Why a CAM plant ZLD train is its own process category
When a listed battery materials company commissions a CAM plant, the typical ZLD train in 2026 is a five-stage sequence: chemical precipitation (lime/CaCl₂) for F⁻ removal, two-pass RO, MVR evaporation, forced-circulation crystallization, and ATFD finishing. For a 50 GWh plant processing 2,500–3,500 m³/day of mixed CAM effluent, this train delivers 95–99% water recovery, ≥98% NaCl/Li₂CO₃-grade solids, and a turnkey CAPEX of $11M–$25M, with NMP and lithium recovery offsetting 25–40% of net OPEX.
CAM wastewater is not generic battery wastewater, and it is not upstream lithium refinery brine. It is its own process category because the four streams that drain off a CAM building carry a contaminant load that no off-the-shelf ZLD design is sized for. The four streams, in order of mass load:
- NMP condensate from electrode-coating dryer condensers: 2,000–10,000 mg/L NMP, often 15–25% of total flow by volume.
- Cathode precursor washwater from co-precipitation reactors and washing filters: Ni/Co/Mn/Li salts at 500–5,000 mg/L, the single largest dissolved-solids load.
- Electrolyte-mixing scrubber water from LiPF6 salt dissolution rooms: F⁻ at 1,500–2,500 mg/L plus NH₄⁺ and trace POF₃ hydrolysis products.
- Formation/aging wastewater from first-cycle cell formation drains: lower volume but high in TOC and residual LiPF6 hydrolysis products.
Fluoride is the contaminant that disqualifies biology. Activated sludge and MBR biomass lose activity at roughly 50 mg/L F⁻; CAM scrubber water runs 30–50× that number, so any train built on a biological front end fails the moment electrolyte-mixing starts (HydropureWater field data, 2026). NMP is technically biodegradable but slow at 40–60 day HRT, and it is worth $4–6/kg recovered, so the design bias is toward recovery, not oxidation. Lithium, cobalt, and nickel salts sit at concentrations 10–100× typical industrial brine, putting generic ZLD economics under stress and forcing a resource-recovery line into the train rather than treating it as an add-on.
Demand context is now structural. The IEA's 2024 Global EV Outlook baseline projects lithium-ion demand above 2,000 GWh by 2030, with 40+ new gigafactories in planning or construction by 2026 — each requiring a ZLD system before first production runoff. The Springer 2021 four-process ZLD comparison (Able et al., Clean Technologies and Environmental Policy, 2021) remains the academic baseline for thermal brine trains, but it models seawater and oilfield brines with no F⁻ or NMP. Battery plants need a customized train.
The five-stage ZLD train a commissioned CAM plant runs
The defensible 2026 process flow for a CAM plant ZLD system is a five-stage train that takes raw CAM effluent from the equalization basin to dry salt stack with 95–99% water recovery and 0 m³/day liquid discharge. Each stage has a specific load-reduction target; the numbers below are the sizing basis for a 3,000 m³/day plant.
Stage 1 — Equalization + DAF. A HydropureWater ZSQ DAF system brings TSS from 200–800 mg/L down to <30 mg/L and oil to <5 mg/L, stripping graphite and PVDF binder carry-over that would otherwise foul downstream RO membranes. Flow balancing is held to ±10%, with each DAF unit rated 4–300 m³/h. This stage is not optional — NMP condensate and precursor washwater both carry suspended solids that blind membranes within hours if not removed upstream.
Stage 2 — Chemical precipitation + MBR polish. Lime and CaCl₂ dosing through a HydropureWater chemical dosing system drops F⁻ from 2,000 mg/L to <50 mg/L as CaF₂ sludge, and drops COD from 800–1,500 mg/L to <60 mg/L after an HydropureWater MBR system polish. The MBR is doing COD work, not fluoride work — that distinction matters when a regulator audits the PFD. The CaF₂ sludge goes to a HydropureWater plate and frame filter press for dewatering to <65% dry solids before landfill or, increasingly, recovery to HF acid or to a fluoropolymer feedstock.
Stage 3 — Two-pass RO. A HydropureWater industrial RO system configured as BWRO at 75–85% recovery followed by a high-pressure pass at 80–90 bar pushes combined recovery to 95%, with permeate conductivity <50 µS/cm. The high-pressure second pass is what makes brine volume small enough to feed the thermal train economically — without it, MVR capex roughly doubles.
Stage 4 — MVR mechanical vapor recompression. Concentrate at 5–25 wt% TDS is concentrated to 25–30 wt% TDS at 0.02–0.05 kg steam per L evaporated and a 60–70% steam reduction versus MEE. The compressor and tube bundle are the largest single CAPEX line in the train. Titanium Grade 2 or Duplex 2205 metallurgy is required for F⁻ service; expect 10–15 year service life with annual tube cleaning.
