Revision 2 keeps the founding idea, one plant that accepts co-mingled waste and returns three independently sellable materials, and rebuilds the process around it. The bioreactor, the zeolite hydrolysis bed, the acid leach and the H₂S-to-acid loop are removed. A monomer-purification cascade, a vacuum-vented compounding line and a multi-sensor front end are added. The cost model is restated with a realistic Lang factor, an itemized opex, an energy balance and a market-checked product price, and the plant is staged so that the risky chemistry is built last.
NPMAI ECOSYSTEM is an research and development company that publishes across chemistry, process engineering, and applied AI. It does not treat "research" and "shipped tooling" as separate departments — the belief underneath both is that serious technical work should be published in the open, checked by whoever wants to check it, and built on by the next person for free.
This paper is part of the organisation's ongoing programme on circular-economy reactor design — taking mixed, difficult-to-sort post-consumer waste streams and building a single plant capable of recovering more than one saleable material from them at once, rather than requiring feedstock to already be sorted into a single clean stream before it can be processed at all.
Mixed post-consumer waste — co-mingled plastic, rubber, and carbon-filled material arriving together rather than pre-sorted — is the norm for almost every real waste stream, and the norm for almost every recovery process is to treat it as a problem rather than a feedstock. Single-stream routes solve one fraction and send the rest to landfill or incineration.
This paper reports a plant designed around the opposite assumption. A shared front end — bale breaking, metal removal, multi-sensor sorting, float–sink and ballistic separation — splits mixed waste into three lines that share sorting, water, heat, nitrogen, vacuum and control infrastructure. The PET line washes flake in hot caustic, depolymerizes it by glycolysis to bis(2-hydroxyethyl) terephthalate (BHET), purifies the BHET by filtration, carbon and ion-exchange polishing and crystallization, and repolymerizes it to rPET chips. The rubber line devulcanizes crumb thermo-mechanically in a twin-screw extruder with process oil and captures its own off-gas in a dry scavenger. The carbon line washes and compounds carbon-black-filled polyolefins into black pellets in a vacuum-vented twin-screw extruder.
Revision 1 of this paper proposed a different PET route (acid leach, zeolite hydrolysis and an Ideonella sakaiensis bioreactor), a DPDS-mediated rubber route, and an internal H₂S-to-H₂SO₄ acid loop. A line-by-line review found that its strongest claims did not hold: the organism consumes the monomers, no purification step existed between hydrolysis and repolymerization, the acid loop was worth about $68k/yr, and the fiber price had no market comparable. Revision 2 replaces those elements and restates the economics. Table 1 lists every change and the reason for it.
| Rev 1 element | Rev 2 decision | Reason |
|---|---|---|
| 36 % H₂SO₄ acid leach (108) with Alloy 20 / PTFE-lined vessels, applied to all feed | Replaced by a hot caustic wash for the PET and polyolefin lines only; rubber sees no acid | The acid leach would strip ZnO from rubber and forces corrosion-resistant alloys plus neutralization. Hot caustic wash is the standard PET flake wash and runs in stainless steel. |
| H-ZSM-5 zeolite fluidized-bed hydrolysis (112), 165 °C at 2.5 bar | Replaced by glycolysis to BHET (118): 190–200 °C, atmospheric, EG as reagent and solvent | Polymer chains cannot enter zeolite micropores, and the catalyst never appeared in the kinetic model. Saturated steam at 165 °C is about 7 bar, so 2.5 bar meant superheated steam and a low water concentration. Hydrolysis also leaves EG dilute in a large water volume. |
| DFB quench and fines trap (114, 116) | Removed; hot filtration (120) takes over the solids-removal role | Cooling to 30 °C existed only to protect the bioreactor. Nothing downstream needs it. |
| Ideonella sakaiensis immobilized-frame bioreactor (118), wiped-film dehydrator (120), disc-stack centrifuge (122) | Removed | The organism assimilates TPA and EG, so it consumes the product [1]. It needs pH 7.2 against an acidic hydrolysate, an 18 h residence time (hundreds of m³), sparging, CIP and biomass removal, and it does nothing for nylon. |
| No monomer purification between hydrolysis and repolymerization | Added: EG evaporation (122), aqueous carbon + ion-exchange polishing (124), BHET crystallization (126–128), drying (130) | Spinnable-grade polymer needs polymer-grade monomer. Dyes, catalyst residue and oligomers must be removed and EG recovered. |
| Joint PET / nylon 6,6 hydrolysis and separate repolymerization (124, 126) | Nylon removed from the chemical line; sorted and baled (108) | Mixed monomer streams cannot be repolymerized cleanly. Nylon 6,6 also needs nylon-salt crystallization and a pressurized autoclave stage (it melts near 265 °C and HMDA is volatile), for a small share of feed. |
