Recycling and Circularity: A Practical Reference

Curbside collection puts a cart at the end of a driveway and a truck on a route, and a material recovery facility turns what the truck brings back into baled commodities. Most US programs are single-stream, meaning every recyclable goes into one cart, which raised participation and raised contamination at the same time. A modern single-stream MRF runs 15–50 tons an hour through a fixed sequence: a presort line where people pull bags, film and hazards off the belt, a cardboard screen, glass breaker screens, disc screens that separate flat paper from containers, a magnet, optical sorters for plastics, an eddy-current separator for aluminum, a quality-control sort, and a baler. A large plant on two shifts processes 100,000–200,000 tons a year. Everything downstream depends on what comes off that line, because a bale's price is set by its purity: baled aluminum cans sell for $1,100–1,800 a tonne while the same aluminum inside a mixed residual stream is worth nothing. Residual, the fraction that leaves as trash, usually runs 15–30% of what enters, and the operator pays the local tipping fee of roughly $55–60 a ton on all of it.

Single-stream collection is the reason tonnage exists at all. It roughly doubled what households set out compared with the source-separated bins it replaced, and it cut collection cost, since one truck making one pass with one compartment is cheaper than two. The weakness shows up at the plant, where the material arrives mixed, wet and broken. Glass shatters in the truck, and the fragments abrade screens and lodge in the paper, so a single-stream MRF typically delivers 5–15% contamination in its bales while mechanical plastics recycling usually needs under about 2%. The operator also carries a commodity position it did not choose, buying feed on one date and selling bales on another: after China's National Sword restrictions in 2018, mixed paper went from about $75 a ton to negative and hundreds of US programs cut materials.

Use single stream when the goal is tonnage and the buyers are steel mills, aluminum smelters and paper mills, all of which tolerate a 5–15% contaminated bale. If the goal is food-grade PET or bottle-to-bottle HDPE, do not start here; a deposit return system produces a far cleaner stream and is the only thing that reliably does, so run one alongside the MRF rather than expecting the MRF to reach that grade. If the local tipping fee is under about $40 a ton, assume the MRF needs a municipal contract or an extended producer responsibility payment to cover its costs, because the commodity revenue alone will not. Dual-stream collection is still the right answer where paper quality matters more than participation, and a few Northeast programs kept it for exactly that reason. Before adding sorting equipment, check what fraction of the residual is genuinely unsortable rather than badly collected, since a collection change is usually cheaper than a new optical sorter.

Line throughput 15–50 tons per hour, 100,000–200,000 tons a year on two shifts · residual 15–30% of incoming tonnage at a $55–60 a ton tipping fee · bale contamination 5–15% against under about 2% needed for mechanical plastics recycling · aluminum cans roughly 2% of incoming weight and 30–50% of commodity revenue · capital roughly $15–30 million for a 30–50 ton per hour plant · processing cost $70–110 a ton against $60–120 a ton of commodity revenue

Single-stream MRFs operated by Republic Services and Waste Management across the US; Eureka Recycling in Minneapolis, which publishes its own composition and residual data; Recycle BC, the producer-funded collection system in British Columbia that reports packaging recovery near 78%; the dual-stream programs still running in parts of the Northeast, which produce cleaner paper at lower tonnage.

A MRF has two revenue lines and the split decides everything. Commodity sales run $60–120 a ton of incoming material in normal markets, and processing costs $70–110 a ton, so the gate fee or the municipal contract closes the gap in most years and is the whole business in bad ones. Capital is roughly $15–30 million for a 30–50 ton per hour single-stream plant, which is a fixed cost spread over throughput, so a plant at half its nameplate has close to double the unit cost. Labor is the largest operating line after that, typically 30–40% of operating cost, and sorter turnover above 100% a year is normal, which is the actual reason robotic cells get bought. The revenue is far more concentrated than the tonnage: aluminum cans are about 2% of incoming weight and 30–50% of commodity revenue, so a MRF is in practice a metal recovery business that also handles paper. Residual is the cost nobody models properly, because 15–30% of every incoming ton goes back out at a $55–60 a ton tipping fee, and that fee rises faster than commodity prices do. Whoever holds the commodity price risk in the municipal contract is the question to ask first, since the 2018 paper collapse moved dozens of these contracts from profitable to loss-making without anything changing at the plant.

5 minSingle Stream Recycling – Tour a Material Recovery Facility (MRF)Van Dyk Recycling Solutions · 500k+ views

3 minHow does a Material Recovery Facility (MRF) work?RecycleRightNC · 100k+ views

The U.S. Recycling System (US EPA) · Techno-economics of hydrocarbon fuel production and recyclables recovery from landfill-destined municipal solid waste: AI-enhanced materials recovery facility design (Journal of Cleaner Production)

Shredding cuts a complex product into pieces small enough that its materials come apart and small enough that a separator can act on each piece. The dominant machine in metals recycling is the hammermill: a rotor two to three meters across swinging free hammers at 500–700 rpm inside a grate, driven by 3,000–10,000 hp (roughly 2–7.5 MW). A car body goes in whole and leaves as fist-sized fragments in under a minute, at 60–450 tons an hour depending on the machine, using 15–40 kWh per ton. For an end-of-life vehicle the output splits roughly 68–75% ferrous, about 5% nonferrous, and 20–25% shredder residue, the mixed plastic, foam, glass and dirt fraction that is usually landfilled. Elsewhere in recycling the same job is done by slower machines: low-speed shear shredders at 20–100 rpm and high torque for e-waste, tires and refuse-derived fuel, and granulators cutting plastics to 4–20 mm. The word that matters is liberation, meaning the fraction of pieces that are a single material, because a piece that is still two materials joined together cannot be sorted by anything downstream.

Nothing else delivers this much throughput per dollar, and shredding tolerates feed that no other process will take. It is also the step that makes magnetic, eddy-current and sensor-based separation possible at all, since all three act on individual particles. The weakness is that the mixing is irreversible. Once a wiring harness goes through the rotor, its copper is spread through the ferrous fraction, which is why shredded steel scrap typically carries 0.2–0.4% copper and there is no economic way to remove it from liquid steel afterward. Finer shredding improves liberation but costs more energy and creates more fines below about 5 mm, which no separator catches and which leave with the residue. Shredders also catch fire: fuel tanks, aerosol cans and, increasingly, lithium-ion cells embedded in products cause deflagrations, and battery-driven fires at shredder yards and MRFs have become one of the industry's largest insurance costs.

Shred when the value of the product is in its bulk metal and the assembly is too complex to take apart by hand. Dismantle first when the value sits in a component: a printed circuit board is worth $2,000–6,000 a tonne intact and far less once it is spread through a mixed shredded fraction, and the same argument applies to motors containing rare-earth magnets and to EV modules. Set the shred size by the smallest thing you actually need to separate rather than the finest the machine can reach, because every step down in size costs energy and makes fines. Use a low-speed shear shredder where dust, fire risk or fume matters, which is most e-waste and all battery feed, and a hammermill where tons per hour is the whole point. If the downstream separation is magnets and eddy current only, a coarse shred is enough; if it is optical or X-ray sorting, the pieces have to be small and flat enough to present a single surface to the sensor.

Hammermill drive 3,000–10,000 hp (roughly 2–7.5 MW) at 500–700 rpm · throughput 60–450 tons per hour · energy 15–40 kWh per ton · vehicle output roughly 68–75% ferrous, about 5% nonferrous, 20–25% shredder residue · shredded steel carries 0.2–0.4% copper · fines below about 5 mm are not recovered · wear parts $3–8 a ton against $1.20–3.20 a ton of electricity

Texas Shredder and Lindemann hammermills at US and European shredder yards; Wendt and Danieli Centro Recycling megashredder installations feeding EAF steelmakers; Untha and Vecoplan low-speed shredders for e-waste and refuse-derived fuel; Herbold and Erema granulators cutting washed plastics to flake; the shredder residue stream at US yards, which runs into millions of tons a year and has no established buyer.

There is no revenue from shredding itself, only a lower cost for everything after it, which is why the margin is thin and why the machine is always owned by whoever also sells the output. Electricity is smaller than it looks: at 15–40 kWh a ton and roughly $0.08 a kWh, power is $1.20–3.20 a ton, less than the $3–8 a ton spent on hammers, grates and liners, so wear parts and the downtime to change them are the operating cost that matters. Capital runs $20–50 million for a large shredder plant with its downstream magnetic and eddy-current separation, and the fixed cost only works at 300,000–600,000 tons a year, which means a secure feed radius rather than a better machine is what makes the site viable. The gross margin is the spread between what the yard pays for incoming scrap and the $300–450 a tonne that shredded ferrous sells for, and it moves with the steel cycle rather than with anything the operator controls. Shredder residue is the line that has quietly gotten worse: it is 20–25% of every ton processed, it goes to landfill at a rising tipping fee, and the alternatives to landfilling it have been studied for thirty years without one becoming standard.

16 minHow Millions Of Old Cars Are Recycled Into New Steel | Massive Car Recycling LineThe Factoran · 1m+ views

2 minInside an Industrial Scrap Metal Shredder | Full Process Tour & Metal SeparationMetro Group · 10k+ views

Circular Economy for Automotive Shredder Residue (Oak Ridge National Laboratory) · Automobile shredder residues in Italy: characterization and valorization opportunities (Politecnico di Torino)

These are the two separation steps that need no sensor and almost no operating cost. An overhead belt magnet or a magnetic head pulley lifts ferrous metal out of a passing stream and recovers 90–98% of it, using permanent ferrite or rare-earth magnets that consume no power at all. An eddy-current separator handles the nonmagnetic metals immediately afterward: a rotor carrying alternating permanent magnets spins at 3,000–4,000 rpm inside a stationary non-conducting shell at the end of a conveyor, and the rotating field induces circulating currents in any conductive particle passing over it. Those induced currents create an opposing field, so the particle is repelled forward off the end of the belt while non-conductors drop straight down, and the two fractions land in separate bins with a splitter between them. The force scales with electrical conductivity divided by density, which is why aluminum flies furthest, copper and brass travel less far, and stainless steel barely moves. A 1.5–2 m wide unit handles 5–30 tons an hour and recovers 90–95% of the aluminum above about 5 mm.

The operating cost is close to zero. Permanent magnets need no excitation, the only power draw is the drive motor and belts, and there are no consumables and no compressed air, which is what separates these two steps from every sensor-based sorter downstream. The weaknesses are all about what falls outside the physics. Recovery drops sharply below about 5 mm because the induced force scales with particle size, so fine aluminum leaves with the residue. Flat thin items like foil and laminated pouches do not develop enough force against their own air resistance and land unpredictably. Stainless steel and other low-conductivity alloys barely respond and report to the reject, so they need an induction or sensor-based sorter instead. Two operational failures are common: long wire and strapping wrap the rotor and stop the line, and an eddy-current rotor runs hot, so a fire or a heavy fines buildup can demagnetize it.

Put a magnet on every line that could contain steel. It is the cheapest recovery step available and it also protects the shredders, screens and sorters downstream from tramp metal. Put an eddy-current separator immediately after the magnet whenever the stream contains more than roughly half a percent aluminum, which covers essentially every MRF, shredder line and incinerator bottom-ash plant. Screen the feed into two or three size fractions and run a separate unit on each rather than one unit on everything, because an eddy-current separator works over a narrow particle size band and needs a single layer of material spread across the belt. Do not expect either device to sort metal from metal: if the job is separating aluminum from copper, or cast from wrought aluminum, use dense media separation or a sensor-based sorter instead. If the target is stainless, skip both and go to an induction-based metal sorter.

Ferrous recovery 90–98% from an overhead or drum magnet · eddy-current rotor at 3,000–4,000 rpm · throughput 5–30 tons per hour on a 1.5–2 m unit · aluminum recovery 90–95% above about 5 mm and poor below it · installed cost roughly $100,000–250,000 per eddy-current separator · baled aluminum $1,100–1,800 a tonne against $800–1,400 for mixed nonferrous zorba · derived payback under a year on a 10 ton per hour MRF line

Eddy-current separators from Eriez, Steinert, Bunting, Goudsmit and Wagner Magnete on MRF and shredder lines worldwide; drum magnets recovering steel cans at every US single-stream MRF; the mixed nonferrous "zorba" fraction from auto shredders, which is produced by this equipment and then sold on for further sorting; municipal incinerator bottom-ash plants, where eddy-current separation is the main source of revenue from the ash.

This is one of the few steps in recycling where the recovered metal covers the capital outright, and the arithmetic is short enough to check. A 2 m eddy-current separator handling 10 tons an hour costs roughly $100,000–250,000 installed. Aluminum is about 2% of what enters a MRF, so at 10 tons an hour and 90% recovery the unit delivers 0.18 tons an hour, or about 360 tonnes over a 2,000-hour year. At $1,100–1,800 a tonne for baled cans that is $400,000–650,000 of metal against an operating cost of a few kilowatts and a belt. That calculation is done here rather than taken from a published figure, but it is why almost every shredder line and MRF has one and why payback is measured in months rather than years. The magnet ahead of it is cheaper still and recovers steel worth $300–450 a tonne shredded. What these two steps do not do is upgrade metal to metal, so the nonferrous fraction leaves as mixed zorba at $800–1,400 a tonne rather than as a sorted alloy, and closing that gap is the entire argument for the sensor-based sorters that follow.

1 minEddy current Non ferrous separator working principle 3D animationGoudsmit Magnetics - Driven by magnetism · 50k+ views

2 minK&J Magnetics - Eddy Current SeparatorK&J Magnetics · 10k+ views

An analytic model for eddy current separation (Minerals Engineering) · Sensor based optimisation of eddy current separation in bottom ash recycling (TU Delft Repository)

A sensor-based sorter looks at every object on a fast belt, decides what it is, and knocks the ones it wants off the end with a puff of compressed air. The workhorse sensor is near-infrared spectroscopy over roughly 1,000–2,500 nm, where the overtones of C–H and O–H bonds give each polymer a distinct absorption pattern, so PET, HDPE, PP, PS and PVC look different to a line-scan spectrometer under halogen light. The belt runs at 2.8–3.5 m/s, the classifier decides in a few milliseconds, and a bank of valves at 6–12 mm pitch fires as the object crosses the discharge. A 2 m unit handles 4–8 tons an hour of containers and reaches 90–98% purity in one pass. The same machine frame takes other sensors for other jobs: color cameras for glass and for natural against colored HDPE, X-ray transmission for atomic number, which is how PVC is picked out by its chlorine and how cast aluminum is separated from wrought, and laser-induced breakdown spectroscopy or X-ray fluorescence for specific alloys.

One machine replaces roughly six to ten hand sorters and holds a purity a person cannot sustain across a shift, which is why a modern MRF has four to eight of them. The failure modes are specific and worth knowing by name. Carbon black absorbs across the whole near-infrared band, so a black item returns no usable spectrum and is invisible to NIR; the detectable black pigments that fix this exist but are not widely adopted. A full-sleeve PET bottle reads as the sleeve polymer, because NIR sees only the surface. Wet or food-soiled surfaces cut the signal, and a thin film lying flat on the belt is often read as the belt. There is also a fixed trade-off between purity and yield: tightening the classifier to raise bale purity always drops recovery, and reaching above about 95% purity usually takes a second pass or a dedicated cleanup unit rather than better settings.

Use optical sorting whenever the price gap between a sorted bale and a mixed one is larger than the cost of running the pass, which is most of the time in plastics. Natural HDPE sells for $600–1,400 a tonne against $200–400 for colored, so a color camera is one of the fastest paybacks in a MRF. Use it when a rule sets the output spec rather than a buyer: the EU requires 25% rPET in PET bottles from 2025, food-grade rPET has to come from a positively sorted stream, and no amount of negative sorting reaches that grade. Go to X-ray transmission, XRF or LIBS instead of NIR when the job is metal against metal, since NIR reads nothing useful off a metal surface. Below roughly 1 ton an hour the fixed cost per ton is hard to carry, so hand sorting or a robotic cell is usually cheaper at that scale. If black plastic or sleeved bottles are a large part of the stream, fix that upstream in packaging design and treat the loss as a design problem rather than buying more sensors.

NIR band roughly 1,000–2,500 nm · belt speed 2.8–3.5 m/s with valves at 6–12 mm pitch · throughput 4–8 tons per hour on a 2 m unit · purity 90–98% in a single pass · installed cost $150,000–400,000 per unit · compressed air 200–600 cfm, commonly 20–40% of a MRF's electricity bill · natural HDPE $600–1,400 a tonne against $200–400 colored, and sorted PET $200–500 against $0–100 for mixed rigids

TOMRA AUTOSORT, Steinert UniSort, Pellenc ST, Sesotec and Bühler units on MRF, plastics-wash and e-waste lines; X-ray transmission sorters upgrading mixed nonferrous zorba by separating cast from wrought aluminum; LIBS sorters grading specific aluminum alloys at shredder yards, a route PNNL and Argonne have published test data on; color sorters at glass cullet plants separating flint, amber and green.

Capital is $150,000–400,000 per installed unit, and the return comes entirely from the spread between a sorted bale and a mixed one. Take a 2 m sorter running 4 tons an hour for 2,000 hours a year, which is 8,000 tonnes through the machine. If it converts 1,000 of those tonnes from mixed rigids at $0–100 a tonne into sorted PET at $200–500, the gain is roughly $200,000–400,000 a year against a machine costing $150,000–400,000, so payback lands somewhere between one and two years. That calculation is done here rather than quoted, and the sensitive input is not the sensor but how much of the throughput actually changes grade. Operating cost is dominated by compressed air, which at 200–600 cfm per large unit is commonly 20–40% of a MRF's electricity bill and is the line most feasibility models leave out. In metals the same equipment is justified by a bigger spread: sorted wrought aluminum typically sells $300–600 a tonne above mixed zorba, which is enough to fund an X-ray or LIBS sorter at a shredder yard on volume that a plastics line could never reach. The regulatory demand matters as much as the price spread, because a recycled-content mandate with a date creates a buyer for high-purity output that would otherwise have none.

5 minSorting of plastic waste streams by near-infrared spectroscopy | Online course - MOOCKU Leuven · 10k+ views

1 minAUTOSORT Optical Sorting Sensor Detailed AnimationVan Dyk Recycling Solutions · 10k+ views

Alloy Selective Optical Sorting of Mixed Post-Consumer Aluminum Scrap Streams (Pacific Northwest National Laboratory) · SpectralWaste Dataset: Multimodal Data for Waste Sorting Automation (arXiv)

A robotic sorting cell is a camera over a belt, a neural network that labels every object it sees, and a delta or gantry robot with a suction gripper that reaches down and drops the targets into chutes. The vision stack usually pairs an RGB line-scan camera with a near-infrared sensor, because color, shape and printed branding identify a format while the spectrum identifies the polymer. A cell picks 60–80 items a minute in normal operation, with the fastest two-arm units quoted up to 120, against 30–40 a minute for a person who cannot hold that rate across a full shift. Installed cost is roughly $200,000–400,000 a cell, or about $5,000–7,000 a month under a robots-as-a-service contract. The difference from an air-jet optical sorter is that the robot picks individual objects rather than ejecting a whole class, which is what makes it useful on quality-control lines and on residue where the target is a small share of a dirty stream.

Robotic cells fit the two places air jets do not: the end-of-line quality check, where a handful of items have to come out of an almost-clean flow, and the residue line, where the target is a few percent of what passes. They also produce a per-item record of everything that went by, which is the first real composition data most MRFs have ever had and is directly useful for extended producer responsibility reporting. The binding limit is pick rate. A belt carrying 4 tons an hour of containers presents thousands of items a minute, and a cell touching 70 of them is a cleanup device rather than a bulk sorter. Uptime is the second limit and the one that decides the money: suction cups wear and foul on film and liquid, wet cardboard and tangled material defeat the gripper, and a jam stops the cell entirely. Recognition accuracy is the number vendors lead with and the least binding of the three, because moving a model from 95% to 99% changes bale purity slightly while a cell running six hours of a ten-hour shift changes whether the capital comes back at all.

Put the first cell on quality control, where the item count per minute is closest to what a robot can actually pick, and where a small purity gain moves a whole bale into a higher grade. Add a second on the residue line only after the composition data shows recoverable material at more than roughly 5% of that stream. Do not buy a robot for a job an air-jet sorter does: if the target is a large share of the stream by count, an optical sorter at $150,000–400,000 handles far more tons per dollar. Justify the purchase on labor you cannot hire rather than on purity, since sorter turnover above 100% a year is normal and an unfilled position is a line running slower. When you evaluate a vendor, ask for picks per hour over a month of real production and the uptime that produced it, and treat benchmark accuracy figures as the least informative number in the quote.

Pick rate 60–80 items a minute typical and up to 120 quoted, against 30–40 for a person · installed cost roughly $200,000–400,000 a cell, or $5,000–7,000 a month as a service · fully loaded US sorter position $35,000–55,000 a year, so $70,000–110,000 across two shifts · derived simple payback 2–4 years at high uptime · published techno-economics put AI-based sortation at $28.7 a tonne against $28.0 for optical sortation · sorter turnover above 100% a year is normal

AMP Robotics cells in US MRFs, plus the company's own secondary sortation facilities built around them; Machinex SamurAI, sold as a retrofit onto an existing sorting line; ZenRobotics, the Finnish system now owned by Terex, which started on construction and demolition waste; Recycleye, Glacier and EverestLabs selling cells and the composition data that comes with them.

The case rests on labor arithmetic that is easy to check and easy to get wrong. A fully loaded US sorter position costs $35,000–55,000 a year, and a two-shift plant staffs each position twice, so removing one position saves $70,000–110,000 a year. Against $200,000–400,000 for an installed cell that is a simple payback of roughly two to four years, derived here rather than quoted, and it holds only at high uptime, because annual picks scale directly with hours run. Robots-as-a-service at $5,000–7,000 a month works out to $60,000–84,000 a year, which is close to one two-shift position, so the vendor has priced the cell at the labor it displaces and taken the downtime risk onto its own balance sheet. That pricing is a fair signal of where the margin sits. Published techno-economic work on AI-driven municipal waste sortation puts it at $28.7 a tonne against $28.0 for optical sortation, meaning the cost per ton is roughly a wash and the argument has to be won on output purity feeding downstream processing, not on cheaper sorting. The composition data is a real second revenue line for producer responsibility reporting and commodity negotiation, and it is the part of the offering that does not depend on the gripper working.

