-32 per cent higher power consumption cost relative to Option 4 which utilizes energy efficient autogenous grinding technologies. Grinding media and wear lining costs range between 0.41 $/t and 1.82 $/t. Option 3 has much higher media and wear lining cost as two ball mills of 8.8 MW installed power are required to grind 8 Mtpa of RMS concentrate from P80 2.3 mm to P80 75 µm. OPEX for Option 3 is the highest due to the high media and wear lining cost.

Figure 2. Operating Cost Comparison Carbon Tax A carbon tax is expected to be introduced in the near future and would add a significant cost to the operation. For this exercise a simplified estimate of the effect of carbon tax is considered. It was assumed that the Carbon Tax would be applied to total circuit energy and steel consumption relating to media and comminution equipment wear liners. The following criteria were applied for the carbon tax estimate: CO2 emission, 5 t per 1 t of steel media (Price et al, 2002) CO2 emission, 1.0 kg per kWh of electricity, CO2 tax, $23 per t of CO2 (Australian Government, 2008) Table 3 shows a summary of calculations related to carbon emission and carbon tax effect on OPEX. It can be observed that the introduction of carbon tax at 23 $/t would increase OPEX in the order of 9-11 per cent. The majority of carbon emission is from electrical energy consumption while indirect contribution from steel consumption (dominated by grinding media) is in the order of 5-16 per cent for Options 2 and 3 that utilise ball milling. Table 3. Carbon Emission and Carbon Tax Summary Power CO2 t/a Steel CO2 t/a CO2 Tax $/t Opex $/t (no CO2 tax) CO2 tax % Opex

Option 1 329,503 5,804 0.77 6.17 11.1

Option 2 315,768 18,256 0.77 6.42 10.7

76

Option 3 248,757 37,300 0.66 6.66 9.0

Option 4 238,328 8,306 0.57 5.38 9.3

CAPITAL COST (CAPEX) The scope of the estimates follows the Work Breakdown Structure developed specifically for the study and considers each flowsheet from the COS reclaim feeders to either the final magnetic separator concentrate discharge or the magnetic separator tailings discharge. The CAPEX estimate is developed based on the premise that the process is located inland in West Australia. All costs are estimated in Australian dollars and are presented as 1st quarter 2009 costs. They are judged to have an accuracy of ±35% which is commensurate with the accuracy requirements for a high level options study of this nature. The details of the cost estimate can be found in McNab et al, 2009. The total capital cost was as follows: Option 1 – $ 346.6M Option 2 – $ 356.9M Option 3 – $ 321.3M Option 4 – $ 312.6M Total estimated CAPEX for each circuit are within 14 per cent which does not infer any one option is a standout from a capital cost perspective at accuracy level for this study. In comparison, the Boddington copper gold project CAPEX (Seidel et al. 2006) for the HPGR circuit option was 7 per cent higher than the SAG option. Therefore, it appears that there may not be significant CAPEX “penalty” for adoption of more energy efficient grinding technology. FINANCIAL COMPARISON Applying a 10 per cent discount rate over 12 years of operation, high level, pre tax, net present value (NPV) determinations were calculated for Options 1 to 3 relative to the base case, Option 4, which returned the lowest capital and operating cost and therefore NPV. Options 1 and 3 have a similar NPV outcome ranging between negative $115-122 M relative to Option 4. Option 2 shows the least favorable outcome with a $149 M NPV deficit relative to Option 4. This option is disadvantaged by both high capital and operating cost. The conclusion drawn from this financial evaluation is that highly energy efficient autogenous processing routes can offer significant financial advantage for competent magnetite ores requiring fine grinding.

Figure 3. Project Net Present Value (NPV) Comparison.

77

CONCLUSIONS In this study it was found that highly energy efficient autogenous processing routes can offer significant benefits for fine grained competent magnetite ores. The traditional AG mill and pebble mill style comminution circuit or those requiring significant steel grinding media to operate have been found to be less effective from a pure economic perspective. Circuit options utilising multistage magnetic separation and with energy efficient autogenous comminution equipment, although more complex, are more likely to add project value. For the ore type evaluated, the application of HPGR and stirred mill technology is indicated to reduce energy consumption by up to 25 per cent compared to conventional flowsheets with wet tumbling mills. There are many other flowsheet selection drivers that can become relevant, however, operating cost associated with power draw and grinding media will always remain critical, even more so with expected introduction of carbon tax. A “synergy” of HPGR, pebble and stirred milling can result in a very effective circuit from a capital and operating point of view. It can be expected that highly energy efficient autogenous processing routes would be further developed and increasingly applied in practice. ACKNOWLEDGEMENTS The authors acknowledge the contribution of Brian McNab from AMEC Minproc. REFERENCES Clout, J M F, Trudu, A, Zhu, D, Holmes, R J, Young, J, 2004. Australian magnetite resources and pellet plants in Proceedings of 2004 Pelletizing Conference, Dalian, China, August 1922. David, D, 2007, The Importance of Geometallurgical Analysis in Plant Study, Design and Operational Phases’, 9th Mill Operator’s Conference 2007, Fremantle, Western Australia Dowling, E C, Corpi, P A, McIvor, R E, Rose, D J., 2001. Application of High Pressure Grinding Rolls in an Autogenous – Pebble Milling Circuit. in Proceedings of SAG 2001 Conference, Vancouver B.C., Vol III, pp 194-201. Gao, M, Laine, G, Schwartz, P, Holmes, R, Energy Efficent technologies for Fine and Ultrafine Grinding. AUSIMM Journal, July/August 2003, pp 36-40. Dunne, R, 2006. HPGR – the Journey from Soft to Competent and Abrasive. in Proceedings of SAG 2006 Conference, Vancouver B.C., Vol IV, pp 190-205. Koivistoinen, P., Virtanen, M., Eerola, P., Kalapudas, R., 1989. A Comminution Cost Comparison of Traditional Metalic Grinding, Semiautogenous Grinding (SAG) and Two Stage Autogenous Grinding. in Proceedings of SAG 1989 Conference, Vancouver B.C., L. Price, J. Sinton, E. Worrell, D. Phylipsen, X. Huc, J. Li., 2002. Energy use and carbon dioxide emissions from steel production in China. Energy 27 (2002) 429–446 McNab, B, Jankovic, A, David, D, Payne, P. (2009) Processing of Magnetite Iron Ores – Comparing Grinding Options. Proceedings of Iron Ore 2009 Conference, Perth, Australia, 27-29 July, pp. 277-288. Pforr, B, 2001. Fine screen oversize grinding at Hibbing Taconite Company, SME Annual Meeting, Denver, CO, February 2001 Rajala, G, Suardini, G, Walqui, H, 2007. Improving secondary grinding capacity at Empire Concentrator, SME Annual Meeting, Denver, CO, February 25-28. Seidel, J, Logan, T C, LeVier, K M, Veillette, G, 2006. Case study – investigation of HPGR suitability for two gold/copper prospects in Proceedings of SAG 2006 Conference, Vancouver B.C., Vol IV, pp 140-153 The Australian Goverment, 2008. Carbon Pollution Reduction Scheme: Australia's Low Pollution Future - White Paper. Volume 1, December Valery, W., Jankovic, A., 2002. The Future of Comminution, Proceedings of 34th International Octobar Conference on Mining and Metalurgy .October, pp 287-298

78