| Ti(OBu)₄ at 300 ppm Ti; Sb₂O₃ rejected | Corrected to 10–20 ppm Ti plus a phosphorus stabilizer (Sb₂O₃ kept as a fallback for non-food fibre) | 300 ppm is antimony-level dosing; at that level a Ti catalyst causes yellowing and thermal degradation. |
| Coaxial spinning with pH 4 / pH 9 coagulation baths (128) | Deferred to Phase 3 as conventional melt spinning; base plant sells rPET chips | Melt-spun PET and nylon are not coagulated. A spinning line is also a separate business with its own qualification and customers. |
| DPDS devulcanization autoclave (130), solvent wash (132), Banbury (134), re-vulcanization press (136) | Replaced by a twin-screw thermo-mechanical devulcanization extruder (142) with process oil, a two-roll mill (146) and a QC lab (148) | DPDS at 2.5 phr is about 495 t/yr at 200 t/d, a large reagent bill against the rubber revenue line. The extruder route is documented in the patent literature [11], and customers re-vulcanize themselves. |
| H₂S combustion, V₂O₅ SO₂ converter and SO₃ absorber (144–150) feeding acid back to the leach | Removed; dry scavenger bed + caustic polish scrubber + VOC control (144) | Rev 1's own benchmark credit was about $68k/yr, roughly 0.2 % of opex, for three pieces of exotic equipment. With no acid leach there is no acid sink. Off-gas is still contained, sized for 100 % of flow. |
| N₂ purge at the carbon extruder | Replaced by two vacuum vents | Vacuum venting removes moisture and volatiles, improving odour and mechanical properties at lower cost. |
| NIR + density sorter (104) | Upgraded to multi-sensor sorting, float–sink and ballistic separation, with an explicit reject stream | NIR cannot see black plastics; PVC must be removed; nylon and rubber overlap in density. |
| Lang factor 2.2; fiber at $3,000/t; 50 / 30 / 20 feed split with no rejects; mass balance only | Restated: Lang 3.1 with 15 % contingency; rPET chips at $1,500/t; 45 / 28 / 17 split with 3 % nylon and 7 % rejects; full energy balance and itemized opex | Rev 1 understated installed cost, used a fiber price with no market comparable, and had no reject stream or energy balance. |
Eight stages carry mixed waste to three finished materials. Each is specified with its governing reaction or physical principle and its operating window. Windows are design targets drawn from the published glycolysis, reclaim and compounding literature and must be confirmed at bench and pilot scale.
No chemistry occurs here, but the plant's product quality is set here. Bales are opened, metals are removed by overband magnet and eddy current, and a multi-sensor sorter combines near-infrared for PET, PP, PE and nylon, mid-infrared or hyperspectral imaging for dark polymers that near-infrared cannot see, and an XRF or chlorine signal that ejects PVC. A float–sink hydrocyclone bank separates polyolefins (density below 1.0 g/cm³) from PET, rubber and nylon (all above 1.0). Because rubber and PET both sink, an elastic/ballistic separator splits resilient rubber from rigid fragments. Nylon is baled for sale and the ≈7 % unsortable residue is routed to RDF or disposal. Design targets for PET flake entering the glycolysis reactor are PVC ≤ 100 ppm and metals ≤ 50 ppm.
PET is granulated to ≤ 8 mm and washed in friction washers at 85–90 °C with 1–2 wt% NaOH and about 0.3 wt% surfactant for 15–20 minutes, then rinsed to neutral in a three-stage counter-current train and dried to below 0.5 wt% moisture. This removes label adhesive, oils and food residue. It is the standard PET flake wash and needs only stainless steel. The dark polyolefin fraction is washed in a similar but smaller line (150). Rubber is not acid- or caustic-washed; it is cleaned mechanically (Section 2.7).
Dry PET flake is depolymerized in ethylene glycol (EG : PET ≈ 3 : 1 by mass) with zinc acetate at about 0.5 wt% of PET, at 190–200 °C and atmospheric pressure, for about two hours. The vessel is agitated, condensed and blanketed with nitrogen: EG is combustible, and oxygen exclusion prevents discolouration. Moisture is held below 0.5 wt% to limit hydrolytic side reactions.
Glycolysis replaces Revision 1's hydrolysis for three reasons. EG is both reagent and solvent, so no large water inventory has to be heated and later evaporated. The step runs at atmospheric pressure in 316L. And the product, BHET, can be purified by crystallization and fed to polycondensation without an esterification stage.
This block did not exist in Revision 1 and is what makes spinnable-grade polymer possible.
Molten BHET (melting point ≈110 °C) enters a prepolymerizer and then a finisher at 265–280 °C, with pressure stepped down to below 1 mbar over about 2.5–3 hours. Titanium catalyst is dosed at 10–20 ppm (Ti basis), for example titanium(IV) butoxide or a chelated Ti, with a phosphorus stabilizer and an organic toner for colour. Antimony trioxide at 200–250 ppm Sb remains the low-cost, colour-stable fallback where the end use is non-food fibre or strapping.