Stage 5 — Forced-circulation crystallizer + ATFD. A 30–50 wt% slurry is seeded and held for 4–8 h residence to grow NaCl or Li₂CO₃ crystals at ≥98% purity. An agitated thin film dryer takes the crystal paste to <0.5% moisture as free-flowing salt ready for bagging or bulk shipment.
Mass balance for a 50 GWh plant: 3,000 m³/day influent compresses to 30–80 m³/day wet solids and zero liquid discharge. The Nature Communications 2021 solar crystallizer field result of 48.0 kg/m²/day (Zhang et al., Nat. Commun., 2021) is the energy-efficiency benchmark driving new hybrid MVR-solar designs in 2026 RFPs, though solar alone still cannot meet year-round duty in northern gigafactory sites (Shanxi, Saxony, Nevada).
| Stage | Unit operation | Influent load | Effluent target | Metallurgy / sizing note |
|---|---|---|---|---|
| 1 | Equalization + DAF | TSS 200–800 mg/L; oil variable | TSS <30 mg/L; oil <5 mg/L | 4–300 m³/h per unit; flow balance ±10% |
| 2 | Lime/CaCl₂ precipitation + MBR polish | F⁻ ~2,000 mg/L; COD 800–1,500 mg/L | F⁻ <50 mg/L; COD <60 mg/L | CaF₂ sludge to filter press; MBR is COD polish only |
| 3 | Two-pass RO (BWRO + high-pressure) | TDS 1,000–3,000 mg/L | Permeate <50 µS/cm; recovery ~95% | High-pressure pass at 80–90 bar |
| 4 | MVR mechanical vapor recompression | 5–25 wt% TDS | 25–30 wt% TDS concentrate | Titanium Gr. 2 or Duplex 2205; 10–15 yr life |
| 5 | Forced-circulation crystallizer + ATFD | 25–30 wt% slurry | Dry salts <0.5% moisture, ≥98% purity | Seeded slurry; 4–8 h residence |
Selecting the thermal train: MVR, MEE, or ATFD
Choosing the thermal train is the single largest CAPEX/OPEX decision in a CAM ZLD design. The decision is driven by brine TDS profile, electricity price, and whether lithium/NMP recovery is in scope.
The operating rule for 2026 designs: MVR is the workhorse for stable 5–25 wt% brine at sites with electricity below $0.06/kWh; MEE is the fallback for variable feed or steam-rich sites where gas or biomass steam is cheaper than grid power; ATFD handles the final 30–50 wt% pasty concentrate and produces free-flowing salts at <0.5% moisture. MVR requires stable power because the compressor is the steam source; MEE tolerates intermittent power because it is steam-driven. MVR specific energy is 1.2–1.8 kWh per kg water evaporated; MEE sits lower per kWh but adds a steam-plant line to the CAPEX stack.
Metallurgy rule: rubber-lined carbon steel is acceptable for the ATFD shell because vapor-phase F⁻ activity is negligible at <0.5% moisture, but titanium Grade 2 or Duplex 2205 is required for MVR and MEE tubes that see liquid brine with F⁻ in solution. Skipping the alloy upgrade is the most common cause of tube failure in battery ZLD service.
Reference example: Hyper Filteration commissioned a 200 KLD RO + MEE ZLD at Tata Autocomp GY Battery, Pune in May 2025, demonstrating the MEE fallback path on a real commissioned battery site (Hyper Filteration, 2025-05). That project is the closest published analog for a CAM-adjacent ZLD and is the bench against which most 2026 RFQs are being cut.
Resource recovery lines that change the economics
Two value-recovery lines move a CAM ZLD plant from cost center to partially revenue-positive. NMP reclamation at $4–6/kg recovered is the larger lever on plants that process more than 50 tons/day of electrode coating; the wiped-film evaporator upstream of the DAF can cut net OPEX by 25–40% on those sites. Plants below 50 tons/day of coating usually do not see positive NMP economics and run the condensate through the biological polish instead.
Lithium recovery is the second lever. Cathode washwater at LiOH/Li₂CO₃ spot prices of $13–22/kg (2026) can generate $2M–$5M annual revenue from a 50 m³/day Li-bearing stream (HydropureWater field data, 2026). Payback math closes in 3–5 years at Li₂CO₃ above $13/kg and Li concentration above 500 mg/L. Below 500 mg/L, the crystallizer seed and reagent cost structure does not support a credit and the lithium stays in the NaCl byproduct stream.
Cost of inaction is now material. Combined water purchase and wastewater surcharges in water-stressed hubs (Shanxi, Nevada, Saxony) reach $4–$8/m³, putting ZLD break-even at 4–7 years before lithium credits are counted. A HydropureWater plate and frame filter press handles the CaF₂ cake from the precipitation stage and the Li₂CO₃ cake from the crystallizer, so solids handling does not become a hidden bottleneck at the 50 m³/day wet-solids output of a 3,000 m³/day CAM train.
