China's Massive Iron-Tailings Plant Claims a 98% Conversion Rate—Here's What That Number Actually Means

One of the key metrics for evaluating iron ore as steelmaking feedstock is its iron content, or Fe grade. Lump ore and pellets used in blast furnaces need a grade of at least 60%, ideally close to 65%. Lower-grade ore contains more gangue (mainly silica and alumina), which generates large amounts of slag during steelmaking and reduces thermal efficiency. But naturally occurring iron ore doesn't always start out at a sufficiently high grade.

A large-scale ore-processing plant has now begun full operation in Anshan, Liaoning Province, China. According to an announcement on September 1, 2026, by the Institute of Process Engineering (IPE) of the Chinese Academy of Sciences, a fluidized-bed magnetization roasting plant designed to process hard-to-beneficiate iron ore and leftover tailings passed an evaluation conducted by an expert panel convened by the Metallurgical Mines' Association of China. The plant began production in January 2026 and has reportedly been running stably at its designed processing capacity.

The facility processes ultra-low-grade tailings that had previously been discarded—material containing only about 11% iron—and reportedly extracts a high-grade iron concentrate of about 65% iron from it. The published figures are eye-catching, but understanding what they mean requires distinguishing between the volume of raw material that can be processed, the extent of the chemical reaction, and the actual amount of iron that can be recovered as product.

Sources: Reports based on Chinese customs statistics (SteelOrbis), USGS's "Mineral Commodity Summaries 2025," and materials published by the Chinese Academy of Sciences.

The developers describe the plant's annual processing capacity of 5.56 million tonnes as among the largest in the world for a single facility of this type. However, comparing the raw numbers alone, this amounts to only about 0.44% of China's iron ore imports in 2025. Moreover, the capacity to process low-grade raw material and the volume of imported iron ore are not directly interchangeable figures. This ratio is not an import-substitution rate—it's merely a reference point for gauging relative scale.

5.56 Million Tonnes a Year Is "Feedstock Processing Capacity," Not "Iron Output"

Some coverage and headlines have implied that this facility produces 5.56 million tonnes of iron from waste every year. That reading is incorrect. The published figure of 5.56 million tonnes annually (about 6.13 million short tons) refers to the amount of hard-to-beneficiate iron ore and tailings that can be fed into the plant for processing.

The facility consists of three independent production lines, each processing about 1.85 million tonnes (about 2.04 million short tons) of raw material annually. This is the first phase of a plan by Angang Group Mining Co. to reprocess 24 million tonnes of iron ore beneficiation tailings per year.

The technology licensing agreement with IPE was signed in November 2023. Major construction began in March 2024, followed by a full-process trial run in September 2025. Production started in January 2026. The recent announcement reports the results of an expert evaluation based on roughly eight months of operational data.

This evaluation was conducted by industry experts convened by the Metallurgical Mines' Association of China. The technology was assessed as being "at an internationally advanced level," but an industry association's expert evaluation differs from peer review of an academic paper. This announcement is not a peer-reviewed report of operational data from a commercial plant.

The announcement also does not disclose the annual output of concentrate. If we assume the average Fe grade of all 5.56 million tonnes of feedstock is 11% and the resulting concentrate has an Fe grade of 65%, the feedstock would contain about 610,000 tonnes of iron. Even with 100% iron recovery, the theoretical upper limit for concentrate output would be about 940,000 tonnes per year.

This calculation—5.56 million tonnes × 11% ÷ 65%—does not represent actual production figures. Still, it illustrates just how different feedstock processing capacity is from actual iron or concentrate output.

Making Magnetically Weak Iron Minerals Easier to Recover Through Roasting

Why is it necessary to heat and roast the ore at all? The key lies in mineral magnetism.

Magnetic separation, which uses magnetic force to separate target minerals from gangue, is widely used in ore beneficiation. Strongly magnetic magnetite ($\mathrm{F}{\mathrm{e}}_3{\mathrm{O}}_4$) is relatively easy to recover this way. In hard-to-beneficiate iron ore and tailings, however, iron is often present in weakly magnetic hematite ($\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_3$) or limonite ($\mathrm{FeO}(\mathrm{OH})\cdot \mathrm{n}{\mathrm{H}}_2\mathrm{O}$). When these become fine particles smaller than several tens of micrometers, their magnetic force becomes weak relative to the drag forces from water flow, making them difficult to recover with conventional magnetic separators.

