Multi-Stream Plastic Recycling Plant Design: One Site, Multiple Resins

Updated June 2026 · Reviewed by the Kitech technical team

Good multi-stream plastic recycling plant design is less about buying machines and more about deciding which polymer streams share a line, where they split apart, and how you stop them contaminating each other. A plant that takes PET bottles, HDPE and PP rigids, and PE film off the same tipping floor has to separate those materials by density and polymer type before a single flake get washed. Get the sequence wrong and you ship low-grade regrind; get it right and you sell bottle-grade rPET.

What Is a Multi-Stream Plastic Recycling Plant?

A multi-stream plant process three or more distinct polymer streams on shared or parallel lines, while a single-stream plant is dedicated to one resin. That’s the whole distinction, and it matters because the word “stream” gets used two different ways. Upstream, a material recovery facility (MRF) sorts commingled household recyclables into separate commodities. Inside a plastics plant, “multi-stream” means your line is built to run several of those commodities, bottles, rigids, film, rather than one. Multi-stream recycling earns its place when one site has to turn several incoming waste streams into separate, saleable products.

Here’s the part most equipment brochures skip: multi-stream isn’t automatically the better choice. It’s a feedstock-driven decision. When recyclables are collected commingled (single-stream collection), cross-contamination goes up and the value of the recovered material go down; separated or dual-stream collection typically yields cleaner, higher-value output, according to practitioner fact sheets from organisations like the Institute for Local Self-Reliance. So before you specify a single tonne of equipment, answer one question: how variable is your input?

In the field across 500+ installations in 80+ countries, the plants that struggle are almost never the ones that chose “too simple.” They’re the ones that bought a single line and then started accepting mixed bales, because the moment a second polymer enters, contamination shows up in the pellets. If your supply contract already mixes PET, HDPE, PP and film, design for multi-stream from day one, retrofitting sorting into a single-stream layout is far more expensive than planning for it.

For a fuller view of how these lines fit a site, see our overview of turnkey plastic recycling solutions and how rigid and film streams are configured.

The Polymer Streams a Multi-Stream Plant Must Handle

Each polymer stream behaves differently in water, under heat, and in the shredder, which is exactly why one universal line is a myth. Density is the most useful number in plant design, because float-sink separation is the cheapest, highest-throughput cut you can make. PET sinks; polyolefins float. Everything downstream follows from that split.

10-Stream Polymer Processing Matrix

Density gives you the big split, but it doesn’t fully purify PET. Published research notes that PET density runs roughly 1.33–1.37 g/cm³ and overlaps some contaminants, so float-sink alone won’t deliver bottle-grade material, you need NIR or optical sorting afterward to remove the look-alikes. That single fact is why a credible multi-stream design always pairs a sink-float stage with sensor sorting rather than relying on one or the other.

Plant Layout and Material Flow

A multi-stream plant flows through eight stages, and the layout’s job is to keep clean material moving forward while contaminants drop out early. That sequence run reception → pre-sort → size reduction → sorting → washing → dewatering and drying → pelletizing → storage. Where you place the splits and buffers is the real art. Kitech engineers a custom configuration for each factory, and most plants combine a shared sorting front end with dedicated wash lines to improve the cleanliness of every output stream.

Two layout rules save more money than any single machine choice. First, keep the drying section adjacent to the extruder. If washed flake travels far before it’s dried and pelletized, it re-absorbs moisture in transit, and wet feed degrades pellet quality. Second, film and rigid can’t share a wash line: PE film wraps around conveyors and feeders and holds water, so film lines need larger wash areas, longer drying, and forced feeding, while rigids need stronger crushing and shorter drying. Trying to run both through one wash circuit is the classic cause of conveyor jams and inconsistent output.

A practical example: a contractor commissioning a 1,000 kg/h mixed-rigids-plus-film plant placed the film dryer 30 metres from the film extruder to fit the building. Within a week, film pellets showed moisture streaks and the line was throttled back to 700 kg/h to compensate. Relocating the dryer beside the extruder restored rated output. That lesson is cheap to learn on paper and expensive to learn in steel, which is why layout comes before equipment selection, not after. For component-level choices once the layout is fixed, our guide to the industrial plastic shredder and the rigid plastic recycling line go deeper.

