Split README into topic documents

- README reduced to overview with linked document index
- docs/equipment/pressing-line.md: fleet strategy, Newbilt plan, site reqs
- docs/suppliers/vendor-landscape.md: researched vendor narratives
- docs/sustainability.md: solar, zero-discharge water, materials
- docs/worker-safety.md: training, clean rooms, PPE
- docs/facility-security.md: access control, CCTV, archive/audio protection,
  fire and life safety, plant monitoring, Thai compliance anchors

docs/sustainability.md

@@ -0,0 +1,73 @@
+# Sustainability
+
+Two commitments are fixed design requirements for this plant, not aspirations
+to be traded away during engineering: run on as much renewable energy as
+possible, and discharge no contaminated water. Both are decisions we can
+actually make, because the two biggest utility systems (steam and cooling)
+are explicitly excluded from the Newbilt scope and will be sourced and
+installed locally in Thailand. That gives us a free hand to spec them green
+from day one.
+
+## Renewable energy: large solar installation
+
+The pressing machinery itself is a modest electrical load: 32 kVA connected,
+~21 kW average for a full Duplex line. The real energy demand sits in the
+utilities around it, above all the steam generator (400 kg/h saturated steam
+at 10 bar / 180 °C) and the 250 kW cooling system.
+
+- A large rooftop/ground PV installation is planned. Northern Thailand's solar
+  irradiance is favorable, and daytime-heavy production shifts align well with
+  PV output.
+- The chosen hardware direction: multiple large **Deye** three-phase hybrid
+  inverters (30 to 80 kW class, parallelable up to 10 units) paired with a
+  **Gobel Power** high-voltage LiFePO4 battery stack at **271 kWh** (15
+  stacked 18 kWh packs, LFP 314 Ah cells, rated 8000+ cycles). Deye is an
+  officially supported protocol on the Gobel BMS, and Deye has established
+  Thai distribution. Two engineering checks are open: matching the battery
+  rack's DC voltage window to the specific Deye model (a full 15-pack rack
+  can exceed the 800 V window of the smaller Deye series), and obtaining
+  Gobel's safety certifications (not published) for a Thai commercial
+  install. At 271 kWh the stack is a peak-shaving and ride-through buffer,
+  not full off-grid autonomy, against loads like the 250 kW cooling system.
+- The machine load and cooling system are straightforward PV targets.
+- Steam is the open engineering question. Options under evaluation include an
+  electric boiler (maximizes the solar share), a biomass boiler (locally
+  abundant fuel), and solar-thermal pre-heating of boiler feedwater. No
+  technology choice has been made yet; it will be driven by honest numbers on
+  load profiles, storage, and grid backup, not by what sounds greenest.
+- System sizing (kWp, storage, grid-connection model) is open until the site
+  is chosen.
+
+## Zero contaminated water
+
+The reference process schematic shows a conventional open "dirty water"
+cooling loop. We will not build it that way.
+
+- Cooling (up to 250 kW, 220 l/min peak, roughly 98% of it mold cooling) will
+  run as a closed loop with a cooling tower/chiller, so process water is
+  recirculated rather than discharged.
+- Steam condensate will be recovered and returned to the treated feedwater
+  cycle instead of being dumped.
+- The planned in-house galvanics line (silvering and stamper electroforming)
+  involves plating chemistry: silver solutions and nickel baths. This is the
+  hardest part of the zero-discharge commitment. Rinse water will need
+  closed-loop treatment and metal recovery, designed in from the start rather
+  than retrofitted.
+- Solvent handling (flagged in the site requirements) will be contained and
+  disposed of as controlled waste, never via wastewater.
+- Target: no process-water discharge to the local environment. Detailed water
+  balance and treatment design are pending site selection.
+
+## Materials
+
+The plant design already includes a PVC recycling loop: an in-house grinder
+returns trimmings and rejects to the extruder, with the standard feed mix at
+80% virgin / 20% recycled compound. Record scrap becomes feedstock, not
+landfill.
+
+For the virgin compound itself, the preference is **local (Thai/regional)
+PVC sourcing if the quality meets vinyl-record specifications**. The current
+supplier list (2021) is almost entirely European, with one US option;
+qualifying a local compound would cut shipping distance, cost, and the
+carbon footprint of the feedstock. European suppliers remain the fallback
+where quality demands it.