Energy Demand from Mine to Metal
Energy Demand from Mine to Metal
PAYNE INSTITUTE COMMENTARY SERIES: EXPLAINER
October 19, 2026
The energy cost of these metals depends less on the state of mining technology than on what is mined and how it is processed, because grade, mineralogy and processing route are the primary determinants. Energy demand falls into two streams, a fuel stream dominated by diesel haulage, which grows as pits deepen and has few near-term substitutes, and an electrical stream dominated by grinding mills and electrolytic cells downstream. This short paper considers several metals from the mine to their first saleable metal or chemical. The supply chain stages that consume most of the energy for aluminium, nickel and lithium lie beyond the mine gate.
The six commodities examined here each have reasonably good process-level energy data. Copper and gold show the effect of grade, iron ore the effect of mineralogy, nickel and lithium the effect of processing route, and aluminium the effect of stage, since nearly all of its energy is used after the ore leaves the mine.
A paper in Tribology International, calculated that mineral and rock mining together make up 6.2 percent of world total energy consumption, and estimated that friction absorbs about 40 percent of the energy used in mineral mining. Mining Energy Consumption 2021, estimate mining as 3.5 percent of global final energy. Azadi, Northey, Ali and Edraki drew a wider boundary in Nature Geoscience, estimating that greenhouse gas emissions from primary mineral and metal production, smelting and refining included, equalled about a tenth of global energy-related emissions in 2018. Holmberg’s group identifies grinding (32 percent of mineral mining energy) and haulage (24 percent) as the largest uses, and the Weir review, covering five of the six commodities examined here, finds that diesel burned in mobile equipment supplies 46 percent of final energy at an average site and comminution electricity 25 percent, with ventilation and other electrical loads making up the remainder. It also estimates that crushing and grinding may take up to 1 percent of global final energy.
In a survey of Australian gold and copper mines presented at the AusIMM Mill Operators’ Conference in 2012, Ballantyne, Powell and Tiang found that comminution used 52 percent of site electricity but only 21 percent of total reported energy, a gap that arises because diesel is reported at its calorific value whereas metered electricity excludes losses in generation and transmission. Adjusting electricity for those losses lifted comminution’s share to 36 percent, and the surveyed circuits accounted for 1.3 percent of Australian electricity use. The U.S. Department of Energy’s Mining Industry Energy Bandwidth Study applies a similar correction, multiplying onsite electricity by 3.17. The industry-funded Coalition for Energy Efficient Comminution puts the share at up to 3 percent once quarrying and cement grinding are included, according to CIM Magazine, and a 2012 International Mineral Processing Congress paper cites 3 to 4 percent from Daniel and Lewis-Gray (2011), but neither figure isolates mining.
Electricity and fuel at the mine site
Cochilco’s 2024 update of its annual energy survey, covering 49 mines, smelters and refineries that account for 98 percent of Chilean copper output, is the most complete process-level energy account published by any major mining country. It records total consumption of 199,452 terajoules in 2024, 4.1 percent more than in 2023, divided almost evenly between electricity (102,307 terajoules, or about 28.4 terawatt hours) and fuels (97,145 terajoules). The previous survey put the sector at about 14 percent of Chile’s energy use in 2023 and found that in 2022 it consumed the equivalent of 32.7 percent of national gross electricity generation and 21.1 percent of the country’s diesel, up from 11.5 percent in 2006. The two carriers serve almost entirely different stages, with open-pit mining taking 42 percent of all energy and 81 percent of all fuel, of which diesel made up 92 percent, while concentrators, where ore is crushed, ground and floated, used 56,970 terajoules of electricity (about 15.8 terawatt hours), or 56 percent of the sector’s power. Leaching, solvent extraction and electrowinning took about 16 percent.