11 minHow Robots Could Save $6+ Billion Worth Of Recyclables A Year | AI In ActionBusiness Insider · 1m+ views

2 minSamurAI™ Robotic Sorter in Single Stream Recycling FacilitiesMachinexInc · 1k+ views

Artificial intelligence-driven municipal solid waste sortation and its significance on downstream waste valorization in the United States: Techno-economic and life cycle assessment (Resources, Conservation and Recycling) · ZeroWaste Dataset: Towards Deformable Object Segmentation in Cluttered Scenes (arXiv)

Steel is magnetic, so pulling it out of mixed waste costs almost nothing, and an electric arc furnace will melt nearly any grade of it back into liquid steel. A modern furnace holds 80–150 tonnes, strikes an arc from three graphite electrodes down into a cold scrap charge, and taps 40–60 minutes later, drawing roughly 350–450 kWh of electricity per tonne. The IEA puts the whole scrap-based EAF route at 2.1 GJ of final energy per tonne of crude steel against 21.4 GJ/t for the blast furnace and basic oxygen furnace, and at 0.04 tonnes of direct CO2 per tonne against 1.2 tonnes. Check which energy accounting a comparison uses before quoting it, because worldsteel counts electricity in primary-energy terms at 9.8 GJ per MWh and gets 5.2 GJ/t against 22.7 GJ/t for the same two routes, which is about a quarter rather than a tenth. Blast-furnace shops use scrap as well: a basic oxygen furnace charge is typically 15–25% scrap, which absorbs the heat the decarburization reaction gives off. Worldwide about 650 million tonnes of scrap goes into 1,869 million tonnes of crude steel, so roughly 30% of the metallic input to steelmaking is already recycled metal. The US share is much higher, and the arithmetic is worth doing: USGS reports 82 million tonnes of raw steel in 2025 against only 21 million tonnes of pig iron, and a BOF charge is 70–85% hot metal, which puts electric furnaces at close to 70% of US output.

The emissions gap is the largest of any material on this sheet. Counting grid electricity at the world average carbon intensity, the scrap EAF route emits 0.3 tonnes of CO2 per tonne of crude steel against 2.2 tonnes for BF-BOF, and it gets cleaner every year the grid does, with no change to the plant. The capital gap matters just as much to whoever is building: the IEA puts annualized capital at $34–58 per tonne of crude steel for a scrap EAF against $52–94 for BF-BOF, and a minimill is economic at 0.5–1.5 million tonnes a year where an integrated mill needs three to five times that. Two weaknesses are structural. Copper and tin cannot be removed from liquid steel at any sensible cost, so scrap chemistry caps which products a mill can roll, which is the subject of the next entry. And scrap supply is fixed by what was built decades ago rather than by today's demand, because the average steel product lasts about 40 years, so a mill cannot buy its way out of a tight scrap market the way it could order more iron ore.

If you are building new steel capacity and scrap is available within a few hundred miles, build an EAF: it is cheaper per tonne of capacity, faster to permit, and it can be idled when power prices spike, which an integrated mill cannot. If the product is rebar, structural sections, merchant bar, or plate, scrap chemistry is a non-issue and 100% scrap is the normal charge. If the product is exposed automotive sheet, tinplate, or anything deep-drawn, plan on diluting with direct-reduced iron or pig iron and price that in from the start, because sorting alone will not get copper low enough. If you are forecasting how much of a national steel industry can go electric, look at scrap arisings and the product mix rather than at furnace capacity, since the constraint binds on chemistry before it binds on melting. And if the site has no firm grid connection at 100+ MW, the EAF question is settled before the metallurgy is discussed.

EAF heat 80–150 t, tapped in 40–60 minutes at roughly 350–450 kWh/t · scrap-based EAF 2.1 GJ/t of final energy against 21.4 GJ/t for BF-BOF · 0.3 t CO2 per tonne including grid power against 2.2 t · BOF charge typically 15–25% scrap · about 650 Mt of scrap into 1,869 Mt of world crude steel · US No. 1 heavy melting composite averaged $319/t delivered in 2025 · annualized capital $34–58/t of crude steel against $52–94/t for BF-BOF

Nucor is the largest US steelmaker and has never operated a blast furnace; Steel Dynamics and Commercial Metals run the same model, and Big River Steel in Arkansas showed that an EAF can hold automotive sheet tolerances. Cleveland-Cliffs and U.S. Steel are the only two companies still running integrated mills in the United States, at eight locations between them. Scrap is bought and sold against published grade specifications: the ReMA (formerly ISRI) specifications in North America and the EU-27 Steel Scrap Specification in Europe.

Scrap is the whole cost story. US purchases of iron and steel scrap were worth $19.7 billion in 2025, the No. 1 heavy melting composite averaged $319 per tonne delivered, and a heat takes roughly 1.1 tonnes of scrap per tonne of liquid steel after melt loss and slag. The IEA puts raw materials and energy at 60–80% of the cost of making steel by any route, and for an EAF that is mostly one purchase order. Everything else is small by comparison: 350–450 kWh/t of electricity is $11–40 per tonne at $30–90/MWh, plus graphite electrodes, refractories, ferroalloys, and oxygen. Operating cost lands at $31–54 per tonne of crude steel against $48–87 for BF-BOF on the IEA's numbers, so the operator's margin is the spread between the finished product price and the scrap price, and both move. That spread is what makes the business cyclical rather than steady: scrap and steel prices usually move together, and the mills that survive downturns are the ones whose contracts index the two to each other. Carbon policy is now adding a second revenue line on top, since the EU Emissions Trading System with its border adjustment prices the 1.9-tonne CO2 gap per tonne directly into what a European buyer pays for primary steel.

6 minA Detailed Explanation of the Electric Arc Furnace - What It is and How It WorksJames Sword Engineering · 100k+ views

16 minHow Millions Of Old Cars Are Recycled Into New Steel | Massive Car Recycling LineThe Factoran · 1m+ views

Iron and Steel Scrap (U.S. Geological Survey Mineral Commodity Summaries 2026) · Iron and Steel Technology Roadmap (International Energy Agency)

A tramp element is anything that arrives with the scrap and cannot be taken back out, and copper is the one that decides how much of the steel industry can run on recycled metal. Steelmaking strips carbon, silicon, manganese, phosphorus and sulfur by oxidation, because all of them oxidize more readily than iron and report to the slag. Copper is nobler than iron, so blowing oxygen puts iron into the slag before it touches the copper, and there is no economical way to remove it from a liquid bath. Vacuum distillation, sulfide fluxing, and chlorination have all been demonstrated since the 1970s and none of them has reached a production plant. The damage shows up later, in hot rolling: reheating a slab to about 1,200 °C oxidizes surface iron into scale while the copper stays behind, and since copper melts at 1,085 °C the enriched layer under the scale is liquid, wets the austenite grain boundaries, and tears the surface open as the slab is rolled. Tin and antimony make it worse by lowering that layer's melting point, and nickel helps by raising it, which is why the classic fix is a nickel addition of roughly half the copper content. Nickel costs several times what the steel does, so almost nobody uses it.

Two things actually work, and both have a hard ceiling. Sorting works up front: pulling motors, wiring harnesses and radiators before the shredder is why shredded auto scrap runs 0.2–0.4% copper instead of the roughly 1.3% a whole car would give, since a combustion car carries about 20 kg of copper against perhaps 1,500 kg of steel. Grade segregation works too, and No. 1 busheling from stamping plants runs around 0.02–0.06% copper, which is why a sheet mill will pay up for it. Dilution is the third tool and the only one that has no technical limit, but it consumes primary iron units, which is the input scrap was meant to displace. The weakness is that both trends are moving the wrong way at once: an electric vehicle carries 60–83 kg of copper against 20 kg for a combustion car, so future shredded scrap will be dirtier, and as ore-based ironmaking shrinks the pool of clean diluent shrinks with it.

If you are making rebar, merchant bar, or structural sections, ignore this entirely and buy the cheapest shredded scrap available, because those grades tolerate 0.4% copper and more. If you are making exposed automotive sheet, tinplate, or anything deep-drawn, budget for a low-residual iron source before you commit to the mill: a charge of 0.30% copper scrap diluted with copper-free iron has to be one part scrap to two parts iron to reach 0.10% copper, and one part to four to reach 0.06% (that ratio is arithmetic on the numbers here, not a published figure). If you are buying sorting equipment, price it against the scrap grade spread rather than against the copper content, since the only thing that pays is moving a bale from one published grade into a better one. If you are modeling how far a national industry can decarbonize on scrap, model the product mix, because flat products are where the constraint binds and long products are where it does not.

Shredded auto scrap 0.2–0.4% copper · deep-drawing and exposed sheet needs well under 0.1% · No. 1 busheling roughly 0.02–0.06% · rebar and merchant bar tolerate 0.4% and above · one part 0.30% scrap to two parts copper-free iron gives 0.10%, one part to four gives 0.06% (derived here) · copper melts at 1,085 °C against a 1,200 °C slab reheat · about 20 kg of copper in a combustion car against 60–83 kg in an electric one

The ReMA specifications in North America and the EU-27 Steel Scrap Specification write copper and tin limits into the grade definitions, which is how the constraint gets priced. Three plants exist mainly to supply low-residual iron units to electric furnaces: Nucor's direct-reduced iron plant in St. James Parish, Louisiana, Cleveland-Cliffs' hot-briquetted iron plant in Toledo, Ohio, and voestalpine's HBI plant near Corpus Christi, Texas. The European Union's End-of-Life Vehicles Directive requires depollution and component removal before a car is shredded, which is one reason European shredded scrap is cleaner than the US equivalent.

The market prices this constraint openly, as the spread between scrap grades. Against a No. 1 heavy melting composite that averaged $319 per tonne delivered in the US during 2025, prime busheling usually trades tens of dollars a tonne higher, and the gap widens whenever sheet mills are running hard, because that is exactly when the clean scrap is scarce. Imported pig iron and hot-briquetted iron normally cost roughly $100–200 a tonne more than shredded scrap, and a mill buying them is paying that premium purely for the absence of copper. Who pays depends on the product: a rebar mill is indifferent and a flat-rolled mill is not, which is why the two ends of the industry have completely different views on whether scrap supply is tight. Sensor sorting of shredded fractions is the technology most often proposed as the fix, and the test it has to pass is simple, since it only pays if it moves a bale into a higher published grade rather than merely lowering copper by some amount inside the same grade. The long-run consequence is worth stating plainly: a steel industry running entirely on scrap could not make every product it makes today, so some primary ironmaking has to survive, and the argument is about how much rather than whether.

Scrap use in the steel industry (World Steel Association) · Copper contamination in end-of-life steel recycling, developing a new strategy from million-tonnes to milligrams (University of Cambridge)

Recycling aluminum skips the Hall-Héroult cell, which is where essentially all the energy in primary aluminum goes: electrolysis draws 13–15 MWh per tonne of metal, and remelting scrap takes about 5% of the energy needed to refine the same tonne from bauxite. The plant is a casthouse rather than a smelter. Clean heavy scrap goes into a reverberatory side-well furnace, dross and contaminated material go into a rotary furnace under a molten salt cover that keeps oxygen off the metal, and painted or oily stock goes through a delacquering kiln first because organics burning in the melt raise metal loss. Chemistry is then adjusted with primary metal and master alloys, and the heat is cast into ingot, billet, or rolling slab. In the United States in 2025, 3.6 million tonnes of aluminum was recovered from purchased scrap, 56% of it new scrap from factories and 44% old scrap from discarded products, against just 660,000 tonnes of primary metal from the six remaining domestic smelters. Aluminum recovered from old scrap alone was equivalent to about 28% of US apparent consumption.

One loop genuinely closes and the rest does not. A beverage can is a 3xxx-series body with a 5xxx-series end, and melting the two together lands close enough to the body alloy that cans are made almost entirely from recycled metal. Everything else in the shredded post-consumer stream, sold as Zorba and Twitch, mixes cast and wrought pieces whose chemistries are incompatible: a cast alloy such as 356 carries about 9% alloying elements including 7% silicon, while a sheet alloy such as 6111 carries about 3% total and at most 1.1% silicon. Silicon and iron cannot be removed from aluminum at any sensible cost, so the mixed stream can only flow downhill, and US cast parts run about 60% recycled content against about 25% for wrought parts. The other recurring loss is physical: melt loss runs 2–5% on clean heavy scrap and much higher on thin painted stock, and the salt cake from rotary furnaces has to be reprocessed or, in the EU, handled as hazardous waste.

If the feed is a single known alloy from a factory floor, remelt it in a closed loop with the customer who generated it and skip the merchant market entirely, because that scrap is worth nearly ingot price. If the feed is baled used beverage cans, remelting into can sheet works and is the one high-volume closed loop in the industry. If the feed is shredded mixed scrap, plan on making cast alloys and price the output accordingly, since no current sensor-based method sorts shredded aluminum by individual alloy at commercially viable rates. If you are evaluating a sorting technology, get the throughput before the purity: LIBS and XRF are precise but run at 3–8 tonnes an hour, X-ray transmission reaches about 30 tonnes an hour and only splits cast from wrought while leaving roughly 15% cast in the wrought fraction, and optical sorters reach 35 tonnes an hour but aluminum alloys all look the same. If a project's economics depend on selling wrought-grade scrap into wrought products, treat that as the risk in the model rather than an assumption.

Remelting takes about 5% of the energy of primary aluminum, which draws 13–15 MWh/t · US 2025: 3.6 Mt recovered from purchased scrap against 660 kt primary, 56% new and 44% old · old scrap equal to about 28% of US apparent consumption · cast alloy 356 has 9% alloying elements including 7% silicon against 3% and 1.1% max for sheet alloy 6111 · US cast parts about 60% recycled content, wrought parts about 25% · melt loss 2–5% on clean heavy scrap · sorting throughput 3–8 t/h for LIBS and XRF, about 30 t/h for XRT, 35 t/h optical

Novelis is the largest aluminum recycler and runs closed-loop contracts that take stamping scrap straight back from automakers, most visibly for Ford's aluminum-bodied F-150. Constellium and Real Alloy cover the merchant end, remelting purchased scrap into rolling slab and cast ingot. Hydro sells CIRCAL billet with a stated post-consumer recycled content, which is what a carbon-driven customer is actually buying. The six remaining US primary smelters produced 660,000 tonnes in 2025, less than a fifth of what the country's secondary industry recovered.

The margin is the spread between what a remelter pays for scrap and what the ingot or billet sells for, minus melt loss, energy, and dross and salt handling, and melt loss is usually the largest thing the operator controls. US market ingot averaged 180 cents a pound in 2025, up 39% from 2024, so a percentage point of yield is real money on a large heat. The siting logic is different from primary aluminum in a way that matters to anyone building: because remelting uses about 5% of the electricity, a secondary plant does not need the dedicated power contract a potline needs, so it can be put next to the scrap instead of next to cheap power. The awkward fact in the US numbers is that 2.2 million tonnes of aluminum scrap was exported in 2025 against 0.89 million tonnes imported, meaning the country ships out more scrap than it produces primary metal, and the value added by remelting is captured somewhere else. Carbon policy is now pushing back on that, since the EU Emissions Trading System with its border adjustment prices the primary-to-secondary gap directly and low-carbon billet sells at a premium. The forward risk worth watching is the cast-alloy sink. Engine blocks, cylinder heads, and transmission cases were the traditional home for dirty secondary alloy, electric vehicles do not have them, and the large structural castings that replaced them are made to tighter iron and silicon limits, so they cannot absorb the same feed.

6 minHow Are Aluminium Cans Recycled? | How Do They Do It?DCODE by Discovery · 5m+ views

6 minThe Aluminum Can Recycling LoopThink Cans · 10k+ views

Secondary Aluminum Operations, AP-42 Section 12.8 (U.S. Environmental Protection Agency) · Alloy selective optical sorting of mixed post-consumer aluminum scrap streams (Pacific Northwest National Laboratory)

Copper scrap splits into two routes that have almost nothing in common. Clean scrap of known alloy is direct-melted: it goes straight into a brass mill or a wire-rod mill without ever being refined, which is where about 80% of the copper recovered from scrap in the United States ends up. Everything mixed, coated, or low-grade goes to a secondary smelter instead, where a blast or rotary furnace makes black copper at 70–80%, a converter takes it to 80–90%, fire refining in a reverberatory furnace reaches about 99%, and electrolysis finishes at 99.99% cathode that is indistinguishable from primary metal. Pretreatment sits in front of both routes and is most of the labor: insulated wire goes through a chopping line where granulators and air tables separate copper granules from plastic, oily turnings are dried in a rotary kiln, and mixed metals are separated by sweating, where furnace temperature is staged so each metal liquefies in turn. Copper recovered from old scrap in the US was about 160,000 tonnes in 2025 and from new manufacturing scrap about 760,000 tonnes, together supplying roughly 30% of US copper supply.

Nothing is lost in the loop, which is the reason this industry is older than most of the metals on this sheet. Cathode made from scrap sells at the exchange price alongside primary metal, and copper's value per tonne is high enough to pay for hand sorting, long-distance freight, and the theft losses that come with it. The constraint is alloys rather than purity. Brass and bronze scrap carries zinc, tin, and lead that electrical copper cannot have, wire-drawing copper needs 99.90% minimum and close to 100% IACS conductivity, and trace bismuth, antimony, and selenium at parts-per-million levels are enough to make rod crack during drawing. So brass scrap goes back into brass and stays there, and only the smelting-and-electrolysis route can pull those elements out. Capacity is the other weakness: the US operates two secondary copper smelters and four secondary refineries, so low-grade material either travels a long way or leaves the country.

If the scrap is clean and you know the alloy, sell it for direct melt and never let it enter a smelter, because refining charges and metal loss are pure deductions from something already at specification. If the material is insulated wire, buy a chopping line rather than paying a smelter to burn the plastic off, since granulation recovers the copper mechanically and does not create a combustion emissions problem. If the feed is mixed, plated, or full of tramp alloys, secondary smelting is the only route that gets back to cathode, and you should price it as a fee-for-service tolling arrangement rather than a sale. If you are considering building capacity, note that the two ends of this business are entirely different: a chopping line costs single-digit millions and hundreds of companies run one, while a secondary smelter costs hundreds of millions and a handful of companies worldwide operate them.

US 2025: 160 kt of copper recovered from old scrap and 760 kt from new scrap, together about 30% of US copper supply · brass and wire-rod mills take roughly 80% of copper recovered from scrap · secondary smelting path is 70–80% black copper, 80–90% after converting, about 99% after fire refining, 99.99% cathode after electrolysis · electrical copper needs 99.90% minimum and close to 100% IACS · US has 2 secondary smelters and 4 secondary refineries · LME grade A cash averaged 440 cents a pound in 2025 and COMEX high grade 480 cents

Aurubis runs the reference plants: Lünen in Germany is the long-standing secondary copper smelter, and its Richmond, Georgia site is the first multimetal recycling plant of that type built in the United States. Boliden's Rönnskär smelter in Sweden and Aurubis Beerse in Belgium take similar complex feed. Scrap is bought against ReMA grade names that the trade still uses daily, including Barley for bare bright wire, Birch and Cliff for No. 2, and Honey for yellow brass. On the direct-melt side, US brass mills and wire-rod mills are the volume buyers, which is why the grade a yard can produce decides which customer it can sell to.

Copper scrap is priced as a discount to the exchange, and that discount is the entire business model. Clean No. 1 bare bright typically settles in the mid-90s as a percentage of the COMEX or LME price, No. 2 grades in the high 80s, and insulated wire on assayed recoverable copper content less the cost of chopping. With LME grade A cash averaging 440 cents a pound in 2025 and COMEX high grade 480 cents, a few points of discount is tens of dollars a tonne, so accurate grading is where a yard makes or loses money. It is also a working-capital business rather than a processing one: material is bought on one date and sold weeks later, so an unhedged yard is carrying a copper position it did not choose, and the operators that survive price declines are the ones that hedge on purchase. Secondary smelters run the same fee model as primary ones, deducting treatment and refining charges and returning a payable percentage of contained metal, which means the smelter's revenue is largely independent of the copper price while the collector's is not. Two forces are pulling more capital into this stream: the metal price itself, and government interest in domestic supply, since scrap is the cheapest new tonne available to anyone able to organize collection.

5 minMini copper scrap melting smelting refining furnace and eletro refining processMetalcess · 100k+ views

16 minHow Millions of Feet of Copper Wire Are Recycled In Factory – Massive Copper Recycling LineThe Factoran · 100k+ views

Copper (U.S. Geological Survey Mineral Commodity Summaries 2026) · Secondary Copper Smelting, AP-42 Section 12.9 (U.S. Environmental Protection Agency)

A starting-lighting-ignition battery weighs 15–20 kg and holds 8–12 kg of lead, roughly 40% of it as cast grid alloy and 60% as lead oxide and sulfate paste, inside a polypropylene case filled with sulfuric acid. Recycling starts in a hammer mill that breaks the whole battery, after which a series of water baths separates the pieces by density: grid metal sinks, paste settles as a slurry, polypropylene floats, and the acid is drained off. The paste is desulfurized with sodium carbonate, which converts lead sulfate to lead carbonate and produces sodium sulfate that is sold to detergent and glass makers. Smelting follows in a rotary or reverberatory furnace with coke and scrap iron as reductants, giving lead bullion that then goes to refining kettles where copper is drossed off, and arsenic, antimony, and tin are oxidized out with air and caustic before the metal is alloyed and cast. Recovery is 98–99% of the lead in the feed, and the polypropylene is washed, ground, pelletized, and molded into new cases, so the case closes the loop alongside the metal.

This is the highest recycling rate of any consumer product anywhere, about 99% in the US, and none of it is because the metallurgy is clever. One material, one format, a known return path through the shop that sold it, a core charge that makes the customer bring it back, and enough lead value per unit to pay for the trip. The result is that the US has had no primary lead refinery since the last one closed in 2013, and about 1.0 million tonnes of secondary lead in 2025 covered 70% of domestic consumption. The weakness is the metal itself. Lead is a potent neurotoxin, smelting it releases fume and dust, and the industry carries a permitting and liability burden that has nothing to do with whether the process works. When the EPA tightened its air standard for secondary lead smelting in 2012 the US industry consolidated to roughly a dozen plants, and two closures left contaminated sites behind: Exide's Frisco, Texas smelter in 2012 and its Vernon, California plant in 2015. Where the same batteries are broken by hand in the informal sector, which is common across South Asia, Africa, and Latin America, the result is measurable blood-lead elevation in the surrounding population.

If you are designing a producer responsibility scheme for anything, study this one first and copy the mechanism rather than the target, because the core charge and the retailer take-back obligation are what produce a 99% return rate and no collection target does that on its own. If you are underwriting a secondary lead smelter, treat the air permit, the closure bond, and the long-tail soil liability as the main risks and the furnace as the easy part. If you are looking at lithium-ion recycling economics, use this as the benchmark and note which of the five conditions are missing, since lithium packs differ in format, chemistry, hazard class, and return path all at once. If you are sourcing lead, understand that in the US you are buying secondary metal whether you asked for it or not, because there is no domestic primary alternative.

