Iron Ore Mining
From the Pit to the Ship's Hold

The overwhelming majority of global iron ore reserves come from Banded Iron Formations, BIF, deposited during the Archean to Paleoproterozoic eras, 2.5 billion to 1.8 billion years ago. At the time Earth's atmosphere had almost no free oxygen. Large quantities of dissolved ferrous iron sat in the oceans. Oxygen released by early cyanobacteria through photosynthesis oxidized the ferrous iron into ferric iron, which precipitated out and alternated with siliceous layers to form bedded ore bodies.

After 1.8 billion years ago, atmospheric oxygen levels rose, dissolved iron in the oceans was gone, and this type of deposition stopped. The Earth will not produce a second batch.

This geological fact has one direct implication for understanding the industry: in BIF deposits, the supergene enrichment zone where grades were upgraded from the original 25%-35% Fe to 60%-67% Fe through hundreds of millions of years of weathering and leaching has limited thickness. In the Pilbara most mines have enrichment zones between 50 and 150 metres. Rio Tinto's 2023 annual report disclosed that average mined grade across the Pilbara has slipped from around 63% Fe a decade ago to around 62% and change. Looks like less than one percentage point. The effect on a blast furnace is not linear. Each 1% drop in grade means a relative increase in gangue content in the furnace feed, slag volume goes up, coke rate follows, carbon emissions follow. 62% and 63% ore are not the same thing to a steelmaker.

The four main iron-bearing minerals used in industry are hematite, magnetite, goethite, and siderite. Hematite dominates the export ores of Australia and Brazil, DSO, Direct Shipping Ore, 58% Fe and above, crushed and screened and loaded onto ships. Magnetite is the main ore type in China, Russia, and Sweden, grades typically 15% to 35% Fe, has to be ground very fine to separate the iron minerals from quartz gangue. China's import dependency exceeding 80% is not because China lacks iron ore. It is because the vast majority of China's iron ore is this kind of low-grade magnetite, and the mining and processing costs are far higher than Australian and Brazilian DSO.

The vanadium-titanium magnetite deposit at Panzhihua is a different situation. The ore contains around 10% TiO₂. When this titanium enters the blast furnace it forms titanium carbonitride, which has a very high melting point and severely degrades slag fluidity. Pangang's blast furnace utilization coefficient has been persistently below the industry average. Smelting this type of ore requires operating practices that diverge significantly from other Chinese steelmakers, almost a standalone technical system.

About 95% of global iron ore production comes from open-pit mines. Underground mining is too expensive. Iron ore is not gold. The per-tonne price cannot support that cost. LKAB's underground sublevel caving operation at Kiruna in Sweden survives because 67% grade magnetite concentrate gets turned into DR-grade pellets in LKAB's own pellet plants, and the pellet premium covers the extra cost of underground mining. Globally this is a special case.

The on-paper logic of open-pit mining is straightforward. Stripping ratio times unit waste-moving cost plus ore mining and hauling cost equals mining cost. Pilbara stripping ratios run 1:1 to 2:1 or so, rising as the pit deepens. The problem is this on-paper logic only explains the broad cost framework. At the operational level, marginal swings in profit are often controlled by factors that do not look very "strategic."

Tyres are one of them. A 793F or 930E haul truck tyre is about four metres tall, weighs a bit over five tonnes, costs roughly sixty thousand US dollars each. Lifespan depends heavily on haul road conditions. Well-maintained roads get over ten thousand hours, poor roads mean replacement at eight thousand. Only three companies in the world make tyres this size: Bridgestone, Michelin, and Goodyear. If they want to add capacity, a new production line takes three to four years to build. During the 2021-2022 global shortage of mining tyres, some mines could not get enough and had to throttle truck speeds or park some of their fleet. A large mine's annual tyre procurement budget easily exceeds thirty million dollars. In brokerage industry reports this money typically gets lumped into "other operating costs" and passed over in a sentence.

