Seawater Desalination: Thermal vs Membrane and the Energy Physics

Thermal or membrane: which route and why?

There are two families of seawater desalination. Thermal processes evaporate seawater and condense the vapour, leaving salt behind: multi-stage flash (MSF) flashes brine across successively lower-pressure stages, while multi-effect distillation (MED) boils thin films across a train of effects. Membrane desalination — seawater reverse osmosis (SWRO) — applies hydraulic pressure above the osmotic pressure to force water through a semi-permeable polyamide membrane while salts are rejected.

Thermal plants tolerate high-fouling, high-temperature feed and produce very pure distillate (<10 mg/L TDS), which is why they persisted in the Gulf where cheap co-generation heat was available. But thermodynamically, boiling water is an expensive way to separate a dilute solute. The energy of separation is small; the latent heat of vaporisation (~2.3 MJ/kg) is enormous, and only partly recovered through the GOR (gain output ratio) of the effects. SWRO instead spends work only against the osmotic pressure, so once high-rejection SWRO membranes and energy-recovery devices matured, its energy demand fell to a level thermal routes cannot approach.

Thermal "equivalent" energy converts fuel heat to an electrical basis; the exact figure depends on the value assigned to low-grade steam, but the ranking holds.

Why is seawater so hard to push through a membrane?

The obstacle is osmotic pressure. Water spontaneously flows toward the saltier side of a semi-permeable membrane; to reverse that flow you must apply a hydraulic pressure exceeding the osmotic pressure difference. For a dilute electrolyte the van’t Hoff relation gives a good first estimate:

Because you must beat this osmotic pressure and supply a net driving pressure to generate useful flux, SWRO trains run at a high 55–70 bar. And as you extract permeate, the retentate concentrates: at 50% recovery the reject roughly doubles to ~70,000 mg/L, pushing its osmotic pressure past 50 bar. That rising back-pressure — not the membrane area — is what caps single-pass SWRO recovery at about 40–50%, versus 75–85% for brackish RO. Push higher and you need impractical pressures, risk scaling, and gain little permeate.

Worked example: osmotic pressure and the minimum energy

Take standard seawater and compute both the osmotic pressure and the thermodynamic floor on energy.

• Molarity: treating 35,000 mg/L as NaCl (M = 58.44 g/mol): C = 35 / 58.44 = 0.599 mol/L.
• Osmotic pressure: π = 2 × 0.599 × 0.083145 × 298 = 29.7 bar (real seawater ~27 bar; van’t Hoff slightly over-predicts at this strength).
• Reversible minimum energy at recovery r: SEC_min = π₀ · ln[1/(1−r)] / r. At r = 0.45: SEC_min = 27 × ln(1/0.55) / 0.45 = 27 × 0.598 / 0.45 = 35.9 bar of work per m³ permeate.
• Convert to electrical units (1 bar·m³ = 0.02778 kWh): 35.9 × 0.02778 = ≈1.0 kWh/m³.

So ~1 kWh/m³ is the absolute thermodynamic floor for a reversible seawater separation at 45% recovery. No real process reaches it because it assumes infinite membrane area, zero flux, and zero friction — but it anchors every efficiency claim. A modern SWRO train delivering ~3–4 kWh/m³ is operating at roughly 30% of the reversible limit; thermal plants sit at a few percent.

How do energy-recovery devices cut SWRO energy?

The single biggest lever is what you do with the concentrate. In an SWRO train the reject leaves the last pressure vessel at nearly full applied pressure — typically only 2–3 bar below feed. Dumping that stream to drain throws away most of the pumping energy. Early SWRO plants that did exactly this used 5–7 kWh/m³.

Worked comparison at P = 65 bar, r = 0.45, η_pump = 0.85 (feed = 2.22 m³ per m³ permeate; concentrate = 1.22 m³):

• Without ERD — the HP pump pressurises all 2.22 m³ of feed: SEC = 65 × 2.22 / 0.85 = 170 bar·m³ → 4.7 kWh/m³.
• With an isobaric pressure exchanger (~96% efficient) — the device pressurises the 1.22 m³ concentrate-matched feed for almost free; the HP pump handles ~1.0 m³ at 65 bar and a small booster tops up ~3 bar across the PX stream: SEC = (65 × 1.0 / 0.85) + (3 × 1.22 / 0.85) = 76.5 + 4.3 = 80.8 bar·m³ → 2.2 kWh/m³.