Because Cochilco reports energy per tonne of ore as well as per tonne of copper, its data separate the effects of geology and equipment more cleanly than any other public dataset. Between 2001 and 2023 fuel use per tonne of copper rose from 8.2 to 18.1 gigajoules and electricity from 10.0 to 18.4 gigajoules, and by 2024 the combined figure of 36.2 gigajoules was roughly double its 2001 level. Over the same period the fuel burned per tonne of ore extracted from open pits rose 54 percent, from 44.2 to 68.1 megajoules, as pits deepened and haul roads lengthened, while the tonnage of ore moved from open pits rose 129 percent. Concentrator electricity per tonne of copper in concentrate more than doubled, from 6,112 to 13,073 megajoules, but electricity per tonne of ore processed rose only 10 percent and averaged 78.8 megajoules (about 21.9 kilowatt hours) over the past decade. Grinding a tonne of rock takes roughly as much energy as it did two decades ago, but many more tonnes must now be ground for each tonne of copper. Underground mines drew 35.6 megajoules of electricity per tonne extracted in 2024, about eight times the open-pit figure and 85 percent more than in 2010, which Cochilco attributes to compressed air, ventilation, pumping and other auxiliary services and to projects still under construction.
Desalination and the pumping of seawater to inland sites used 9,443 terajoules in 2023, 14.4 times the 2012 level, with pumping responsible for 83 percent of the load, and 11,308 terajoules in 2024, about 11 percent of sector electricity. Its projection for 2025 to 2034 has sector demand rising from 27.6 to 33.2 terawatt hours, with concentration reaching 18.1 terawatt hours (about 55 percent of the total) and seawater supply 5.4 terawatt hours, more than four-fifths of it for pumping. Over that period electricity demand grows 20.2 percent against 8.3 percent for copper output. Cochilco expects renewables to supply 98.6 percent of the sector’s electricity by 2030, and it puts copper mining at 33.8 percent of national electricity consumption in 2025.
The DOE bandwidth study shows a similar structure in the United States across coal, metal and industrial mineral mining. Built on 2000 production data and counting electricity at its primary-energy equivalent, it estimated sector consumption at 1,246 trillion Btu a year, with diesel supplying 34 percent, electricity 32 percent and natural gas 22 percent. Grinding used 494 trillion Btu, or 40 percent, more than twice the 211 trillion Btu for materials handling, which is dominated by diesel haulage, and ventilation came third at 122 trillion Btu. The study dates from 2007 and its totals are a generation old, but its process breakdown is still the most detailed published for the United States. Its two largest categories call for different remedies, since electric grinding can be decarbonized through the power supply, whereas diesel haulage requires changing the fleet itself through trolley assist, battery trucks or in-pit crushing and conveying.
Comminution as the dominant electrical load in processing
Ore must be broken finely enough for valuable mineral grains to separate from waste rock, a condition called liberation, which for non-ferrous ores usually means particles a few tens of microns across. An engineering review drawing on Norgate and Haque notes that lower-grade ores need still finer grinding, which cuts mill throughput and adds to the energy and emissions carried by each tonne of metal. A typical circuit passes run-of-mine ore through a primary crusher, a semi-autogenous grinding (SAG) mill and a ball mill before the slurry goes to flotation.
The DOE study cites 1 percent as the most common literature estimate of grinding efficiency and calculates a theoretical minimum of about 2 trillion Btu. Measurements on three SAG mills and four ball mills at the Goldex, Canadian Malartic and New Afton operations, published by the Canadian Institute of Mining, Metallurgy and Petroleum in 2019, found that only about 9 percent of the electrical energy supplied to the mills went into breaking rock, while roughly four-fifths was absorbed as heat by the slurry and most of the rest was lost in the drives. Hardness adds a further source of variation, with work indices, a laboratory measure of resistance to grinding, ranging in the DOE study from 1.43 kilowatt hours per ton for calcined clay to 134.5 for mica.
According to the Weir review, comminution takes 30 to 40 percent of total energy at copper mines regardless of configuration, between about 22 and 30 percent at gold mines, and about 27 percent at a hard-rock lithium site, where crushing and regrinding sit within a concentrator that uses about 60 percent of site energy. At sulfide nickel operations energy divides roughly equally between mining and processing, and at magnetite iron ore operations the grinding circuit alone draws about 33 kilowatt hours per tonne of ore. Ballantyne and Powell’s benchmarking of mines producing 24 percent of world copper and 15 percent of world gold produced the comminution energy curves now used to compare circuits. Norgate and Haque’s 2010 life cycle analysis in the Journal of Cleaner Production found that crushing and grinding generated 46.8 percent of the greenhouse gas emissions of copper concentrate production, whereas loading and hauling produced about half of the emissions for iron ore and bauxite, with explosives contributing 1 to 8 percent across the three ores.