US 2025: about 1.0 Mt of secondary lead, 70% of apparent consumption, worth $2.4 billion · no primary lead refinery in the US since 2013 · lead-acid batteries about 67% of US lead consumption and roughly 99% recycled · an SLI battery is 15–20 kg with 8–12 kg of lead, about 40% grid alloy and 60% paste · 98–99% of the lead recovered · North American lead averaged 106 cents a pound in 2025 · spent batteries trade at roughly $300–700 per tonne of battery

Clarios, Ecobat, and Gopher Resource run the largest North American secondary lead operations, and Clarios both makes and recycles batteries, which is as close to a genuinely closed industrial loop as this sheet contains. About 40 US states ban lead-acid batteries from landfill and require retailers to accept a return, which is the legal half of the mechanism the core charge covers commercially. The Basel Convention's technical guidelines on used lead-acid batteries exist because the informal recycling problem is a transboundary one, with spent batteries exported to countries where breaking them by hand is legal or unpoliced.

The economics are the simplest on this sheet: a spent battery is a bag of lead with a known assay, and it trades at roughly $300–700 per tonne of battery against a North American lead price that averaged 106 cents a pound in 2025. A tonne of batteries carries around 600 kg of lead, so the scrap price is a discount of half or more to contained metal value, and that gap pays for breaking, desulfurization, smelting, refining, and the emissions controls. Byproducts matter more here than in most metal recycling: polypropylene granulate, sodium sulfate, and antimony recovered from the bullion are all sold, and a smelter that flares them away gives up real margin. The barrier to entry is regulatory rather than technical, since the furnace is old technology and the permit is not, which is why capacity has consolidated into a small number of large plants rather than fragmenting the way scrap yards do. Two things would change the picture. Lead demand is tied to the 12-volt battery, which every vehicle still has including electric ones, so the feed is stable; and if that ever stops being true, the collection system that makes the 99% rate possible loses the economics holding it together.

3 minLead Battery RecyclingHowStuffWorks · 100k+ views

5 minHow It's Made, Recycling Car Batteries.Moe Sal · 100k+ views

Lead (U.S. Geological Survey Mineral Commodity Summaries 2026) · Lead acid battery recycling for the twenty-first century (Royal Society Open Science)

This is the shared back end that spent autocatalysts, electronic scrap, spent petroleum and chemical catalysts, jewelry sweeps, and plating solutions all funnel into, and it exists because separating six chemically similar platinum-group elements from each other is hard enough that only a handful of plants do it. A lot arrives, is crushed, milled, and blended until it is homogeneous, and is then sampled, because the material has no specification and the assay is what both sides settle against. Recovery goes one of two ways: smelting with an iron, copper, or lead collector that dissolves the precious metals into a metallic phase while everything else reports to slag, or direct leaching in aqua regia or another chloride medium. The resulting solution or concentrate then goes through a separation train of solvent extraction, selective precipitation, and ion exchange that pulls the elements apart one at a time, ending in metal sponge at 99.95% or better. Gold takes a shorter path, chlorinated to about 99.5% by the Miller process and finished to 99.99% by Wohlwill electrolysis. Grades are what make any of this worth doing: spent autocatalyst runs roughly 1,000–2,000 grams per tonne of combined platinum-group metal against 3–6 g/t in the ore that primary refineries are fed.

Recycled metal is chemically identical to primary metal and sells into the same market at the same price, which is why this is one of the few streams on the sheet that has never needed a policy behind it. The volumes are real: Johnson Matthey put 2025 secondary supply at 48.1 tonnes of platinum against 172.9 tonnes primary, 97.4 tonnes of palladium against 205.0 tonnes, and 9.7 tonnes of rhodium against 21.8 tonnes, so roughly a fifth to a third of world supply of each metal already comes from scrap. The weakness is that the feed is procyclical in a way the refiner cannot control. Collectors hold material back when prices fall and release it when they rise, so secondary supply is smallest exactly when the market is tightest, and a refinery sized for peak flow runs below capacity for years at a stretch. The second weakness is time: four to eight weeks typically pass between delivery and settlement on a platinum-group lot, and somebody is financing the metal for all of it.

If you have precious metal in a waste stream, get the assay protocol agreed before the metallurgy, because the sampling method and who witnesses it decide the settlement more than the recovery percentage does. If your lots are small, sell to an aggregator instead of a refiner, since refinery lot minimums and fixed treatment charges make a small parcel uneconomic no matter what it assays. If you are choosing between smelting and leaching for a given feed, take smelting when the matrix is a ceramic or a mixed oxide and the metals are dispersed, and leaching when the material is already fine and the metals are near the surface, because leach kinetics fall off badly on a dense monolith. If you are building recovery capacity, build collection and preprocessing rather than a separation train, since the refining step is a commodity service with a handful of incumbents and the margin sits upstream of it. And if you are modeling precious-metal supply, treat secondary supply as price-elastic and lagging, not as a stable base.

2025 secondary supply against primary: platinum 48.1 t versus 172.9 t, palladium 97.4 t versus 205.0 t, rhodium 9.7 t versus 21.8 t · about 140 t of platinum and palladium recovered globally from scrap in 2025 · US autocatalysts alone gave about 50 t of palladium and 8.6 t of platinum · spent autocatalyst runs 1,000–2,000 g/t combined platinum-group metal against 3–6 g/t in ore · refined sponge at 99.95%, gold at 99.5% after the Miller process and 99.99% after Wohlwill electrolysis · four to eight weeks from delivery to settlement · 2025 price averages about $1,200/oz platinum, $1,100/oz palladium, and $5,800/oz rhodium

Sabin Metal in New York is the reference independent refiner for spent petroleum and chemical catalyst, a feed with a different matrix from autocatalyst and much larger lot sizes. Metalor, Valcambi, and Asahi Refining cover the gold and silver side, and the LBMA Good Delivery list is the standard a bar has to meet before a bank will accept it without re-assay. The International Precious Metals Institute publishes the sampling and settlement practice that most of this trade runs on. On the platinum-group side the separation capacity sits with a short list of integrated producers who refine primary and secondary feed through the same trains, which is why a recycler's access to refining is usually a contract rather than an asset.

Refiners charge for the service rather than taking a position in the metal: a treatment and refining charge per lot, plus a returnable percentage that hands back 90–98% of assayed content depending on the feed and the lot size, with the balance kept as refining loss. That structure means a refiner's revenue tracks throughput and is largely indifferent to the metal price, while a collector's revenue is entirely exposed to it, and the two therefore behave completely differently in a downturn. The working capital is the other half of the business. Metal is locked up between delivery and settlement, so refiners advance funds against assay or lease metal back to the customer, and the financing terms often decide who wins a contract rather than the recovery percentage. Capital intensity is high and the permits are harder than the plant, so new separation capacity is rare and the incumbents have been the incumbents for decades. For anyone entering, the useful conclusion is that the margin is upstream: assay accuracy, settlement speed, and inventory financing are what collectors and aggregators compete on, and the metallurgy at the far end is a purchased service with published terms.

6 minHow to Refine Precious Metals - Step One: ConcentrationInternational Precious Metals Institute (IPMI) · 50k+ views

4 minHow to Refine Precious Metals - Hydrometallurgy: Part 1- LeachingInternational Precious Metals Institute (IPMI) · 50k+ views

PGM market report, May 2026 (Johnson Matthey) · Platinum-Group Metals (U.S. Geological Survey Mineral Commodity Summaries 2026)

A retired electric-vehicle pack weighs 300–600 kg, holds 40–100 kWh, and moves as class 9 dangerous goods, so the first real cost in this chain is freight rather than chemistry. Packs reach a collector three ways: warranty and crash returns through dealers, fleet retirements, and consumer cells through take-back points. The first two supply most of today's volume, because the vehicles built in the late 2010s are mostly still on the road. Diagnostics decide what happens to a pack once it arrives. State of health is present capacity divided by original capacity, and a pack usually comes out of a vehicle somewhere between 70% and 80%; measuring that properly means a full charge and discharge at roughly C/3, which takes six to ten hours per module, so grading a pack module by module ties up a test rig for days unless the operator can read the battery management system's own logged history or use a faster impedance or pulse measurement. Second life takes the modules that grade well, rebuilds them into a stationary system, and sells the remaining capacity into applications where weight and volume do not matter.

A module at 75% of its original capacity still holds real energy, and stationary storage has no weight or volume limit, so the material genuinely has value left in it. Four things get in the way. Grading costs money before anything is sold, commonly $10–30/kWh in published estimates, and packs arrive in dozens of incompatible mechanical and communication formats, so integration is one-off engineering work every time. The repurposer becomes the manufacturer for liability purposes, which is what UL 1974 exists to structure, and an insurer pricing fire risk on cells with an unknown thermal history charges for that uncertainty. The fourth problem is the competition: Chinese LFP cells sold under $50/kWh in 2025 against an average pack price near $115/kWh in 2024, and a new cell comes with a cycle warranty behind it that a used module cannot carry.

If you have a large fleet of identical packs with logged battery-management data and one buyer for the output, second life is worth doing, because the two costs that usually kill it, grading and integration, both fall with sameness. If the packs are mixed models bought at auction, do not: you will pay to test each one and then build a one-off system around it. If the chemistry is LFP, the case is stronger, since an LFP pack has almost no recoverable metal value and sending it to a recycler will cost you a gate fee rather than pay you. If the chemistry is NMC or NCA, run the comparison against recycling the pack now, because the nickel and cobalt are worth money today and second life defers that revenue by five to ten years while you carry storage, insurance, and commodity risk. As a default, assume a second-life system has to land under roughly $50–80/kWh installed to beat a new LFP system with a warranty.

Pack 300–600 kg and 40–100 kWh · retired at 70–80% state of health · a full capacity test at about C/3 takes 6–10 hours per module · grading commonly $10–30/kWh before any repackaging · Chinese LFP cells under $50/kWh in 2025 against an average pack price near $115/kWh in 2024 · a second-life system needs to land under roughly $50–80/kWh installed to compete · a 500 kg NMC pack yields roughly 150 kg of black mass.

The Johan Cruijff ArenA in Amsterdam runs a 3 MW storage system built from 148 Nissan Leaf packs, and B2U Storage Solutions operates a site in Lancaster, California assembled from used Honda and Nissan packs. Connected Energy in the UK and 4R Energy, Nissan's joint venture with Sumitomo, are the longest-running operators; UL 1974 is the standard a repurposer is evaluated against.

Second life is a spread business: take or buy a pack, spend money grading and repackaging it, and sell stationary capacity. Both ends of that spread moved the wrong way over the last five years, because new cell prices fell fast while grading and integration are labor and did not fall with them. The feed side is thinner than the forecasts assumed as well, since most packs available today come from crashes and warranty claims rather than age-out, and a crashed pack is exactly the one nobody wants to repurpose. A recycler, meanwhile, pays cash now: a 500 kg NMC pack yields roughly 150 kg of black mass, which at $2,500–4,000 a tonne is $375–600, plus a few hundred dollars of copper, aluminum, and steel. That is a small number, but it is certain, which is why a pack worth repurposing is usually also worth recycling. The carbon argument is real, since making a new cell emits roughly 60–100 kg CO2e per kWh and reusing a module avoids almost all of it, but outside a few European procurement rules nobody pays for that directly. If you are underwriting one of these businesses, ask whether identical packs are contracted at volume and whether a buyer has signed for the output at a price, because without both, the diagnostics cost alone consumes the margin.

10 minSecond Life Batteries: The Solution for Sustainable Mobility?DW REV - Mobility & Innovation · 5k+ views

6 minWhy Second Life Battery Projects Are So Important For EveryoneTransport Evolved · 10k+ views

Electric-vehicle battery second-life and recycling pathways: How economics depend on chemistry, processing, and application (National Laboratory of the Rockies) · Lithium-ion battery second life: pathways, challenges and outlook (Frontiers in Chemistry)

Mechanical preprocessing is the shared front end of battery recycling, and its job is to turn a pack into a powder that costs a fraction as much to ship and can be fed to a chemical plant. A pack arrives at some unknown state of charge, so the first step is discharge: a resistive load takes cells below about 2 V each in thirty minutes to several hours, while soaking cells in brine, which small operators still use, takes days and leaves a salty contaminated effluent behind. Next the pack is opened and the cooling plates, wiring harness, bolted structure, and modules come out, which is manual work running roughly two to eight hours per pack because no two pack designs are alike. Modules or bare cells are then shredded, either under an inert atmosphere or wet, both of which exist to keep residual charge from starting a fire and to capture the electrolyte and the hydrogen fluoride it forms with moisture. Screening, magnetic separation, eddy-current separation, and air classification then pull out the steel casing, the copper and aluminum foil and busbars, and the plastics. What is left is black mass: cathode and anode powder, binder, and residual electrolyte, at roughly 25–40% of the original pack mass.

The logistics saving is the whole argument. A pack is mostly steel, aluminum, plastic, and void, so shredding it near where it was collected and shipping only the powder to a central refinery is the single largest cost reduction available in the chain, and it is what makes a network of regional shredders feeding one chemical plant work. The weaknesses are labor, safety, and the fact that nothing is actually separated chemically. Manual disassembly is the most expensive step in the front end and resists automation because pack designs change every model year. Shredding a charged cell can start a fire, so inert or wet processing plus dust and offgas capture is mandatory rather than optional. And blending chemistries into one powder is irreversible: aluminum and fluorine carried into the black mass both cause trouble in a downstream leach, and a mixed-chemistry powder can no longer be fed to a direct recycling process at all.

Shred almost always, and do it ahead of any hydrometallurgical or direct route. If you are building a network, put the shredders close to the collection points and the refinery in one place, because black mass concentrates roughly three to four times the value per tonne of the pack it came from. If your feed is production scrap from a single gigafactory line, keep the streams segregated by chemistry from the start, since that is the only realistic way to preserve the option of direct recycling. If the packs are damaged, wet, or of unknown state of charge, skip manual disassembly and use wet or inert shredding at the module level, which is safer and much cheaper per pack. If you can sell whole modules to a smelter that will take them, mechanical preprocessing can be skipped entirely, and for hazardous or unsortable feed that is often the right answer.

Discharge below about 2 V per cell in 30 minutes to several hours · manual pack disassembly roughly 2–8 hours per pack · black mass is 25–40% of pack mass and about three to four times the value per tonne · NMC black mass runs roughly 20% nickel, 5% cobalt, and 3.5% lithium, worth about $6,000–8,000 a tonne contained · payables 60–75% of contained nickel and cobalt with lithium often unpaid, so NMC black mass sells at roughly $2,500–4,000 a tonne · LFP black mass $0–500 a tonne, and LFP packs usually move on a gate fee of $1,000–3,000 a tonne · US shredding capacity around 155,000 t in 2023 against roughly 90,000 t of available feed.

Li-Cycle's spoke plants shred submerged in liquid and were built specifically to decouple shredding from refining; Redwood Materials takes whole packs and consumer cells at its Nevada site; Cirba Solutions, Ecobat, Altilium in the UK, Tozero in Germany, and SungEel HiTech in Korea run comparable front ends. Duesenfeld in Germany is the process worth knowing for one detail: it recovers the electrolyte by vacuum distillation instead of burning it, which almost no other route does. Metso and Andritz sell the shredding and separation equipment as a packaged line.

Black mass trades as a payable percentage of contained metal, not at a headline price, so the seller carries the metal cycle whether it wants to or not. Take a tonne of NMC black mass at roughly 20% nickel, 5% cobalt, and 3.5% lithium: that is about 200 kg of nickel, worth roughly $3,200 at $16,000 a tonne, plus 50 kg of cobalt and 35 kg of lithium, which is around 185 kg of lithium carbonate equivalent. Add those up at recent prices and contained value lands near $6,000–8,000 a tonne, of which the seller typically receives 60–75% of the nickel and cobalt and frequently nothing for the lithium, so the check is roughly $2,500–4,000. Now do the same for LFP: about 3% lithium, or 160 kg of carbonate equivalent worth $1,500–2,500, and iron and phosphate worth almost nothing, which is why LFP black mass trades at $0–500 a tonne and LFP packs usually move on a gate fee of $1,000–3,000 a tonne instead. That gap is the largest change in this industry's economics in the past five years, because LFP went from roughly 15% of global electric-vehicle cell output in 2020 to more than half by 2025. The second economic fact is that shredding is cheap and refining is not, so the shredding side got built first and got overbuilt: US shredding capacity was around 155,000 t in 2023 against roughly 90,000 t of feed actually available, which compressed gate fees and put several operators out of business. Policy now moves the price as well, since the EU classified black mass as hazardous waste effective 5 March 2025, restricting export outside the OECD, while China lifted its import ban on 1 August 2025.

16 minHow Millions Of Batteries Are Recycled Every Day - Inside Battery Recycling Mega FactoryThe Factoran · 100k+ views

7 minRecycling of Lithium ion Battery: Shredding and Separating LineHenan Recycle Env Protection Equipment Co., Ltd · 50k+ views

Review on Preprocessing Strategies, Deactivation, Thermal Safety, and Future Perspectives in Lithium-Ion Battery Recycling (National Energy Technology Laboratory) · Lithium-Ion Battery Recycling (US EPA)

Pyrometallurgy feeds whole cells, modules, or black mass into a shaft or electric furnace at 1,300–1,500 °C and lets high-temperature metallurgy do the sorting. The separator, binder, and electrolyte burn off, the graphite acts as a reductant and part of the fuel, and the charge splits into two phases: a metallic alloy or matte carrying nickel, cobalt, copper, and iron, and a slag carrying aluminum, manganese, and lithium. The alloy goes on to a conventional leach and solvent-extraction refinery, which is why the back half of this route is ordinary base-metal hydrometallurgy and produces the same nickel and cobalt sulfates a mine-fed refinery does. Recovery is 90–98% for nickel, cobalt, and copper, and 0% for lithium unless a separate step is added to leach it back out of the slag, which typically returns 40–70%. Graphite and electrolyte are destroyed by design, so nothing comes back from either. The organics in the cells supply a large share of the process heat, but the fluorine released from the electrolyte and binder makes offgas scrubbing mandatory, so a smelter without an existing gas-cleaning train cannot simply start taking cells.

Feed tolerance is the reason this route exists. A smelter takes mixed chemistries, unsorted packs, wet cells, damaged cells, and modules nobody can economically take apart, with no discharge step and no disassembly line, which removes the most labor-intensive and most dangerous work in the front end. It is also the only route that will accept material pulled from crashed or burned vehicles. What it costs is everything that reports to the slag. Lithium, aluminum, and manganese go there, and lithium in slag is a low-grade feed of roughly 1–3% Li2O against about 6% in a spodumene concentrate, so recovering it is a second hydrometallurgical plant treating poor ore. Energy use and direct CO2 are also several times the ambient-temperature route's, and the failure mode to watch is regulatory rather than technical: the EU battery regulation requires 50% lithium recovery from 2027 and 80% from 2031, and a smelter without slag treatment cannot meet either.

Pick pyrometallurgy when the feed is heterogeneous and the value is in nickel and cobalt. If you are taking end-of-life packs from a mixed vehicle fleet, packs of unknown state of charge, or anything with fire risk, this is the route that works, and the discharge and disassembly labor you avoid is real money. Pick hydrometallurgy instead when the feed can be sorted and you need the lithium, because 80–95% lithium recovery is where that route pays and 0% is not. If the feed is LFP, a smelter is the worst answer available, since LFP's only real value is the lithium that reports to slag, so treat it as a disposal service priced on a gate fee. The strongest case for pyro is not a greenfield plant at all: if you already own a nickel or copper smelter with permits, fluorine-capable gas cleaning, and a refinery behind it, adding a battery line is an incremental project, and the marginal cost per tonne is far below what a new plant would need to charge.

1,300–1,500 °C in a shaft or electric furnace · nickel, cobalt, and copper recovered at 90–98% · lithium recovery 0% without slag treatment and 40–70% with it · slag runs roughly 1–3% Li2O against about 6% in spodumene concentrate · graphite and electrolyte destroyed, so 0% recovery on both · EU battery regulation requires 50% lithium recovery from 2027 and 80% from 2031 · adding a battery line to an existing smelter costs tens of millions against $200–500M for a greenfield hydrometallurgical refinery.

Umicore's Hoboken plant in Belgium is the reference operation, with a dedicated battery line rated near 7,000 t/yr, roughly 20,000 electric-vehicle packs, running alongside the smelter that already processes electronic scrap and spent autocatalysts. Glencore's Sudbury complex in Ontario feeds battery material into an existing nickel smelter and refines the alloy at Nikkelverk in Norway; Nickelhütte Aue in Germany and Accurec, which pairs a thermal step with hydrometallurgy, are the other European operators.

This is the economics of an existing asset rather than of a process. A smelter's fixed costs are already carried by its primary business, so battery feed is judged on marginal terms: does the payable metal in the charge cover fuel, reagents, gas cleaning, and the throughput given up on other material. That test is far easier to pass than the one facing a greenfield chemical plant, which has to repay $200–500M against a feed stream that does not yet exist at scale, and it is why adding a battery line to an existing smelter costs tens of millions instead. It also means capacity does not grow smoothly, because it depends on somebody else having spare furnace hours. Revenue comes almost entirely from nickel, cobalt, and copper, so the route is exposed to exactly the same prices as primary supply and is most profitable when a mine would also be profitable. For low-value or hazardous feed the gate fee is the business rather than a supplement to it, typically $1,000–3,000 a tonne, and for LFP it is the entire revenue line. The strategic argument for keeping this capacity is not that it recovers metal cheaply, because it usually does not; it is that nothing else will take the material at all, and that capability has a price.

2 minUmicore FAQ - How does the battery recycling process work?Umicore · 1k+ views

8 minPyro vs Hydro - Recycled EV ContentRecycled EV Content · under 1k views

Pyrometallurgical Approach to Extracting Valuable Metals from a Combination of Diverse Li-Ion Batteries' Black Mass (ACS Sustainable Resource Management) · Global Regulations for Sustainable Battery Recycling: Challenges and Opportunities (Argonne National Laboratory)

Hydrometallurgy dissolves black mass in sulfuric acid with hydrogen peroxide as a reductant at roughly 60–90 °C for one to four hours, which puts better than 95% of the cathode metals into solution. Iron, aluminum, and copper are then dropped out as impurities, and the pregnant liquor runs through a solvent-extraction train, usually D2EHPA to pull manganese and Cyanex 272 to make the cobalt-nickel split. Each separated stream crystallizes as a battery-grade sulfate. Lithium stays in solution through the whole circuit and comes out last, precipitated from the raffinate as carbonate with soda ash or converted to hydroxide. Recovery is 95–98% for nickel, cobalt, and manganese and 80–95% for lithium, with copper and aluminum recovered as byproducts; graphite usually leaves in the leach residue and the electrolyte is already gone by this point. The products are the same chemicals a mine-fed refinery makes, so they drop into an existing precursor plant with no qualification beyond the usual purity spec, which is the main reason most new Western capacity is being built on this route.

