Membrane Distillation for ZLD: 2026 Engineering Guide to Brine Minimization

Why Membrane Distillation Fits a ZLD Train

Membrane distillation is a thermal, vapor-pressure-driven process that uses a hydrophobic membrane and a temperature difference to vaporize water from RO concentrate or brine, producing high-purity distillate. In a ZLD train it sits between reverse osmosis (which tops out near 70 g/L feed salinity and ~50% water recovery) and a crystallizer, pushing overall freshwater recovery to ~95% and reducing brine volume sent to evaporation ponds by more than 90%.

Zero Liquid Discharge is defined as a system that produces no liquid effluent: every kilogram of water is recovered as distillate and every kilogram of salt exits as a solid or a marketable byproduct. The original ZLD installations were commissioned by power plants in Colorado, U.S., in response to the escalating salinity of the Colorado River (per S3/S4, 2025), and the same plant-scale logic still drives adoption in 2026 across thermal power, textile dye houses, and lithium-battery chemical processing.

The bottleneck is RO. A standalone RO unit can handle feedwater with a maximum salinity of 70 g/L and achieve a freshwater recovery rate of up to 50% (per S3/S4). That means roughly half of the incoming flow leaves the RO skid as concentrate still containing recoverable water, and that concentrate has to go somewhere. OARO extends the ceiling to 140 g/L and ~72% recovery (per S3/S4), but even that leaves a quarter of the flow as brine.

MD closes that gap using a hydrophobic membrane that admits only water vapor: the liquid phase and dissolved salts are physically rejected. Because the driving force is vapor pressure, not hydraulic pressure, MD can concentrate above the osmotic limit of any RO/OARO stage. It complements a thermal crystallizer better than OARO does because MD already strips the bulk water out thermally, so the crystallizer only has to handle the final water-of-hydration and a small residual flow — slashing both crystallizer size and energy per kilogram of salt.

MD Module Types and How They Behave in a ZLD Duty

Four classical MD configurations dominate the literature and the bid list: DCMD (Direct Contact MD), AGMD (Air Gap MD), VMD (Vacuum MD), and SGMD (Sweeping Gas MD). They differ in how the vapor is collected on the permeate side, and that difference cascades into flux, energy demand, and wetting risk. The 2026 module-and-application review by Nthunya and Mamba (Membranes, 2026-07) is the cleanest recent taxonomy and forms the basis of the comparison below.

Across all four, the authors identify three engineering levers that govern stable long-term performance: module design, hydrodynamics, and operating-condition optimization. Their August 2025 follow-up on serial versus parallel module connections shows the field is now optimizing at the skid level, not just the membrane level (per S2 similar-articles list, Nthunya & Mamba, 2025-08). For ZLD duty, hollow-fiber modules are common because of their high packing density (often 1,000–3,000 m²/m³) and easy scale-up from bench to commercial skid, with the trade-off being higher shell-side pressure drop and more pronounced temperature polarization than flat-sheet or spiral-wound geometries.

Wetting is the failure mode that defines every MD spec. Once liquid water intrudes into a membrane pore — through surfactant carryover, oil ingress, or dry-out — salt rejection collapses from >99.9% to single-digit percent, and the only fix is membrane replacement, not cleaning. That single fact drives almost every operating-discipline and pretreatment rule in the section below.

*Flux ranges are typical engineering envelopes from peer-reviewed MD module studies (per S2, 2026-07); industrial values depend on feed TDS, ΔT, and module geometry. LMH = L/m²·h.

Performance Envelope: Recovery, Rejection, and Energy

Standalone RO delivers up to 50% recovery on a 70 g/L feed ceiling, OARO stretches to ~72% on a 140 g/L feed ceiling, and MD operates as the final concentrator to near-saturation before the crystallizer (per S3/S4). When these are hybridized, the MLD/ZLD train can reclaim as much as 95% of the inlet water and release less than 10% of wastewater to adjacent evaporation ponds (per S3/S4). That 95% / <10% split is the benchmark a ZLD spec should be measured against in 2026.

MD's salt rejection is consistently reported as high — >99.9% on intact, unwetted membranes — but the same 2026 review makes clear that scale-up is limited by fouling, scaling, heat/mass transfer resistance, and module design (per S2, Nthunya & Mamba, 2026-07). The May 2025 interfacial-heating review in Environmental Science & Technology (Hu et al., 2025-05) treats thermal efficiency as the active engineering frontier: heating the feed only at the membrane interface rather than the bulk reduces conductive losses and is one of the more promising paths to lower specific energy.