At the Huckleberry copper mine in British Columbia, a comparative study reported in Engineering and Mining Journal found that high-pressure grinding rolls ahead of a ball mill would need 21 percent less energy than the existing SAG-based circuit to reach the mine’s 160 micron target grind. Uptake of such equipment has been slow, and Tim Napier-Munn argued in Minerals Engineering that the industry treats comminution efficiency as a low priority even though grinding is frequently a mine’s largest energy user and a large cost, and he proposed a coordinated ten-year effort by producers and researchers to reduce it. Energy is also a modest line in mine budgets, since electricity made up about 9 percent of operating costs, including depreciation, at large Chilean copper mines in 2018 and fuels about 5 percent, according to Cochilco’s Observatorio de Costos.
Ore type and energy intensity
Grade is the first of the three variables, and in the DOE dataset recoverable product averaged 82 percent of mined material for coal, 19 percent for iron, 0.16 percent for copper and 0.001 percent for gold and silver, so that the metals sector as a whole moved about 22 tonnes of material for each tonne of product. Mineralogy then determines the processing route, and routes differ by orders of magnitude in energy demand. Table 1 collects indicative intensities from the sources discussed in this section.
Table 1. Indicative energy intensity by ore type and processing route
| Ore type or route | Indicative energy intensity | Boundary | Main energy carrier | Source |
| Hematite iron ore, crushed and screened | Less than 0.15 GJ per tonne ore (mining and processing) | Site: mining and processing | Diesel (about 90%) | Weir review (2021) |
| Magnetite iron ore, processing | About 0.23 GJ per tonne shipped product; grinding about 33 kWh per tonne ore | Site: processing only | Electricity | Weir review (2021) |
| Copper concentrator, Chile | 78.8 MJ (about 21.9 kWh) per tonne ore, decade average | Site: concentrator only | Electricity | Cochilco (2024) |
| Copper, Chile, all stages | 36.2 GJ per tonne copper (2024) | Mine to cathode, including smelting and refining | Diesel and electricity, about half each | Cochilco (2025) |
| Copper, average across 400-site database | About 24 to 25 GJ per tonne copper | Mine site | Diesel and electricity | Weir review (2021) |
| Gold, underground / open pit | About 134,000 / 372,000 GJ per tonne gold (134 / 372 GJ per kg) | Mine site | Diesel and electricity | Weir review (2021) |
| Gold, non-refractory ore at 3.5 g/t (life cycle) | About 200,000 GJ per tonne gold | Cradle to gate, life cycle | Mixed | Norgate and Haque (2012) |
| Nickel from laterite (RKEF, HPAL) | About 230 to 570 GJ per tonne nickel | Ore to metal, processing routes | Carbon-based fuels and electricity | Manzoor et al., Nature (2025) |
| Alumina refining (Bayer process) | About 10.2 GJ per tonne alumina (2024 world average) | Refinery only | Heat, coal about 50% of fuel (2024) | ICSOBA (2024); IAI statistics (2026) |
| Primary aluminium smelting only, world average 2024 | 14.0 MWh (about 50 GJ) per tonne aluminium; Australia 14.49 MWh | Smelter only (electrolysis) | Electricity | IAI statistics (2026); Australian Aluminium Council |
Boundaries differ between sources (site, mine-to-cathode, or cradle-to-gate life cycle), so values show relative scale only.
In Chile, oxide copper ores are heap leached and recovered by solvent extraction and electrowinning, a route that needs no fine grinding and used about 12,956 megajoules of electricity per tonne of cathode in 2023, whereas sulfide ores pass through concentrators and then smelters and refineries. As oxide deposits have been worked out, concentrate’s share of national production rose from 67.5 to 74.6 percent between 2001 and 2023 and electrowon cathode output fell, a shift Cochilco identifies as a principal driver of rising electricity intensity. Because Cochilco’s survey includes smelters and refineries, its 36.2 gigajoules per tonne covers more stages than the Weir review’s mine-site average of about 24 to 25 gigajoules per tonne, which draws partly on a database of more than 400 copper operations.
