Oberon — a recycled-plastic module that replaces gravel in a permeable pavement

Three failures meet in the same square metre of pavement. Concrete and asphalt block infiltration, so monsoon rain runs off instead of recharging the ground. Urban flooding costs Indian metros ₹15,000–20,000 crore a year, while groundwater tables in Jaipur, Bengaluru and Chennai fall 1–3 metres a year ².

Plastic accumulates faster than sorting absorbs it, and the leftover is mixed, exactly the fraction a sorted-polymer process rejects. Compression moulding and polyurethane bonding are decades old; what changed is the waste stream, and a method that accepts unsorted feedstock reaches tonnage sorted recycling cannot.

The project began strength-first. The first geometry was a tetrapod, four arms at tetrahedral angles, chosen because that shape carries the most load per unit of material, with structural performance treated as the governing requirement for a road surface.

A hand calculation of rotary-die cycle times corrected it: a part that cannot be made at road quantity does not solve a problem measured in tonnes of waste. I moved manufacturing speed from a check run after the geometry to a constraint the geometry had to satisfy first, which disqualified the tetrapod and then the tri-lobe dome that replaced it.

The bet is that solving at the smallest unit of the road makes mixed, unsorted plastic usable as a primary material, and that manufacturing speed is a design input from the first sketch. Polymer sorting is the step that keeps most of India's plastic out of any recycling route, and every stronger geometry depended on it.

Metal, glass and organic matter are still removed before granulation, so what goes is polymer sorting, not sorting. A shape derived from a tetrahedron carries more load per unit of material than the current form. I chose the weaker shape at specification, accepting that a design decision rather than a measured limit sets the use case: pedestrian and light-vehicle surfaces.

Phenolic compression moulding normally runs at 155–170 °C with mould closure from 30 s to 600 s ⁸. The design runs 11.127 modules per second across 334 cavities, giving each charge 30 s in the die at 130–150 °C: the floor of the time band, and below the temperature band, where melt viscosity rises and cure slows ¹.

The failure mode is specific. A 25.0 mm part heated from its surface does not bring its core to cure temperature in that window, and quenching removes the residual heat that would advance cure after ejection, leaving an under-cured core inside a hard skin that visual inspection would miss.

The test moulds 70 modules on a single-cavity heated soft tool over 10 weeks: three dwell settings at two quench conditions, ten replicates each, plus ten controls on a conventional 60 s in-die hold. It passes at a degree of cure of 95% or more by residual exotherm, water absorption at or below 6% by mass after 24 hours to match IS 15658 ⁹, and crush strength within 10% of the control. Below 90% cure, a heated in-die hold replaces the quench, and rate, cavity count and capex are rebuilt before tooling.

Square metres sold can be bought with discounting; repeat specification cannot, and it is the only evidence available before durability data exists. Three stage metrics connect first use to referral: demonstration installs that become a first paid order, sites whose infiltration after the first monsoon holds at or above the 100 mm/hr CPHEEO recommends for stormwater design ¹², and inbound enquiries traced to an existing site.

The counter-metric is patch rate: square metres removed and replaced per 1,000 m² laid in the first 12 months. Fresh modules drain perfectly, so a site patched every few weeks can still return a satisfied buyer and a healthy infiltration reading. Above 20 m² per 1,000 m² a year, the geometry or the bond is failing. Commercial work stops if fewer than two of the first six demonstration installs convert to a paid order within 12 months.

The first buyer is a private township or residential developer, reached through the project's architect or landscape consultant at specification stage. Public procurement runs through GeM and state tender portals against a named standard ¹³, and IS 15658 does not cover permeable blocks ⁹. I resequenced the buyer order to put the developer first, accepting a smaller first order over a wait on BIS.

Several Indian states grant FAR or FSI incentives, tax rebates or faster approvals to IGBC-certified projects ¹⁴, and a floor-space bump across a residential plot is worth multiples of the paving premium.

A new building material is specified by being walked on, so the entry cost per account is a subsidised demonstration. Against ₹821/m² of gross profit at the base margin, that cost is repaid by 299.88 m², below the 500 m² volume-discount threshold the pricing already sets ³.

The first method is to solve at the system's crudest element. A road's smallest element is gravel, so the module replaces gravel, not a paver or a slab, and at fragment scale polymer type stops mattering. Recycled-plastic construction elsewhere works at component scale, which keeps the polymer sorted, the parts large and the mould slow.

The second came out of the record. Promoting manufacturing speed to a governing input produced the current geometry, but as a direction, not a threshold: no cycle time, cavity count or quote was attached, so three geometries were judged against the same hand calculation. I resequenced the process: when a constraint becomes a design input, its first artefact is the number it must hit. Under that rule, 11.127 modules per second and 334 cavities would have existed at the first generation and met a toolmaker before the third.