The melt is filtered, pelletized underwater, crystallized and dried below 50 ppm moisture. Spinning is not part of the base plant. If built (Phase 3), it uses conventional melt spinning: dried chips, extruder, spin pack, cross-flow quench air, godets and winder to give POY or FDY yarn. Melt-spun PET is quenched in air, not coagulated in a bath.
Rubber is granulated to ≤ 2 mm, freed of fibre and steel, and screened. It is devulcanized in a co-rotating twin-screw extruder (L/D ≥ 40) at a peak barrel temperature of 200–220 °C, 150–300 rpm and 2–4 minutes residence, with 3–5 phr of aromatic-free process oil. A twin-screw route to reclaim is described in the patent literature [11]. Compared with Revision 1's batch autoclave with DPDS at 12 rpm and 30 minutes, the extruder is continuous, needs no expensive reagent, and shares its technology base with the carbon line.
| Bond | Dissociation energy | Role |
|---|---|---|
| S–S (network) | ~240 kJ/mol | Primary crosslink — target for cleavage (polysulfidic links are weaker still) |
| C–S | ~272 kJ/mol | Secondary crosslink |
| C–C (backbone) | ~347 kJ/mol | Must survive to preserve polymer properties |
| C–H | ~413 kJ/mol | Not cleaved in the operating window |
Off-gas (144). A hood at the die and vent feeds a dry scavenger bed (ZnO or iron oxide), a caustic polish scrubber sized for 100 % of off-gas flow including upsets, and VOC control (carbon or thermal oxidation, sized after assay).
The product is sheeted and cooled on a two-roll mill (146) and baled. The plant contains no re-vulcanization press; customers compound and cure reclaim themselves. A QC laboratory (148) measures cure behaviour, Mooney viscosity, tensile properties and crosslink density.
Washed, dried dark polyolefin is compounded in a co-rotating twin-screw extruder with a four-zone profile of 180 / 200 / 220 / 220 °C at 120 rpm and about 90 seconds residence. Two vacuum vents at 50–100 mbar absolute remove moisture and volatiles, which replaces Revision 1's nitrogen purge, cuts operating cost and improves odour. A continuous screen changer (156) filters the melt and a strand or underwater pelletizer (158) finishes the pellets. This is a physical step; the carbon black stays in the compound.
The apparatus is specified with a reference numeral and a defined function for every component. Solid boxes are in the reference plant; dashed boxes are optional or supporting units.
FIG. 1 — Revision 2 process flow. Sorted streams leave the shared front-end on the right-hand trunk and enter the PET, rubber and carbon lines. Dashed red: optional Phase 3 spinning and the QC laboratory.
| No. | Component | Function |
|---|---|---|
| 100 | Feed receiving hopper & bale breaker | opens bales and meters mixed waste onto the sorting line |
| 101 | Overband magnet + eddy-current separator | removes ferrous and non-ferrous metal before size reduction |
| 102 | Multi-sensor sorter | NIR for PET / PP / PE / nylon; MIR or hyperspectral imaging for dark polymers; XRF or chlorine signal to eject PVC |
| 104 | Float–sink hydrocyclone bank | splits polyolefins (< 1.0 g/cm³) from PET, rubber and nylon (> 1.0 g/cm³) |
| 106 | Elastic / ballistic separator | separates resilient rubber pieces from rigid plastic fragments |
| 108 | Nylon-bale and reject bunkers | sorted nylon / engineering plastics baled for sale; ≈7 % residue routed to RDF or disposal |
| 110 | PET granulator + pre-wash | sizes PET to ≤ 8 mm flake and removes loose dirt |
| 112 | Hot caustic wash | 85–90 °C, 1–2 wt% NaOH + surfactant in friction washers; strips labels, adhesive and food residue |
| 114 | 3-stage counter-current rinse + water loop | rinses flake to neutral pH and recycles rinse water |
| 116 | Flake dryer | dries PET flake to < 0.5 wt% moisture ahead of glycolysis |
| 118 | Glycolysis reactor | agitated, N₂-blanketed, 316L; 190–200 °C, EG : PET 3 : 1, zinc acetate; converts PET to BHET and oligomers |
| 120 | Hot filter | pressure-leaf or candle filter; removes undissolved solids, PVC char and metal fines |
| 122 | EG / water multi-effect evaporator | vacuum evaporation recovers excess EG to 118 and 132 and concentrates the BHET |
| 124 | Aqueous dissolution + polishing | BHET concentrate dissolved in hot water; activated carbon removes colour bodies; cation exchange removes Zn²⁺ |
| 126 | BHET cooling crystallizer | crystallizes BHET from water; mother liquor recycled to 118 |
| 128 | Pusher / peeler centrifuge | separates and washes BHET crystals |