Blast design is another one. Blasting in an open-pit iron ore mine is not about smashing rock. It is about achieving a specific size distribution. Fragments too large and the downstream crusher jams, liner wear accelerates, unplanned shutdowns increase. Fragments too fine and the proportion of fines in the product goes up, lump yield goes down. Lump ore goes straight into the blast furnace and sells at a premium of three to five dollars per tonne over fines. At Rio Tinto's Pilbara shipment rate of about 330 million tonnes per year, a one percentage point swing in lump yield translates to ten to fifteen million dollars annually. Adjustments a blast engineer makes to hole pattern design can produce measurable effects on that number.

Truck dispatching is a topic that could fill its own article. One point: a haul truck returning empty to the loading face burns nearly as much diesel and wears nearly as much rubber as it does loaded, while generating zero revenue. In a mine running multiple pits simultaneously, a dispatch algorithm that reduces empty haul distance by 5% translates to tens of millions of dollars in annual savings. The autonomous haulage systems deployed by Rio Tinto and BHP in the Pilbara, their most important benefit is not saving driver wages (that portion is actually limited). It is that autonomous systems execute the dispatch algorithm's optimal solution on every single trip, 24 hours a day. No driver route preferences, no shortcutting across poorly maintained roads, no idle time during shift changes. The utilization improvement from autonomy runs 15% to 20%, and the bulk of that gain comes from eliminating behavioural randomness.

DSO does not need beneficiation. Crush it, screen it, done. The complexity is in magnetite.

The core step in magnetite processing is grinding. Iron minerals and gangue, mostly quartz, are interlocked very tightly in magnetite ore. Without grinding to sufficient fineness, magnetic separation cannot achieve clean separation. Most cases require grinding to 80% passing 75 microns, some ores need 45 microns or finer. The SAG mill plus ball mill grinding circuit accounts for 50% to 70% of the entire processing plant's electricity consumption. Magnetite projects are acutely sensitive to power prices. This is why those magnetite projects in Western Australia's Mid-West region have been stalled for over a decade. The upfront investment required for power infrastructure is too large.

Pelletizing is another layer of processing. Concentrate powder plus bentonite, balled on a disc, fired at 1200 to 1300 degrees Celsius. Vale has pellet plant capacity of over 35 million tonnes per year across Brazil and Oman combined. Pellets have one critically important physical property that does not get much outside attention: compressive strength. The sea voyage from Brazil to China takes about 40 days. Pellets at the bottom of the ship's hold bear the weight of over ten metres of ore stacked above them. If the pellets are not strong enough they crumble into powder by the time the ship reaches port. The pellet premium is gone, and unloading becomes a mess. A significant part of Vale's pellet R&D effort, recipe adjustments, firing regime optimization, is organized around the objective of making sure the pellets are still pellets after 40 days on the ocean being compressed.

A couple more words on sinter. Sinter accounts for over 70% of blast furnace burden in China. The sintering process is the second largest emission source in a steel plant after the blast furnace itself, dust, SO₂, dioxins. Multiple Chinese cities have been restricting sinter capacity. If this policy direction continues, demand for fines will be compressed and demand for pellets will rise. The relative pricing structure across different iron ore products will be rearranged.

Line up every iron ore mine in the world by CFR China cash cost from low to high, and you get the industry cost curve. The most widely used tool for understanding the competitive landscape.

The left end, about 15 to 25 dollars per tonne CFR, is Rio Tinto, BHP, FMG's core Pilbara operations, plus Vale's Carajás S11D. S11D ore body is virtually at the surface, no blasting required, mobile crushers and conveyor belts replace haul trucks. The middle section, 25 to 50 dollars, is Vale's Southeastern system, FMG's lower grade product lines, Kumba in South Africa, Indian export ore. The right end, 50 to 90 dollars or higher, is Chinese domestic mines. China's raw ore output looks big at 9 to 10 billion tonnes per year, but converted to 62% Fe equivalent it is only about 2.5 to 3 billion tonnes. These are the marginal supply. When prices drop below 60 dollars, Chinese mines shut down in large numbers. When prices recover above 80, they restart. This on-off elasticity is the most sensitive short-term price regulator.