The heat-source opportunity is what changes MD's economics. MD runs on 40–80 °C temperature gradients — the same range as condenser reject heat, boiler blowdown heat, geothermal return loops, or solar thermal. A plant that already wastes low-grade heat at 60 °C can in principle feed that energy directly into the hot side of an AGMD or VMD module and offset most of the operating thermal load. Flux declines as feed TDS climbs, but a single canonical flux number for industrial ZLD brine does not exist — flux is governed by ΔT, vapor-pressure gap, flow velocity, and module geometry, so an engineer must pilot on the actual brine before committing to a nameplate. For membrane element supply planning, spare RO and UF membrane elements should be staged alongside the MD membrane order to keep both trains online during scheduled replacements.

Fouling, Scaling, and Wetting: The Real Failure Modes

Fouling and scaling are the two dominant long-term performance killers in MD (per S2, 2026-07), and scaling risk grows sharply as the brine approaches saturation because CaCO₃, CaSO₄, and silica all have retrograde solubility that worsens in the hot-side boundary layer. The December 2025 hollow-fiber MD fouling study by Cho et al. (Membranes, 2025-12) is the most recent peer-reviewed evidence on the exact failure mode ZLD operators face — it analyzes fouling under brine-reduction duty and confirms that flux decline tracks with feed-concentration factor and with cold-side temperature control.

Membrane wetting is the third failure mode, and it is qualitatively different from fouling. Fouling reduces flux; wetting destroys rejection. Surfactant carryover from upstream wash streams, oil ingress from a failing DAF, or membrane dry-out during a shutdown all let liquid water enter the pores, and once that happens the only fix is element replacement. The practical feed-spec rule is therefore: oil and grease below detection limit (typically <1 mg/L), surfactants controlled upstream, and a documented wet/dry storage protocol for every standby module.

Operating discipline is the cheapest insurance. Avoid thermal shocks greater than ~5 °C/min on the hot side, control cold-side temperature within ±1 °C to keep the vapor-pressure gap stable, and treat permeate conductivity as the wetting early-warning signal — a rise from ~5 µS/cm into the hundreds means liquid breakthrough and an immediate shutdown. The pretreatment chain should run as DAF pretreatment for oil and suspended solids, followed by multi-media filtration for turbidity polishing, with softening or UF in front of the MD module to drop hardness and silica below the scaling threshold. Engineers sizing the DAF stage should consult the 2026 DAF design parameters guide for hydraulic and air-to-solid ratios appropriate to textile or chemical feeds.

MD vs OARO vs Brine Concentrator: A 2026 Decision Framework

Choosing between MD, OARO, and a thermal brine concentrator is the decision that defines a ZLD project's economics. The table below sets the head-to-head on the metrics a plant manager will actually weigh — feed-TDS ceiling, achievable recovery, energy form, brine output, and capex/opex intensity — anchored to the 70 g/L / 140 g/L thresholds and the ~95% / <10% hybrid-recovery benchmark from S3/S4.

MD is the right choice when (a) low-grade heat is available on site, (b) feed TDS is above 100 g/L where OARO is already at or past its operating limit, and (c) the operator wants electrical-light operation or faces a constrained grid connection. OARO is the right choice when the plant already has high-pressure pumping capacity and feed is comfortably below 140 g/L (per S3/S4). MVC and MED crystallizers are the final step, not a competitor to MD — they handle the last few percent of water and the salt solidification, and they sit downstream of the MD stage in a properly designed ZLD train. The industrial RO unit that fronts the train must therefore be specified for the discharge pressure and recovery target the downstream MD stage expects, not for maximum freshwater production alone.

Putting It Together: A Reference MD + RO + Crystallizer Flowsheet

A reference flowsheet for a 100 m³/h influent reads as: equalization → DAF pretreatment → multi-media filtration → UF/softening → industrial RO unit (to ~50% recovery on a 70 g/L feed ceiling) → MD polish (to ~95% train recovery) → crystallizer (salt solidification, <10% residual to evaporation pond).