| 130 | BHET dryer | dries BHET to polymer-grade monomer |
| 132 | Polycondensation train | prepolymerizer + finisher, 265–280 °C, pressure stepped to < 1 mbar; Ti catalyst (10–20 ppm) + phosphorus stabilizer |
| 134 | Melt filter + underwater pelletizer | filters the melt and produces rPET chips; chip crystallizer / dryer downstream |
| 136 | Melt-spinning line (Phase 3, optional) | chip dryer, extruder, spin pack, quench air, godets and winder for POY / FDY yarn |
| 140 | Rubber shredder / granulator + screens | reduces to ≤ 2 mm crumb; fibre and steel removal |
| 142 | Twin-screw devulcanization extruder | thermo-mechanical devulcanization, 200–220 °C peak, process oil 3–5 phr |
| 144 | Off-gas capture | hood at die and vent, dry ZnO / iron-oxide scavenger bed, caustic polish scrubber, VOC control |
| 146 | Two-roll mill / sheeter / baler | homogenizes, cools and packs reclaimed rubber |
| 148 | QC laboratory | curemeter, Mooney viscometer, tensile testing, crosslink-density (swelling) test |
| 150 | Friction washer + sink-float tank | washes the dark polyolefin fraction |
| 152 | Polyolefin flake dryer | dries washed flake |
| 154 | Twin-screw compounding extruder | 180 / 200 / 220 / 220 °C, 120 rpm, ≈90 s; two vacuum vents replace an N₂ purge |
| 156 | Continuous melt filter | screen changer for uninterrupted filtration |
| 158 | Pelletizer | strand or underwater pelletizing |
| 160 | Wastewater treatment & water recycle | treats caustic-wash and rinse water; returns recycled water to 112–114 and 150 |
| 162 | Steam / thermal-oil heating plant | supplies 118, 122, 132; recovers heat from 122 and 132 vapours |
| 164 | N₂, compressed air, cooling water, vacuum utilities | inert gas for 118 / 132; vacuum for 122 and 132 |
| Zone | Materials of construction | Basis |
|---|---|---|
| Hot caustic wash, rinse, polyolefin wash (112, 114, 150) | 304 / 316L stainless steel | 1–2 wt% NaOH at 85–90 °C; no acid duty, so no high-alloy or PTFE-lined vessels |
| Glycolysis reactor, hot filter, EG evaporator (118–122) | 316L; 2205 duplex if chloride in the filtrate is measured above trace | 190–200 °C hot glycol; N₂-blanketed; chloride from PVC carry-over is the corrosion risk; no copper alloys |
| Polishing, crystallizer, centrifuge, dryer (124–130) | 316L | Aqueous BHET, 20–80 °C |
| Polycondensation train (132) | 316L with thermal-oil jackets | 265–280 °C under deep vacuum |
| Devulcanization extruder (142) | Nitrided or hard-faced barrel and screws | Abrasive fillers and sulfur species at 200–220 °C |
| Off-gas capture (144) | Carbon-steel scavenger vessels with lining; 316L or polypropylene-lined caustic scrubber | H₂S, mercaptans and VOC; wet caustic duty in the scrubber |
| Compounding extruder (154) | Carbon steel with hardened barrel liner | Standard compounding duty, 220 °C maximum |
The individual reactions are, by design, not the novel element. What this section proposes as new is the integration of one sorting-led front end, one utility system and three purpose-built lines with closed internal loops.
| Element | Structural novelty proposed | Functional effect |
|---|---|---|
| Shared multi-sensor front-end and shared utilities (100–108, 160–164) serving three chemically distinct lines | One sorting, water, heat, nitrogen and vacuum system for PET, rubber and carbon lines, with the sorter acting as the plant's quality control | Removes duplicated infrastructure; utilities and offsites and the front end are the largest shared capex lever |
| Glycolysis–filtration–evaporation–polishing–crystallization cascade with a closed EG and mother-liquor loop (118–130, 132) | Mixed-colour, post-consumer PET flake is converted to polymer-grade BHET in a plant that also handles rubber and polyolefin; EG released in polycondensation returns to glycolysis | Virgin-grade monomer from unsorted colour; no water-heavy hydrolysis and no biological stage |
| Dry sulfur-capture chain on the rubber extruder (142 → 144) | Hood, scavenger bed, 100 %-rated caustic polish and VOC control as a containment chain, not a recovery loop | Contains H₂S and mercaptans at the source at low capex |
| Vacuum-vented carbon compounding (154) | Two-vent vacuum degassing in place of an N₂ purge | Lower operating cost and better odour and mechanical properties |
Every figure in Sections 5 and 6 traces to an equation below or to a stated assumption in Table 7a. Parameters marked as measured (Ea, De, KF, χ, K, a) are determined in bench work, not assumed from generic literature values.