The recoveries are the best of any commercial route, the plant runs near ambient temperature so energy use is a fraction of a smelter's, and lithium comes back rather than going to slag. The costs are reagents, salt, and water. Every tonne of black mass consumes roughly 1.5 tonnes of sulfuric acid, a few hundred kilograms of peroxide, and enough caustic and soda ash to neutralize what the acid did, and it leaves behind 10–20 cubic meters of process water and several tonnes of sodium sulfate per tonne of metal produced. Sodium sulfate sells for $50–100 a tonne into a low-value market if it sells at all, so it is frequently a disposal line rather than a revenue line, and effluent treatment is often the permit-limiting step for a new site. The circuit is also tuned to a nickel and cobalt grade, so an LFP or sodium-ion feed does not just reduce the margin, it inverts it. Fluorine from residual LiPF6 attacks equipment and has to be managed, and aluminum carried through from sloppy shredding consumes acid for nothing.

If the feed is nickel-cobalt production scrap or sorted end-of-life NMC and NCA black mass, this is the right route and the nickel and cobalt pay for the plant. If the feed is mixed, wet, or damaged, a smelter will take it and a hydrometallurgical plant will not, so send it there instead. If the feed is LFP, do the arithmetic before committing capital, because there is no nickel or cobalt to recover and the lithium alone rarely covers the reagent bill. If a recovery mandate applies to you, this route is the straightforward way to meet the EU battery regulation's 50% lithium recovery from 2027 and 80% from 2031, which is a real reason to pick it over pyrometallurgy in Europe. Before sizing anything, contract the feed: the sector's write-offs came from plants built against forecasts of end-of-life packs that have not arrived, not from chemistry that did not work.

Leach in sulfuric acid with peroxide at 60–90 °C for 1–4 hours, dissolving better than 95% of the cathode metals · recovery 95–98% for nickel, cobalt, and manganese and 80–95% for lithium · roughly 1.5 t of sulfuric acid per tonne of black mass, plus peroxide, caustic, and soda ash, for about $600–900 a tonne in reagents · 10–20 cubic meters of process water per tonne of black mass and several tonnes of sodium sulfate per tonne of metal, worth $50–100 a tonne if it sells · all-in processing roughly $1,500–3,000 per tonne of black mass · a commercial refinery costs $200–500M at 20,000–50,000 t/yr · EU targets of 50% lithium recovery from 2027 and 80% from 2031.

Redwood Materials in Nevada and South Carolina is the largest US operation and takes whole packs rather than buying black mass; Ascend Elements runs the Apex plant in Kentucky and a lithium line in Covington, Georgia; Cirba Solutions and American Battery Technology Company are the other US names. Li-Cycle built the shredding network but ran out of money mid-construction on its Rochester refining hub, which Glencore took control of. In Europe, Altilium, Tozero, and Librec are the credible startups, and in Asia SungEel HiTech, CATL's Brunp, GEM, and Huayou Cobalt run most of the world's operating capacity.

Reagents are the operating cost that distinguishes this route, and they are worth pricing out rather than waving at. Per tonne of black mass, roughly 1.5 tonnes of sulfuric acid at $100–150 a tonne is $150–225, a few hundred kilograms of peroxide at around $600 a tonne is another $180, and the caustic and soda ash needed to neutralize the circuit and precipitate lithium add $250–450, so the reagent bill lands near $600–900 a tonne before labor, power, or effluent handling. Add those and all-in processing typically runs $1,500–3,000 a tonne of black mass. Against that, NMC black mass sells at roughly $2,500–4,000 a tonne, which tells you immediately that this route only works on nickel-cobalt feed and only when the refiner captures the difference between payable and contained metal. Capital is the second problem: a full refinery is a chemical plant with a solvent-extraction train, a crystallization circuit, and effluent treatment, at $200–500M for 20,000–50,000 t/yr, and fixed costs dominate, so a plant running at half utilization has roughly double the unit cost. The sodium sulfate stream deserves a line in the model of its own, because several tonnes per tonne of metal at $50–100 a tonne is either a small credit or a disposal charge depending on whether a buyer exists within trucking distance. If you are underwriting one of these plants, the three questions that matter are where the tonnes come from, what chemistry they are, and who takes the salt.

15 minRecycling Lithium-Ion Battery: How Does It Work?Engineering with Rosie · 50k+ views

4 minMetso Battery black mass recycling processMetso · 1k+ views

Primary material supply configurations and domestic recycling for cost-effective battery material production in the US (Lawrence Berkeley National Laboratory) · Financial viability of electric vehicle lithium-ion battery recycling (iScience)

Direct recycling keeps the cathode crystal structure intact instead of dissolving it back to elements, so the product is cathode powder rather than metal salts. The cathode coating is first separated from its aluminum foil, either by dissolving the PVDF binder in a solvent, by a thermal step that burns it off, or by froth flotation that also splits the cathode powder from the graphite it was shredded with. The recovered powder has lost lithium and some of its layered order during service, so it is relithiated, typically hydrothermally in a lithium hydroxide solution at 180–220 °C or in a molten lithium salt, and then annealed at 700–900 °C to restore the structure. Better than 90% of the active material comes through, and a well-run relithiation returns capacity within a few percent of virgin material. Because the separation and re-synthesis steps of a hydrometallurgical circuit disappear, energy and reagent cost run roughly 30–50% below that route. What does not disappear is the feed requirement: the process cannot fix a powder that is two chemistries mixed together, because there is no single anneal that produces a defined cathode from a blend.

The output is worth several times what black mass is worth, since cathode active material sells at roughly $10,000–25,000 a tonne for nickel-rich NMC against $2,500–4,000 a tonne for the black mass it came from, and the emissions and reagent savings are genuine rather than accounting. Two things hold it back. The first is feed: it needs a single chemistry of known composition and low contamination, which in practice means production scrap from one factory line, and the current vehicle fleet supplies mixed-chemistry packs that a shredder blends into one powder. The second is obsolescence. Cathode formulations change every few years, so material recovered from a 2016 cell comes back as NMC111, which no cell maker is buying in 2026, and a recovered powder still has to pass 12–24 months of qualification at a customer before it is a product. Upcycling, which adds nickel during relithiation to convert an older cathode to a current one, is the research answer to that and is not yet commercial.

If your feed is trim scrap and reject electrodes from one gigafactory line, direct recycling is the right route and it is the only one that captures the full value of the cathode you already paid to make. If your feed is end-of-life packs from a mixed fleet, do not plan on it, because the sorting the process assumes does not exist and paying for it separately usually costs more than the value uplift. If the chemistry is LFP, look here first: hydrometallurgy cannot pay on LFP and a smelter will not recover anything from it, so relithiating LFP back to LFP is the one route with an economic case. If you are modeling this, count the sorting cost inside the process cost rather than assuming somebody else pays it, and check whether a cell maker has actually qualified the output rather than tested it. As a default, treat published direct-recycling economics as valid only for the fraction of the feed that is genuinely single-chemistry scrap, and ask what that fraction is.

Relithiation at 180–220 °C hydrothermally or in molten salt, then annealing at 700–900 °C · better than 90% of active material recovered, at capacity within a few percent of virgin · energy and reagent cost roughly 30–50% below the hydrometallurgical route · cathode active material sells at roughly $10,000–25,000 a tonne for nickel-rich NMC against $2,500–4,000 a tonne for black mass · 12–24 months of qualification at a customer before the powder is a product · production scrap runs 5–10% of gigafactory output at maturity and 20–30% during a ramp.

The US Department of Energy's ReCell Center, led by Argonne National Laboratory, is the reference program and published most of the relithiation and cathode-healing work the startups build on. Princeton NuEnergy runs a plasma-assisted separation and regeneration process at pilot scale in Texas; OnTo Technology in Oregon has worked on direct cathode recovery longer than almost anyone; Ascend Elements' Hydro-to-Cathode process is the closest commercial relative, though it dissolves the cathode and rebuilds precursor rather than preserving the crystal.

The value case is straightforward and the feed case is why nobody has closed it yet. Cathode active material at $10,000–25,000 a tonne against black mass at $2,500–4,000 means a direct route keeps value that the hydrometallurgical route passes downstream to whoever makes the precursor, and it does so on roughly 30–50% less energy and reagent cost. That is the largest per-tonne prize in battery recycling. The problem is that the addressable feed today is manufacturing scrap, which is 5–10% of gigafactory output once a line is running well and 20–30% while it is ramping, and that number falls as cell makers get better, so the feed shrinks exactly as the plants scale. End-of-life packs will eventually be the larger stream, but they arrive mixed, and nothing in the collection system sorts them by cathode chemistry. So the economics currently rest on a bilateral arrangement with a cell maker plus public funding, which is what the ReCell Center and the DOE grant programs have supplied, and no plant is yet running on the value of its output alone. The one place the arithmetic could flip on its own is LFP: there is no metal value to chase, so no other route pays, and returning a $5,000–8,000 a tonne cathode from material a recycler would otherwise charge a gate fee to take is a real business if the qualification holds.

5 minRelithiation: The Future of Lithium Battery Recycling?Electrified · 1k+ views

2 minA Novel Way to Recycle Lithium-ion BatteriesWPI · 1k+ views

A Review of Direct Recycling Processes for Lithium-Ion Battery Cells (Materials) · EverBatt: A Closed-loop Battery Recycling Cost and Environmental Impacts Model (Argonne National Laboratory)

Preprocessing is everything between a collected device and a feed that a metallurgical plant will buy. It opens with depollution: batteries, capacitors, mercury-containing backlights, toner cartridges and cathode-ray tubes come out by hand, because any one of them contaminates or ignites what follows. What is left goes down one of two paths. Manual dismantling separates the device into housings, power supplies, cables, drives and boards, which is slow and recovers the most. Shredding runs the whole device through a primary shredder and a granulator, then pulls steel out with drum magnets, aluminum with eddy-current separators, and the rest apart on density and air tables, with sensor or hand picking to lift the board fraction. The world generated 62 million tonnes of e-waste in 2022 and documented 22.3% of it as formally collected and recycled, so most of the material never reaches a line like this at all. ITU and UNITAR project 82 million tonnes by 2030 with the documented rate slipping to 20%, and they put the recoverable material left outside the formal system at roughly $62 billion a year.

Shredding is fast, and magnetic and eddy-current separation cost almost nothing per tonne once the line is built, so a mixed pile becomes four or five salable commodity grades in a single pass. The weakness is that shredding destroys the concentration it is supposed to preserve: gold sits in plated connectors and bond wires at a few grams per tonne of whole device, and once those grams are smeared through dust and plastic they cannot be recovered at any sensible cost. The rare earths in speaker and motor magnets go the same way, which is part of why recycled sources meet only about 1% of rare earth element demand. Lithium cells that survive depollution are the other failure mode, since a punctured cell in a shredder starts a fire, and shredder fires are now a routine cause of plant losses. Manual dismantling avoids both problems and costs several times more per tonne, which is why the highest-recovery hand operations sit in countries with low labor costs. The same arithmetic produces the informal sector: open cable burning and open-vessel acid leaching in Accra, Guiyu and Delhi recover a fraction of what an integrated smelter would, and the difference is paid in lead, dioxin and mercury exposure to the people doing the work, children included.

Dismantle by hand when the feed is high value and consistent, which means servers, phones, laptops and medical equipment, where boards are a large share of the mass and the labor pays for itself. Shred when the feed is mixed household equipment and the value sits in steel, aluminum and copper, because hand-sorting a microwave oven costs more than the metal inside it. Either way, treat depollution as the step that cannot be skipped, since one missed cell can take the line out for a week. Sell the board fraction to an integrated smelter rather than leaching it yourself: recoveries above 95% happen at that scale and small-vessel acid recovery is where they do not. If the numbers only close by exporting the residue to a jurisdiction with weaker enforcement, price that as regulatory risk rather than as margin.

62 Mt of e-waste generated in 2022 · 22.3% documented as formally collected and recycled · 82 Mt projected for 2030 at a 20% documented rate · roughly $62 billion of recoverable material a year outside the formal system · gold at a few grams per tonne of whole device · integrated smelter recoveries above 95% · about 1% of rare earth demand met from recycled sources

Sims Lifecycle Services and ERI in North America; the WEEE Directive's Annex VII list of components that must be removed before any further treatment; the R2v3 and e-Stewards certifications that downstream buyers increasingly require; Agbogbloshie in Accra and Guiyu in Guangdong as the informal counterexample.

There are two revenue lines, and for most consumer e-waste the gate fee or extended producer responsibility payment is the larger one. Material revenue comes mostly from steel, aluminum and copper by weight and from the board fraction by value, and the boards are usually a few percent of the mass. Costs are labor, freight, and the fractions nobody wants: leaded CRT glass and mixed plastics with brominated flame retardants both carry a disposal charge rather than a price. A shredding line is a few million dollars of capital and runs on throughput, so the binding constraint is collection rather than processing capacity. That is what makes the ITU and UNITAR figures an investment argument: at 22.3% documented collection, roughly $62 billion of recoverable material a year sits outside the formal system, and recycled sources still meet only about 1% of rare earth element demand. The margin goes to whoever controls the collection contracts, because preprocessing itself is a commodity service that anyone with a shredder can offer.

10 minHow 6 Million Pounds Of Electronic Waste Gets Recycled A Month | Big BusinessBusiness Insider · 1m+ views

4 minWatch Your Dead Tech Get Demolished at an E-Waste Recycling PlantGizmodo · 50k+ views

The Global E-waste Monitor 2024 (ITU and UNITAR) · Children and digital dumpsites: e-waste exposure and child health (World Health Organization)

Circuit boards are not processed in a dedicated gold plant. They are charged into a copper smelter, where the epoxy and glass-fiber laminate, roughly 30% of the board by weight, burns as fuel and reductant while molten copper collects the precious metals into a matte. The matte is converted, cast into anodes, and electrorefined; gold, silver, palladium, platinum, selenium and tellurium do not dissolve and drop out as anode slime under the cathodes, which goes to a precious-metals refinery. A tonne of high-grade boards, meaning phone and server boards, carries roughly 150 to 400 grams of gold. A tonne of mined ore carries roughly 1 to 5 grams, and the USGS put the average feed grade at a typical gold operation at 0.9 grams. Even whole shredded PC scrap, where the boards are diluted by steel and plastic, runs about 10 grams per tonne, which the USGS noted is more gold than is recovered from 17 tonnes of ore. Boards are about 40% metal, 30% polymer and 30% ceramic by weight, with copper alone at 20 to 40%.

One process recovers a dozen metals from a feed nobody can sort by hand, at recoveries above 95% for copper, gold, silver and palladium, and the organics in the board supply part of the heat instead of coke. The weakness is that everything which oxidizes reports to slag: aluminum, iron and tantalum are lost, and so are the rare earths in the board's magnets and capacitors, which sit at parts-per-million levels and are far below any economic cutoff. Brominated flame retardants make the off-gas halogen-rich, so the plant needs gas cleaning and bromine capture that a smelter running mineral concentrate does not. Capital is the other constraint. An integrated copper and precious-metals smelter costs on the order of a billion dollars, only a handful in the world take this feed, and a seller therefore faces long freight, waiting lists, and payable terms written by the buyer.

If you hold board scrap, sell it rather than trying to leach it; recoveries above 95% belong to plants at this scale. Grade the boards first, because the sorting decides the buyer. High-grade phone, server and telecom boards, which trade around $2,000 to $6,000 a tonne, go to an integrated precious-metals smelter, and low-grade power supply, television and appliance boards are essentially a copper feed, so a secondary copper smelter is the right home and paying precious-metal refining charges on them destroys the value. If you need aluminum, tantalum or rare earths back, this is the wrong process and those parts have to be removed before the boards are charged. If your volume is under a few hundred tonnes a year, aggregate through a broker rather than negotiating your own smelter contract, since assay terms on small parcels are usually worse than the freight saving.

150–400 g of gold per tonne of high-grade boards · 1–5 g per tonne in mined ore, 0.9 g at a typical operation · about 10 g per tonne in whole shredded PC scrap, more than 17 tonnes of ore yields · boards roughly 40% metal, 30% polymer, 30% ceramic · copper 20–40% of board weight · $2,000–6,000 per tonne for high-grade boards · above 95% recovery for copper, gold, silver and palladium

Umicore's Hoboken plant in Belgium, which runs e-scrap alongside mining concentrates and byproducts; Boliden's Rönnskär smelter in Sweden and its Kaldo furnace built for electronic scrap; Aurubis at Lünen in Germany; Dowa's Kosaka smelter in Japan; Glencore's Horne smelter in Quebec.

The plant is a copper smelter first, and electronic scrap is a high-margin side feed added to an existing furnace rather than the reason anyone builds one. Sellers are paid on assay against treatment and refining charges, so the commercial argument is about sampling and payable percentages rather than about the process, and a large parcel with a disputed assay can move six figures. High-grade boards trade around $2,000 to $6,000 a tonne, which is what makes collection and grading worth doing even when the rest of the device is close to worthless. Because capacity is scarce and capital-intensive, the buyer sets the terms, and margin concentrates at the two ends of the chain: whoever controls collection, and whoever owns the smelter. The accessible business for a new entrant is grading and aggregation in the middle, which is real but thin. Supply security is now a second driver, since gold, palladium and copper recovered inside a country do not need an import license.

13 minHow Scrappers Cash In On Gold From Your Old Computer | World Wide Waste | Business InsiderBusiness Insider · 5m+ views

13 minCan CPU Gold Be Direct Smelted? Electronics Recycling & Urban Mining For Precious Metals!mbmmllc · 1m+ views

Obsolete computers, "gold mine", or high-tech trash? Resource recovery from recycling (U.S. Geological Survey) · Recycling of Printed Circuit Boards to Recover Critical Materials (Ames National Laboratory)

Sintered NdFeB magnets are about 30% rare earth by weight, mostly neodymium and praseodymium, with dysprosium or terbium added when the magnet has to hold its field hot, and about 65% iron. They sit inside assemblies that were never meant to be opened: roughly 2.5 g in the voice coil actuator of a 2.5-inch hard drive, about 3 kg spread across an electric vehicle, and around 500 kg in a direct-drive wind generator. Recovery begins by getting the magnet out, which means demagnetizing it by heating past the Curie point near 310 °C, then defeating the adhesive and the nickel-copper-nickel plating. Two routes follow. Hydrogen decrepitation lets the magnet absorb hydrogen, which expands the lattice and crumbles the magnet into a powder that can be milled, blended to grade and re-sintered into new magnets, a short loop that skips mining and separation entirely. The long loop dissolves the magnet in acid, precipitates the rare earths as oxalates and calcines them to oxides, then re-separates them by solvent extraction, which yields a saleable oxide but repeats most of the chemistry primary production already does.

The feed grade is the strong part: 30% rare earth by weight against roughly 1 to 10% in the ores and clays that primary supply comes from, and the short loop avoids the cracking, separation and metal reduction that dominate primary energy use. Everything upstream of the process is the weak part. Magnets are glued and press-fit into rotors, plated, and mixed in grade, so extraction is manual labor performed on an object worth a few dollars. A Carnegie Mellon and DOE study modeled whole plants and found the hard drive case needed a cost of recovery of $229 per kg against a neodymium oxide price of $42 per kg, more than five times, driven by how scattered the feed is; the same modeled plant fed on electric and hybrid drive motors reached $30.8 per kg against a product worth $58.62 per kg, which is profitable. The price a recycler has to beat is set by Chinese primary supply and moves a long way: NdPr oxide averaged $124 per kg in 2022, $55 per kg in 2024, and $69 per kg in 2025, so a project underwritten at one end of that range is underwater at the other. The USGS still records only limited quantities of rare earths recovered from permanent magnets, and recycled sources meet about 1% of demand.

Use short-loop hydrogen decrepitation when you can get a concentrated single-source stream: a motor maker's own production scrap, a fleet's returned drive units, or data center hard drives that arrive already sorted. In those cases the disassembly cost is spread over kilograms rather than grams. Avoid it when the feed is mixed consumer electronics, because the labor to pull 2.5 g out of a hard drive costs more than the magnet is worth. Choose the hydrometallurgical long loop when magnet grades are mixed or the coating and adhesive will not come off cleanly, since the chemistry tolerates a dirty, multi-grade feed that re-sintering cannot. If the reason for the project is dysprosium and terbium supply outside China rather than the margin, put that in the business case explicitly and size the support it needs, because at current prices the margin is not there on most feeds.

NdFeB roughly 30% rare earth and 65% iron · 2.5 g per 2.5-inch hard drive, about 3 kg per EV, about 500 kg per direct-drive turbine · Curie point near 310 °C · cost of recovery $229/kg from hard drives against a $42/kg neodymium oxide price · $30.8/kg from EV motors against a $58.62/kg product · NdPr oxide $124/kg in 2022, $55/kg in 2024, $69/kg in 2025 · about 1% of rare earth demand met from recycling

HyProMag and the hydrogen processing of magnet scrap route developed at the University of Birmingham; Noveon Magnetics in Texas, which sinters finished magnets from recovered feed; Cyclic Materials in Ontario; the DOE Critical Materials Innovation Hub at Ames National Laboratory; the EU Critical Raw Materials Act, which sets a benchmark of meeting 25% of annual consumption of strategic raw materials from recycling by 2030.

Cost is dominated by feed acquisition and disassembly, not by the metallurgy, which is the opposite of most recycling businesses. Work the arithmetic on a car: 3 kg of magnet at roughly 30% rare earth is about 0.9 kg of contained rare earth, worth on the order of $60 at a $69 per kg NdPr oxide price, and that $60 has to cover collecting the vehicle, pulling the traction motor, extracting and demagnetizing the magnets, and processing them. That figure is derived here rather than published, and it is the whole reason this stream is marginal while copper and gold streams are not. Capital is modest by metallurgical standards, so the constraint is throughput of sorted feed rather than plant cost. The realistic buyers today are magnet makers who need a non-Chinese source and governments willing to pay for one, which is why the EU benchmark and DOE program funding show up in every credible business plan. Watch the NdPr price before anything else, because a swing from $124 to $55 per kg has repriced this industry twice in five years.

6 minCan rare-earth elements be recovered through recycling? | DW NewsDW News · 10k+ views

2 minRecycling Magnets and Rare Earth ElementsESCI - European Science Communication Institute · 5k+ views

Design and optimization of processes for recovering rare earth elements from end-of-life permanent magnets (AIChE Journal) · Rare Earths, Mineral Commodity Summaries 2026 (U.S. Geological Survey)

A crystalline silicon module is 76% glass by weight, 10% polymer, 8% aluminum frame, 5% silicon, 1% copper and under 0.1% silver. None of those materials is difficult to recycle on its own. The difficulty is that the cells are laminated between the glass and the backsheet in cross-linked EVA, which will not release without heat or solvent, so the module behaves as one composite object rather than six materials. The standard commercial process is mechanical: remove the junction box and cables, shear or unclip the aluminum frame, then shred the laminate and screen glass cullet away from a mixed fines fraction. Glass and aluminum are 84% of the weight, so that route recovers most of the mass while leaving the silicon, silver and copper together in a low-value powder. Higher-recovery lines delaminate first, by hot knife, by thermal treatment that burns the encapsulant off near 500 °C, or by solvent, and then leach the silver and copper off the cell.

