Forward Osmosis: How the Osmotically-Driven Membrane Process Works

What is forward osmosis and how does it differ from reverse osmosis?

Forward osmosis exploits the natural tendency of water to move across a semipermeable membrane from a region of low solute concentration (the feed) to a region of high solute concentration (the draw solution). The transport is spontaneous: it is driven by the difference in osmotic pressure across the membrane, so unlike pressure-driven membrane filtration the process needs no high-pressure pump to force water through the barrier. In reverse osmosis (RO) the applied hydraulic pressure must exceed the feed osmotic pressure to reverse this natural flow; in FO the osmotic gradient is the driving force.

That single distinction reshapes the whole engineering problem. In RO the energy is spent at the membrane, pressurising the feed. In FO the membrane step is nearly energy-free, but the solutes you added to the draw are now diluted and must be recovered so the product water can be released and the draw reused. The real capital and operating cost of an FO plant therefore lives almost entirely in the draw-regeneration step, not the FO stage itself. Understanding FO means understanding where that energy goes.

Because the feed is never pressurised, FO membranes operate at low fouling propensity and the fouling that does occur tends to be loosely bound and largely reversible by osmotic backwashing or a simple flow reversal. This makes FO attractive for feeds that would rapidly foul or scale an RO membrane: high-turbidity waters, digestate, landfill leachate, produced water, and the concentrated brines encountered in zero-liquid-discharge trains. The trade-off is throughput. FO fluxes are modest, and the process is only as good as the osmotic pressure of the draw and the ease with which that draw can be regenerated.

A useful mental model: RO is a brute-force process that pushes water against its will, while FO is a patient process that lets water follow its chemical potential and then charges the energy bill downstream. Both move water across the same class of dense polyamide or cellulose-triacetate membrane; they differ in where the thermodynamic work is done and, consequently, in what fouls, what scales, and what costs money.

What is the water-flux equation in forward osmosis?

To a first approximation, the water flux in FO is proportional to the osmotic-pressure difference across the active membrane layer, governed by the pure-water permeability coefficient of the membrane:

This is deceptively clean. The critical caveat is that π_draw and π_feed in the equation are the local osmotic pressures at the membrane active layer, not the bulk values you measure in the tanks. Concentration polarisation drives the local values sharply apart from the bulk: the draw is diluted right at the membrane surface while the feed is concentrated there, so the effective driving force is a fraction of the bulk difference. In real modules the observed flux can be less than a third of what the bulk osmotic-pressure difference would predict. Any FO design that uses bulk values naively will badly over-predict throughput.

The osmotic pressure itself is estimated from the van’t Hoff relation for dilute ideal solutions:

Worked example — osmotic pressures. Consider seawater feed at roughly 0.6 mol/L NaCl-equivalent at 25°C (298 K). Using van’t Hoff: π_feed = 2 × 0.6 × 0.083145 × 298 ≈ 29.7 bar, which matches the familiar ~28–30 bar figure for seawater. A 2 mol/L NaCl draw gives π_draw = 2 × 2.0 × 0.083145 × 298 ≈ 99 bar (before the osmotic-coefficient correction, which would push it higher). The bulk driving force is therefore about 99 − 30 = 69 bar. With a membrane permeability A = 1.0 LMH/bar, the ideal flux would be 69 LMH — but after concentration polarisation the achievable flux might be only 12–20 LMH. That gap is the entire story of FO membrane engineering.

Why is internal concentration polarisation the dominant flux killer?

Concentration polarisation (CP) is the build-up or depletion of solute at the membrane surface caused by the water flux itself. FO membranes are asymmetric: a thin dense active layer sits on a thick porous support. This geometry creates two distinct CP phenomena, and the one inside the support layer is the more damaging.

External concentration polarisation (ECP) occurs in the flowing boundary layers on either face of the membrane. On the feed side, water leaving the feed concentrates rejected solute against the active layer (concentrative ECP), raising π_feed locally. ECP is a hydrodynamic effect and can be suppressed by increasing cross-flow velocity, using spacers, or promoting turbulence — it is the same boundary-layer physics familiar from RO.

