Why Optical Film Wastewater Is a Special Case for Zero Liquid Discharge
BOPET, OCA, and anti-reflective (AR) coating rinsewater carry a 2,000–8,000 mg/L TDS load with only 200–800 mg/L COD, pH 5–9, and trace solvents (NMP, DMF, IPA) — orders of magnitude lower in organics than textile or FGD wastewater but with conductivity that rules out direct reuse. Optical film plants have no published discharge limits to meet; they close the loop because the coating line cannot tolerate conductivity excursions. Generic ZLD case studies (shale gas at 80,000–120,000 mg/L TDS, textile at COD 1,500–3,000 mg/L, FGD at high hardness) over-size the biological step and under-size the membrane/crystallizer pair when copy-pasted onto a BOPET matrix. A 2026 optical film ZLD train is membrane-led, not biology-led, and that distinction drives every equipment choice downstream.
The reuse-quality bar is set by the coating line, not the regulator. Optical-grade PET/BOPET rinse loops demand permeate at less than 50 µS/cm conductivity, less than 5 mg/L TSS, and less than 1 NTU turbidity; a single particle or ionic spike above these thresholds causes pinholes, haze, or delamination defects on the coated web. Because the spec is driven by defect cost rather than compliance, the ZLD system has to be over-engineered relative to a typical industrial RO — typically two-pass RO or DTRO with a polishing stage, not single-pass. Treating optical film rinsewater as "just another industrial ZLD job" is the single most common scoping error, and it shows up as 8–15% lower water recovery and 20–30% higher membrane replacement cost in operating plants.
The Four-Stage Process Train Used in 2026 Optical Film ZLD Plants
A working 2026 optical film ZLD train runs in four sequential stages, each sized to protect the unit operation downstream. Skipping pretreatment or under-sizing the crystallizer turns a 97% recovery system into an 80% recovery system within six months of fouling.
Stage 1 — Pretreatment. A multimedia filter (sand + anthracite + garnet, 10–30 m/h service flow) drops TSS below 5 mg/L, followed by a 0.1 µm PVDF ultrafiltration skid that brings SDI below 3 and TSS below 1 mg/L. Dissolved air flotation is added only when FOG or anti-stat surfactant loading pushes the UF recovery below 85%. Stage 1 typically achieves 90–95% water recovery and the rest of the train is sized from its permeate.
Stage 2 — Membrane concentration. Two-pass RO or DTRO concentrates the UF permeate from roughly 5,000 mg/L TDS up to 70,000–80,000 mg/L, recovering 75–90% as permeate that already meets the less than 50 µS/cm reuse spec. DTRO is selected over spiral-wound RO when influent TDS exceeds 5,000 mg/L, when silica scaling potential is above 150 mg/L SiO₂, or when the feed carries suspended fines a cartridge filter cannot catch.
Stage 3 — Thermal evaporation. The RO concentrate (typically 1–3% of the original feed flow) goes to an MVR or TVR evaporator. MVR steam economy runs 3.5–6 kg water evaporated per kg steam equivalent — a 65–75% thermal reduction versus single-effect live-steam evaporation. The Onishi 2018 ZLD desalination framework still anchors the mass/energy balance for this stage in most 2026 EPC designs.
Stage 4 — Crystallization. A forced-circulation crystallizer takes the evaporator bottoms at 25–30% weight salts and produces a Na₂SO₄/NaCl cake at less than 8% moisture for landfill or, where permitted, off-spec sale. Overall plant liquid recovery lands at 97–99% and solid output is below 1% of feed volume.
| Stage | Equipment | Inlet Spec | Outlet Spec | Stage Recovery |
|---|---|---|---|---|
| 1 — Pretreatment | MMF + UF (0.1 µm PVDF) ± DAF | 2,000–8,000 mg/L TDS, TSS 20–100 mg/L | SDI less than 3, TSS less than 1 mg/L | 90–95% |
| 2 — Membrane | Two-pass RO or DTRO | 4,000–6,000 mg/L TDS | Permeate less than 50 µS/cm; concentrate 70,000–80,000 mg/L | 75–90% |
| 3 — Evaporation | MVR or TVR | 70,000–80,000 mg/L TDS, 1–3% of feed | Condensate less than 10 µS/cm; bottoms 25–30% salts | 95–98% |
| 4 — Crystallization | Forced-circulation crystallizer | 25–30% salt solution | Cake less than 8% moisture; mother liquor recycled | less than 1% liquid loss |
Equipment Selection: RO vs DTRO, MVR vs TVR, and When to Add Ozone
Equipment choices for a 2026 optical film ZLD plant are driven by three variables: feed TDS, available utilities (steam vs electricity), and trace solvent loading. Get any of these wrong and either OPEX climbs 30–60% or the membranes foul within 60 days.