Basis: 1,000 kg (1 tonne) of mixed feed entering the front-end (100). Unlike Revision 1, the balance carries a nylon bale and a front-end residue, and it includes an energy balance.
| Stream | Feed in (kg) | Yield chain | Out (kg) | Product |
|---|---|---|---|---|
| PET train (110–134) | 450 | 95 % wash and flake × 84 % BHET recovery (mother-liquor recycled) × 96 % polycondensation and pelletizing = 76.6 % | 344.7 | rPET chips |
| Rubber train (140–148) | 280 | 92 % crumb clean-up × 93 % devulcanization and screening = 85.6 % | 239.6 | Reclaimed rubber |
| Carbon train (150–158) | 170 | 95 % wash × 95 % melt filtration and pelletizing = 90.3 % | 153.4 | Black polyolefin pellets |
| Nylon / engineering plastics | 30 | Sorted and baled (not credited in TEA) | 30.0 | Nylon bale |
| Front-end residue | 70 | Moisture, dirt, labels, PVC, metals, unsortable items | 0.0 | RDF / disposal |
| Total | 1,000 | Blended product yield 73.8 % | 767.7 | incl. 30.0 kg nylon bale |
Scaled to the 200 t/day reference case over 330 operating days (66,000 t/yr): rPET chips ≈ 22,753 t/yr, reclaimed rubber ≈ 15,811 t/yr, black polyolefin pellets ≈ 10,126 t/yr, nylon bale ≈ 1,980 t/yr and residue ≈ 4,620 t/yr. In the PET train, 105.3 kg of each 450 kg is lost: about 22.5 kg in washing and flake preparation, about 68.4 kg in BHET purification purges, and about 17.2 kg in polycondensation and pelletizing.
| PET-line thermal duty | GJ per t PET in |
|---|---|
| Heating glycolysis charge to 195 °C (after hot-EG recycle) | 1.65 |
| EG recovery: 2.68 t excess EG × ≈0.85 GJ/t | 2.28 |
| BHET crystallization and drying | 0.20 |
| Polycondensation melt heating and hold | 0.55 |
| Caustic wash heating and flake drying | 0.60 |
| Gross thermal duty | 5.28 |
| Heat-integration credit (≈34 %, multi-effect EG recovery and vapour heat recovery) | −1.78 |
| Net purchased thermal energy | 3.50 |
| Area | Electricity (kWh/t feed) | Thermal (GJ/t feed) |
|---|---|---|
| Shared front-end (100–108) | 45 | — |
| PET train (110–134) | 202.5 | 1.575 |
| Rubber train (140–148) | 126.0 | — |
| Carbon train (150–158) | 59.5 | 0.119 |
| Utilities and offsites (160–164) | 10.0 | — |
| Total per tonne of mixed feed | 443.0 | 1.694 |
The thermal duty in Table 6a is dominated by EG recovery. Revision 1's hydrolysis route evaporates water instead. As an illustration, at an assumed 10 wt% monomer loading the hydrolysate carries about 9 t of water per tonne of PET, and evaporating it costs about 2.4 GJ per tonne of water, above 20 GJ per tonne of PET before any heat recovery. Loading was not stated in Revision 1, so this is illustrative, but the order of magnitude is why the water-based route was dropped.
The analysis uses the six-tenths-rule scaling of Eq. 25–27 on the 31-item equipment list of Tables 2 and 9. Unlike Revision 1, opex is itemized, prices are checked against 2026 market assessments where they exist, and results are compared with Revision 1 on a like-for-like basis. All figures are AACE Class 5 concept-screening estimates [12], not vendor quotations.
| Item | Assumption | Basis and note |
|---|---|---|
| Operating days, feed | 330 d/yr; 200 t/d reference (66,000 t/yr) | As Rev 1 |
| rPET chips | $1,500/t base; sensitivity $1,200–1,800 | Argus (6 Feb 2026) assessed food-grade rPET pellets at about $1,225/t (Asia), $1,521/t (US West Coast) and $1,654/t (NW Europe), and colourless flake at about $810–1,130/t [13]. $1,500/t is in the upper half of that range and is defensible only if BHET-derived chips earn a chemical-recycling premium over mechanical and virgin material. Fibre demand was reported weak in Q1 2026 [14]. |
| Washed PET flake (Phase 1) | $700/t | Below the colourless-flake range above, since flake is mixed-colour |
| Reclaimed rubber | $450/t; sensitivity $300–700 | Carried over from Rev 1; no public price benchmark was located |
| Black polyolefin pellets | $750/t; sensitivity $600–900 | Argus lists transparent rLDPE pellets at about $735–1,160/t [13]; pellets from dark carbon-filled feed price at or below the low end |