The cost curve only shows cash operating cost. It does not include sustaining capital expenditure. Pilbara pits typically last 15 to 25 years. To maintain annual output of over 300 million tonnes, Rio Tinto spends over 2.5 billion dollars per year on replacement mine development and infrastructure maintenance in the Pilbara alone. Spread across tonnes shipped, this adds 15 to 20 dollars per tonne. Once you add this in, the flat left end of the curve is not so flat.

The other issue is that the cost curve is a point-in-time snapshot that does not reflect change over time. Pits are getting deeper, haul distances are getting longer, grades are gradually declining, environmental compliance is getting more expensive. In ten years this curve will have shifted upward across the board. The rate of upward shift is not uniform. Mines on the left end, because their highest quality ore bodies are being depleted faster and replacement ore bodies typically have less favourable conditions, may see their costs rise no slower than mines in the middle segment.

1966 to 2010, annual benchmark pricing. Miners and Japanese, Korean, and European steelmakers negotiated an annual contract price each year. From 2003 onward Chinese crude steel production exploded. Annual negotiations became increasingly unresolvable. In 2010 Vale switched to quarterly pricing and the entire annual benchmark system fell apart.

Current pricing centres on the Platts IODEX 62% Fe CFR China daily index. The Dalian Commodity Exchange launched iron ore futures in 2013 and opened to international participants in 2018. SGX iron ore swaps are also an active derivatives market.

This system has a structural characteristic that needs to be understood. Nominal global seaborne iron ore trade volume is about 1.6 billion tonnes per year. Long-term contracts between the four major miners and large steelmakers lock up most of that volume. What actually circulates and trades on the spot market is probably only 20% to 30% of total seaborne volume. The Platts index is compiled based on that 20% to 30%. Small pool. A few cargoes changing hands can push prices by a large amount. This is why iron ore price volatility runs high among commodities, with intra-year swings frequently exceeding 40%. The price spike after the 2019 Brumadinho disaster and the crash in the second half of 2021 when China imposed production curbs both moved far beyond what the underlying fundamental variables alone could explain. Thin liquidity amplifies every marginal shift in supply and demand.

Grade is only one dimension. Alumina, phosphorus, sulphur, silica each carry their own pricing weight. Each 1% increase in Al₂O₃ adds roughly 20 kilograms per tonne of hot metal to blast furnace slag volume, and pushes coke rate up by 8 to 12 kilograms. Phosphorus is more troublesome. Once it enters the hot metal it stays almost entirely in the steel, causing severe cold shortness. Steelmakers producing automotive outer panels and pipeline steel have phosphorus tolerance precision at the 0.01% level. Exceed it and the product is either downgraded or rejected. Two cargoes both labelled 62% Fe, differing in Al₂O₃ and P content, can be worth 5 to 10 dollars per tonne apart to the same steelmaker. The vast majority of investor-facing analysis compresses iron ore quality into a single grade number, which is forcing a multi-dimensional problem into one dimension.

The blending trade at Chinese ports is another layer of market structure, operating outside the miners' brand system. On the stockyards at Rizhao, Qingdao, and Caofeidian, traders blend ores from different origins and different compositions according to recipes, producing mixed ores that meet the specific chemical specifications required by individual steelmakers' blast furnaces, and sell directly to them. This blending trade accounts for roughly 15% to 20% of China's imports. When inventory of a particular brand at port drops low, PB Fines or Newman Fines for example, traders are forced to adjust their blending recipes and may pile into purchasing substitute brands, generating short-term price jumps that have nothing to do with the macro supply-demand picture. Without understanding this micro-level market structure, short-term iron ore price movements look inexplicable.

Vale, Rio Tinto, BHP, FMG. Combined they control about 70% of seaborne volume.