RO versus DTRO. Spiral-wound RO is cheaper per m² of membrane area and works at TDS up to 5,000 mg/L, but its narrow feed spacer (28–34 mil) fouls rapidly on silica, BOD residual, or barium/strontium scaling. DTRO uses a disc-tube geometry with 8 mm feed channels and tolerates suspended fines up to 50 mg/L with 3–5× higher cross-flow velocity. For optical film streams where silica regularly sits at 50–120 mg/L and feed TDS runs 5,000–8,000 mg/L, DTRO is the lower-risk default.
MVR versus TVR. MVR uses a mechanical vapor compressor driven by electricity, so OPEX tracks the local kWh tariff. TVR uses a thermocompressor driven by live steam, so OPEX tracks the local steam tariff and capital cost drops 15–20% because no compressor is needed. The decision rule is straightforward: if your plant is grid-constrained and steam-rich (typical of Korean petrochemical complexes, many Chinese BOPET lines co-located with a PTA plant), choose TVR; if it is steam-poor and grid-stable (typical of Taiwan display-film fabs), choose MVR. Below 500 m³/day the MVR compressor becomes disproportionately expensive and TVR usually wins on lifecycle cost.
Ozone polishing. Residual NMP, DMF, and IPA pass straight through UF/RO and raise the COD of the RO concentrate, which then fouls the evaporator. A pre-RO ozone oxidation polisher (O₃ dose 50–200 mg/L, 15–30 min contact) breaks these solvents and lifts COD removal by 30–50%, but only on moderate loads. Above 1,000 mg/L solvent COD a wet-air oxidation or Fenton step is needed first.
| Decision | Choose If | Avoid If | Typical 2026 Cost Premium |
|---|---|---|---|
| DTRO over spiral-wound RO | Feed TDS greater than 5,000 mg/L, SiO₂ greater than 100 mg/L, or suspended fines greater than 5 mg/L | Feed TDS less than 3,000 mg/L and no silica risk | +15–25% CAPEX, –20–30% membrane replacement OPEX |
| MVR over TVR | Electricity less than 0.08 USD/kWh and no surplus steam | Live steam available at less than 8 USD/ton | +20% CAPEX, –40% thermal OPEX |
| Pre-RO ozone polisher | Solvent COD 100–1,000 mg/L, downstream evaporator fouling history | Solvent COD less than 100 mg/L or feed already past biological treatment | +5–8% CAPEX, +0.02–0.05 USD/m³ OPEX |
2026 Cost Benchmarks: CAPEX, OPEX, and Where the Money Goes
For a 1,000–3,000 m³/day optical film ZLD plant commissioned in 2026, turnkey CAPEX runs USD 1.8M–6.5M with the cost split heavily weighted to thermal equipment. The MVR or TVR evaporator accounts for roughly 35% of CAPEX, the forced-circulation crystallizer about 20%, RO/DTRO skids about 18%, pretreatment about 12%, and BOP / electrical / instrumentation the remaining 15%.
OPEX at 2026 utility tariffs in coastal China or Korea sits at USD 0.35–1.20 per m³ treated, broken down approximately as 55% thermal energy, 20% electrical, 15% membrane and media replacement, and 10% labor and chemicals. A multi-effect evaporator running on live steam would push OPEX to USD 1.5–2.50 per m³, which is why MVR or TVR is the single largest OPEX lever in any optical film ZLD design. The savings come from steam-economy improvements of 65–75% versus single-effect evaporation, directly cutting the 55% thermal share by 40–60%.
Compared against a disposal-only baseline of tanker off-site at USD 8–25 per m³ (typical for coastal China and Korea where direct discharge is no longer permitted for new capacity), a 2,000 m³/day ZLD plant has a payback of 2.5–4.5 years, and that excludes avoided liability and reuse-water credit (per m³ of permeate at less than 50 µS/cm replacing demineralized water at USD 1.50–3.00 per m³ in 2026). Payback tightens to under 3 years when the plant displaces a meaningful fraction of purchased DI water.
| Cost Component | 2026 Range (USD) | % of Total | Key Driver |
|---|---|---|---|
| Evaporator (MVR or TVR) | 0.6M–2.3M | 35% | Steam economy, compressor sizing, material of construction |
| Crystallizer (forced circulation) | 0.4M–1.3M | 20% | Worst-case salt loading, not average |
| RO / DTRO skids | 0.3M–1.2M | 18% | Feed TDS, recovery target, membrane type |
| Pretreatment (MMF + UF + DAF) | 0.2M–0.8M | 12% | FOG and surfactant load |
| BOP / electrical / I&C | 0.3M–0.9M | 15% | Utility tie-in, automation scope |
| OPEX per m³ treated | 0.35–1.20 | — | 55% thermal, 20% electrical, 15% membranes, 10% labor/chem |
| OPEX per m³ (multi-effect, live steam) | 1.50–2.50 | — | Reference baseline without MVR/TVR |
A 2026 plant is typically built around an industrial RO system for ZLD pre-concentration feeding an MVR evaporator, with a multi-media filter for RO feed protection sized to keep SDI below 3 and protect the membrane CAPEX line item from premature fouling.