| Nylon bale | Not credited | Upside only |
| Feedstock | $60/t delivered | Assumption; may be negative for some streams (sensitivity $0–150) |
| Energy | Electricity $0.10/kWh; heat $10/GJ | Site-dependent |
| Labour | 14 + 0.3 × t/d full-time staff at $45,000 loaded (74 FTE at 200 t/d) | Includes a minimum crew |
| Maintenance; insurance, G&A and QA | 3.5 % and 2.0 % of installed capex per year | Typical process-industry ranges |
| Installed-cost factor | Lang 3.1 × 1.15 contingency | Eq. 26 |
| Residue disposal; packaging and logistics | $90/t of residue; $25/t of product | Assumption |
| t/day | t/yr | Capex | Revenue | Opex | EBITDA | ROI | Payback (yr) |
|---|---|---|---|---|---|---|---|
| 1 | 330 | $4.52M | $0.24M | $0.97M | −$0.72M | -16.0% | n/a |
| 10 | 3,300 | $17.05M | $2.44M | $2.44M | $0.00M | 0.0% | n/a |
| 30 | 9,900 | $35.09M | $7.33M | $5.17M | $2.15M | 6.1% | 16.29 |
| 50 | 16,500 | $49.15M | $12.21M | $7.69M | $4.52M | 9.2% | 10.87 |
| 75 | 24,750 | $64.28M | $18.31M | $10.69M | $7.62M | 11.9% | 8.44 |
| 100 | 33,000 | $77.79M | $24.42M | $13.61M | $10.80M | 13.9% | 7.20 |
| 150 | 49,500 | $101.85M | $36.63M | $19.29M | $17.34M | 17.0% | 5.87 |
| 200 | 66,000 | $123.35M | $48.84M | $24.83M | $24.01M | 19.5% | 5.14 |
At 1 t/day EBITDA is negative and at 10 t/day it is about zero, because a minimum crew and equipment cost floors are spread over too little product. The plant earns a positive EBITDA above roughly 10 t/day, but return on capital only becomes attractive at scale: ROI is 13.9 % at 100 t/day and 19.5 % at 200 t/day, with payback falling from 7.2 to 5.1 years. Small-scale deployment is not recommended.
Revenue is $48.8M: rPET chips $34.1M (70 %), reclaimed rubber $7.1M and polyolefin pellets $7.6M. The PET train supplies most of the revenue and about 42 % of equipment cost, so the plant's economics are the PET chip price.
| Cost line (200 t/day) | $/yr | Share |
|---|---|---|
| Feedstock | $3.96M | 16.0% |
| PET chemicals & consumables | $3.24M | 13.1% |
| PET electricity | $1.34M | 5.4% |
| PET thermal energy | $1.04M | 4.2% |
| Rubber chemicals & media | $1.12M | 4.5% |
| Rubber electricity | $0.83M | 3.3% |
| Polyolefin chemicals & consumables | $0.25M | 1.0% |
| Polyolefin electricity + thermal | $0.47M | 1.9% |
| Front-end electricity & consumables | $0.56M | 2.3% |
| Utilities, offsites (electricity, N2, air, water) | $0.26M | 1.1% |
| Residue disposal | $0.42M | 1.7% |
| Product packaging & logistics | $1.22M | 4.9% |
| Labour | $3.33M | 13.4% |
| Maintenance | $4.32M | 17.4% |
| Insurance, G&A, QA | $2.47M | 9.9% |
| Total opex | $24.83M | 100.0% |
| Area | Purchased equipment (DEC) | Installed capex (× Lang 3.1 × 1.15) |
|---|---|---|
| Shared front-end (100–108) | $8.1M | $29.0M |
| PET train (110–134) | $14.4M | $51.4M |
| Rubber train (140–148) | $4.2M | $15.0M |
| Carbon / polyolefin train (150–158) | $1.8M | $6.5M |
| Utilities and offsites (160–164) | $6.0M | $21.4M |
| Total | $34.6M | $123.4M |
| Case (200 t/day) | Capex | Opex | EBITDA | ROI | Payback (yr) |
|---|---|---|---|---|---|
| Rev 1 as published (fiber $3,000/t, Lang 2.2) | $109.7M | $33.2M | $52.6M | 48.0% | 2.09 |
| Rev 1 restated (fiber $1,500/t, Lang 3.1, opex as published) | $154.6M | $33.2M | $18.0M | 11.6% | 8.60 |
| Rev 2 — full plant (base case) | $123.4M | $24.8M | $24.0M | 19.5% | 5.14 |
| Rev 2 — Phase 1 only (PET sold as washed flake at $700/t) | $79.9M | $18.1M | $16.3M | 20.4% | 4.90 |
The headline ROI falls from Revision 1's 48.0 % to 19.5 %, and that is the honest figure. Revision 1's result rested on a $3,000/t fiber price about twice the level of recycled PET pellet assessments, and on a Lang factor of 2.2. Restated to a $1,500/t product and a Lang factor of 3.1, Revision 1 returns 11.6 %. Revision 2 is 20 % cheaper to build than that restated case and returns about 8 points more, with no biological unit and no acid loop to prove.