The pressure exchanger more than halves the membrane-unit energy. Add intake pumping, pretreatment, a boron second pass and product transfer, and a real plant lands at the familiar 3–4 kWh/m³. Isobaric ceramic-rotor and piston exchangers (~95–98% efficient) have displaced older Pelton turbines precisely because those last few points of efficiency dominate the whole-plant number.

What does seawater pretreatment have to remove?

Polyamide SWRO membranes are intolerant of particulates, biofouling and oxidants, so intake and pretreatment define plant reliability. Open surface intakes carry algae, plankton and seasonal blooms; subsurface (beach-well or gallery) intakes deliver naturally filtered, low-turbidity water at higher capital cost. Downstream, the target is a silt density index (SDI₁₅) below ~3 and often below 2 before the membranes.

• Screening and coagulation to remove suspended solids and destabilise colloids.
• Media or ultrafiltration (UF) as the fouling barrier — UF now dominates new builds for its robust, algal-bloom-tolerant SDI.
• Cartridge filtration (5 µm) as the final guard before the high-pressure pumps.
• Antiscalant and dechlorination — any residual chlorine must be quenched (e.g. with bisulphite) because it oxidises polyamide.

Boron is the signature seawater challenge. Seawater carries ~4–5 mg/L boron, present at natural pH (~8) as uncharged boric acid (pK_a ≈ 9.2), which a first-pass membrane rejects poorly (~75–90%). To meet the WHO guideline of 2.4 mg/L — or stricter irrigation limits — plants add a partial second pass at elevated pH (>9.5). Raising pH converts boric acid to the charged borate ion, which is rejected like any other salt, so the second pass strips boron efficiently.

How is the brine (concentrate) managed?

At ~45% recovery, roughly 55% of the intake leaves as concentrate at ~65,000–75,000 mg/L — nearly twice ambient salinity — plus antiscalant residuals and any coagulant. Because this brine is denser than seawater, it sinks and can form a hypersaline layer that smothers benthic habitat if discharged carelessly. Responsible design uses multiport diffusers to inject the brine as high-velocity jets that entrain seawater and dilute the plume rapidly to within a few percent of background salinity at the edge of a defined mixing zone. Co-discharge with power-station cooling water or treated effluent is another dilution route.

Where discharge is constrained — inland desalination, sensitive receiving waters, or a regulator demanding near-zero liquid effluent — the alternative is to concentrate the brine further toward solids using zero liquid discharge (ZLD) or minimal liquid discharge (MLD). These pair high-pressure or high-recovery membranes with thermal evaporators and crystallisers to recover additional water and leave a dry salt cake, trading energy and capital for the elimination of a liquid discharge. Read our detailed treatment of ZLD and MLD concentrate strategies for where that balance makes sense.

When is SWRO the right choice &mdash; and when not?

SWRO is the default for new seawater capacity: lowest energy, modular scale-up, electrical (not thermal) drive, and rapid technology gains. It is the clear pick where power is the binding constraint and the feed can be economically pretreated. Thermal MSF/MED still earns a place where waste heat is genuinely free (integrated power-and-water complexes), where feed fouling is extreme, or where ultra-pure distillate is required. For engineers moving from brackish to seawater duty, the governing shift is the jump in osmotic pressure and the collapse of achievable recovery — the same membrane physics as our reverse osmosis system design guide, but pushed to the pressure and pretreatment extremes. The design that succeeds is the one that respects the thermodynamic floor, recovers concentrate energy aggressively, and plans brine disposal from day one.

How to specify a seawater desalination (SWRO) train

1. Characterise the source water. Establish TDS, temperature range, turbidity/SDI, algal-bloom risk and boron level. These set osmotic pressure, pretreatment and second-pass needs.
2. Select intake and pretreatment. Choose open vs subsurface intake, then coagulation plus media or ultrafiltration to reach SDI15 below 3, with dechlorination ahead of the membranes.
3. Set recovery and applied pressure. Pick recovery around 40–50%; check that concentrate osmotic pressure plus net driving pressure stays within a 55–70 bar operating envelope.
4. Size the energy-recovery system. Specify isobaric pressure exchangers (95–98% efficient) and a booster pump so the HP pump only pressurises the permeate-equivalent flow.
5. Add a boron second pass if required. Where product boron must meet WHO/irrigation limits, add a partial second pass at elevated pH to reject borate.
6. Design brine discharge or ZLD. Model the hypersaline plume and specify multiport diffusers, or adopt ZLD/MLD where liquid discharge is not permitted.