How to Scope an Optical Film ZLD Project in 2026
Four steps take a 2026 optical film ZLD project from whiteboard to commissioned plant without the usual under-sizing errors.
Step 1 — 30-day wastewater characterization. Run 24-hour composite sampling across every rinse stream (coating bath overflow, post-rinse, IPA drag-out, cleaning-in-place) for at least 30 days, capturing TDS, COD, BOD, hardness, silica, FOG, and solvent species by GC-MS. Average values are useless; the worst-case TDS spike and the solvent peak dictate the evaporator and crystallizer sizes. Skipping this step is the single most common reason 2026 ZLD plants run below their design recovery.
Step 2 — Pilot the RO or DTRO on-site for 60–90 days. Vendor brochures quote flux at 25°C and clean feed; an on-site pilot at 30–50% design recovery confirms real flux (typically 20–35% below brochure), cleaning frequency (CIP every 2–6 weeks is normal for DTRO on optical film), and whether silica antiscalant alone prevents fouling. Pilot data goes directly into evaporator and crystallizer sizing, so don't shortcut it.
Step 3 — Match evaporator steam economy to local utility tariffs. Use the 2026 MVR/TVR decision rule from the equipment section. At electricity above 0.10 USD/kWh and steam below 8 USD/ton, TVR wins. At electricity below 0.08 USD/kWh and no surplus steam, MVR wins. Size the crystallizer to the worst-case RO concentrate composition, not the average — under-sizing it by 20% is a six-figure retrofit.
Step 4 — Engage a single-train vendor with process warranties. Mixing brands at the brine interface (one vendor's RO, another's evaporator, a third's crystallizer) creates a finger-pointing problem when recovery slips. A vendor that supplies the full train from ozone oxidation system for NMP/DMF breakdown through pretreatment, RO, evaporator, and crystallizer can issue a single-point process warranty on water recovery and salt purity. For broader context on the vendor landscape, the 2026 ZLD and nutrient-recovery vendor landscape survey maps the players; the microelectronics TMAH ZLD case study shows how the same train architecture applies to a sister industry.
Frequently Asked Questions
What water recovery can optical film ZLD achieve in 2026? A properly designed four-stage train (MMF + UF + two-pass RO or DTRO + MVR + forced-circulation crystallizer) returns 97–99% of the feed as less than 50 µS/cm permeate, with the remaining 1–3% leaving the plant as a less than 8% moisture salt cake.
How much does a 1,000 m³/day optical film ZLD plant cost in 2026? Turnkey CAPEX runs USD 1.8M–3.0M depending on effluent complexity, with the MVR/TVR evaporator and crystallizer together making up roughly 55% of the total. OPEX at 2026 utility tariffs in coastal Asia is USD 0.35–1.20 per m³ treated.
When should I choose DTRO over spiral-wound RO for an optical film ZLD? Choose DTRO when feed TDS exceeds 5,000 mg/L, silica is above 100 mg/L as SiO₂, or the stream carries suspended fines a cartridge filter cannot catch. The disc-tube geometry tolerates 3–5× higher cross-flow velocity and avoids the spacer fouling that cuts spiral-wound recovery within 60–90 days on optical film matrices.
Is ZLD economically justified for optical film coating rinsewater? Yes, at most Asian sites commissioned in 2026. Tanker off-site disposal runs USD 8–25 per m³ in coastal China and Korea, and ZLD OPEX runs USD 0.35–1.20 per m³, giving a 2.5–4.5 year payback at 2,000 m³/day before counting avoided DI water purchases at USD 1.50–3.00 per m³.
What is the difference between MVR and TVR for optical film ZLD? MVR uses a mechanical vapor compressor driven by electricity, with 65–75% thermal reduction versus single-effect evaporation. TVR uses a thermocompressor driven by live steam, has lower CAPEX, and tracks the local steam tariff instead of the electricity tariff. Below 500 m³/day TVR usually wins on lifecycle cost; above 500 m³/day with cheap electricity, MVR wins.