The mechanical route is cheap, already running at commercial scale, and recovers the two heaviest materials, which is what a mass-based recycling target measures. It recovers very little of the value, because the silver and silicon are where the value density sits and they stay in the fines. IRENA and IEA-PVPS put the technically recoverable raw material value at up to $450 million cumulatively by 2030 and over $15 billion by 2050, against cumulative panel waste of 1.7 to 8 million tonnes by 2030 and 60 to 78 million tonnes by 2050; dividing the 2050 pair gives roughly $190 to $250 a tonne, so about $4 to $5 for a 20 kg panel, and that arithmetic is done here rather than published. Burying the same panel at a US average landfill tipping fee near $55 a ton costs about $1.20 plus haulage, which is also derived. When the recoverable material is worth a few dollars and disposal costs a few dollars, any process costing more than that only happens where a rule requires it. In the US there is a regulatory cost before any of this: TCLP testing to determine whether a module is hazardous waste runs $550 to $3,000 per module, and results for PV vary by more than 50% between tests, so the determination is expensive and not repeatable.

If you operate in the EU, the decision has been made for you, since PV panels came into scope of the WEEE Directive in the 2012 recast and the producer pays for collection and treatment. In the US, recycle where a state law requires it or where a buyer for frame aluminum and glass cullet is close enough that haulage does not consume the value, and otherwise plan on landfill and price it honestly rather than assuming a recycler will take the panels for free. If you are decommissioning a utility-scale plant, make the hazardous-waste determination early, because testing cost and accumulation time limits drive the schedule more than the tear-down does. If you are building a recycling business, model the gate fee as the revenue and recovered material as upside. And if you specifically need silver and silicon back, budget for a delamination step, because the mechanical route will not give them to you in a usable form.

76% glass, 10% polymer, 8% aluminum, 5% silicon, 1% copper, under 0.1% silver · 1.7–8 Mt of cumulative panel waste by 2030 and 60–78 Mt by 2050 · recoverable material value up to $450 million by 2030 and over $15 billion by 2050 · roughly $190–250 a tonne, about $4–5 per 20 kg panel (derived) · about $1.20 a panel to landfill at a $55 a ton tipping fee (derived) · TCLP testing $550–3,000 per module

The EU WEEE Directive, which brought panels into scope in its 2012 recast; PV Cycle's collection network in Europe; Veolia's dedicated PV line at Rousset in France; First Solar's in-house recycling for its own CdTe modules; SOLARCYCLE in Odessa, Texas; California's classification of PV modules as universal waste and Washington State's producer stewardship law for modules.

Revenue is a gate fee plus aluminum and glass, and the gate fee is the larger line almost everywhere. The investment case is the volume wave rather than today's margin: cumulative panel waste goes from 1.7 to 8 million tonnes in 2030 to 60 to 78 million tonnes by 2050, which is a tenfold increase or more in twenty years. The catch is the lag. Panels last 25 to 30 years, so the tonnage arriving now reflects installations from around 2000, when almost nothing was installed, and anyone building capacity today is building well ahead of the feed. Silver is where the economics could change, since it is under 0.1% of the weight and a large share of the recoverable value, so a delamination and leach step that gets cheap enough would reprice the whole stream. The carbon argument is separate and real: recovered aluminum avoids roughly the primary smelting energy, and recovered silicon avoids polysilicon production, which is the most energy-intensive step in making a panel. Until one of those changes, this is a business funded by producer responsibility payments and mandates, which is why the EU has a functioning collection system and most of the world does not.

13 minWhat REALLY happens to used Solar Panels?Undecided with Matt Ferrell · 500k+ views

16 minThe race to solve solar energy's recycling problemDW Planet A · 100k+ views

End-of-Life Management: Solar Photovoltaic Panels (IRENA and IEA-PVPS) · Knowledge-Based Hazardous Waste Determinations for Solar Photovoltaic (PV) Modules: A Foundational Study (National Laboratory of the Rockies)

A wind turbine blade is 80 to 90% composite by mass, and that composite is 60 to 70% reinforcing fiber, almost always glass with carbon in the spar caps of the largest blades, and 30 to 40% thermoset resin, plus balsa or foam in the core, steel root fasteners and a copper or aluminum lightning conductor. The resin is cross-linked, so it cannot be remelted. Heating it decomposes the polymer rather than softening it, which rules out the melt-and-repelletize route that every metal and thermoplastic stream depends on. Blade mass runs at roughly 12.5 tonnes per megawatt of rating, so a 2.5 MW turbine carries about 31 tonnes of blade and a single blade weighs around 10 tonnes. The options, ordered by how much material they move today, are cutting into segments up to 30 m for landfill, grinding to 1–3 cm shreds for filler or landfill, cement kiln co-processing, and pyrolysis or solvolysis to recover fiber. In a cement kiln the resin burns and displaces coal, cutting kiln carbon dioxide by up to 16%, and the glass fiber goes into the clinker as silica and alumina in place of sand and clay, so both fractions get used and nothing comes back as a blade.

Cement co-processing takes a whole shredded blade with no separation of fiber from resin, runs in existing plants at commercial scale, and uses both fractions, which is why it is the incumbent answer wherever landfill is closed off. Its limits are that the output is clinker, the recovered value is low, it needs a kiln within haul distance, and it suits glass-fiber blades rather than carbon-fiber ones. Pyrolysis at 400–700 °C recovers the fiber but leaves glass at roughly half its virgin strength, so recovered glass fiber is a filler rather than a reinforcement, while carbon fiber comes through largely intact, which is why the chemical routes chase carbon-fiber blades. Landfill is a smaller problem by volume than it is usually presented as: cumulative US blade waste through 2050 is about 2.2 million tons, roughly 1% of remaining landfill capacity by volume and 0.2% by mass, and there is no evidence the material leaches. Cost is the reason nothing changes. Cutting a blade into segments on site runs about $25 a US ton and grinding to shreds about $90, against a US average landfill tipping fee near $55 a ton, so cutting and burying a 10 tonne blade comes to roughly $900, while the crane work to get the rotor down is $26 per kW, about $22,000 a blade on a 2.5 MW turbine, and is incurred whatever happens next. Both per-blade figures are derived from those rates rather than published. Total end-of-life cost lands at $19 to $39 per kW against roughly $1,000 per kW of turbine capital cost.

If you are decommissioning in Europe, plan on cement kiln co-processing, because several member states already prohibit landfilling untreated composite and the UK median landfill gate fee is £113 a tonne including an £89 landfill tax. Distance to a kiln that will take the shred is the binding constraint, so check that before committing to a route. In the US, landfill is legal and costs around $65 a tonne, so any recycling route needs a state rule or a corporate commitment behind it; modeling from Cambridge puts pyrolysis of glass-fiber blade waste at a loss of over $140 a tonne, which no operator absorbs voluntarily. If the blades are carbon fiber, look at pyrolysis or solvolysis, since recovered carbon fiber is worth enough to pay for the process where recovered glass is not. If you are buying turbines now, ask what the resin system is, because a chemically separable blade changes this decision twenty years out and costs nothing to specify today.

80–90% composite, of which 60–70% fiber and 30–40% resin · roughly 12.5 tonnes of blade per MW, so about 10 tonnes a blade on a 2.5 MW turbine · on-site cutting $25 a US ton, grinding $90, US average tipping fee $55 · roughly $900 a blade to cut and bury, plus about $22,000 a blade of crane work (both derived) · total end-of-life cost $19–39/kW against about $1,000/kW of turbine capital · UK landfill £113 a tonne against about $65 in the US · cement kiln co-processing cuts kiln carbon dioxide by up to 16%

Veolia's Missouri plant, which shreds GE blades into cement kiln feed; Holcim's Geocycle co-processing operations in Europe; Vestas' CETEC chemical process for standard epoxy blades; Siemens Gamesa's RecyclableBlade; landfill bans on untreated composite waste in Germany and the Netherlands; WindEurope's call for an EU-wide ban on landfilling decommissioned blades.

The largest cost in blade disposal has nothing to do with recycling. Rotor teardown at $26 per kW is about $65,000 on a 2.5 MW turbine, roughly $22,000 a blade, and the owner pays it whatever the destination, so every route is priced from the point where the blade is already on the ground. After that the numbers are small: cutting and burying a 10 tonne blade is around $900, and total end-of-life cost per turbine ran from about $15,000 to $100,000 in NREL's modeling, or $19 to $39 per kW against roughly $1,000 per kW of turbine capital. Disposal has therefore never been expensive enough to force a change in practice on its own. Where a recycling route does get chosen, a landfill tax or a brand commitment is the reason, not the value of the recyclate: the UK's £89 a tonne tax brings alternatives into range, and at the US price near $65 a tonne it does not. Haul distance is the lever an operator actually controls, since semi-trailer transport at about $14 a mile carries two segments and passes the tipping fee within a couple of hundred miles, which is what makes on-site grinding worth its higher processing cost on remote sites.

7 minWhy Wind Turbine Blades Are So Hard to Recycle | World Wide WasteBusiness Insider · 1m+ views

12 minFINALLY! Fully recyclable wind turbine blades!Just Have a Think · 50k+ views

Wind turbine blade material in the United States: Quantities, costs, and end-of-life options (National Renewable Energy Laboratory) · Wind turbine blade end-of-life options: An economic comparison (University of Cambridge Repository)

Mechanical recycling of PET takes baled bottles and gives back PET, with the polymer chemistry unchanged. Bales are broken open, sorted again by near-infrared and by color, and ground to flake roughly 6–12 mm across. The flake goes through a sink-float tank, where PET at about 1.38 g/cm³ sinks while the polypropylene caps and polyethylene label sleeves float off, and then through a hot caustic wash at 80–90 °C that removes adhesive, food residue and surface ink. After drying to under about 50 ppm moisture, the flake is either sold as flake or extruded to pellet and finished by solid-state polycondensation, which holds the pellet at 200–220 °C under vacuum or nitrogen for several hours. That last step is there because every melt pass hydrolyzes ester bonds and drops intrinsic viscosity by roughly 0.02–0.03 dL/g, while blowing a bottle that holds carbonation pressure needs about 0.78–0.84 dL/g; solid-state polycondensation builds the chains back up and strips the volatile contaminants that a food-contact clearance is written around. Yield from bale to food-grade pellet is usually 65–80%.

Three things make this the one plastic stream with real closed-loop economics. A bottle is a single polymer in a single format, so the sorting problem is small; deposit systems produce a stream clean enough to work with, returning about 98% of PET bottles in Germany and about 92% in Norway against 20–30% collection where there is no deposit; and the food-contact route is settled, because FDA reviews a specific process against a surrogate-contaminant challenge test and issues a letter of no objection, which took the industry years of testing to establish and now takes a new entrant roughly 12–24 months. The weaknesses are all feed weaknesses. One PVC bottle in a load carbonizes at PET processing temperature and specks the whole batch, thermoform trays carry different additives and a different intrinsic viscosity than bottles, and opaque or heavily pigmented PET cannot be sorted back into clear bottle stock. The chain also shortens on every pass, so material cycles a finite number of times before it drops out to fiber or strapping, and the honest question is how many passes rather than whether the loop is closed.

If your feed is deposit-system bottles, mechanical recycling is the route and nothing else comes close on cost. If your feed is curbside single-stream PET, expect to sell fiber and sheet rather than bottle resin, because the residual PVC, the sleeve labels and the mixed thermoforms will cost you the food-grade clearance more often than not. If the feed is colored, opaque or heavily degraded PET, stop here and look at depolymerization instead, which does not care what color the input was. Before committing capital, check that a specific buyer has qualified your specific process rather than rPET in general, since the letter of no objection is written against a process and not a material. As a rule of thumb, if you cannot name the collection system your feed comes out of, you do not yet have a bottle-to-bottle business.

Flake 6–12 mm, PET density about 1.38 g/cm³ · caustic wash 80–90 °C, drying to under about 50 ppm moisture · solid-state polycondensation 200–220 °C · intrinsic viscosity about 0.78–0.84 dL/g for bottles, falling 0.02–0.03 dL/g per melt pass · 65–80% yield from bale to food-grade pellet · bales $200–500 a tonne, food-grade rPET pellet roughly $1,100–1,800 · EU requires 25% rPET in PET bottles from 2025 and 30% in all beverage bottles from 2030

Indorama Ventures, ALPLA, Veolia PET Germany, Plastipak and Evergreen run bottle-to-bottle plants at commercial scale; the equipment is largely from Starlinger recoSTAR, EREMA VACUREMA and Krones MetaPure, all of which sell their decontamination step as a qualified process rather than a machine. CarbonLite ran the largest US bottle-to-bottle capacity and filed for bankruptcy in 2021, which is the clearest evidence that clean feed and a working process do not by themselves make the margin safe.

The unit economics are a spread between two prices the recycler does not set. Bales cost $200–500 a tonne, and at 70% yield the bale content of a tonne of pellet is $290–715; wash, extrusion and solid-state polycondensation add roughly $300–500 a tonne, so delivered cost lands near $600–1,200 against food-grade rPET at roughly $1,100–1,800 a tonne. That chain is arithmetic done here rather than a published figure, and the sensitive input is the bale price, which moves with collection volume and with the virgin resin price at the same time. Capital runs roughly $1,000–2,000 per annual tonne of capacity, so a 30,000 tonne a year bottle-to-bottle line is a $30–60 million project, and utilization matters more than throughput because the fixed costs dominate. rPET traded $300–600 a tonne above virgin through the tight stretches, then virgin PET prices fell in 2023 and 2024 on new Asian capacity and the premium compressed, which idled plants on both sides of the Atlantic. What holds the floor under the business is regulation rather than the material value: the EU requires 25% rPET in PET bottles from 2025 and 30% in all beverage bottles from 2030, and a rule with a date and a penalty behind it does far more for the business than a brand commitment does.

5 minGo behind the scenes at a PET plastic recycling plant!PETCORE EUROPE · 100k+ views

7 minCarbonLite: Inside the World's Largest Plastic Bottle Recycling Plant | SoCal Connected | KCETKCETSoCalConnected · 1m+ views

Guidance for Industry: Use of Recycled Plastics in Food Packaging (Chemistry Considerations) (FDA) · Recycling of Plastics in the United States: Plastic Material Flows and Polyethylene Terephthalate (PET) Recycling Processes (ACS Sustainable Chemistry & Engineering)

Polyolefins are HDPE, LDPE and PP, and mechanically recycling them follows the same sequence as PET with two differences that decide the outcome. All three float, at 0.90–0.96 g/cm³ against water at 1.00, so a sink-float tank separates them cleanly from PET and PVC but barely separates them from each other, and PP at 0.90–0.91 against HDPE at 0.94–0.96 is too small a gap to split reliably at plant scale. The wash runs cooler than PET, usually 60–80 °C with caustic, because these polymers soften and because the contamination is grease and product residue rather than adhesive. Melt filtration then does the work that sorting could not: screen changers or laser filters at 80–150 µm pull out paper, wood, aluminum and unmelted foreign polymer, and a vacuum devolatilization zone in the extruder strips the odor compounds that come off personal-care and food residues. The polymers degrade in opposite directions during that melt pass, which is worth knowing by name: PP undergoes chain scission under shear and its melt flow index climbs while impact strength falls, and PE branches and crosslinks instead, so its melt flow index falls and gels appear. Both need antioxidants added back, since the original stabilizer package was consumed during first use. Yield from bale to pellet is usually 65–80%, the same band as PET.

The equipment is cheap and mature, and where the sorting has already been paid for the arithmetic works: natural HDPE bales sell for $600–1,400 a tonne against $200–400 for colored, so a color camera at the material recovery facility pays for itself quickly. The weakness is that the output is almost always one rung down. A 2% polypropylene contamination in HDPE is enough to cut environmental stress crack resistance sharply, which is the property a pipe or a detergent bottle is specified on, so mixed polyolefin recyclate goes into drainage pipe, decking, pallets and crates rather than back into packaging. Odor is the other blocker for consumer goods, because limonene and aldehydes from the original contents survive washing and only partly survive devolatilization. Flexible film is the extreme case and the reason the overall polyolefin numbers look bad: a few grams per piece, usually a printed multilayer of PE with another polymer, US recycling around 2–4% for decades, and bales worth $0–300 commercial and negative for residential collection.

If you have natural HDPE bottles, color-sort them and run them mechanically, because that is the one polyolefin stream where the price spread pays for the process on its own. If your feed is mixed rigid polyolefins from curbside, plan the business around pipe, lumber and crates rather than around packaging, and price the output against virgin PP or HDPE minus a discount rather than plus a premium. If you need food-contact PP or HDPE, do not assume the PET playbook transfers, since there is no deposit stream feeding it and a controlled-collection scheme has to be built first. If the feed is flexible film, mechanical recycling only works behind a gate fee or a producer responsibility payment, and dissolution is the route that actually handles multilayer structures. As a default, decide the question of what fraction of your feed is a single polymer before buying any equipment, because the melt filter and the devolatilizer improve the pellet and cannot change what polymers are in it.

Polyolefin density 0.90–0.96 g/cm³ against water at 1.00, PP 0.90–0.91 versus HDPE 0.94–0.96 · caustic wash 60–80 °C · melt filtration at 80–150 µm · about 2% foreign polymer is enough to cut environmental stress crack resistance · 65–80% yield from bale to pellet · natural HDPE $600–1,400 a tonne against $200–400 colored · flexible film 2–4% US recycling rate, bales $0–300 commercial and negative residential

Berry Global, KW Plastics and Envision Plastics run large post-consumer HDPE and PP compounding operations in the US; EREMA, Starlinger and Coperion supply most of the extrusion and melt-filtration lines. Trex makes composite decking from post-consumer polyethylene film and wood flour, which is the largest single US outlet for a stream nobody else wants; the Association of Plastic Recyclers publishes the design guide that decides whether a given bottle or label is considered recyclable in this route at all.

This is a commodity conversion business with a structural problem: unlike rPET, recycled polyolefin usually sells at a discount to virgin rather than a premium, because the buyer is a pipe or pallet maker with no recycled-content obligation. Post-consumer PP and HDPE pellet typically runs a few hundred dollars a tonne below the corresponding virgin resin, conversion cost is roughly $250–450 a tonne, and the feed is bought at $200–1,400 a tonne depending entirely on color and polymer. That leaves the processor squeezed from both ends when virgin resin falls, which is exactly what happened in 2023 when new polyethylene capacity came online and post-consumer resin demand softened at the same time. Policy is the main source of demand that does not move with the oil price: the EU requires 30% recycled content in all beverage bottles from 2030, which reaches HDPE milk bottles, and the UK plastic packaging tax charges over £200 a tonne on packaging with less than 30% recycled content, which sets a ceiling on what a converter will pay for recyclate. Brand commitments matter less than they appear, because most are pledges rather than contracted volumes at a stated price, and the 2018 cohort of 25%-by-2025 commitments mostly landed near 12–14%. If you are underwriting one of these plants, the two questions are whether the color sorting is inside your cost or somebody else's, and whether the offtake is a signed volume.

1 minWaste Plastic Recycling Machine | How to recycle PP, PE into plastic pellets and granules?Shuliy Global Recycling & Eco-Machinery · 100k+ views

14 minHow to Make Recycled BEAMS from Plastic Waste at HomeBrothers Make · 1m+ views

APR Design Guide Overview (Association of Plastic Recyclers) · Defining quality by quantifying degradation in the mechanical recycling of polyethylene (Nature Communications)

Dissolution leaves the polymer chain intact and takes everything else off it, which is what separates it from the two chemical routes below it on this sheet. A solvent chosen for the target polymer dissolves it at temperature and pressure, typically 80–200 °C, while pigments bound to filler, other polymers, aluminum foil, paper and dirt stay solid and are filtered out. The dissolved contaminants that come along, mostly dyes, plasticizers and low-molecular-weight additives, are then pulled out of solution by activated carbon or clay adsorption, and the clean polymer is precipitated by cooling or by adding an antisolvent and devolatilized to strip residual solvent down to the parts-per-million range. Because no bond in the backbone is broken, the recovered material has the molecular weight distribution the feed had, so the mechanical properties come back close to virgin and the color and odor do not come back at all. Selectivity is the design lever: PureCycle dissolves polypropylene in butane above its critical point, CreaSolv uses formulated solvents for polystyrene and polyethylene, and the STRAP process developed at the University of Wisconsin-Madison runs a multilayer film through toluene, then tetrahydrofuran, then dimethyl sulfoxide in sequence to take off the polyethylene, the ethylene vinyl alcohol barrier and the PET one layer at a time. Polymer recovery is usually 85–95% of what was in the feed.

The output is the strongest argument: same polymer, chain length unchanged, and free of the color and additive history that caps what mechanical recyclate can be sold into, which is why the target markets are pigmented PP, carpet fiber and multilayer film. The feed tolerance is real too, since a laminate that no sink-float tank can separate comes apart in a solvent that only one of its layers dissolves. Two things hold it back. The first is that dissolution cannot repair anything: a chain shortened by three previous heat histories dissolves and precipitates just as short, so this route fixes contamination and does not fix degradation. The second is solvent recovery, which is most of the cost and most of the risk, because the process needs above 99% of the solvent back through distillation, the latent heat of vaporizing it dominates the energy bill, and the losses show up as both an operating cost and a permitting problem. Plants also have to hit residual-solvent limits in the parts-per-million range before the pellet is saleable into food or personal-care packaging.

Use dissolution when the polymer is fine and the contamination is the problem, which is the case for pigmented polypropylene, post-consumer carpet face fiber and printed multilayer film. If the feed is a clean single polymer, do not use it, because mechanical recycling produces a comparable pellet at a fraction of the operating cost and is the route to beat. If the feed is degraded or the buyer needs a genuinely virgin specification with no thermal history, go to depolymerization instead, since dissolution preserves whatever chain length arrives. If the feed is mixed polyolefin film with dirt and food residue and no single target polymer, pyrolysis is the more honest comparison, and the question becomes whether you want a polymer back or an oil. When you evaluate a project, ask for solvent loss per tonne of product and the cost of makeup solvent, because that number rather than the dissolution chemistry is what has sunk plants.

Dissolution typically 80–200 °C, with a solvent selective for one polymer · contaminants removed by carbon or clay adsorption, polymer recovered by cooling or antisolvent precipitation · 85–95% of the polymer in the feed recovered · solvent recovery above 99% needed for the economics to work · residual solvent in the pellet down to parts-per-million levels · processing cost roughly $500–1,000 a tonne against $250–500 for mechanical

PureCycle Technologies runs a supercritical-butane polypropylene purification plant at Ironton, Ohio with a nameplate near 48,000 tonnes a year, which started up in 2023 and has spent a long time below that rate; APK AG operates its Newcycling solvent process at Merseburg, Germany on multilayer film; Fraunhofer IVV's CreaSolv process has been demonstrated for polystyrene and for flexible sachets. The STRAP work at the University of Wisconsin-Madison, published in Science Advances, is the reference for sequential-solvent separation of multilayer packaging.