Sorting, Separation and Cross-Contamination Control

Mixed plastics are separated by running density float-sink, near-infrared (NIR) polymer identification, optical colour sorting and metal removal in sequence, no single technology does the whole job. This is the heart of multi-stream design, and it’s where most yield is won or lost. Think of it as a cascade of purity gates, each one removing a different class of contaminant before the material reach the wash line. Sensor-based sorting machinery, from established suppliers such as TOMRA and others, reads each flake and ejects off-spec polymers at the polymer gate.

5-Gate Stream-Separation Cascade

Now the finding that reshapes the whole cascade. NIR sorters can’t reliably identify carbon-black or dark plastics. The carbon-black pigment absorbs the infrared beam, so the sensor see nothing to classify, a black PP tray reads the same as empty belt. The Association of Plastic Recyclers’ own sorting guidance warns that dark colours containing carbon black attenuate reflected NIR energy and are at risk of being mis-sorted. Worse, a full-wrap shrink sleeve made of a different polymer can mask the container underneath and defeat sorting entirely.

“The carbon black in dark packaging absorbs near-infrared light, so the sorter simply can’t read the polymer. We design the cascade assuming a fraction of black and sleeved items will slip through, a manual quality check and a tolerance for that loss is part of an honest plant design, not an admission of failure.”

What do you do about it? Three things. Plan for residual loss rather than promising 100% capture; add manual quality control after NIR for high-value streams; and where the feedstock is yours to influence, push suppliers toward detectable (non-carbon-black) packaging. AI-assisted sensor sorting helps on the rest, peer-reviewed work reports that integrating AI with sensor sorting can lift separation accuracy and throughput by up to 95%, and emerging mid-wave infrared (MWIR) imaging is being developed specifically to crack the black-plastic blind spot. The underlying near-infrared sorting method is long established, see, for example, USPTO patent US 6,610,981 B2 for an early NIR plastic-sorting apparatus.

In Kitech’s field experience commissioning sorting lines, the most expensive mistake is trusting a single NIR pass: because dark and sleeved items slip through, we design in a manual quality-control station and an eddy-current stage for non-ferrous metal, and we size the pre-sort for the dirtiest expected bale rather than the cleanest. That tolerance for a known loss, at roughly 5–10% of a contaminated stream, is what separates a plant that hit its purity target from one that chases it.

Vendor brochures often quote 98–99% purity. Treat those as supplier-stated and directional; anchor your own acceptance specs to documented test methods and the contamination ranges in independent recycler guidance instead. Honest purity targets, backed by a quality-check stage, age better than marketing numbers.

Size Reduction and Washing by Stream

Size reduction and washing are where each stream finally go its own way, and where the design must respect polymer chemistry. Most lines use a single-shaft shredder for the coarse cut (roughly 20–80 mm), followed by a granulator for the fine cut (around 4–12 mm) before washing and extrusion. Smaller flake washes more thoroughly but costs more energy and produces more fines, so the screen aperture you choose is a direct trade between cleanliness and yield.

What size shredder do I need for a given throughput?

As a rough starting point, match installed shredder motor power to target throughput and material: light film at 1,000 kg/h might need on the order of 75–110 kW, while dense mixed rigids at that throughput can demand more because they resist cutting. Honestly, exact sizing depends on bulk density, contamination and flake target, so request a trial with your actual feedstock rather than sizing from a chart. Our deep dive on shredder screen size selection works through the flake-size trade-off, and the shredder blade material guide covers wear on contaminated feed.

Washing is the stage that separates a credible PET line from a hobby operation. PET bottle streams need a hot caustic wash, European PET recycling lines typically run around 85 °C with caustic soda and detergent in multiple steps, and published protocols cite roughly 80 °C with 1–2% sodium hydroxide and a short residence time to dissolve glues and remove labels. That hot caustic step is what frees the adhesive so the label floats away and the flake come out clean enough for food-contact recycling.

For design, the takeaway is simple: if your input includes both PET and film, you’re building at least two wash circuits, not one. That’s the single biggest reason a true multi-stream plant costs more than the sum of its machines, and the reason the layout has to plan the split before the wash, not after.

Pelletizing and Output Quality

Pelletizing converts washed flake into a uniform, saleable pellet, and the choice of pelletizing method and the grade you target decide who buys your output. The common routes are strand pelletizing, water-ring, and underwater pelletizing; film and lightly contaminated streams usually need a degassing extruder to pull out volatiles and residual moisture before the melt is cut. Some buyers want clean washed flake and pelletize it themselves; others pay a premium for finished pellets. Clean rPET feeds bottle-to-bottle lines and polyester fibre, turning post-consumer scrap into a branded input.

