Reverse Osmosis Sizing: Pressure Balance, Transport Models and What Element Specs Really Tell You
Reverse osmosis looks simple on a flow sheet: pressurise the feed, push water through a membrane, collect permeate on the low-pressure side and concentrate on the high-pressure side. The engineering difficulty sits in the gap between that description and what actually happens inside a spiral-wound element over a three-year run. Anyone specifying an RO skid has to reconcile three things at once: the osmotic pressure of the feed, the transport model that explains rejection, and the commercial membrane options that exist for a given water chemistry.
Why Pressure, Not Pore Size, Sets the Ceiling
Put fresh water and seawater on either side of a semi-permeable membrane and the solvent migrates toward the concentrated side until the head difference balances the chemical driving force. That head difference is the osmotic pressure, and it depends on what is dissolved, how much, and at what temperature - not on the membrane itself. For a dilute solution the relationship is written as pi = iCRT, where i counts the ions released when a solute molecule dissociates, C is molar concentration, R is the molar gas constant and T is absolute temperature. Apply more than that pressure to the concentrated side and the flow reverses.
The flux equation that falls out of this is N = Kh(delta p - delta pi), with Kh the hydraulic permeation coefficient, which creeps upward as temperature rises. Two practical consequences follow. Feed salinity sets a hard floor on operating pressure, and no amount of extra membrane area recovers that. And a plant commissioned in summer will show different permeate flow in winter at the same gauge reading, which is expected behaviour rather than a fault.
Three Models, and the One That Reconciles Them
Lonsdale and co-workers treated the active skin of an RO membrane as dense and pore-free. Solute and solvent both dissolve into that homogeneous layer, then diffuse under a chemical-potential gradient set by concentration and pressure. Adsorption at the feed surface and desorption at the permeate surface are assumed fast, so the middle step - molecular diffusion - controls the rate. Rejection then comes from the difference in solubility and diffusivity between the two species inside the membrane phase.
The competing picture is preferential adsorption with capillary flow. Dissolved organics such as alcohols, acids, aldehydes and esters depress surface tension, while some inorganic salts raise it slightly, because solute concentration at the surface differs from that in the bulk. When an aqueous solution meets a polymeric porous membrane whose chemistry rejects solute and prefers water, a thin pure-water layer forms at the interface, and pressure pushes that layer through the surface capillaries. In cellulose acetate the mechanism is described through hydrogen bonding: water binds to carbonyl oxygen atoms, the first adsorbed layer loses entropy and takes on an ice-like structure, and molecules then hop from one activation site to the next as bonds form and break. Crystalline regions stay inaccessible; transport happens through the disordered amorphous regions where bound water occupancy is low.
Both models have a problem. Under electron microscopy a dry membrane shows no pores at all, yet the capillary model needs them. Deng Yu and colleagues resolved the contradiction with what became known as the dry-closed, wet-open description, proposed through the 1980s-1990s. Dry the membrane and the pores contract and close, which is why micrographs show nothing. Wet it and the polymer swells, the solvent forces the pores open, and capillary flow becomes possible. The same membrane simply behaves differently in the two states, which is also why the two classic models each explain part of the evidence.
A related idea, the non-pressurised adsorption-osmosis or forward-osmosis route, appeared in the 1990s. Water permeates forward into a draw solution or a super-absorbent solid whose salt concentration exceeds seawater, so no external pressure is needed; the trade is that the draw solute then has to be recovered, usually by modest heating. Solid-salt desorption costs less energy than the liquid-salt route.
What the Membrane Actually Holds Back
RO membranes reject inorganic ions, colloidal matter and macromolecular solutes, which is why the product is described as purified water rather than filtered water. The same property makes RO usable for pre-concentrating macromolecular organic solutions before evaporation or further separation. On the rejection side, the practical rule of thumb is that single-stage RO suits feed with conductivity below 500 uS/cm, while a properly running unit delivers effluent conductivity in the 1-10 uS/cm band.
Reading Element Specifications Without Being Misled
Membrane selection has moved on three fronts. Ultra-low-pressure elements gained share from 1999 onward because lower pressure ratings let the whole mechanical train - pumps, vessels, piping - be specified down a class, cutting both power and capital. They started in small systems built around 4-inch elements and moved up; the largest unit running them reached an output of 650 t/h.
Low-fouling elements address the failure mode that dominates RO operating cost, trading a longer service life and less frequent cleaning against a modest premium. Surface charge is the third lever. Nitto Denko's positively charged ES10C produced 10-15 MOhm resistivity water in three-stage semiconductor service; three Hyundai Electronics plants in Korea run three-stage systems with a combined final output of 800 t/h at 8-9 MOhm; and a 170 t/h three-stage system at a Shanghai semiconductor plant hits the same figures. At the smaller end, two-stage systems at several domestic pharmaceutical plants, at 5-20 t/h scale, achieve permeate resistivity of 1.7-3 MOhm. High-temperature, food-grade and sanitary elements cover sterilisation duties - standard water-treatment membranes run 0-45 deg C, while sanitisable ones tolerate 90 deg C.
The Pre-treatment Train Around the Membrane
Raw water is collected, pressurised by a booster pump, and passed through a quartz-sand filter, an activated-carbon filter and a cation softener before a cartridge or security filter protects the elements. The RO main unit does the desalination; product then goes through a water-air mixer and on to filling. Getting the front end right pays off downstream: RO used as pre-desalination cuts the load on ion-exchange resins by over 90% and reduces regenerant consumption by the same margin, which is why resin life extension is often the strongest economic argument for the membrane step.
Where RO Earns Its Keep
Application splits into boiler make-up and industrial pure water at the top, drinking water second, then electronics, semiconductors, pharmaceuticals, medical care, food, beverage, liquor, chemicals and environmental protection. Growth has been steep: RO membrane use has expanded at roughly 20% a year since 1995, and by 1999 the industrial element market was supplying 6,000 eight-inch and 26,000 four-inch elements, with 2000 and 2001 stronger still. Applications across the water-treatment sector are estimated to have generated more than 1 billion RMB of annual output value.
Desalination capacity tells its own story. Plants producing 100,000 tonnes of water per day already operate abroad, with large spiral-wound units at 6,000 tonnes/day; domestic plants built or under construction produce 350-1,000 tonnes/day. Single-pass water utilisation reaches 45% abroad against roughly 35% domestically, and fishing-boat units mostly use small 2.5-inch elements. Fewer than ten domestic companies mass-produce desalination plants, and an 18,000-tonne/day facility under construction in Hebei is the largest domestic RO plant using seawater membranes. The structural weakness remains supply: around 95% of the RO membrane market is imported, with domestic product holding about 5%. Independent domestic production technology was only mastered in the early 21st century, after a state-supported programme run through the Hangzhou Water Treatment Development Centre under the State Oceanic Administration and its subsidiary Hangzhou Beidouxing Membrane Products Co., Ltd.