| Case (200 t/day) | Capex | EBITDA | ROI | Payback (yr) |
|---|---|---|---|---|
| Base case | $123.4M | $24.0M | 19.5% | 5.1 |
| rPET chips $1,200/t | $123.4M | $17.2M | 13.9% | 7.2 |
| rPET chips $1,800/t | $123.4M | $30.8M | 25.0% | 4.0 |
| Reclaimed rubber $300/t | $123.4M | $21.6M | 17.5% | 5.7 |
| Reclaimed rubber $700/t | $123.4M | $28.0M | 22.7% | 4.4 |
| Polyolefin pellets $600/t | $123.4M | $22.5M | 18.2% | 5.5 |
| Polyolefin pellets $900/t | $123.4M | $25.5M | 20.7% | 4.8 |
| Capex +30 % | $160.4M | $22.0M | 13.7% | 7.3 |
| Lang factor 3.63 instead of 3.1 | $144.4M | $22.9M | 15.8% | 6.3 |
| Opex +15 % | $123.4M | $20.3M | 16.4% | 6.1 |
| Feedstock $0/t (vs $60/t) | $123.4M | $28.0M | 22.7% | 4.4 |
| Feedstock $150/t | $123.4M | $18.1M | 14.7% | 6.8 |
| All product yields −10 % | $123.4M | $19.3M | 15.6% | 6.4 |
| H₂S at 8 kg/t rubber (Rev 1 benchmark, disposable media) | $123.4M | $21.0M | 17.1% | 5.9 |
| Combined downside: rPET $1,200, capex +30 %, opex +15 %, yields −10 % | $160.4M | $7.1M | 4.4% | 22.7 |
| Combined upside: rPET $1,800, capex −10 %, feedstock $0 | $111.0M | $35.5M | 32.0% | 3.1 |
The plant remains above 13 % ROI across single-variable shocks except a capex overrun of 30 %, but the combined downside (a low chip price, a capex overrun, higher opex and lower yields together) collapses the return to about 4 %. The chip price and the capex factor dominate. The 8 kg/t H₂S case costs about 2 points of ROI, so it does not by itself break the design.
A Phase 1 plant, built without the glycolysis, purification and polycondensation units (118–134) and selling washed flake at $700/t, costs $79.9M and earns $16.3M EBITDA (20.4 % ROI). Phase 2 adds $43.4M of capex and $7.7M of EBITDA, an incremental return of 17.7 % at $1,500/t chips. That return is very sensitive to the chip price.
| rPET chip price ($/t) | Incremental ROI, flake $700/t | flake $500/t | flake $900/t |
|---|---|---|---|
| $1,200 | 2.0% | 15.0% | -11.0% |
| $1,300 | 7.3% | 20.3% | -5.7% |
| $1,400 | 12.5% | 25.5% | -0.5% |
| $1,500 | 17.7% | 30.7% | 4.7% |
| $1,800 | 33.5% | 46.5% | 20.5% |
Phase 2 is therefore a market-gated decision, not a foregone conclusion. It clears 15 % incremental ROI only if chips sell at about $1,450/t or more, so signed offtake at or above that level should be a condition of committing its capital.
| Route | Feedstock flexibility | Product quality | Relative capex | Technology risk | Circularity |
|---|---|---|---|---|---|
| Mechanical recycling (grind and reprocess) | Needs sorted, single-stream input | Degraded molecular weight, limited cycles | Low | Low | Basic reuse |
| Chemical recycling alone (glycolysis or methanolysis) | Plastics only, needs reasonable sorting | High (virgin-like monomer) | High | Moderate; commercially practised | Closes the PET loop |
| Enzymatic depolymerization (engineered PET hydrolases) | Plastics only, needs amorphized PET | High | High | Higher; scaling up | Closes the PET loop |
| Conventional devulcanization (reagent-free) | Rubber only | Moderate | Moderate | Low | Rubber reuse |
| Landfill / incineration | Accepts anything | None | Lowest, ongoing liability | Low | Negative |
| Revision 1 design | Mixed waste, one front-end | Unproven; no monomer purification | High (understated) | High: biology, zeolite, acid loop | Internal acid loop worth ≈0.2 % of opex |
| This work — Revision 2 | Mixed waste, one shared front-end; PVC and nylon sorted out | Virgin-grade BHET → rPET chips; reclaim; black pellets | High, shared across three revenue streams | Moderate: glycolysis and extruder devulcanization are established unit operations | Closed EG loop; contained off-gas; three saleable outputs |
Revision 1 built its environmental case on the H₂S-to-H₂SO₄ loop. Removing that loop does not remove the H₂S; it changes how the plant handles it, and it removes an unrelated wastewater-treatment gap.
| H₂S generation (kg / t rubber) | H₂S (t/yr) | ZnO-equivalent media (t/yr, 30 % utilization) | Media cost ($/yr at $2.7/kg) |
|---|---|---|---|
| 0.5 | 9.2 | 74 | $0.20M |
| 3 | 55.4 | 442 | $1.19M |
| 8 | 147.8 | 1,178 | $3.18M |
At the 0.5 kg/t design basis, 9.24 t/yr of H₂S is generated. A dry scavenger bed sized on Eq. 24, backed by a caustic polish scrubber for the mercaptan and VOC fraction, is inexpensive at this scale and needs no exotic alloys. If bench assay instead confirms Revision 1's 8 kg/t benchmark, spent-media cost rises to about 3.18$M/yr (Table 13) and a wet caustic scrubber sized for the higher load, or a switch back to a reagent-free devulcanization process, should be reconsidered. Either way, the off-gas is contained rather than converted to a low-value credit.