Magnetization roasting is a process that partially reduces these weakly magnetic iron oxides in a reducing gas atmosphere, converting them into strongly magnetic magnetite. When carbon monoxide or hydrogen is used, the reaction can be expressed as follows:

By removing some of the oxygen, the trigonal hematite crystal structure is converted into magnetite, which has a spinel structure. This makes it easier to recover iron-bearing particles even with a relatively weak magnetic field.

However, controlling the reaction is not easy. If reduction proceeds too far, weakly magnetic wüstite ($\mathrm{FeO}$) or fayalite ($\mathrm{F}{\mathrm{e}}_2\mathrm{Si}{\mathrm{O}}_4$, formed through reaction with silicates) can form, lowering the efficiency of magnetic separation. Excessively high reaction temperatures or overly long residence times can lead to over-reduction.

The basic principle of the fluidized-bed method is to feed a mixture of reducing gas and inert gas from the bottom of the furnace, causing fine ore powder to behave like a boiling liquid. Unlike shaft furnaces, it isn't limited to lump ore and can handle fine powder, while also more easily avoiding the wall buildup that plagues rotary kilns. The large contact area between particles and gas allows for efficient heat and mass transfer.

According to IPE, the Anshan plant combines a mechanism for precisely controlling particle residence time, technology that promotes the reaction at lower temperatures, and a system for recovering heat in stages for effective reuse. However, the published materials do not disclose the actual composition or source of the reducing gas used, the furnace's operating temperature, or the average residence time.

"98% Conversion Rate" Is Not the Same as "98% Iron Recovery"

Two figures have drawn particular attention as operational results: a "98% iron mineral conversion rate" and "total energy consumption of 0.82 GJ/t." Evaluating these requires confirming exactly what each number measures.

According to explanations from outlets such as Science and Technology Daily, the 98% conversion rate indicates that 98% of the weakly magnetic iron minerals in the feedstock—such as hematite and limonite—were converted into magnetite. In other words, this is a measure of how far the targeted chemical reaction proceeded. It is not the same as "Total Iron Recovery," which measures the proportion of iron in the feedstock that is actually recovered as product.

Even with a high phase-conversion rate, if fine particles are subsequently lost as slime during magnetic separation, or if iron minerals are not sufficiently separated from gangue, the proportion of iron actually recovered as product can be lower. The published materials do not disclose the following key metrics:

1. Total iron recovery: The proportion of iron contained in the feedstock that ends up recovered in the concentrate.
2. Concentrate yield: The ratio of the mass of concentrate obtained to the mass of feedstock input.
3. Impurity concentration in the concentrate: The amount of silica ($\mathrm{Si}{\mathrm{O}}_2$) and alumina ($\mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3$) remaining in the concentrate at roughly 65% Fe grade.

Regarding the energy consumption figure of "0.82 GJ per tonne of ore," IPE states this is about 35% lower than comparable existing technologies. However, the energy consumption, furnace type, and feedstock conditions of the comparison technology are not specified. China Daily also reports that production costs have dropped significantly, but does not publish specific processing costs (yuan/tonne) or the iron ore price assumptions used to evaluate profitability.

Converting low-grade tailings into concentrate requires many processing steps, including crushing, classification, roasting, cooling, multiple stages of magnetic separation, and dewatering. Unless the concentrate can be supplied at a price competitive with imported ore—factoring in all these costs—securing commercial viability will be difficult. The currently available public information is not sufficient to fully assess the cost conditions underlying this assumption.

What Six Validation Stages Reveal, From Basic Research to Commercial Operation

Evaluating the results from the Anshan plant requires distinguishing between basic research, pilot testing, and commercial-scale operation. IPE researchers have published peer-reviewed papers on fluidized-bed magnetization roasting over more than a decade.