In the Pilbara, hematite ores are typically crushed, screened and shipped, and the Weir review puts total energy use at hematite operations at less than 0.15 gigajoules per tonne, about 90 percent of it diesel for mobile equipment. Magnetite ores, such as the Minnesota taconites and the deposits at CITIC Pacific’s Sino Iron project in Western Australia, must be finely ground and magnetically separated before sale, and processing alone takes about 0.23 gigajoules per tonne of shipped product. Cleveland-Cliffs’ technical report summary for United Taconite, filed with the U.S. Securities and Exchange Commission, describes ore of about 23 percent magnetic iron ground until 76 to 86 percent passes 325 mesh, and an operating load of 75 megawatts at the Fairlane processing plant against 3.9 megawatts at the mine. Part of that grinding energy is recovered during pelletizing, because magnetite releases heat as it oxidizes in the induration furnace. The DOE’s profile of the iron mining industry estimates induration heat at 300,000 to 450,000 Btu per ton of pellets for magnetite concentrate after heat recuperation, compared with about 750,000 Btu for hematite. The oxidation releases about 498 kilojoules per kilogram of magnetite, according to Eriksson and colleagues in ISIJ International, and a 1993 Bureau of Mines study carried out with Cleveland-Cliffs found that the reaction can supply more than half of the heat input in commercial firing.
Because gold grades are measured in grams per tonne, energy per unit of gold exceeds that of the base metals by several orders of magnitude. The Weir review gives averages of about 134 gigajoules per kilogram at underground mines and 372 gigajoules per kilogram at open pits. Norgate and Haque’s 2012 life cycle assessment estimated about 200,000 gigajoules per tonne of gold from non-refractory ore grading 3.5 grams per tonne, roughly half as much again for refractory ores that require extra processing, with mining and comminution the largest contributors. At a Nevada operation that roasts double-refractory ore, a 2023 assessment in Science of the Total Environment assigned 34.8 percent of global warming impact to grinding and 56.1 percent to off-gas treatment.
Sulfide nickel ores can be upgraded by flotation before smelting, which discards most of the rock, but pyrometallurgical treatment of laterite melts the entire ore, as Benchmark Mineral Intelligence explains. Laterites contain about 60 percent of the world’s nickel, according to a 2025 paper in Nature by Manzoor and colleagues, which puts laterite processing energy at roughly 230 to 570 gigajoules per tonne of nickel. The same paper gives emissions of about 45 tonnes of carbon dioxide equivalent per tonne of nickel for rotary kiln electric furnace (RKEF) and blast furnace routes, 14 tonnes for high-pressure acid leaching (HPAL) and 6 tonnes for sulfide processing. A review in China Geology finds that RKEF, which handles about 80 percent of laterite nickel production, uses 2.18 times as much energy as HPAL. Indonesia, which holds the world’s largest nickel reserves according to the U.S. Geological Survey, has tied much of this demand to coal. Global Energy Monitor and the Centre for Research on Energy and Clean Air counted 19.3 gigawatts of operating captive coal capacity as of July 2025, with growth concentrated in nickel processing hubs, and about 31 gigawatts once projects under construction and in planning are added.
Aluminium makes the heaviest electrical demand of any metal in this study, yet almost none of it arises at the mine, where Norgate and Haque estimated that mining and processing bauxite takes only about 55 megajoules per tonne of ore. The Bayer process for refining bauxite into alumina is heat-intensive, with calcination above 1,000°C, as the Australian Aluminium Council describes, and the International Aluminium Institute puts the 2024 world average at 10,247 megajoules per tonne of alumina, down from 10,986 in 2021. Its fuel statistics record 1.24 million terajoules of reported refining energy in 2024, about half of it from coal, 32 percent from gas, 7 percent from oil and 8 percent from electricity, with China accounting for 63 percent of the total and meeting 65 percent of its own refining energy with coal. Smelting adds a far larger electrical load, with electrolysis using a world average of 14,004 kilowatt hours per tonne of aluminium in 2024, according to the Institute. With world primary output at 72.8 million tonnes in 2024 according to the U.S. Geological Survey, the Institute’s power statistics record 984,212 gigawatt hours of reported smelting power, or about 1,023 terawatt hours including its estimate of unreported consumption, roughly 36 times the electricity used by Chile’s copper sector that year. Coal generated about half of this power and hydro about a third, and Chinese smelters accounted for 59 percent of the reported total. Recycled metal needs a small fraction of that energy, 8.3 gigajoules per tonne against 186 gigajoules for primary aluminium from mine to cast house in 2019, according to the Institute.