The hot caustic wash and rinse trains (112–114, 150) generate a continuous wastewater stream carrying surfactant, dissolved organics and suspended solids. Revision 1 specified no wastewater treatment. Revision 2 adds a treatment and water-recycle unit (160): screening, dissolved-air flotation for oil and grease, then biological treatment (or discharge to a municipal system where permitted) before recycling water back to 112–114 and 150. This is a compliance requirement, not an option, at 200 t/day.
| Hazard | Location | Control |
|---|---|---|
| Hot caustic (1–2 % NaOH, 85–90 °C) | 112, 150 | Closed washer housings, eyewash and safety-shower stations, PPE, SOP for charging and draining |
| Ethylene glycol — combustible, toxic by ingestion | 118, 122, 132 | N₂-blanketed vessels, bunded storage, LEL monitoring, no ignition sources in the glycolysis area |
| Zinc acetate, activated carbon dust | 118, 124 | Dust control, PPE, contained dosing |
| Hot BHET / polymer melt (up to 280 °C) | 118, 132 | Insulated lines, thermal-burn PPE, interlocks on access panels |
| H₂S and mercaptans from devulcanization | 142, 144 | Fixed H₂S detectors, hooded capture at the die, forced ventilation, permit-to-work for confined-space entry near 144, SCBA available on site |
| Process-oil mist and heat | 142 | Local exhaust ventilation, guarding, burn PPE |
| PVC / chlorine carry-over | 102, 118 | XRF or chlorine-signal rejection at the sorter; chloride monitoring on the glycolysis filtrate protects stainless and duplex equipment |
| Dust and fines (front-end, extruders) | 100–106, 154 | Dust collection, ATEX-rated equipment in dusty zones, housekeeping SOPs |
| Confined space (vessels, filters, evaporators) | 118, 120, 122, 160 | Permit-to-work, atmospheric testing, rescue plan |
The SO₃ absorption step (Eq. 30 in Revision 1) absorbed SO₃ directly in water, which forms a corrosive acid mist rather than the 98 % acid used industrially, and it required an Incoloy-lined converter for a credit worth about 0.2 % of opex. Revision 2 avoids that hazard by not building the unit. Revision 1 also did not specify a wastewater treatment system despite a continuous caustic and rinse discharge; Section 9.2 closes that gap.
The central claim of this revision is narrower than Revision 1's. Revision 1 argued that a single reactor system could recover fibre, rubber and carbon at a 48 % ROI, underwritten in part by an internal acid loop. That loop turns out to be worth about 0.2 % of opex, its PET route had no path from hydrolysate to spinnable polymer, and its bioreactor consumed the product it was meant to protect. None of that means the founding idea, one shared front-end serving three chemically distinct, independently sellable product lines, is wrong. It means the chemistry inside two of the three lines needed replacing, and the third needed a purification block it never had.
What remains after that replacement is closer to conventional process engineering than to novel chemistry: hot caustic washing, PET glycolysis, BHET crystallization, melt polycondensation and twin-screw devulcanization are all documented, commercially practised unit operations. The contribution this design makes is architectural, sizing and integrating three trains under one sorter and one utility system, and staging capital so the proven, low-risk lines are built first and the market-dependent PET chemistry is gated on signed offtake.
The economics are honest but less exciting than Revision 1's headline: 19.5 % ROI at 200 t/day against a $1,500/t chip price that sits at the upper half of current market assessments, falling to 13.9 % at $1,200/t. That sensitivity, not a single point estimate, is the number to carry into an investment decision, and it is the reason Section 11 treats offtake price as a gate rather than an assumption.
Revision 2 keeps the shared-front-end, three-train concept and replaces the parts of Revision 1 that did not hold up: the acid leach, the zeolite hydrolysis bed, the Ideonella sakaiensis bioreactor and the H₂S-to-H₂SO₄ loop are removed; a glycolysis-to-BHET route with a full monomer-purification cascade, a corrected polycondensation catalyst window, a twin-screw devulcanization line, vacuum-vented compounding and a multi-sensor front end are added. The techno-economic model is restated with an itemized opex, an energy balance, a realistic Lang factor and market-checked prices. At 200 t/day the result is $123M of capex against $155M for Revision 1 restated to the same Lang factor, and 19.5 % ROI with an 18-case sensitivity table rather than a single number. The plant is staged in three phases so the cheapest, lowest-risk lines fund operations while the PET chemistry is validated and gated on offtake before its capital is committed.