Processing cost runs roughly $500–1,000 a tonne, against $250–500 for mechanical recycling of the same polymer, and almost all of the difference is solvent inventory, solvent makeup and the steam to distill it back. Against that, the product sells at a premium a mechanical pellet cannot reach, because a food or personal-care buyer will pay for a resin with no color and no odor, and a pigmented PP bale that costs $200–400 a tonne becomes a pellet competing with virgin PP. The capital is a chemical plant rather than a wash line, with solvent handling, distillation columns and hazardous-materials permitting, so it lands well above $2,000 per annual tonne of capacity and takes 12–36 months of permitting before construction. First-of-a-kind risk has been the actual story here: Ironton is the largest working example and it has demonstrated that a solvent process on real post-consumer feed behaves differently from the same process on pilot feed, which is the standard failure mode for a plant whose input changes composition weekly. Revenue depends on brand offtake and on recycled-content rules rather than on commodity value, so treat a project without a signed volume at a stated price as a research program with a plant attached. The one structural advantage is that the sorting requirement is lower than for mechanical recycling, which means part of the cost that a mechanical plant pushes onto the material recovery facility sits inside this process instead.

2 minCreaSolv® Process AnimationCreaCycle GmbH · 5k+ views

6 minInside PureCycle Episode 2: PurificationPureCycle Technologies · 1k+ views

Recycling of multilayer plastic packaging materials by solvent-targeted recovery and precipitation (Science Advances) · Solvent-based plastic recycling technologies (Nature Chemical Engineering)

Depolymerization breaks the polymer backbone at the bond that formed it and returns the monomer, so the product has no thermal history and the resin made from it meets a virgin specification. It works on condensation polymers because the linking bond is a heteroatom bond that can be reversed by putting the eliminated small molecule back in. PET is depolymerized by glycolysis in excess ethylene glycol at 180–240 °C over a zinc, manganese or titanium catalyst to give BHET, by methanolysis in methanol at 180–280 °C under pressure to give dimethyl terephthalate and ethylene glycol, by hydrolysis in acid, base or neutral water to give terephthalic acid, or enzymatically with an engineered cutinase at 65–72 °C, which is the Carbios route and runs far cooler than the rest. Nylon 6 goes back to caprolactam by steam or acid hydrolysis at roughly 250–300 °C, and polystyrene is the one chain-growth polymer that cooperates, because above about 350 °C it unzips to styrene rather than fragmenting randomly. Polyethylene and polypropylene have nothing but carbon-carbon bonds in the backbone and no reversible linkage, so heating them gives random scission, which is pyrolysis rather than depolymerization. Yield to monomer is typically 70–90%, and the monomer then has to be purified by distillation or crystallization and repolymerized in a separate plant, each of which costs a few more percent.

The output is the real advantage: a purified monomer carries no color, no additive and no chain-length history, so the feed can be colored, opaque, thermoformed or textile PET that mechanical recycling cannot use, and the resin still qualifies for food contact on the same basis as virgin. The limits are chemical and economic. Chemically, the route is confined to PET, nylon and polystyrene, which is a minority of the plastic waste stream, and PET depolymerization still needs the feed sorted well enough that polyolefins and PVC do not end up in the reactor. Economically, you are paying to reverse a polymerization you already paid for once, and then paying to run it forward again, which is why the energy and capital per tonne land far above mechanical recycling. Enzymatic hydrolysis is the most interesting answer to that because it runs near 70 °C rather than above 200 °C, but it needs the PET amorphized and micronized first, and no plant has yet proved the economics at industrial scale.

Use depolymerization when the feed is PET or nylon that mechanical recycling has to reject, and when the buyer needs a virgin-equivalent specification rather than a good recyclate. If the polymer is polyethylene or polypropylene, the answer is no, and any pitch that says otherwise is describing pyrolysis. If the contamination is the problem but the chain is intact, dissolution is cheaper and keeps the polymer, so compare against that first rather than against mechanical recycling. If you are modeling one of these plants, price the monomer at monomer value and check whether the business closes without a recycled-content premium, because in most cases it does not. As a rule of thumb, treat depolymerization as the route you pick for output quality rather than for cost, and make sure somebody has contracted to pay for that quality before the plant is built.

PET glycolysis 180–240 °C, methanolysis 180–280 °C under pressure, enzymatic hydrolysis 65–72 °C · nylon 6 to caprolactam at roughly 250–300 °C · polystyrene unzips to styrene above about 350 °C · 70–90% yield to monomer, before purification and repolymerization losses · processing cost roughly $700–1,500 a tonne against $250–500 for mechanical · terephthalic acid roughly $700–950 a tonne and ethylene glycol roughly $500–700 · rPET has traded $300–600 a tonne above virgin in tight periods

Eastman runs a methanolysis plant at Kingsport, Tennessee designed for roughly 110,000 tonnes a year of hard-to-recycle polyester, and was selected in 2024 for a US Department of Energy industrial demonstration award of up to $375 million toward a second plant at Longview, Texas. Carbios is building the first industrial enzymatic PET plant at Longlaville, France for roughly 50,000 tonnes a year of PET waste, delayed repeatedly by financing; Aquafil's ECONYL process has depolymerized nylon 6 from fishing nets and carpet back to caprolactam commercially for over a decade, which makes it the clearest working example on this list; Loop Industries and Agilyx have run PET and polystyrene depolymerization at demonstration scale for years without reaching commercial volume.

The arithmetic that decides this route is that a monomer is worth monomer money. Terephthalic acid runs roughly $700–950 a tonne and ethylene glycol roughly $500–700, so a process costing $700–1,500 a tonne to run does not close on the commodity value of what it makes, and the gap has to come from somewhere else. That somewhere else is the recycled-content premium, which reached $300–600 a tonne over virgin PET during tight periods and compressed when virgin prices fell in 2023 and 2024, plus public capital: DOE demonstration awards, EU innovation funding and brand prepayments are financing most of the plants now under construction. Capital intensity is the other constraint, since this is a chemical plant with reactors, distillation and monomer purification, running well above $2,000 per annual tonne, and the depolymerization unit still has to sit next to or ship to a polymerization line. The one place the economics stand up without help is nylon 6, where caprolactam is worth roughly twice what PET monomers are worth and the feed, carpet and fishing net, arrives with a gate fee attached. If you are underwriting a PET project, the question is not whether the chemistry works, because glycolysis and methanolysis have been industrial for decades; it is whether the recycled-content rule that pays the premium survives the life of the plant.

4 minCHEMICAL RECYCLING #1 - Synthesis of terephthalic acid from waste PET water bottleZodaChem · 5k+ views

6 minChemical recycling – depolymerisation of polystyrene waste to styrene monomerINEOS Styrolution · 5k+ views

Industrial and Laboratory Technologies for the Chemical Recycling of Plastic Waste (ACS Catalysis) · Polymer Deconstruction and Redesign Strategies for Plastics Recycling (Advanced Materials Technologies)

Pyrolysis heats mixed plastic to 400–550 °C with no oxygen present, so the carbon-carbon backbone breaks at random points rather than unzipping, and the fragments distribute across three products: 10–20% non-condensable gas that is usually burned to heat the reactor, 60–80% condensable oil, and 5–20% char and heavy wax. Reactors are rotary kilns, heated screws, fluidized beds or molten salt baths, and the choice mostly changes heat transfer and coking rather than the product slate. The oil is a naphtha-to-gas-oil range liquid, rich in olefins and aromatics, and it is not a drop-in feedstock: a steam cracker will typically co-feed it at only 5–10% of the total and demands chlorine down to single-digit parts per million, silicon near 1 ppm and metals at parts per billion, so the oil has to be hydrotreated first. PVC and PET have to be kept out of the feed entirely, because PVC releases hydrogen chloride that corrodes the plant and poisons downstream catalysts, and PET releases carbon dioxide and terephthalic acid that sublimes and plugs lines. Gasification is the higher-temperature alternative, running partial oxidation at 700–1,500 °C to break everything down to carbon monoxide and hydrogen; it takes almost any feed, including PVC and wet mixed waste, but it destroys all molecular structure, so the product is syngas headed for methanol, ammonia or Fischer-Tropsch rather than a polymer. Both routes make a feedstock, not a plastic, which is the distinction that matters when comparing them with the routes above.

Feed tolerance is the whole argument. Mixed polyolefin film, multi-material laminates and sorting-line residue all pyrolyze, and the alternative for that material is landfill or an incinerator, so the route addresses the stream nobody else can take. What it gives back is much less than the headline suggests. Work the chain through: 100 kg of mixed plastic gives roughly 70 kg of oil, hydrotreating and fractionation cut that to about 60 kg of cracker-grade feed, a steam cracker converts roughly 50–55% of a naphtha-range feed to ethylene and propylene, and polymerization recovers most of that, so about 30 kg comes back as new plastic. That arithmetic is worked here rather than quoted, and it is why the honest number to ask for is plastic-to-plastic yield rather than oil yield. The other weakness is operational: these are chemical plants running on a feed whose composition changes weekly, and the commercial record is a long list of plants that never reached nameplate.

Use pyrolysis when the feed is mixed polyolefins that no mechanical or dissolution route will take, and when there is a gate fee or a landfill ban making the alternative expensive. If the feed is a single polymer, this is the wrong route by a wide margin, because you are spending energy to undo a polymerization and then spending more to redo it. If the feed contains meaningful PVC or PET, either sort them out or go to gasification, which tolerates them and gives you syngas instead. When you evaluate a project, ask three things: the plastic-to-plastic yield rather than the oil yield, whether a specific cracker has accepted the oil specification, and how much of the revenue is the gate fee. As a default, treat any claim of "certified circular" resin as a bookkeeping statement about a shared cracker rather than a statement about the molecules in the pellet, and read the allocation rule before pricing the premium.

Pyrolysis at 400–550 °C without oxygen, gasification by partial oxidation at 700–1,500 °C · products roughly 10–20% gas, 60–80% oil, 5–20% char and wax · steam crackers co-feed pyrolysis oil at about 5–10%, needing chlorine at single-digit ppm and silicon near 1 ppm · about 30% of input mass returns as new plastic, derived here from 70% oil, 60% cracker-grade feed and 50–55% cracker conversion · pyrolysis oil sells near naphtha parity, roughly $500–700 a tonne · gate fees of $50–150 a tonne are often the larger revenue line

Plastic Energy operates commercial pyrolysis lines in Spain and has licensed the technology to petrochemical partners; Brightmark's Ashley, Indiana plant was designed for about 100,000 tonnes a year and financed with roughly $260 million of bonds, and has run far below nameplate; Agilyx and its Regenyx joint venture in Tigard, Oregon pyrolyzed polystyrene back toward styrene at demonstration scale. Fulcrum BioEnergy's Sierra plant near Reno gasified municipal solid waste to fuel and the company filed for bankruptcy in 2024, which is the reference case for how hard the gasification version of this has been to finance and operate.

Revenue has two lines and most models get the ratio backwards. Pyrolysis oil sells near naphtha parity at roughly $500–700 a tonne, and a tonne of plastic gives about 0.7 tonnes of oil, so material revenue is $350–500 per tonne of feed; a gate fee of $50–150 a tonne is smaller but far more reliable, and in regions with high landfill taxes it is what makes a site bankable. Costs are a chemical plant's costs, with capital typically above $2,000 per annual tonne of capacity and operating costs dominated by the energy to heat the reactor and the hydrogen to hydrotreat the oil, and both are worse in practice than in the model because uptime on variable feed has been the recurring failure. The premium that makes projects pencil comes from mass-balance certification under schemes like ISCC PLUS, which lets a cracker running 5% pyrolysis oil sell a small share of its output as certified circular resin at a premium; the allocation method is what buyers and critics argue about, since free attribution concentrates the entire recycled credit into a fraction of the output rather than spreading it, and the certified pellet contains no more recycled carbon than any other pellet from that cracker. That premium exists because brands and the EU packaging rules created demand for a recycled-content claim, so the revenue is a policy artifact and should be modeled as one. Against landfill at a US tipping fee of $55–60 a ton, none of this is close; against a European incinerator gate of €80–130 a tonne plus a landfill ban, it is arguable, which is why the demonstration plants are where they are.

4 minPyrolysis and Catalytic Conversion of PlasticsPlast2Recycle · 100k+ views

6 minChemical recycling: the end of plastic waste? | Rethink SustainabilityFinancial Times · 100k+ views

Techno-Economic Analysis of Biomass, Municipal Solid Waste, and Mixed Plastic Waste Gasification (National Energy Technology Laboratory) · Chemical Feedstock Recovery Through Plastic Pyrolysis: Challenges and Perspectives Toward a Circular Economy (ChemSusChem)

Container glass is soda-lime glass: silica sand, soda ash and limestone melted together at about 1,500 °C in a regenerative furnace that runs continuously for 12–15 years between rebuilds and pulls 200–400 tonnes a day. Cullet is crushed recycled glass charged into that furnace with the raw batch. It is already vitrified, so it melts at a lower temperature and skips the carbonate decomposition that soda ash and limestone have to go through, and each 10% of cullet in the batch cuts furnace energy by roughly 2–3%. A container furnace uses roughly 4–6 GJ per tonne of glass pulled and emits around 0.5 tonnes of CO2 per tonne, about a fifth of which is process CO2 released when the carbonates break down, so cullet removes that share proportionally. Furnace-ready cullet has to be sorted by color and nearly free of ceramics, stone and porcelain, because a single unmelted particle becomes a stress inclusion that fails the bottle. Beneficiation plants get there with screens, magnets, eddy-current separators, optical color sorting and X-ray or laser detection for the ceramic fraction, and the resulting spec is usually under about 25 grams of ceramic, stone and porcelain per tonne. US container furnaces typically run 25–35% cullet against a European average above 50%, and several European countries run above 70%.

Glass does not degrade in the loop the way polymers and paper fibers do, so a bottle can become a bottle indefinitely, and the furnace energy saving is real and measurable batch by batch. The problem is that glass is heavy and worth almost nothing. Mixed cullet leaves a material recovery facility at $0–30 a ton, and bulk trucking runs roughly $0.10–0.20 per ton-mile, so a 200-mile haul costs $20–40 a ton and consumes the entire value of the load. Single-stream collection makes it worse by shattering the glass in the truck, and the small mixed-color fragments that result abrade screens, contaminate the paper bales and are expensive to color-sort, which is why a large share of curbside glass ends up as landfill cover or road aggregate. Color is the other hard constraint: flint (clear) glass tolerates only a few percent of green or amber in the cullet, while green glass takes almost anything, so a market with mostly clear bottles needs much better sorting than the tonnage alone suggests.

If there is a container plant within roughly 150–250 miles, cullet is worth collecting and beneficiating, and the furnace operator will usually pay for it because it displaces both energy and purchased soda ash. Past that radius, stop calling it glass recycling and plan for aggregate, abrasive or filtration sand, which are honest uses at honest prices. If the goal is high cullet rates in flint glass, run a deposit return system: it delivers whole, color-separable bottles with almost no ceramic contamination, which is what Germany's and Scandinavia's rates are built on. If you only have single-stream feed, budget for a beneficiation plant at $25–50 a ton of throughput and expect a meaningful fraction to fail the ceramic spec anyway. Where the furnace is far away and no deposit exists, dropping glass from the curbside program and taking it at a drop-off point is usually the cheaper answer, and several US municipalities have made exactly that trade.

Melting at about 1,500 °C, 200–400 tonnes a day, 12–15 year furnace campaign · roughly 4–6 GJ and 0.5 t CO2 per tonne of glass, about a fifth of it process CO2 · each 10% cullet cuts furnace energy roughly 2–3% · ceramic, stone and porcelain spec under about 25 g per tonne · mixed cullet $0–30 a ton at the MRF against $0.10–0.20 per ton-mile freight · US furnaces 25–35% cullet, EU average above 50% · beneficiation $25–50 a ton

Strategic Materials, the largest US cullet processor, now part of Sibelco; O-I Glass and Ardagh container plants buying furnace-ready cullet on long-term contracts; the FEVE "Furnace for the Future" hybrid electric container furnace in Germany; Michigan and Oregon deposit systems, which produce color-sortable whole bottles; European countries above 70% cullet in container glass.

This is a freight business with a melting furnace attached. Furnace-ready cullet delivered to a container plant typically sells for $60–120 a ton, beneficiation costs $25–50 a ton, and the raw material leaves a MRF at $0–30 a ton or negative, so the margin exists only when the haul is short. Bulk trucking at $0.10–0.20 per ton-mile puts a hard radius on the business, which is why cullet processors site plants inside a couple hundred miles of a furnace and why the same plant is uneconomic 400 miles away. On the furnace side the value is easy to compute: cullet displaces roughly 1.2 tonnes of raw batch per tonne charged, soda ash is the expensive component of that batch, and the energy saving is 2–3% per 10% cullet, so a plant moving from 30% to 60% cullet cuts melting energy by roughly 6–9%. Carbon pricing strengthens that case, since about a fifth of container glass CO2 comes from carbonate decomposition and cullet is the only way to avoid it without changing the glass chemistry. Capital for a beneficiation plant runs roughly $5–15 million for 100,000–200,000 tons a year, which is small by recycling standards, and the risk in the model is feed quality rather than capital: a stream that fails the ceramic spec has to be sold as aggregate at a fraction of the price.

5 minHow do we recycle glass?Strategic Materials · 100k+ views

13 minWhat *REALLY* happens to 'Recycled' Glass?! - (you might be surprised)JerryRigEverything · 5m+ views

Glass: Material-Specific Data (US EPA) · Supply Chain Energy and Greenhouse Gas Analysis Using the Materials Flows through Industry (MFI) Tool: Examination of Decarbonization Technology Scenarios for the U.S. Glass Manufacturing Sector (NREL)

Paper recycling starts in a hydrapulper, a large tub that drops baled paper into water at 4–6% consistency and agitates it until the sheet falls apart into individual fibers. No chemistry is needed for that step, which is why recycled paper mills are far simpler than kraft mills. What follows is cleanup: coarse screening for plastic and string, centrifugal cleaners for sand, staples and grit, and slotted fine screens down to about 0.15 mm for adhesives. Graphic grades add a deinking loop, where alkaline chemistry and a surfactant detach ink particles and flotation cells carry them off on air bubbles. Yield from old corrugated containers to new containerboard runs 80–92%, while deinked pulp yields 65–80% because the ink, coating clay and fines leave as sludge. The fiber itself gets shorter and stiffer each pass, mostly through hornification, an irreversible collapse of the pores in the cell wall on drying, so the fiber swells less and bonds less every time it goes around.

This is one of the few recycling streams that has paid for itself for a century, and the volumes are large: US recovery is about 71–76% for old corrugated containers and 66–68% across all grades. A recycled containerboard mill also skips the digester and the chemical recovery island entirely, which cuts total energy roughly 40–50% against making the same tonne from wood. The honest qualifier is that a kraft mill burns its own black liquor, so it buys less fossil energy per tonne even though it uses more energy overall, and a recycled mill's purchased electricity and gas can be higher. The structural weakness is fiber length. Tensile strength typically falls 10–20% per cycle and the fines wash out with the process water, so most fiber survives roughly four to seven passes before it is only good for low-grade board or is lost entirely. Recovered fiber supplies close to 60% of the world's papermaking furnish, and the remaining share has to be new fiber, which is not a policy choice but an arithmetic one. Contamination is the day-to-day problem: wet-strength resins, polyethylene-coated cups, laminated packaging and pressure-sensitive adhesives all survive the pulper, and adhesives in particular cause sheet breaks that cost a mill hours of production.

Commercial back-of-store cardboard is worth collecting almost anywhere, because it is clean, concentrated at a loading dock and sells for $60–180 a ton. Residential mixed paper is a different business: it runs $0–60 a ton, it can go negative, and it only works where a domestic mill will take it, so check the mill map before designing a program around it. If a paper stream carries wet strength, a polyethylene coating or heavy adhesive, route it to a mill that has the screening capacity for it rather than assuming any mill can take it, since one bad bale can shut a machine down. If the goal is repeated cycles of the same fiber, plan for a downgrade cascade instead: office paper into newsprint or tissue, corrugated into more corrugated, and eventually out. And if the buyer is an export broker rather than a mill, price in the possibility that the market closes, because it did in 2018 and the recovery took several years of new domestic capacity.

Repulping at 4–6% consistency, fine screens down to about 0.15 mm · yield 80–92% for corrugated to containerboard, 65–80% for deinked pulp · fiber tensile strength falls 10–20% per cycle, roughly four to seven usable passes · recovered fiber is close to 60% of world papermaking furnish · US recovery 71–76% for corrugated, 66–68% all grades · corrugated $60–180 a ton, mixed paper $0–60 · recycled mill total energy 40–50% below a kraft mill

Pratt Industries, which runs 100% recycled containerboard mills in the US; Green Bay Packaging's recycled containerboard mill in Wisconsin; Cascades' Bear Island mill in Virginia, built to absorb East Coast recovered fiber; ND Paper's conversions of former kraft capacity; China's 2018 restrictions and 2021 import ban on recovered paper, which pushed US mixed paper prices below zero before domestic capacity caught up.

The mill buys fiber on a spot market and sells board on a contract, so the spread between old corrugated container prices and containerboard prices is most of the margin. Fiber is usually 30–40% of cash cost at a recycled mill, which makes the operator a price taker on its main input: when corrugated ran above $200 a ton in 2021 the recycled mills were squeezed while integrated kraft mills were not. Capital for a modern recycled containerboard machine is roughly $500 million to $1.5 billion for 400,000–700,000 tons a year, so the plants are few, large and built where the fiber is, which in practice means near cities. On the collection side the economics split sharply by source: commercial cardboard is clean, concentrated and profitable to haul, while residential mixed paper often costs more to collect and sort than it sells for and survives on the rest of the municipal recycling contract. Sludge is the cost that gets left out of models, since 10–25% of the input leaves as a wet residue that is landfilled or burned, and the mill pays a tipping fee on it. Regulation matters less here than in most streams, but recycled-content rules for packaging and landfill diversion targets both add a floor under demand that the commodity price alone would not provide.

4 minHow is Paper Recycled? Learn About the Recycling Process at Pratt IndustriesPratt Industries · 100k+ views

9 minHow Is Paper Recycled?History of Simple Things · 10k+ views

Paper and Paperboard: Material-Specific Data (US EPA) · Paper & Cardboard Recycling Facts (AF&PA)

Construction and demolition debris is the largest waste stream in the US by weight, around 600 million tons a year, more than twice municipal solid waste, and concrete is roughly 70% of it. The process is crushing. A demolition contractor separates the concrete from wood, gypsum and metal, feeds it to a primary jaw crusher, pulls the rebar out with an overhead magnet, sends the product through a secondary impact or cone crusher, and screens it into sizes. Mobile crushing spreads run 100–400 tons an hour and are moved onto the demolition site itself, which is the point: the crusher exists so the material never leaves. The output is recycled concrete aggregate, and almost all of it goes into road base, sub-base, fill and pipe bedding rather than back into structural concrete. The reason is the mortar still stuck to the old aggregate, which makes the particle porous: recycled aggregate absorbs 3–8% water against under 1% for natural stone, and a mix using it at full coarse replacement typically loses 10–25% of its compressive strength.