CAPEX and OPEX bands by plant scale
Procurement teams need a defensible cost envelope before issuing RFQs. The 2026 numbers below are turnkey installed prices for a complete five-stage ZLD train including civil works, instrumentation, and commissioning. Western sites carry a 40–60% premium over China-domestic pricing for labor and UL/CE certification, but the same NMP and lithium offsets apply, so the absolute savings are higher even at the premium.
| Plant scale | Influent flow | Turnkey CAPEX (2026) | Key offset |
|---|---|---|---|
| 10 GWh (pilot / single line) | ~500 m³/day | $4M–$9M | Modest NMP recovery; no Li line |
| 50 GWh (standard gigafactory) | ~3,000 m³/day | $11M–$25M (40–60% premium on Western sites) | NMP + Li credits cut OPEX 25–40% |
| 200 GWh (mega-plant cluster) | ~10,000 m³/day | $25M–$55M | Economies of scale; full Li line |
OPEX is dominated by electricity (45–55%), chemicals (15–20%, antiscalant and CaCl₂ dosed through a HydropureWater chemical dosing system), labor (10–15%), and consumables including membrane replacement every 3–5 years. The 200 GWh band captures the economies of scale that drop specific CAPEX per m³/day by roughly 35% versus the 10 GWh band, but it also concentrates the dispatch risk on a single MVR compressor train.
The 2026 regulatory envelope shaping every commission
A ZLD design must be defensible against three regulatory regimes simultaneously. The EPA draft of July 29, 2026 places chemical production, lithium battery material manufacturing, and hydrometallurgical operations within the scope of mandatory ZLD application; from October 2027, new and expanded projects in these sectors must achieve either 100% process water reuse or zero liquid wastewater discharge (per EPA draft "Industrial Zero Liquid Discharge Mandatory Implementation Pathway," 2026-07). The draft also names MVR evaporators, forced-circulation crystallizers, high-salinity concentrators, and EDI ultrapure water systems as compliance certification technology pathways — which is why 2026 RFQs are being rewritten to match.
China cathode material limits bind the discharge end of any reuse stream: F⁻ ≤ 15 mg/L, COD ≤ 200 mg/L, NH₃-N ≤ 30 mg/L, SS ≤ 70 mg/L (per China GB cathode material discharge standard). The five-stage train meets all four at the reuse point, with the MBR polish and the high-pressure RO pass doing the work on COD and SS, and the lime/CaCl₂ stage plus the CaF₂ filter press dropping F⁻ well below 15 mg/L after the polishing stage.
EU Battery Regulation 2023/1542 from 2027 requires EV batteries placed on the EU market to carry a carbon-footprint declaration and a quantitative water-footprint disclosure. The 95–99% reuse data the ZLD train generates is direct compliance evidence, and it is increasingly being requested by EU OEMs at the cathode tender stage — not at the discharge permit stage. For a working walkthrough of how the western New York auto and battery corridor is meeting pretreatment limits, the 2026 EV and auto plant pretreatment compliance guide is a parallel reference. A broader process-flow benchmark that maps the same hybrid train logic onto a different industrial wastewater is the 2026 EV battery ZLD process design and cost guide.
Frequently Asked Questions
What is the minimum ZLD train size for a 10 GWh CAM plant?
Approximately 500 m³/day of influent, requiring a turnkey CAPEX of $4M–$9M at 2026 China/EU prices. The five-stage train is identical in unit operations to the 50 GWh case but is sized to a single DAF, a single RO skid, and a smaller MVR compressor.
Can MBR alone treat CAM wastewater?
No. Fluoride at 1,500–2,500 mg/L inactivates biomass at the ~50 mg/L activity threshold; MBR is a polish step after chemical F⁻ precipitation, not a primary treatment. Designing the train around an MBR front end is the most common engineering error in first-draft CAM ZLD PFDs.
How long does an MVR evaporator last in CAM ZLD service?
10–15 years with annual tube cleaning. Titanium Grade 2 or Duplex 2205 metallurgy is required for F⁻ service; 316L stainless fails within 2–4 years in the same duty and should not be specified.
Is lithium recovery from ZLD brine economically viable in 2026?
Yes, at Li₂CO₃ prices above $13/kg and stream concentrations above 500 mg/L Li, payback is 3–5 years. Below 500 mg/L, the reagent and seeding cost structure does not support a credit and the lithium stays in the NaCl byproduct.
Does ZLD eliminate PFAS liability for a CAM plant?
Not automatically. PVDF binder in the feed does not break down to PFAS, but LiPF6 electrolyte hydrolysis does produce trace PFOA/PFNA precursors that must be tracked in the mass balance and routed to thermal destruction if they concentrate in the crystallizer solids. ZLD sidesteps liquid-effluent PFAS limits but does not sidestep solid-phase PFAS management.