Internal concentration polarisation (ICP) occurs inside the porous support layer, where there is no cross-flow to sweep solute away. In the common FO orientation (active layer facing the feed, support facing the draw), draw solute must diffuse through the tortuous, stagnant support to reach the active layer. Water permeating the other way dilutes it. The result is dilutive ICP: the draw concentration at the active layer is far below the bulk draw. Because ICP happens in a region with no achievable mixing, it cannot be fixed hydrodynamically — only by improving the membrane. This is why ICP, not ECP, is the characteristic and dominant flux limitation of FO.

The severity of ICP is captured by the membrane structural parameter S, a length scale that combines the support-layer geometry:

The physical lesson is that in FO the support layer is not inert packaging — it is a mass-transfer resistance in series with the active layer, and it usually dominates. A membrane with an excellent active layer but a thick, dense support (an ordinary RO membrane, for instance) performs terribly in FO. Purpose-built FO membranes use thin, highly porous, low-tortuosity supports, sometimes electrospun nanofibre backings, precisely to drive S down. Modelling flux without an ICP term — typically an exponential correction of the form exp(−J_wS/D) applied to the draw osmotic pressure — will overstate performance by a factor of two to five.

How do you select a draw solute, and why is regeneration the real problem?

The draw solution is the engine of FO, and choosing it is a multi-objective optimisation. An ideal draw solute would simultaneously: (1) generate a high osmotic pressure at modest concentration; (2) be cheap, non-toxic and non-fouling; (3) exhibit low reverse diffusion back through the membrane; and (4) — most importantly — be easy and cheap to separate from the product water so the draw can be regenerated and recycled. No single solute is best on all four axes, so selection is a compromise dictated by the application.

The uncomfortable truth is the draw-regeneration problem. FO moves the energy cost downstream; it does not eliminate it. If the draw is a simple salt regenerated by RO, then the plant is essentially an FO pre-concentrator bolted onto an RO train, and the thermodynamic minimum energy to separate the water is unchanged — you have paid it in the RO stage. FO only wins when either (a) the FO step lets a difficult feed be treated that RO alone could not handle (fouling, scaling, very high salinity), or (b) the draw can be regenerated by a fundamentally cheaper route than pressurised RO — most notably low-grade or waste heat driving a thermolytic ammonia–carbon-dioxide draw. The classic thermolytic scheme uses ammonium bicarbonate/carbamate: heating the diluted draw to about 60°C decomposes it into NH₃ and CO₂ gases, which are stripped off (leaving clean product water) and then recombined and re-dissolved to reconstitute the concentrated draw. Where 40–60°C waste heat is free, this can undercut RO on energy.

So the correct engineering question is never “which draw has the highest osmotic pressure?” but “which draw–regeneration pairing has the lowest total cost of ownership for this feed and this site’s available energy?” A draw that is trivial to regenerate with on-site waste heat beats a stronger draw that needs high-pressure RO every time waste heat is available.

What is reverse solute flux and why does it matter?

No membrane is perfectly selective, so while water permeates from feed to draw, a small amount of draw solute leaks back the other way, from draw to feed. This reverse solute flux (RSF, J_s) is a defining loss mechanism in FO and has three separate consequences: it depletes the expensive draw (an operating cost, as make-up solute must be added continuously), it contaminates the feed/retentate (a problem for brine disposal or ZLD brine concentration), and it accelerates fouling and can drive scaling at the membrane surface.

That last point is important and non-intuitive. Because J_s/J_w collapses to (B/A)/(i·R·T), it is a clean membrane-quality metric: a low value means the membrane is well selective and you lose little draw per litre of water made, whatever the operating concentration. Divalent-salt draws (MgCl₂, with i ≈ 3 and large hydrated ions the membrane rejects well) give low specific reverse flux; small monovalent solutes leak more freely.

Worked example — draw loss. Suppose a membrane has A = 1.0 LMH/bar and B = 0.5 LMH for an NaCl draw at 25°C. With i = 2, R = 0.083145 and T = 298, the term i·R·T = 49.5 L·bar/mol. NaCl molar mass is 58.4 g/mol, so J_s/J_w ≈ (0.5/1.0) × (58.4 / 49.5) ≈ 0.59 g NaCl per litre of water produced. For a plant making 100 m³/day of product, that is 0.59 kg/m³ × 100 m³ = 59 kg/day of NaCl lost into the feed and needing replacement — modest for cheap salt, but potentially prohibitive for an expensive engineered draw solute, which is why exotic draws must pair low RSF with genuine recyclability.