Lithium brines in Chile’s Salar de Atacama are concentrated in solar evaporation ponds before chemical conversion, whereas spodumene concentrate is conventionally calcined at 1,000°C or more to convert alpha-spodumene to its beta form and then roasted with sulfuric acid at about 250°C, as Dessemond and colleagues describe in Minerals. Mowbray and colleagues reported in Science in 2026 a route that dissolves spodumene in aqueous ammonium fluoride at room temperature, without the calcination step. Using operating data from SQM, Kelly and colleagues at Argonne National Laboratory compared Chilean brine with modelled Western Australian spodumene in Resources, Conservation and Recycling and found that the brine route could produce up to 45 percent less greenhouse gas for some cathode chemistries, as Mining Magazine reported. A 2022 review by Chordia and colleagues found that environmental impacts increase as brine grade declines, while spodumene showed no consistent relationship with grade, a pattern the authors attribute to variation in plant design. Lappalainen and colleagues traced most of the impact of the conventional acid roasting route to heat, sulfuric acid and sodium hydroxide.
Declining grades
Calvo, Mudd, Valero and Valero, writing in Resources, found that the average grade of the copper mines they analysed fell by about a quarter between 2003 and 2013, and that over the same decade energy use at the Chilean mines in their sample rose 46 percent while production rose 30 percent. Cochilco’s 2024 update shows the pattern persisting, with the average grade of Chilean ore falling from 0.75 percent in 2010 to 0.62 percent in 2024, and the 2023 survey shows the same gradient across producers. Large private operations, which had the lowest average grade at 0.59 percent copper against 0.64 percent at Codelco and 0.85 percent at mid-sized private mines, included six operations above 50 terajoules per thousand tonnes of copper, a level Codelco’s average stays below.
Since grinding energy per tonne of a given rock is roughly fixed, halving the grade at constant recovery about doubles comminution electricity per tonne of metal. Norgate and Jahanshahi modelled this effect for copper and nickel across smelting, leaching and concentrating routes in Minerals Engineering and found it becomes significant below about 1 percent grade. For copper concentrated and smelted at a 75 micron grind, their estimates rise from 33 gigajoules per tonne of metal at 3 percent grade to 72 at 1 percent, 131 at 0.5 percent and 602 at 0.1 percent, mostly because of the extra waste rock that must be moved and treated.
Company disclosures
BHP’s FY2026 results presentation reports that 80 percent of electricity at its operated assets came from renewable sources, as evidenced by surrendered renewable energy certificates, with Chile and the BMA coal joint venture at 100 percent but Copper South Australia at only 35 percent. The same presentation gives operational emissions of 9.4 million tonnes of carbon dioxide equivalent for FY2026 and a restated 9.3 million tonnes for FY2025, a year BHP had originally reported at 8.7 million tonnes before switching to a national residual mix factor for Scope 2 emissions at its Australian assets, so reported progress depends heavily on how contracted renewable power is accounted for. On the fuel side BHP describes only trials of battery-electric haulage and treats diesel as a cost sensitivity, estimating that each US$10 per barrel change in the diesel benchmark shifts unit costs by about US$0.15 per tonne in Western Australian iron ore and about US$1.10 per tonne at BMA.