Worked mass balance on a 70 g/L feed: RO recaptures ~50% as permeate and sends 50 m³/h of concentrate at ~140 g/L to the MD stage; MD strips another ~45 m³/h of water across the hydrophobic membrane and produces ~5 m³/h of near-saturated brine; the crystallizer then dries that 5 m³/h to solid salt and a small vapor stream. Net system recovery lands at ~95%, with less than 10% of the original feed leaving as liquid brine (per S3/S4). For a broader ZLD cost benchmark, the integrated wastewater plant cost comparison places this hybrid train in the mid-capex, low-opex quadrant relative to evaporator-only designs, and the decentralized ZLD market drivers 2026 analysis shows textile and lithium-battery sites adopting exactly this RO+MD+crystallizer topology.

Instrument the train with: feed conductivity (load tracking), MD permeate conductivity (wetting alarm, hard interlock at ~50 µS/cm), hot-side and cold-side temperature transmitters, differential pressure across each membrane module, and distillate flow totalizer. Integrate the hot side with existing plant utilities — condenser heat, boiler blowdown heat, or solar thermal — rather than installing a new steam source. Sludge from the DAF and from the crystallizer purge is handled in an integrated biological stage such as an MBR-integrated wastewater treatment unit feeding a high-efficiency sedimentation tank, with the dewatered cake pressed in a plate-and-frame filter press for solids disposal.

Cost Drivers and 2026 Specification Checklist

The five variables that move MD project economics are: thermal-energy source and cost, membrane replacement frequency (almost always wetting-driven, not age-driven), pumping energy for hot- and cold-side circulation, heat-exchanger fouling on the hot loop, and pretreatment chemical cost. The single largest OPEX swing is low-grade heat availability — the S2 review explicitly identifies heat-and-mass-transfer limits as the dominant scale-up barrier, which means any improvement in heat-source temperature or in heat-exchanger approach temperature directly reduces specific energy.

The most common procurement mistakes are over-specifying flux (vendors will quote peak flux on a clean module at maximum ΔT; nameplate must be derated for fouling and turndown), ignoring heat-source CAPEX (a $0.02/kWh waste-heat stream changes the project IRR more than a 10% membrane discount), and omitting a DAF/UF pretreatment step that would have prevented the wetting event that killed year-one membranes.

Looking forward, the active 2026 research areas — interfacial heating (Hu et al., 2025-05) and module-connection optimization (Nthunya & Mamba, 2025-08) — directly attack the two biggest MD cost drivers (thermal energy and module-level flux stability), so a plant specifying in 2026 should expect OPEX to keep falling over the next design cycle.

Frequently Asked Questions

Where does membrane distillation sit in a ZLD train relative to RO and the crystallizer?

Membrane distillation sits between reverse osmosis and the crystallizer. RO handles the bulk concentration to ~50% recovery on a 70 g/L feed ceiling (per S3/S4), MD polishes the RO concentrate to near saturation using a 40–80 °C thermal gradient, and the crystallizer then solidifies the remaining brine. The hybrid train reaches ~95% freshwater recovery and sends less than 10% of the original flow to evaporation ponds (per S3/S4).

What feed salinity can RO and OARO handle, and where does MD take over?

Standalone RO is limited to a maximum feed salinity of 70 g/L with up to 50% freshwater recovery. OARO extends that ceiling to 140 g/L feed salinity and up to 72% recovery. MD operates above both, concentrating brine to near saturation because its driving force is vapor pressure, not hydraulic pressure (per S3/S4).

What are the primary failure modes an MD system faces in a ZLD duty?

Three failure modes dominate: fouling, scaling, and membrane wetting. Fouling and scaling reduce flux over time and worsen as brine approaches saturation; wetting — caused by surfactant carryover, oil ingress, or dry-out — destroys salt rejection and is irreversible without element replacement (per S2, Nthunya & Mamba, 2026-07, and Cho et al., 2025-12).

What overall water recovery can a properly designed MD + RO + crystallizer train achieve?

A properly designed hybrid train reclaims as much as 95% of the inlet water as distillate, releases less than 10% of wastewater to evaporation ponds, and converts up to 90% of the waste stream into recovered water (per S3/S4). The remaining solids exit the crystallizer as salt for disposal or resale.

Can MD run on waste heat instead of a dedicated steam source?

Yes — and that is usually where the economics are made. MD operates on a 40–80 °C temperature gradient, which matches condenser reject heat, boiler blowdown heat, geothermal return-loop temperature, or solar thermal collectors. A site that already wastes low-grade heat at 60 °C can in principle feed that stream directly into the hot side of an AGMD or VMD module, offsetting most of the operating thermal load and changing the project IRR more than any membrane or pump efficiency gain.

Further Reading

• DAF design parameters for 2026