A recurring problem Kitech sees on the production floor is operators blaming the pelletizer for poor pellets when the real fault sits upstream: because a pelletizer running at 1,000 kg/h only preserves whatever flake quality the wash line delivered, contamination shows up as black specks and weak melt strength no matter how good the cutter is. For automotive rPP or food-grade rPET, the evidence is in the wash and the sort, not the pellet die.

The design implication is that output quality is set upstream, not at the pelletizer. A pelletizer can’t rescue poorly sorted, poorly washed flake, it just turns contaminated flake into contaminated pellets. If you want bottle-grade rPET, the money goes into sorting and the hot caustic wash; the pelletizer simply preserve what you already achieved. For the equipment itself, see our overview of the plastic pelletizer, and for material-specific routes, our notes on ABS recycling and PLA bioplastic recycling.

Sizing Throughput and Reading the Investment

Throughput is the master variable: it sets equipment size, automation level and operating cost in one decision. Most multi-stream plants fall into three capacity tiers, and which tier fits depends on your feedstock supply and offtake contracts, not on ambition. Oversizing a plant you can’t feed is as costly as undersizing one you can. Across Kitech’s installed base, the most expensive mistake at this stage is oversizing: a Tier 3 line starved of feedstock burns energy and labour for output it never reaches, because much of the connected power draw is fixed and doesn’t fall with throughput. We size to secured tonnes and signed offtake, not to the brochure maximum.

3-Tier Capacity Ladder

Now the operating cost, which is where design decisions show up on the electricity bill. Energy data has to be read carefully because it depends on the process boundary. The EU’s Joint Research Centre estimated around 61.1 kWh per tonne of input waste for granulation-centric processing, and regranulation alone runs roughly 0.18–0.24 kWh/kg. But a full plant that washes, dries and extrudes is much higher: peer-reviewed life-cycle work reports on the order of 883–1,132 kWh per tonne depending on the drying load. That gap between the figures isn’t a contradiction, it’s the difference between one process step and the whole line.

Concretely, the design lesson is this: energy concentrates in washing and extrusion, so those are the stages to engineer for efficiency, not the shredder. Energy consumption and labor costs are the operational drivers that decide whether a full line pay back. Thermal drying in particular is an energy sink, which is one more reason to keep film, which carries the most water, on a dedicated, well-designed dewatering line. For the full cost model, including capex, labour and payback, see our plastic recycling plant cost guide for 2026, which covers capex, opex and payback in detail.

What’s Changing in Multi-Stream Plant Design for 2026

Regulation, not market size, is the biggest force reshaping plant design in 2026. The EU Packaging and Packaging Waste Regulation (PPWR, Regulation 2025/40) entered into force on 11 February 2025 and generally starts applying on 12 August 2026, with the stated aim that all packaging be recyclable in an economically viable way by 2030. For a plant operator that means rising recycled-content obligations and extended producer responsibility pressure, which in turn rewards flexible, multi-polymer lines that can adapt as packaging mixes shift. If you’re specifying a plant now, design in spare sorting passes and modular wash capacity so you can add a stream later without rebuilding.

On the technology side, AI-integrated sensor sorting is the practical advance: peer-reviewed reviews report accuracy and throughput gains of up to 95%, and MWIR imaging is being developed to finally read the black plastics that defeat conventional NIR. Design-for-recyclability, mono-material packaging, detectable inks, removable sleeves, is the clearest packaging trend, and the Association of Plastic Recyclers’ design criteria are moving toward a national framework. These shifts make the input stream cleaner and more sortable over time, which is exactly why over-fitting a plant to today’s contamination is a mistake.

As a manufacturer, Kitech is seeing buyers de-risk against this by specifying modular sorting passes they can populate later: because the cost of retrofitting a second NIR unit into a finished building is far higher than leaving the floor space and power for it now, the challenge is to design the headroom in from day one. For an EU project facing the 2026 deadline, that headroom is what turns a compliance risk into a planned, low-cost upgrade.

Market forecasts do exist, the recycled-plastics and NIR-sorter markets are both projected to grow at high-single-digit annual rates through the early 2030s, but treat those as directional background, not a basis for sizing. Load-bearing decisions here are regulatory timing and sorting capability, not a CAGR. If your 2026 project depends on EU offtake, anchor your design and compliance dates to PPWR’s 12 August 2026 application date and build in the headroom to meet tightening recycled-content rules.

Frequently Asked Questions