Codelco announced that from January 2026 more than 85 percent of its electricity would come from renewable sources under three 15-year contracts with Atlas, Colbún and Innergex totalling about 1.8 terawatt hours a year. Anglo American expected to draw 60 percent of its global electricity from renewables from 2025, once a ten-year agreement with Queensland’s Stanwell Corporation was added to its South American contracts. Freeport-McMoRan’s 2025 Form 10-K states that El Abra in Chile buys renewable power from Engie under a long-term contract and that Cerro Verde in Peru expects to move to fully renewable supply in 2026, while its U.S. mines rely on power purchase agreements, the open market and capacity rights at the Luna Energy Facility in New Mexico. Grasberg in Indonesia generates its own power from coal, and Freeport plans to replace that plant with a gas-fired combined-cycle facility supplied by a floating LNG storage and regasification unit, now scheduled to start in the second half of 2029 after the September 2025 mud rush led the company to defer the investment.
Rio Tinto’s 2025 Form 20-F reports gross Scope 1 and 2 emissions of 31.5 million tonnes of carbon dioxide equivalent on an adjusted equity basis, and its plans for the Boyne smelter in Queensland show that supply for a smelter depends on new generation actually being built. Wind and solar agreements signed in 2024, together with later arrangements with Edify Energy for the Smoky Creek and Guthrie’s Gap solar projects, could eventually meet around four-fifths of the smelter’s annual demand, but every contracted project was still in development. Fortescue is addressing the diesel stream directly through a network it is building in the Pilbara, which its April 2026 ASX release describes as 1.2 gigawatts of solar, more than 600 megawatts of wind and 4 to 5 gigawatt hours of battery storage. The mobile fleet is to move to renewable electricity over two to three years from the end of 2026, and the full grid is due for completion by the end of 2028. Operations on large grids in Chile, Peru and Queensland are cutting Scope 2 emissions through procurement, whereas diesel fleets and self-generating sites such as Grasberg and the Pilbara can change only as fast as investment in new equipment and generation allows.
Conclusions
Across the six commodities we considered, geology and processing affect energy demand far more than mining technology does, and as copper grades fall worldwide and supply shifts toward magnetite, laterite and spodumene, a growing share of the world’s metal will come from ores that need more grinding, more heat or both. Better equipment such as high-pressure grinding rolls can narrow the large gap between actual and theoretical grinding energy, but it does little to offset the extra rock that must be moved and ground as grades decline.
The two energy streams also differ in how quickly they can be decarbonized, because the electrical load of comminution, electrowinning and smelting sits in a small number of very large machines that can draw on renewable supply. Chile’s copper sector raised renewable supply to 78 percent of its electricity in 2024, according to Cochilco, and producers on grids in Peru, Brazil and Queensland are following through long-term contracts. Diesel haulage and self-generated power are harder to change, because they require new trucks, trolley lines, conveyors or dedicated generation, and their share of mine energy rises as pits deepen. The largest emissions sit further downstream, in the coal-fired alumina refineries, smelters and laterite nickel plants of China and Indonesia.
ABOUT THE AUTHORS
Macdonald Amoah
Macdonald is an independent researcher with interests across critical mineral supply chains, advanced manufacturing gaps, the defense industrial base, and the geopolitical risks of the mining sector.
Morgan Bazilian
Director, Payne Institute and Professor of Public Policy, Colorado School of Mines
Morgan Bazilian is the Director of the Payne Institute and a Professor of public policy at the Colorado School of Mines. Previously, he wD.as lead energy specialist at the World Bank. He has over two decades of experience in the energy sector and is regarded as a leading expert in international affairs, policy and investment. He is a Member of the Council on Foreign Relations.
ABOUT THE PAYNE INSTITUTE
The mission of the Payne Institute at Colorado School of Mines is to provide world-class scientific insights, helping to inform and shape public policy on earth resources, energy, and environment. The Institute was established with an endowment from Jim and Arlene Payne and seeks to link the strong scientific and engineering research and expertise at Mines with issues related to public policy and national security.
The Payne Institute Commentary Series offers independent insights and research on a wide range of topics related to energy, natural resources, and environmental policy. The series accommodates three categories namely: Viewpoints, Essays, and Working Papers.
Visit us at www.payneinstitute.mines.edu
FOLLOW US
DISCLAIMER: The opinions, beliefs, and viewpoints expressed in this article are solely those of the author and do not reflect the opinions, beliefs, viewpoints, or official policies of the Payne Institute or the Colorado School of Mines.