Where is forward osmosis actually used?

FO earns its place wherever its strengths — low fouling, tolerance of harsh feeds, no hydraulic pressure, and the option of waste-heat-driven regeneration — outweigh its modest flux. The leading applications are:

• FO–RO hybrids. FO acts as a robust pre-treatment/barrier: a difficult feed (wastewater, produced water) is drawn into a clean saline draw across the fouling-resistant FO membrane, and the diluted draw is then processed by RO to yield product water and re-concentrate the draw. The FO stage protects the RO membranes from the foulants, extending life and cutting cleaning.
• Brine concentration and zero-liquid-discharge. FO can concentrate hypersaline brines beyond the ~70–75 bar practical ceiling of RO, because there is no pressure vessel limit — you simply need a draw of even higher osmotic pressure. This lets FO push brines toward saturation ahead of a crystalliser or evaporator, shrinking the thermal-ZLD duty and its energy bill. See our overview of zero-liquid-discharge systems.
• Osmotic dilution. A concentrated stream that must be diluted anyway (e.g. seawater feed to an RO desal plant, or fertiliser in fertigation) is used as the draw to extract clean water from an impaired source such as treated wastewater. The impaired water is upgraded and the draw is diluted for free — in fertiliser-drawn FO the diluted fertiliser solution is applied directly to crops with no regeneration step at all. This connects naturally to seawater desalination pre-dilution schemes and to industrial water reuse.
• Food, beverage and pharma concentration. Because FO runs at ambient temperature and pressure, it concentrates juices, coffee, dairy and heat-sensitive pharmaceutical or biological products without the thermal damage of evaporation, preserving flavour, aroma and bioactivity. Here an edible draw (sugar or salt) may need no regeneration at all.
• Emergency and portable hydration. Single-use FO “hydration bags” use an edible sugar draw to pull drinkable water from contaminated or brackish sources — the diluted sugar solution is the drink.

For a broader view of pressure-driven alternatives and where each membrane process fits, see our guide to reverse osmosis system design. The recurring theme across all FO applications is the same: FO is chosen not because it is the cheapest way to move water across a membrane — it usually is not — but because it treats feeds that defeat RO, or because it unlocks a cheaper regeneration energy source, or because it avoids the heat and pressure that would ruin a sensitive product.

What are the practical design limits and trade-offs?

Selecting FO over RO (or over a thermal process) comes down to matching the process to the feed, the available energy, and the required recovery. The table summarises the core trade-offs.

Worked example — regeneration energy sets the economics. The thermodynamic minimum work to separate pure water from seawater (~35 g/L, ~28 bar) at 50% recovery is about 1.1 kWh/m³; real RO achieves ~3–4 kWh/m³. In an FO–RO hybrid, the FO stage adds essentially no separation work but the diluted draw still has to be re-concentrated by RO, so the hybrid’s electrical energy is comparable to RO alone — the FO stage buys fouling resistance, not energy savings. By contrast, a thermolytic NH₃–CO₂ draw regenerated with 40–60°C waste heat can move most of the separation duty from electricity to (nearly free) low-grade heat. If a site has, say, 500 kW of otherwise-wasted 55°C process heat, the same duty that would cost several hundred kW of RO pumping power can be met thermally — that, and not the membrane step, is where FO’s economic case is won or lost.

The design rule that follows is blunt: never evaluate an FO stage in isolation. Its flux, its draw, and its regenerator are one coupled system, and the figure of merit is the total cost of delivered product water including make-up draw (from reverse solute flux), regeneration energy, and the value of any RO membranes protected from fouling. FO is a systems play. Get the draw–regeneration pairing wrong and a technically elegant low-fouling membrane becomes an expensive way to do what RO already does; get it right — a harsh feed, a cheap waste-heat regenerator, a high-value recovery target — and FO does what no pressure-driven process can.