Rio Tinto has 16 mines in the Pilbara connected by approximately 1,700 kilometres of proprietary railway to two port complexes. The railway already runs fully autonomous trains. The core output logic of this system is not "16 mines each mining and selling their own stuff." It is taking ore of different chemical compositions from 16 different mines and blending them at precise ratios to produce branded products with highly consistent chemical specifications. Pilbara Blend is the main one. For a blast furnace, consistent feed composition matters more than high grade. When composition fluctuates, furnace conditions become unstable, coke rate rises, and in serious cases you get hanging or channelling. What Pilbara Blend sells is not grade. It is predictability.

That predictability is becoming harder to maintain. Old pits have declining grades. New replacement ore bodies do not have chemical fingerprints that perfectly match the old ones. To keep the final blended product within spec, the blending calculations and ore logistics scheduling have become more complex. Rio Tinto's increased investment in ore tracking and blend optimization systems over recent years is a response to this.

FMG was founded by Andrew Forrest in 2003. By then the good ground in the Pilbara had already been taken by Rio Tinto and BHP. The ore bodies FMG got were lower grade, 56% to 58% Fe. Forrest's bet was that Chinese blast furnace operators had the ability and the motivation to absorb low-grade ore. That bet was right. Chinese steelmakers have the highest blending flexibility in the world. They can pair low-grade ore with high-grade ore and push raw material costs down as far as metallurgical performance allows. FMG went from zero to over 190 million tonnes per year in under twenty years. Its success was not resource-driven. It was driven by a market judgment.

Vale sits on Carajás, the highest-grade large-scale iron ore province on the planet, 65% and above. The sea voyage from Brazil to China, about 40 days, against about 12 from the Pilbara, is a structural burden. This is not just a matter of a few extra dollars per tonne in freight. At equivalent shipment volumes, Vale has nearly one extra month of inventory on the water. At roughly 300 million tonnes per year shipped and an average price of 100 dollars per tonne, ballpark working capital tied up in transit is on the order of 25 billion dollars. This money does not show up in cash operating costs. It does not show up on the cost curve. It consumes return on capital. In a fast-moving price environment, this time lag also means Vale's supply chain is inherently one beat slower in its response.

The blast furnace-BOF route consumes about 72% of global iron ore. In China the proportion is about 90%. Producing one tonne of pig iron consumes roughly 1.5 to 1.7 tonnes of iron ore at 62% grade equivalent and 0.3 to 0.5 tonnes of coke.

The electric arc furnace route consumes scrap steel. About 28% of global crude steel goes through EAFs. In the US and Turkey the share exceeds 70%. In China it is about 10%. EAF expansion will erode iron ore demand over time. This is correct, but the pace needs a discount. Scrap is a resource with a time delay built in. A country's current scrap accumulation depends on its steel consumption volume 20 to 30 years earlier. Large-scale steel consumption in China only started in the early 2000s. The window for bulk scrap to flow back opens gradually between 2025 and 2035. And the quality of recovered scrap varies widely. Scrap from building demolition carries high levels of copper and other tramp elements, only good for low-end rebar. Scrap from automobiles and household appliances is better quality and can make flat products, but the collection, dismantling, and sorting infrastructure and precision in China are not there yet. The constraint on quantity and the constraint on quality together mean that the pace of EAF displacing the blast furnace will be slower than what you get by drawing a trend line and extrapolating.

Hydrogen-based direct reduction, H₂-DRI, using green hydrogen instead of natural gas as the reductant. Sweden's HYBRIT and Germany's SALCOS are the flagship projects. If this route reaches large-scale commercialization, demand for low-grade fines shrinks and demand for high-grade pellets expands.

Hydrogen reduction of iron ore has two inconvenient chemical characteristics. First, the reaction is strongly endothermic. In natural gas-based DRI, methane reforming simultaneously produces reducing gas and heat, and the shaft furnace can sustain its own thermal balance. Switch to pure hydrogen and you have reducing gas but no heat. How to maintain temperature inside the furnace, whether through external heating or preheating hydrogen to very high temperatures, is an engineering problem that needs solving. Second, hydrogen reduction of iron ore produces water vapour, not CO₂. The thermodynamic behaviour of water vapour in a shaft furnace is different from CO₂. It has an unfavourable effect on the reduction equilibrium, requiring a higher hydrogen excess ratio to achieve the same degree of metallization. This translates to hydrogen consumption potentially 30% to 50% above the theoretical stoichiometric ratio. Until green hydrogen drops below 2 dollars per kilogram, H₂-DRI has no cost advantage over natural gas-based DRI. The direction is correct. The timeline is optimistic.