{
  "type": "bar",
  "title": "Mineral conversion rate and total iron recovery are different metrics",
  "unit": "%",
  "categories": ["2023 tailings pilot test (72 hours)", "2026 Anshan commercial plant (reported)"],
  "series": [
    { "name": "Iron mineral conversion rate", "values": [86.46, 98.0] },
    { "name": "Total iron recovery", "values": [48.40, null] }
  ],
  "caption": "In the 2023 test, the conversion rate was 86.46% while total iron recovery was only 48.40%. Total iron recovery at the Anshan commercial plant is undisclosed—this does not mean it is zero. Because feedstock, equipment scale, and test conditions differ, the two cases cannot be simply compared for performance.",
  "source": "DOI: 10.1080/01496395.2023.2189055 / IPE"
}

A 2018 pilot test using fine ore powder reported a total iron recovery of 91.16%. Meanwhile, a 2023 pilot test processing tailings reported a total iron recovery of 48.40% against an iron mineral conversion rate of 86.46%. Because the feedstock and test conditions differ, these results cannot be directly compared. Even so, it's clear that conversion rate and iron recovery rate are distinct metrics.

Organizing what each study and operational result shows—and what conclusions can and cannot be drawn from it—yields the following:

Tracing the research history, the Chinese Academy of Sciences' magnetization roasting research dates back to fluidized-bed research conducted by Mooson Kwauk and colleagues at the Institute of Chemical Metallurgy in the 1950s and 1960s. Research toward industrialization was subsequently stalled for a long period, but Professor Zhu Qingshan and colleagues, who now lead the current project, resumed it in 2004.

A demonstration furnace with an annual processing capacity of 100,000 tonnes built in Yunnan Province encountered funding difficulties in 2008. It reportedly took over a year of addressing on-site issues before continuous operation was achieved in December 2012.

To improve the thermal efficiency issues encountered at the Yunnan demonstration facility, a second-generation fluidized-bed technology with fine-tuned control of particle residence time was developed and adopted at the Anshan plant. The accumulation of academic research and pilot testing supports the underlying principles of the technology and its effectiveness in small-scale trials. However, the 85.85% total iron recovery achieved in the laboratory cannot simply be applied to a commercial plant processing 5.56 million tonnes per year.

How Far Can China Reduce Its Dependence on 1.26 Billion Tonnes of Imports?

Behind China's push to utilize ultra-low-grade tailings lies the challenge of securing a stable supply of iron ore.

China accounts for more than half of global crude steel production and is the world's largest consumer of iron ore. According to materials from the Chinese Academy of Sciences, cumulative iron ore imports over the past decade have reached 11.3 billion tonnes. Reports based on Chinese customs statistics state that imports in 2025 reached 1.26 billion tonnes, a record high.

However, some published materials do not provide the previous year's actual figures needed to verify year-over-year growth rates. The scale of imports and year-to-year fluctuations should be considered separately. Supply sources are also concentrated, with more than 80% of imports depending on specific supplying countries such as Australia and Brazil.

Meanwhile, according to the U.S. Geological Survey's "Mineral Commodity Summaries 2025," China's economically recoverable iron ore reserves stand at 6.9 billion tonnes in metal iron content, the fourth-largest in the world. However, much of the domestic ore has a low Fe grade of around 30%, and nearly half of it is considered difficult to beneficiate economically with conventional technology. Tailings accumulated over many years in tailings dams have also posed a burden in terms of environmental risk and maintenance costs.

IPE anticipates that widespread adoption of this technology could lead to the utilization of approximately 30 billion tonnes of domestic hard-to-beneficiate iron ore and about 10 billion tonnes of iron-bearing tailings. However, this is not a confirmed economically recoverable reserve figure—it is the developers' projection of the total volume of ore and tailings that could potentially be utilized in the future.

In IPE materials from November 2023, hard-to-beneficiate iron ore was estimated at over 20 billion tonnes and iron-bearing tailings at over 4 billion tonnes—figures that differ from the current announcement. The current projections should not be treated as resource volumes with confirmed economic viability.

{
  "type": "bar",
  "title": "Scale and progress of fluidized-bed magnetization roasting technology deployment",
  "unit": "10,000 tonnes/year",
  "categories": ["Anshan plant (operating)", "Total contracted projects (lower bound)", "Laos planned project (2 lines, contract stage)"],
  "series": [
    { "name": "Annual feedstock processing capacity", "values": [556, 1500, 400] }
  ],
  "caption": "The total for contracted projects exceeds 15 million tonnes annually; the figure shown is a lower bound. Contracted capacity differs from operating capacity. The total and individual projects are shown side by side and should not be summed across all three items.",
  "source": "IPE / China Daily / CAS"
}

Regarding the deployment of this technology, licensing agreements totaling more than 15 million tonnes of annual processing capacity are reported to have been signed. However, the number of contracts and the number of projects should be distinguished. The Chinese Academy of Sciences' English-language announcement states 11 licensing agreements, while China Daily's English edition reports 8 projects in addition to Anshan. Since the counting methods for contracts and projects may not align, these two figures alone cannot be judged as contradictory.