The tonnage is enormous and the process is simple, robust and a century old, so nothing here is technically hard. It is also genuinely cheaper than the alternative on most sites, because crushing in place avoids both hauling rubble out and hauling virgin base in. The weakness that matters is what the recycling actually displaces. Virgin crushed stone carries roughly 4–8 kg of CO2 per tonne from quarrying and crushing, while the cement in a concrete mix carries about 0.85–0.9 tonnes of CO2 per tonne of clinker and accounts for close to 90% of the concrete's footprint. Recycling concrete replaces the aggregate and does nothing about the cement, so the carbon saving is single-digit kilograms per tonne, mostly avoided truck-miles. It counts as recovery in national statistics all the same, which is why C&D recovery rates look excellent while the emissions in concrete are untouched. Contamination is the other constraint: gypsum wallboard brings sulfate that attacks new concrete and produces hydrogen sulfide in landfill, and older buildings bring asbestos and lead paint that make selective demolition a permitting matter rather than a preference.

Crush on site whenever the demolition produces more than a few thousand tons and there is room for the equipment, because the saving comes from the two truck movements you avoid, not from the value of the product. If the site is small or in a dense city, haul to a fixed recycling yard, which will usually take clean concrete at $0–15 a ton against $40–60 for landfill. Use the output for base course, fill and pipe bedding as the default; if you want it in structural concrete, keep to the 20–30% coarse replacement that most codes allow, use only the coarse fraction, and adjust the mix water for the higher absorption. Do not count on this route for carbon credit, because the emissions in concrete sit in the cement and this process does not touch them. If reducing the footprint is the actual goal, supplementary cementitious materials, lower clinker factors and longer specified curing ages move far more CO2 per dollar than aggregate substitution does.

US construction and demolition debris around 600 million tons a year, roughly 70% concrete · mobile crushers 100–400 tons an hour · recycled aggregate absorbs 3–8% water against under 1% for natural stone · 10–25% strength loss at full coarse replacement, 20–30% replacement typical in codes · virgin crushed stone roughly 4–8 kg CO2 per tonne against 0.85–0.9 t CO2 per tonne of clinker · recycled aggregate sells $6–12 a ton against $14–20 for virgin crushed stone · clean concrete tipping $0–15 a ton against $40–60 for landfill

Mobile jaw and impact crushers from Rubble Master, Metso and Keestrack working directly on demolition sites; state DOT specifications allowing recycled concrete aggregate in unbound base across most of the US; the Netherlands, which reports C&D recovery near 98%, almost all of it as road base and fill; the EU Waste Framework Directive's 70% recovery target for non-hazardous construction and demolition waste, which most member states meet through backfill.

The value per tonne is low enough that only logistics decides the outcome. Recycled aggregate sells for roughly $6–12 a ton against $14–20 for virgin crushed stone at the quarry, and neither one can travel far, since freight on bulk aggregate exceeds the material value within about 30–50 miles. The money is therefore made on avoided cost rather than on sales: a contractor who crushes on site skips $40–60 a ton of landfill tipping plus the trucking to reach it, and skips buying base course at $14–20 delivered, which usually adds up to $20–30 a ton against an operating cost of roughly $3–6 a ton for the crushing itself. Capital is modest by the standards of this sheet, roughly $400,000 to $1.5 million for a mobile crushing and screening spread, and utilization is the whole model because the equipment moves between jobs. Regulation supplies the floor: landfill bans on clean rubble, disposal charges, and DOT specifications that permit recycled base are what keep the material out of landfill in places where the arithmetic alone is marginal. Anyone modeling this as a carbon business should be careful, because the credit is worth a few kilograms of CO2 per tonne and the revenue is worth a few dollars, so it is a disposal-cost business with an environmental benefit attached rather than the other way around.

5 minContractors don't realize that concrete crushing changes your perception of C&D wasteRUBBLE MASTER Americas Corp · 100k+ views

19 minRecycling Concrete Into Profit at Full Circle Aggregates!Mr. DiGG · 100k+ views

Construction and Demolition Debris: Material-Specific Data (US EPA) · Use of recycled concrete aggregate in concrete: a review (Journal of Civil Engineering and Management)

Textile recycling starts with sorting by hand, because the value of a collected bale depends almost entirely on how much of it can be worn again. A typical sorting house resells 10–20% locally, exports 40–50% for reuse, cuts or shreds about 30% into wiping cloths and low-grade fiber, and throws the rest away. The shredding route, called garnetting, tears fabric back into loose fiber, and it shortens the staple from roughly 28 mm for cotton to 10–15 mm, which is too short to spin on its own, so recycled content is usually capped at 20–30% blended with new fiber. Fiber-to-fiber chemical routes attack the polymer instead: glycolysis or methanolysis takes polyester back to monomer, and dissolution takes cotton into a man-made cellulosic pulp. Both need to know what the garment is made of, and most garments are blends, so near-infrared sorting for fiber composition is now the front end of every serious project. Across the US, textiles are recycled at roughly 15% and fiber-to-fiber recycling accounts for well under 1% of what is collected.

The one part that works is resale, which handles most of the material and needs no technology at all. Everything downstream of resale is where the failures are. Polyester and cotton blends are the majority of apparel and cannot be separated cheaply, since a process tuned to dissolve cellulose leaves polyester behind and a process that depolymerizes polyester has to deal with cotton as a contaminant. Elastane makes it worse: 2–5% in a garment is enough to foul both routes, and there is no cheap way to remove it. Dyes and finishes have to be stripped before a recycled fiber can be dyed to a new color, and durable water repellents add a chemical the buyer may not accept. On top of the chemistry, the resale trade that pays for collection is getting weaker, because cheaper garments wear out faster, so the rewearable fraction of what people donate keeps falling while the tonnage rises. The export leg is the honest failure: the EU exported roughly 1.4 million tonnes of used textiles in 2022 against about 0.55 million in 2000, and a large share of what arrives at markets such as Kantamanto in Accra is unsellable and becomes waste there instead.

Treat resale as the business and recycling as the disposal route for what resale rejects, because that is the actual cash flow. If you are building a fiber-to-fiber plant, secure sorted, single-composition feed under contract before securing the process license, since sorting is the binding constraint and near-infrared lines that can do it are still few. Avoid any feed with elastane unless you have tested the specific process on it. If a brand needs recycled content in polyester today, bottle-derived recycled PET is far cheaper and available now, so be clear that it is a different thing from recycling clothing into clothing and that it does nothing for textile waste volumes. And if the plan depends on selling recycled fiber at a premium, check whether the offtake is a signed volume at a price or a public commitment, because this is the stream where that distinction has already decided company outcomes.

US textile recycling roughly 15%, fiber-to-fiber well under 1% of collections · sorting splits to about 10–20% local resale, 40–50% export, 30% wipers and shoddy · garnetting shortens cotton staple from roughly 28 mm to 10–15 mm, capping recycled content at 20–30% · 2–5% elastane is enough to foul both chemical routes · EU used-textile exports about 1.4 million tonnes in 2022 against 0.55 million in 2000 · rewearable grades $200–500 a tonne, non-rewearable near zero · collection and sorting roughly €300–1,000 a tonne

Renewcell's Circulose plant in Sundsvall, Sweden, a 60,000 tonne a year cellulosic dissolution line that filed for bankruptcy in early 2024 despite signed brand commitments; Refashion, the French producer responsibility scheme that has funded national textile collection since 2007 and remains the longest-running one; near-infrared composition sorting lines such as Fibersort; the EU Waste Framework Directive requirement for separate textile collection from January 2025; Kantamanto market in Accra, which absorbs a large share of European and North American export bales.

Nobody makes money on the recycling half of this stream, which is why the sheet marks it a net cost. Collection and sorting run roughly €300–1,000 a tonne depending on how much hand labor is involved, and the only grade that sells for real money is rewearable clothing at $200–500 a tonne; non-rewearable material is worth close to nothing and shredded fiber often sells below the cost of shredding it. That gap is currently closed by export, which moves the disposal cost to a country with cheaper labor and weaker waste infrastructure, and by producer responsibility fees where a scheme exists. France's scheme has funded collection at national scale for close to two decades and is the only long-run evidence of what that costs. The EU's separate-collection requirement from 2025 increases the tonnage collected without creating a buyer for it, so the near-term effect is more sorted material chasing the same thin markets. For an investor the question worth asking first is what fraction of revenue comes from the gate fee or the EPR payment rather than from fiber sales, because in every operating example it is most of it, and a plant modeled on fiber revenue alone has not been demonstrated at scale.

13 minCan we really recycle our old clothes?DW Planet A · 100k+ views

11 minHow 7.5 Million Pounds Of Donated Clothes End Up At A Market In Ghana Every Week | World Wide WasteBusiness Insider · 1m+ views

Textiles: Material-Specific Data (US EPA) · Textiles and the environment: the role of design in Europe's circular economy (European Environment Agency)

Food is the single largest category of landfilled municipal waste in the US, at roughly 24%, and both routes here exist to keep it out. Composting is aerobic: microbes oxidize the carbon, the pile self-heats to 55–65 °C, which is hot enough to kill pathogens and weed seeds, and the operator manages carbon-to-nitrogen ratio near 25–30 to 1, moisture at 50–60% and oxygen above about 5% by turning windrows or blowing air through a static pile. Roughly half the input mass leaves as CO2 and water vapor over 8–16 weeks plus curing, so finished compost is 40–50% of what came in. Anaerobic digestion seals the same material away from air instead, and a consortium of bacteria hydrolyzes it, ferments it to acids, and finally converts those to methane at 35–40 °C for a mesophilic digester or 50–57 °C for a thermophilic one, with 15–30 days of residence time. A tonne of food waste yields roughly 100–160 m³ of biogas at 55–65% methane, which is 600–1,000 kWh of energy content, or about 200–400 kWh of electricity through a reciprocating engine at around 40% efficiency. What is left is digestate, 85–95% of the input mass and mostly water, which has to be dewatered and land-applied within trucking distance.

The climate case is the strongest one on this part of the sheet: a tonne of food waste sent to landfill generates methane for decades, and diverting it avoids roughly half a tonne of CO2 equivalent, which is far more than the value of anything the process produces. Composting is cheap, forgiving and understood by every municipality, and anaerobic digestion adds an energy product plus a smaller footprint per ton. The weakness both share is that the output is nearly worthless. Bulk compost sells for $10–40 a ton and digestate is usually a disposal problem with a nutrient value attached, so the gate fee is the business. Contamination is the operating headache, because plastic film, cutlery and packaging arrive with the food and have to be screened out, most state rules cap physical contamination near 0.5% by weight, and certified compostable plastics do not reliably break down in the residence times a commercial facility actually runs. PFAS is the newer risk, since it concentrates in the finished product and has already restricted land application in some states. Odor complaints close more organics facilities than economics do.

Compost when the feedstock is mostly yard trimmings and food scraps, the tonnage is under roughly 50,000 tons a year, and there is farmland or landscape demand within a short haul, because the capital is a fraction of a digester's and the operation is simpler. Choose anaerobic digestion when the feed is wet and energy-dense (food processing residue, supermarket waste, fats and greases), when land is tight, or when a renewable gas incentive makes the methane worth several times its commodity value. If the local landfill charges $40 a ton and there is no organics ban, do not expect either route to compete on price, since the gate fee has to sit near the landfill's to win tonnage and neither process runs on $40 a ton without help. Where a digester goes next to an existing wastewater plant, take that option first: the digesters, gas handling and permits are already there, and adding food waste to them is the cheapest capacity in this category. And whatever the route, settle the digestate or compost offtake before construction, because a facility that cannot move its product stops accepting feed.

Food is roughly 24% of landfilled US municipal waste · composting runs at 55–65 °C, C:N near 25–30 to 1, 50–60% moisture, 8–16 weeks plus curing · about half the input mass is lost, so compost is 40–50% of the feed · digesters run 35–40 °C mesophilic or 50–57 °C thermophilic, 15–30 day residence · 100–160 m³ of biogas per tonne of food waste at 55–65% methane, about 200–400 kWh of electricity · digestate is 85–95% of input mass · gate fees $30–70 a ton composting, $40–80 a ton digestion, compost sells $10–40 a ton · diversion avoids roughly half a tonne of CO2 equivalent per ton

California's SB 1383, which requires a 75% cut in organic waste disposal and has driven most recent US capacity; the Massachusetts commercial organics ban in place since 2014; EU rules requiring separate biowaste collection since the end of 2023 and capping municipal landfilling at 10% by 2035; food waste digestion added to existing wastewater treatment digesters, which is the cheapest capacity in the sector; the several hundred livestock manure digesters tracked by EPA AgSTAR, many of them upgrading biogas to pipeline-quality renewable natural gas.

Both routes are gate-fee businesses with a byproduct, and treating the byproduct as the revenue is the standard modeling error. A composting facility costs roughly $2–10 million for 30,000–100,000 tons a year and charges $30–70 a ton, against compost sales of $10–40 a ton that rarely cover more than a quarter of operating cost. A digester handling the same tonnage costs roughly $10–25 million, three to five times as much, and earns it back only where the energy is worth something beyond wholesale power. That is where policy dominates: renewable gas from waste sells at commodity value in most of the US and at several times commodity value under California's Low Carbon Fuel Standard and the federal Renewable Fuel Standard, so two identical plants in different states have entirely different returns. Landfill bans and organic waste mandates are what supply the feed, since a hauler will take material to whichever facility is cheapest and the landfill usually is. The costs that get underestimated are contamination handling, which means depackaging equipment and screening losses, and digestate management, because 85–95% of the input mass leaves the plant and someone has to accept it on land within a short haul.

11 min3.7 The Basics of Anaerobic Digestion of BiowasteMunicipal Solid Waste Management · 100k+ views

4 minCompost Facility TourPhoenix Public Works Department · 50k+ views

How Does Anaerobic Digestion Work? (US EPA AgSTAR) · The Science and Engineering of Composting (Cornell Waste Management Institute)

This is the baseline every other entry on the sheet is priced against. A modern sanitary landfill compacts waste into lifts with daily cover over a composite liner, typically a 60-mil HDPE geomembrane on compacted clay, collects leachate from a drainage layer above it, and drills gas wells into the cell once it is filled; US operators carry 30 years of post-closure care on top. The gas is roughly half methane and half CO2 and comes off for decades, and field collection efficiency usually runs 60–85%, so the uncollected share is the environmental problem with the technology. Waste-to-energy takes the same mixed waste and burns it: a grab crane feeds a reciprocating grate, combustion holds 850–1,100 °C, and a boiler raises steam at roughly 40–65 bar and 400–440 °C, well below a coal plant's conditions because chlorine in the waste corrodes superheater tubes. Flue gas then passes selective non-catalytic reduction for NOx, a dry or semi-dry scrubber for acid gases, activated carbon injection for mercury and dioxins, and a baghouse. Net electrical efficiency is 14–22%, which is about 500–600 kWh exported per tonne burned, and rises to 60–80% total energy recovery when the plant also sells heat into a district network. What comes out the bottom is bottom ash at 20–25% of input mass, which is screened for metals and used as road base in several European countries, plus fly ash and air pollution control residue at 3–5%, which the EU classifies as hazardous waste.

Landfill is cheap, tolerates anything, scales without a technology risk, and can be built by a county. Its weakness is methane, since a tonne of mixed municipal waste generates roughly 60–100 kg of it over the decades that the organic fraction decays, and even a well-run gas collection system leaves 15–40% of that escaping; the leachate is a second liability that does not end with the last load, and PFAS in it has become a real cost at sites designed before anyone measured for it. Waste-to-energy cuts the volume going into the ground by about 90% and destroys the organics outright, so the methane never forms and the metals in the bottom ash are recovered. What it costs is capital, roughly ten times a landfill's per ton of annual capacity, and fossil CO2 up a stack: about 1.0–1.1 tonnes of CO2 leaves per tonne of waste burned, roughly half of it biogenic carbon from food, paper and wood and 0.4–0.5 tonnes fossil carbon from the plastic and synthetic textile fraction. Whether burning mixed waste for electricity is good or bad depends on which boundary is drawn, and both answers are defensible. Judged as a power station it is poor, because 0.4–0.5 tonnes of fossil CO2 divided by the 0.55 MWh exported works out to roughly 700–900 kg of CO2 per MWh, at or above a coal plant. Judged as waste management it looks good, because the same tonne would otherwise have generated 60–100 kg of landfill methane, and at 25–40% escaping collection and a 100-year global warming potential near 28 that is roughly 0.4–1.1 tonnes of CO2 equivalent avoided. The trend argues against new build on both accountings, since better gas capture shrinks the credit on one side and grid decarbonization raises the penalty on the other. A plant financed on a 25-year put-or-pay contract also needs guaranteed tonnage, which pulls against a rising recycling target, and Sweden and Denmark now import waste to keep plants they overbuilt running full.

Use landfill where land is available and the tipping fee is $40–60 a ton, because nothing else competes at that price and no amount of process engineering changes that. Install gas collection early rather than at closure, and if renewable gas credits are available in your market, upgrade the gas to pipeline quality rather than burning it in an engine, since the credit is usually worth several times the energy. Consider waste-to-energy only under one of three conditions: the local disposal floor is already above roughly $80 a ton because of a ban or a tax, land for landfill genuinely does not exist, or there is a district heating customer that turns 20% electrical efficiency into 60–80% total recovery. If none of those hold, a plant will need a subsidy for its whole life. Do not sign a 25-year put-or-pay tonnage commitment while your own diversion targets are rising, because the tonnage you guarantee is the tonnage you have promised not to recycle. And if the reason for building is carbon, sort the plastics out first, since they are the fossil half of the stack emissions and the fraction that other routes on this sheet can actually use.

US tipping fees average roughly $55–60 a ton, about $40 in parts of the South and Midwest and over $100 in the Northeast and coastal West · UK landfill tax over £100 a tonne, German incineration gate €80–130 a tonne · combustion at 850–1,100 °C, 14–22% net electrical efficiency, 500–600 kWh exported per tonne, 60–80% total recovery with district heat · bottom ash 20–25% of input mass, fly ash and air pollution control residue 3–5% · landfill gas collection 60–85% efficient, against 60–100 kg of methane generated per tonne of waste · stack CO2 about 1.0–1.1 t per tonne burned, of which 0.4–0.5 t is fossil · waste-to-energy capital roughly $200,000–400,000 per daily ton, so a 1,000 ton a day plant is $200–400 million · gate fee of roughly $75–85 a ton needed to cover it, derived below

Palm Beach Renewable Energy Facility No. 2 in Florida, a 3,000 ton a day plant completed in 2015 for around $670 million and the only large greenfield US waste-to-energy plant built since the mid-1990s; Copenhagen's Amager Bakke, which sells heat into the city district network alongside power; Singapore's Tuas incineration plants feeding the Semakau offshore ash landfill, built because the island has no room for a conventional one; Germany's 2005 ban on landfilling untreated municipal waste, after which the landfilled share of municipal waste fell to about 1%; EPA's Landfill Methane Outreach Program, which tracks roughly 500 operating landfill gas energy projects in the US.

A landfill is the profitable end of the waste business, which is why the large haulers own disposal. Lined cell construction runs roughly $300,000–700,000 an acre and all-in operating cost including closure and post-closure reserves is roughly $15–30 a ton, against gate revenue of $40–60 in the South and Midwest and over $100 in the Northeast and coastal West, so the margin is wide wherever the fee is high. Waste-to-energy has the opposite structure, and the arithmetic is worth doing rather than asserting. Take a 1,000 ton a day plant at $250 million, which is 330,000 tons a year at 90% availability; financed at 8% over 25 years that capital costs about $23 million a year, or roughly $70 a ton, and operations and maintenance add $30–40 a ton. Against that, 550 kWh a tonne sold at $0.05/kWh brings in about $27 a ton. The gate fee therefore has to be roughly $75–85 a ton for the plant to break even, and that number is derived here rather than quoted. It explains the whole geography of the industry: European gate fees of €80–130 a tonne clear it and US fees of $40–60 do not, which is why Germany built dozens of plants after banning untreated waste from landfill in 2005 and the US has built essentially one since 1995. A landfill ban or a landfill tax is the lever that moves all of this, because raising the price of the cheapest option lifts the floor under every route above it at once; the UK tax went from £7 a tonne in 1996 to over £100, and municipal waste to landfill fell from about 80% to under 10% over that period.

17 minThe Hidden Engineering of LandfillsPractical Engineering · 5m+ views

2 minHow it works - Waste-to-EnergyWaste-to-Energy Facts · 100k+ views

6 minHow Singapore Handles Six Million Pounds of Trash Daily | WSJ A to BThe Wall Street Journal · 1m+ views

Energy Recovery from the Combustion of Municipal Solid Waste (MSW) (US EPA) · Municipal waste statistics (Eurostat)

Extended producer responsibility makes the company that puts a product on the market pay for collecting and recycling it at end of life. The producer does not run trucks. It pays a fee per tonne of material sold into the market to a producer responsibility organization, which tenders collection, sorting and processing on behalf of all its members and reports tonnage to a regulator. Fees are set per tonne by material, and the spread is wide because the underlying cost is wide: Recycle BC's 2026 schedule charges 184 cents a kilogram for plastic film against 97 cents for PET bottles, 58 cents for glass and 48 cents for aluminum food containers, and EU schemes commonly run roughly €20–60 a tonne for glass against €500–900 a tonne for flexible plastics. Fee modulation goes one step further and varies the fee within a material by how recyclable the specific format is, so a carbon-black PET tray that a near-infrared sorter cannot see pays a penalty and a clear bottle does not. The scope reaches well past packaging: 35 states and the District of Columbia have enacted 135 EPR laws across 18 product categories, starting with electronics, paint, mattresses and mercury thermostats.

The clearest result is funding. Recycle BC has run British Columbia's residential packaging system on producer money since 2014 and reports packaging recovery near 78%, against about 32% for US municipal solid waste including composting, and most of that gap is that somebody is paying for the collection rather than any difference in equipment. The weakness is that the fee is small next to the packaging decision it is meant to change. At the BC rates a 25 g PET bottle carries about 2.4 cents of fee (C$0.97 per kg times 0.025 kg) and a 5 g flexible pouch about 0.9 cents, so the pouch is cheaper on the fee as well as on freight even though it is the harder format to recycle, and modulation only reverses that if the penalty is large rather than the few percent most schemes apply. Governance is the second problem, because where producers control the organization that sets their own fees the pressure runs toward the lowest fee a regulator will accept, so independent audit of tonnage and cost is what separates a working scheme from a reporting exercise. The third is that an EPR fee funds supply and not demand: it pays to collect and sort more material without creating a buyer for it, so a scheme run with no recycled-content rule alongside can raise collected tonnage and leave the bales sitting.