Iron ore supply is highly concentrated. Four companies, 70%. Any sudden concentrated event on the supply side hits the price harder than an equivalent-tonnage change on the demand side, because demand adjustments are distributed across hundreds of steelmakers and transmit to prices gradually, while a severed railway, a collapsed tailings dam, or a cyclone landfall transmits as a step change.

After the 2019 Brumadinho tailings dam collapse at Vale, 270 people killed, Vale shut down approximately 90 million tonnes per year of associated capacity. That reduction persisted from 2019 all the way through 2021 and was the key supply-side prop for elevated iron ore prices during those two years.

Every January through March is tropical cyclone season in the Pilbara. Ports and railways suspend operations when cyclone warnings are issued, typically 3 to 7 days each time. During peak periods there can be several in succession, with cumulative seasonal impact on shipments of roughly 20 to 50 million tonnes. Iron ore derivatives traders start watching sea surface temperature data off the northwest Australian coast from November onward, because ENSO state has a statistical correlation with cyclone frequency and intensity for the upcoming season.

Rio Tinto's Pilbara system shipment pace can be tracked at weekly resolution. End-of-quarter surges, pushing to meet quarterly shipment guidance by accelerating dispatches. Start-of-quarter slow ramp-ups. This pulse pattern is predictable, and short-term traders use it for derivatives position management.

Iron ore inventories across roughly 45 Chinese ports, usually fluctuating between 100 and 150 million tonnes. Most people use this headline figure to read market tightness. High inventories, bearish. Low inventories, bullish. Too crude.

Port inventories contain several different components. Some is traders' stranded cargo, cleared customs but unsold, possibly a speculative hold betting on price increases. Some is strategic stockpiling by steelmakers who bought on dips. Some is stranded by product mismatch, a large shipment of a niche brand arrived but nearby steelmakers' blending formulas do not call for that brand, so it just sits there. Same total inventory, different internal composition, completely different price implications.

Geographic distribution of port inventories is also uneven. Northern ports like Caofeidian and Jingtang serve the Hebei steel cluster. Southern ports like Beilun and Zhanjiang serve East China and South China. The same brand being tight in northern ports while plentiful in southern ports is not uncommon. Domestic coastal shipping of iron ore from southern ports to northern steelmakers costs 30 to 50 renminbi per tonne. That cost means inventories across north and south cannot frictionlessly offset each other. The headline number looks adequate. Locally it can be tight.

On the supply side, the oligopoly structure holds. Simandou, if it eventually reaches production after more than twenty years of delays, would be the last single project with the potential to reshape the global iron ore supply landscape. On the demand side, China's crude steel output has very likely peaked or is extremely close to peaking. India and Southeast Asia offer incremental growth that can partially offset, depending on the pace of urbanization and infrastructure investment in those regions.

The product mix will continue shifting toward higher grade and lower impurities. Under decarbonization pressure this trend accelerates. One of the first levers a blast furnace pulls to cut carbon emissions is raising feed ore grade to reduce coke rate. DRI processes inherently favour high-grade pellets. Low-grade fines will not disappear, but their share of the burden structure and their pricing headroom both face long-term compression.

Water is a variable that almost never appears in industry outlooks and may become a hard constraint on specific projects sooner than expected. Magnetite processing requires 1 to 3 cubic metres of fresh water per tonne of raw ore processed. The Pilbara is semi-arid. Mines rely on groundwater extraction and pit dewatering for their water supply. Water permits are getting harder to obtain. For some planned projects in Africa the water constraint may arrive before the financing constraint does.

The cost floor of the entire industry is shifting upward at a rate of less than 1% per year. Enrichment zones are thinning. Grades are slipping. This change is too slow to see in the noise of quarterly data.