Overseas, a two-line limonite roasting project with an annual capacity of 4 million tonnes (2 million tonnes per line) is planned in Laos. However, this is at the contract-signing stage and is not yet an operating facility.

As for the Anshan plant, under the previously mentioned conditions of 11% Fe grade feedstock and 65% Fe grade concentrate, even processing 5.56 million tonnes of feedstock annually would yield only about 940,000 tonnes of concentrate per year at 100% iron recovery. Differences in grade between the feedstock and imported ore must also be considered, and against annual imports of 1.26 billion tonnes, this single facility alone cannot substantially reshape the overall dependency structure.

Undisclosed Data Remain on Reducing Agent Consumption and Environmental Impact

Evaluating this technology in terms of reduced environmental impact requires examining data for the entire plant.

Because magnetization roasting relies on a reduction reaction, it continuously consumes reducing agents. Laboratory and pilot tests have used carbon monoxide ($\mathrm{CO}$), hydrogen (${\mathrm{H}}_2$), coke oven gas, and carbonaceous materials, among others. The type and quantity of reducing agent used at the Anshan plant, the environmental impact of procuring it, and the life-cycle carbon dioxide ($\mathrm{C}{\mathrm{O}}_2$) emissions are not disclosed in the published materials.

A CFD analysis paper involving IPE researchers (DOI: 10.1016/j.ces.2020.116148) calculates that CO2 generated from the fluidized-bed magnetization roasting process ranges from 0.63 to 4.14 grams per kilogram of ore, depending on conditions such as gas temperature, particle size, and CO concentration. However, this is a numerical calculation result for the reaction furnace alone. It is not an emissions intensity figure for the entire plant, including fuel combustion for heating or upstream and downstream equipment.

A review paper published in the Journal of the Southern African Institute of Mining and Metallurgy (SAIMM) also points out that magnetization roasting requires substantial energy and that industrial-scale applications for tailings processing remain few. Furthermore, the fine-powder concentrate obtained cannot be fed directly into a blast furnace as is. Using it as steelmaking feedstock requires an additional granulation process such as pelletizing, which itself requires thermal energy for sintering the pellets.

This project is included among the first batch of projects selected for China's National Development and Reform Commission (NDRC) "Green and Low-Carbon Advanced Technology Demonstration Project" program. It is also listed among advanced technologies useful for resource conservation compiled by the Ministry of Natural Resources. However, policy-level selection differs from a comprehensive environmental impact assessment or third-party emissions audit conducted by an independent institution.

Recycling tailings has the potential to reduce accumulation in tailings dams and land occupation, potentially lowering the risk of dam failure and enabling land reuse. On the other hand, it is not sufficiently disclosed which tailings dam supplies the Anshan plant's feedstock, what moisture content and particle size that feedstock has, or how the silica-dominant residual tailings left after processing will be stored and disposed of to ensure safety. A quantitative environmental impact assessment covering these processes is needed.

IPE has outlined future plans to apply fluidized-bed roasting technology to other hard-to-beneficiate minerals containing manganese, phosphorus, and aluminum, as well as to solid industrial waste. However, these are future research plans, not results demonstrated by the current plant.

The technology for obtaining roughly 65% Fe grade concentrate from ultra-low-grade tailings is underpinned by decades of research in powder engineering and reaction kinetics. However, judging its commercial value requires more than just a 98% conversion rate. It's necessary to confirm how much iron can actually be recovered, how thoroughly impurities can be removed, how much reducing agents and energy cost, and how much environmental impact results.

The key to determining the technology's effectiveness and its competitiveness against imported ore lies in the long-term data obtained from commercial operation. Only once total iron recovery, cost, and environmental impact are presented in a verifiable form can we properly assess how much this achievement can actually contribute to iron ore supply.