Use EPR when collection is chronically underfunded and the municipality is carrying a cost it cannot recover from commodity sales, which is most of the packaging and paper stream at today's prices. If the problem is that the output has no buyer, an EPR fee will not fix it and a recycled-content rule with a date and a penalty will. If the problem is stream quality rather than tonnage, a deposit return scheme buys far more purity per dollar, so run one alongside rather than expecting a producer-funded curbside program to reach food grade. Before assuming any scheme will change packaging design, check three things: whether fees are modulated and by how much, who audits the tonnage independently of the producers, and whether the organization pays the full system cost or a share of it. A scheme covering half the cost is a subsidy to municipal collection, and it is worth calling it that rather than expecting design change from it.

Recycle BC 2026 fees 184 cents/kg plastic film, 97 PET bottles, 58 glass, 48 aluminum food containers · EU scheme fees roughly €20–60 a tonne glass against €500–900 flexible plastics · Recycle BC packaging recovery near 78% against about 32% for US municipal solid waste · 35 US states and DC, 135 EPR laws, 18 product categories · about 2.4 cents of fee on a 25 g PET bottle and 0.9 cents on a 5 g pouch, derived here from the BC rates · BC fees year on year: plastic film up 26%, glass up 21%, aluminum down 19%

Recycle BC, producer-funded since 2014 and the most-cited North American benchmark; Germany's dual system under the VerpackG, with the Zentrale Stelle Verpackungsregister as the registry; France's Citeo, which applies bonuses and penalties on top of the base fee; Maine's 2021 packaging law, the first US packaging EPR statute, followed by Oregon, Colorado, California's SB 54 and Minnesota, with Circular Action Alliance approved as the producer responsibility organization in several of them.

The money is a per-tonne fee on producers that arrives in shelf prices, so households pay for collection through what they buy instead of through property taxes. Inside the scheme the fee is a residual: system cost minus whatever the sorted commodities sell for, which is why the BC schedule moves so much year to year. Between 2025 and 2026 the aluminum container fee fell 18.6% while plastic film rose 26.0% and glass rose 20.8%, and none of that reflects a change in how the material is collected. Administration is usually a small share of the total and the collection contracts are the large one, so the margin goes to the haulers and material recovery facility operators who win the tenders rather than to the producer responsibility organization, which is a cost pass-through. For an operator the real prize is risk transfer: a multi-year contract at a fixed fee per tonne turns a business exposed to commodity prices into a contracted service, which is exactly the exposure that pushed dozens of US programs into losses when mixed paper went from about $75 a ton to negative in 2018. For an investor the question to ask about any EPR-dependent business is who can change the fee, on what notice, and whether the contract survives a change of scheme administrator.

3 minExtended Producer Responsibility for Packaging Law (Explained)Natural Resources Council of Maine · 100k+ views

4 minExtended Producer Responsibility for Packaging | The Ellen MacArthur FoundationEllen MacArthur Foundation · 5k+ views

Extended Producer Responsibility (National Conference of State Legislatures) · Extended producer responsibility: How to unlock the environmental and economic potential of plastic packaging waste? (Resources, Conservation and Recycling)

A deposit return scheme adds a refundable charge to the price of a beverage container and pays it back when the empty comes back. Ten US states and every Canadian province run one, alongside Germany, Norway, the Nordics and a growing list of EU member states. Returns go through reverse vending machines in stores, staffed retail counters, or dedicated redemption centers, and the machine reads the barcode, sorts the container by material and refunds the deposit on the spot. Two things follow from that design. The deposit makes an empty container worth money to whoever is holding it, so it comes back even when the holder is not the person who bought it, and the stream is sorted at the moment of return rather than at a plant fed by a mixed cart. The result is the strongest empirical effect of any policy on this sheet: Germany returns about 98% of PET bottles and Norway about 92%, US deposit states run 60–90%, and the Container Recycling Institute puts US containers under deposit at a 64% recycling rate against about 26% for the same containers where no deposit applies.

Purity is what makes this worth the trouble. A reverse vending machine delivers one material per chute with no food waste, no broken glass in the plastic and no paper fiber, so a deposit stream lands well under the roughly 2% foreign material that mechanical PET recycling needs, while a single-stream material recovery facility bale runs 5–15%. Bottle-to-bottle food-grade rPET is effectively built on this stream, which is why brands with recycled-content commitments lobby for deposits they will also have to fund. The rate depends on the deposit being large enough to notice: Oregon exceeded 90% for the first fifteen years of its law, fell to 75% by 2009 as a nickel lost value to inflation, and doubled the deposit to 10 cents in 2018. The costs are real and land unevenly. Retailers take back dirty containers, give up floor space and are paid a handling fee that is typically 1–3 cents a container, and material recovery facilities lose the aluminum cans that are around 2% of their incoming weight and 30–50% of their commodity revenue, so a deposit law and a curbside program are in direct financial conflict even though both are recycling. The side effect that wins the political argument is litter: seven US states reported beverage container litter down 70–83% and total litter down 30–47% after their laws took effect.

Use a deposit when you need food-contact-grade feedstock, because nothing else produces a stream this clean and no amount of sorting equipment at a MRF closes the gap. Set the deposit at a level people notice: a dime returns more than a nickel, and a nickel set decades ago returns less every year. If the goal is tonnage across many materials rather than purity in one, extended producer responsibility does more per dollar of administration, so run the two together rather than choosing. Expect to pay for the collateral damage, and budget for it explicitly: when the deposit pulls cans and bottles out of curbside, the MRF's revenue falls and its contract usually has to be renegotiated at the same time. If a scheme is being designed from scratch, decide early who keeps unredeemed deposits, because that single line decides whether the system is self-funding at a 70% return rate and insolvent at 95%.

Germany about 98% PET bottle return, Norway about 92%, US deposit states 60–90% · US deposit containers 64% recycled against about 26% for non-deposit · Oregon above 90% for 15 years, 75% by 2009, deposit doubled to 10 cents in 2018 · handling fee typically 1–3 cents a container · deposit stream well under the roughly 2% contamination limit for mechanical PET against 5–15% in a MRF bale · cans about 2% of MRF incoming weight and 30–50% of its commodity revenue · beverage container litter down 70–83% and total litter down 30–47% in seven states after implementation

Germany's DPG system, with a €0.25 deposit on one-way containers; Norway's Infinitum; Oregon's BottleDrop network run by the Oregon Beverage Recycling Cooperative; Michigan's 10-cent deposit, the highest in the US for decades; reverse vending machines from TOMRA, Envipco and Sielaff; the EU Single-Use Plastics Directive's 77% separate collection target for plastic bottles by 2025 and 90% by 2029, which is pushing more member states toward deposits.

Three revenue lines fund a deposit system: unredeemed deposits, the material sold, and a producer fee that covers whatever is left. The first one shrinks as the scheme succeeds, which is the counterintuitive part of the model. At a 70% return rate the unclaimed 30% of deposits is a large pool; at Germany's 98% there is almost nothing left, so producers carry more of the cost through the fee. Against that, the material is worth more than curbside material because it is clean, and food-grade rPET has traded $300–600 a tonne above virgin PET in tight periods. Costs are the handling fee of typically 1–3 cents a container, the reverse vending machines and their servicing, transport from thousands of return points, and counting and reconciliation across retailers. Whoever runs the system is normally a non-profit cooperative owned by producers, so the margin does not sit there; it sits with the equipment vendors, the logistics contractors, and the reclaimers who buy a bale that needs almost no cleaning. For an investor the durable position in this chain is a bottle-to-bottle PET plant with contracted access to deposit feed, since the feed quality is set by law rather than by a supplier who can be replaced.

3 minAre deposit return schemes really the best way to cut litter and emissions? | FT Food RevolutionFinancial Times · 10k+ views

4 minHow do deposit return systems work? | Plastic pollution solutionTOMRA · 1k+ views

Bottle Bills (Container Recycling Institute) · Oregon's Evolving Bottle Bill (Oregon Department of Environmental Quality)

Design for disassembly means laying out a product so that it comes apart in a known sequence, with common tools, without destroying the parts worth keeping. The design levers are few and concrete: fastener type, since a screw can be reversed and an adhesive bond or an ultrasonic weld cannot; the number of tool changes, since every change adds handling time; part count and access order, since the target part should not sit under six others; and material pairing, because a metal insert molded into plastic or a painted polymer housing turns two recoverable materials into one contaminated one. The usual metric is disassembly time to the target part or material, which converts directly into labor cost. Regulation now sets floors on this. EU ecodesign rules that took effect in March 2021 require makers of washing machines, dishwashers, refrigerators and electronic displays to supply spare parts for 7–10 years depending on the part, deliver them within 15 working days, and design so they can be replaced with commonly available tools without permanent damage to the appliance. France has scored products since 1 January 2021 on a repairability index running 0 to 10 across five criteria (documentation, ease of disassembly and fastener type, spare parts availability, spare parts price against product price, and a category-specific criterion), and is replacing it with a durability index that adds reliability, applied to televisions from January 2025 and washing machines from April 2025.

Disassembly protects concentration, which is the thing shredding destroys. Gold in a phone sits in plated connectors and bond wires at a few grams per tonne of whole device, and once those grams are spread through shredder dust they cannot be recovered at any sensible cost, so removing the board whole is worth more than any downstream process improvement. Repair beats recycling outright on carbon, because a repaired appliance avoids the whole of a new one rather than a fraction of its materials. The weakness is that disassembly is labor, and at Western wages the recovered material rarely pays for it: hand dismantling costs several times more per tonne than shredding, which is why the highest-recovery manual operations sit in low-wage countries. Design for repair also fights real requirements, above all sealing and thickness, since an IP68 phone is held together by adhesive for a reason and a glued-in battery is thinner than a latched one. And a repairability score measures declared attributes rather than repairs actually performed, so a product can score well while the binding constraint stays where it usually is, on spare part prices and labor rates.

Design for disassembly when the product is high value per unit and has a service life measured in years, which covers machinery, vehicles, appliances, medical equipment and aircraft, because the same design decisions that make a unit repairable make it remanufacturable and the aftermarket is where the money is. If the product is cheap, light and high volume, put the effort into design for recycling instead, meaning single polymers, no laminates, no carbon-black pigments that a near-infrared sorter cannot see, since nobody is going to hand-disassemble a razor handle at any wage. As a practical target, get the recoverable fraction out within a handful of steps using one tool, and keep every fastener on the path the same type. Never bond two materials that both need recovery, because the joint decides which one you lose. If you sell into the EU, treat spare part availability, delivery time and the repairability or durability score as product specifications with dates attached, not as marketing claims, since they are now conditions of sale.

Spare parts required for 7–10 years and delivery within 15 working days under the EU ecodesign rules in force since March 2021 · French repairability index 0–10 across five criteria since 1 January 2021 · durability index applied to televisions from January 2025 and washing machines from April 2025 · hand dismantling costs several times more per tonne than shredding · gold in a phone at a few grams per tonne of whole device · repair is usually declined once it passes roughly a third of the replacement price

Fairphone, built around user-replaceable modules and repeatedly scored 10 out of 10 by iFixit; Apple's Self Service Repair program and the argument over parts pairing that followed it; the French repairability index labels on laptops, vacuum cleaners, dishwashers and electric mowers; the EU battery regulation's requirement that portable batteries be removable and replaceable by the end user from 2027; the Ecodesign for Sustainable Products Regulation, which extends ecodesign rules to nearly all product groups and adds a digital product passport.

The costs land on the manufacturer and the benefits mostly land elsewhere, which is the reason this is a regulated area rather than a competitive one. On the cost side, screws are slower to place than adhesive on an assembly line, modularity adds part count and connectors to the bill of materials, and a 7-to-10-year spare parts obligation is working capital tied up in slow-moving inventory. On the benefit side, the consumer captures the repair saving and the recycler captures the higher recovery, so a manufacturer selling a one-time product sees cost with no matching revenue. That flips where the maker also sells service: for capital goods the parts and repair business usually carries a much higher margin than the original equipment, so serviceable design is bought for the aftermarket rather than for the environment. A useful rule of thumb on the consumer side is that repair gets declined once its price passes roughly a third of a new unit, which is why a $80–330 phone screen replacement is a real decision against a $400–1,200 handset and a $40 kettle is never repaired at all. France now pays a consumer repair bonus of a few tens of euros per job out of the electronics producer responsibility fee, which is an attempt to move money across that split rather than to change the design.

7 minHow to Design for Disassembly and RecyclingAutodesk Sustainability Workshop · 10k+ views

2 minWhat is Right to Repair, and Why is it Important?iFixit · 10k+ views

Ecodesign for Sustainable Products Regulation (European Commission) · Product design for disassembly and bulk recycling (CIRP Annals)

Remanufacturing is an industrial process that returns a used unit to original specification and sells it with the same warranty as new. It is a different activity from repair, which fixes one fault, and from refurbishment, which cleans and tests without restoring tolerances. The line runs core collection, disassembly to component level, cleaning, dimensional inspection against new-part drawings, replacement of wear parts, machining or resurfacing to bring worn surfaces back inside tolerance, reassembly, and a full functional test. What makes the economics work is that the core still carries the expensive shaping: a forged crankshaft, a machined block, a cast housing and a wound stator all keep their geometry through a service life, and buying that back for a core deposit costs a fraction of forging and machining it again. Reuse is the simpler version of the same idea, where the unit goes back into service without being taken apart, which is how pallets, kegs, reusable transport crates and refillable glass bottles work, and a refillable bottle is typically quoted at 25–50 trips before retirement. The most recent comprehensive US survey, by the International Trade Commission, put remanufactured goods production at $43.0 billion or more in 2011, up from $37.3 billion in 2009, supporting about 180,000 full-time jobs and $11.7 billion of exports.

This is the highest value retention available at end of life, and the output is a priced product rather than a commodity bale, which is why remanufacturing carries better margins than anything else on this sheet. The concentration is in a few sectors: aerospace at $13.0 billion, heavy-duty and off-road equipment at $7.8 billion, motor vehicle parts at $6.2 billion and machinery at $5.8 billion in the ITC's 2011 count, with IT products at $2.7 billion and retreaded tires at $1.4 billion, together about 2% of total sales in those sectors. The binding constraint is core supply, and remanufacturers across almost every sector told the ITC that the availability and price of cores was an extremely important competitive factor, because a line with no cores has nothing to run. Core quality is the second problem, since a core that fails inspection becomes scrap after the disassembly and inspection cost has already been spent, so yield is uncertain in a way a new-parts line never is. The third is obsolescence: remanufacturing works where wear happens faster than technical change, and it collapses where the reverse is true, which is why phones and consumer electronics are refurbished and resold rather than remanufactured. Trade rules are a live constraint too, since Brazil, China and India all restrict imports of remanufactured goods, often on quality or safety grounds.

Remanufacture when the product is capital-intensive and durable, the core holds geometry that is expensive to create, and the technology changes slowly: engines, transmissions, turbochargers, injectors, hydraulic pumps, alternators, aircraft components and medical imaging systems all qualify. Do not remanufacture where a cheaper and more capable new product arrives every eighteen months, because customers will buy the new one. Build the core deposit into the price of the new part from the start, since retrofitting a core return program onto an installed base that never had one usually fails, and a good rule of thumb is that the model does not close unless most of the cores come back. If the whole unit still works, resell or refurbish it instead, because that skips disassembly entirely and disassembly labor is the largest cost in the process. And if you are a buyer rather than a maker, ask whether the reman part carries the same warranty as new, since that one term separates genuine remanufacturing from a cleaned used part.

US remanufactured goods production at least $43.0 billion in 2011, up from $37.3 billion in 2009 · about 180,000 full-time jobs and $11.7 billion of exports · aerospace $13.0 billion, heavy-duty and off-road $7.8 billion, motor vehicle parts $6.2 billion, machinery $5.8 billion, IT products $2.7 billion, retreaded tires $1.4 billion · about 2% of total sales in those sectors · reman parts commonly priced at half to two-thirds of new with the same warranty · refillable glass bottles typically quoted at 25–50 trips

Cat Reman, Cummins ReCon and Detroit Reman in engines and drivetrain; ZF and Bosch eXchange in automotive components; engine overhaul programs at GE Aerospace, Safran and Pratt & Whitney, where a narrowbody shop visit runs into the millions of dollars; Xerox and Lexmark cartridge return programs against the independent cartridge remanufacturers; Michelin and Bandag in tire retreading; CHEP and PECO in pooled pallet reuse; Germany's refillable glass bottle system.

The input is a core bought back with a deposit rather than a billet bought at metal prices, so most of the forging, casting and rough machining cost has already been paid once and does not repeat. Remanufactured units commonly sell at half to two-thirds of the new price with an identical warranty, and gross margins usually hold up at that discount because the operations that were removed are the expensive ones, while the operations that were added are labor. That makes labor the dominant cost line and core yield the dominant risk, since every core scrapped after inspection carries sunk disassembly cost. Aircraft engines are the clearest illustration of where the money sits: engines are often sold at thin or negative margin and the return comes over a 25–30 year service life through shop visits and spare parts, which is a remanufacturing business wearing an equipment badge. Printer cartridges show the same structure from the other side, where original-equipment supplies have long carried gross margins above 50% and independent remanufacturers undercut them at roughly a third to half the price, which is the commercial reason behind chipped cartridges and parts pairing. Margin capture follows control of the core: an OEM that takes cores back through its dealer network and holds the technical data keeps the aftermarket, and an independent remanufacturer without either competes on price alone. For an investor the asset worth owning in this chain is the reverse logistics and the core inventory, not the shop floor.

4 minCaterpillar Remanufacturing OverviewCatRCD · 5k+ views

13 minGiving the World's Largest Truck Engines Second Lives!BuildWitt · 100k+ views

Remanufactured Goods: An Overview of the U.S. and Global Industries, Markets, and Trade (U.S. International Trade Commission) · What is Remanufacturing (Remanufacturing Industries Council)

A recycled content rule requires a stated percentage of post-consumer material in a product by a date, with a penalty for missing it. It works on the demand side, which is the opposite of everything else in this class: collection policy makes material, and a content rule makes a buyer. The EU Single-Use Plastics Directive requires 25% recycled content in PET beverage bottles from 2025 and 30% in all single-use plastic beverage bottles from 2030, and the Packaging and Packaging Waste Regulation, in force since 11 February 2025 and generally applicable from 12 August 2026, extends minimum recycled content to essentially all plastic packaging, with the per-category percentages set in its Article 7 and stepping up in 2030 and again in 2040. The percentages differ by polymer and by whether the packaging is contact-sensitive, so read them off the regulation rather than a summary. It also carries a requirement that all packaging be recyclable in an economically viable way by 2030. California requires 15% post-consumer content in covered plastic beverage containers from 2022, 25% from 2025 and 50% from 2030, and its SB 54 producer responsibility law targets 25% source reduction, a 65% recycling rate for single-use plastic packaging and 100% recyclable or compostable packaging by 2032. The traceability question follows immediately, because a percentage claim on a pellet has to be provable, and for mechanically recycled material it is a physical mass balance across one plant while for chemically recycled material it is a bookkeeping allocation across a steam cracker that cannot tell the two feeds apart.

A rule with a date and a penalty creates a committed buyer, which is what a recycler needs to finance a plant, and it is the difference between a mandate and a pledge. Signatories to the Ellen MacArthur Foundation's Global Commitment promised 25% recycled content in packaging by 2025 and reported roughly 12–14%, and several restated the target to 2030, with no consequence. The weakness is that demand policy without collection policy reprices the existing pool instead of growing it: food-grade rPET traded $300–600 a tonne above virgin PET through the tight stretches, and when virgin prices fell on new capacity in 2023 and 2024 some buyers went back to virgin. Mass balance is the second weakness, because the allocation rule decides what a claim means. Under free attribution a producer can assign all the recycled credit to one premium grade, and excluding the fraction burned as process fuel from the denominator raises the claimed percentage again, so two plants with identical physical inputs can advertise very different numbers. Food-contact clearance is a separate gate that a mandate does not open, since a recycled resin without an EFSA opinion or an FDA letter of no objection cannot go into a bottle no matter what the rule says.

Use a content rule when the supply exists and lacks a committed buyer, and pair it with a collection instrument when it does not, because on its own it bids up a fixed pool. If you are a brand negotiating supply, sign a volume at a price rather than announcing a commitment, since the announcement is what disappears when virgin resin gets cheap. If you are financing a recycler, ask whether the offtake is backed by a statutory obligation with a penalty attached, and check the penalty against the cost of compliance, because a fine smaller than the premium is a price and not a rule. On any recycled-content claim, ask which chain-of-custody model it uses before comparing numbers: segregated and controlled blending are physical, mass balance is an allocation, and book and claim is a certificate with no physical link at all. Then ask the two questions that decide a mass balance figure, which are how credits may be attributed across grades and whether fuel use is excluded from the denominator.

EU 25% recycled content in PET beverage bottles from 2025 and 30% in all single-use plastic beverage bottles from 2030 · PPWR in force 11 February 2025 and applicable from 12 August 2026, with per-category minimum content set in Article 7 stepping up in 2030 and 2040 · California 15% from 2022, 25% from 2025 and 50% from 2030 in covered plastic beverage containers · SB 54 by 2032: 25% source reduction, 65% recycling of single-use plastic packaging, 100% recyclable or compostable · Global Commitment signatories at roughly 12–14% against a 25% pledge · rPET $300–600 a tonne above virgin in tight periods · roughly 0.2–0.5 cents of premium on a 25 g bottle at 30% content, derived here

The EU Single-Use Plastics Directive and the Packaging and Packaging Waste Regulation; California's AB 793 minimum content law and SB 54, whose expanded polystyrene food service ware provision required producers to demonstrate a 25% recycling rate by 1 January 2025 and, when that was not shown, resulted in a prohibition on selling the material in the state; ISCC PLUS and REDcert2 as the certification schemes used for chemically recycled feedstock; ISO 22095, which defines the chain-of-custody models the arguments are about; the digital product passport under the Ecodesign for Sustainable Products Regulation.

The premium is paid by the brand owner and reaches the shelf, and per unit it is small. At a $300–600 a tonne rPET premium, a 25 g bottle at 30% content carries 7.5 g of recycled resin, which at $0.30–0.60 a kilogram of premium works out to roughly 0.2–0.5 cents a bottle. That arithmetic is why these rules survive politically and why the binding constraint is availability rather than cost: a brand will pay half a cent, and there is not enough cleared food-grade material to sell it. The value lands with recyclers holding food-contact clearance and contracted access to clean feed, which in practice means deposit-scheme PET, and it lands with them only while the mandate holds, so the durable asset is the offtake contract rather than the plant. Certification bodies take a small fee per audited tonne and carry outsized influence, because their allocation rules set how much certified output a given quantity of recycled feed produces. The risk to model is regulatory rather than technical: if a deadline slips or a penalty is set below the premium, the price gap closes quickly, and 2023 showed how fast buyers move back to virgin when nothing compels them.

3 minISCC System | Explaining the Mass Balance ApproachISCC System · 10k+ views

Packaging waste (European Commission) · SB 54: Plastic Pollution Prevention and Packaging Producer Responsibility Act (CalRecycle)