Membrane Distillation (MD): Principle & Process

By the Petro Sep engineering team

Published 11 July 2026Updated 28 October 2026

How does membrane distillation work, step by step?

A heated feed flows along one face of the membrane; water evaporates at the entrances of the dry pores, the vapor crosses them and is condensed on the far side or outside the module, and the feed leaves more concentrated.

Heating the feed. The feed is heated, but not boiled. Reviews describe saline feeds heated to 30 to 80 °C before the module, and MD run mostly below 90 °C, under the feed's normal boiling point. Petro Sep's AQUA-SEP® system operates at about 70 °C.
Contact with the membrane. The hot feed flows along a microporous, hydrophobic membrane, and in every configuration it touches the water-repellent face. Because the membrane repels liquid water, the feed does not enter the pores: only vapor passes.
Evaporation at the pore mouths. Water evaporates at the liquid-vapor interface that forms at the entrance of each pore, on the feed side.
Vapor transport. The vapor moves through the pores toward the permeate side, where the vapor pressure is lower.
Condensation. The vapor becomes liquid distillate: in a cold water stream on the far side of the membrane, on a cooled plate behind an air gap, or in a condenser outside the module after a sweep gas or a vacuum has drawn it away. These options are the configurations described below.
Concentration of the feed. Salts and other non-volatile solutes stay in the feed, which leaves the module more concentrated. Salt rejection is 100% in theory, and reported rejections of ions and non-volatile solutes reach 99.9%, even close to saturation. Volatile compounds behave differently: they can evaporate and cross with the water vapor.
Heat. The heat that evaporates the water comes from the feed. It crosses the membrane in two ways: as latent heat carried by the vapor, which does the separating, and by conduction through the membrane material and the vapor in its pores, which is lost.

In a recirculating unit, such as the one in the working-principle diagram of Petro Sep's AQUA-SEP presentation, a pump draws the feed from a heated feed tank, sends it through the membrane module and returns it to the tank, and the concentrate is drawn off from the tank. The terms used here are defined in the membrane glossary.

What is the principle of membrane distillation?

A vapor pressure difference across the membrane: water vapor moves from the feed side, where its vapor pressure is higher, to the permeate side, where it is kept lower.

A reverse osmosis membrane is semi-permeable: a high-pressure pump forces water through it and the salts are held back. A membrane distillation membrane only has to keep liquid out of its pores, and the separation is that of distillation: what evaporates crosses, and what does not evaporate stays in the feed. The vapor pressure difference can be created in three ways.

By temperature. A hot feed on one side and a colder stream, plate or gas on the other: the vapor pressure difference is induced by the temperature difference across the membrane, and the greater that difference, the greater the driving force and the flux. Direct contact, air gap and sweeping gas MD work this way.
By vacuum. In vacuum membrane distillation (VMD), a vacuum on the permeate side, held below the saturation pressure of the water, creates the difference. Petro Sep's AQUA-SEP presentation describes the driving force as the vapor pressure difference between both ends of the membrane pore, due to the vacuum applied on the permeate side. One review gives a vacuum of 5 to 10 kPa.
By concentration. In osmotic membrane distillation (OMD), the driving force is a concentration gradient: the difference between the water activities of the two solutions.

The membrane stays dry only while the pressure difference between the liquid and the vapor side stays below its liquid entry pressure (LEP). Above it, liquid can be pushed into the pores and wet the membrane; in VMD, for example, a stronger vacuum raises the flux, but taking the pressure difference past the LEP may wet the membrane. MD itself runs at or below atmospheric pressure, and needs no feed pressure to overcome the osmotic pressure, as reverse osmosis does.

What are the types of membrane distillation?

There are four main configurations, which differ in how the vapor pressure difference is created and how the vapor is collected and condensed on the permeate side: direct contact (DCMD), air gap (AGMD), sweeping gas (SGMD) and vacuum (VMD).

Table 1. The four membrane distillation configurations, as published reviews describe them.
Configuration	Permeate side	Where the vapor condenses	Strengths	Drawbacks
Direct contact (DCMD)	Cold water flows in direct contact with the membrane	In the cold stream, inside the module	The simplest and oldest configuration, and the most studied: more than 60% of lab-scale studies use it	Heat is lost by conduction through the membrane; it has found little use in pilot-scale systems
Air gap (AGMD)	A thin stagnant air gap, usually 2 to 10 mm, then a cooled condensing plate	On the cold plate, inside the module	Less conductive heat loss; the highest energy efficiency of the single-stage modes, and internal heat recovery is more likely	The air adds resistance to mass transfer, so the flux is usually low
Sweeping gas (SGMD)	A cold, inert gas sweeps the vapor away	In a condenser outside the module	The gas barrier reduces conductive heat loss and improves mass transfer	Needs a big condenser for the large volume of gas; the least studied and least used configuration
Vacuum (VMD)	A vacuum draws the vapor away	In a condenser outside the module	Negligible conductive heat loss and the highest permeate flux of the four	The membrane is more prone to wetting under vacuum; the condenser and vacuum pump add cost, and heat recovery is harder because the vapor condenses outside the module

Variants build on these four. In permeate gap MD (PGMD), the air gap of AGMD is filled with permeate water; in vacuum-assisted AGMD (V-AGMD), a partial vacuum is pulled in the air gap. SGMD and VMD are also used to remove volatile organics or dissolved gases from water.

Petro Sep's membrane distillation technology is VMD, and it supplies no other configuration. Its AQUA-SEP brochure describes VMD as "potentially the most efficient MD configuration", and a better option than direct contact MD for desalting feeds with high salt concentrations, because fouling in VMD is usually low and can be controlled through chemical or water wash. Petro Sep's systems are covered on the vacuum membrane distillation (VMD) systems page.

What membranes does membrane distillation use?

Porous membranes that are hydrophobic, either by nature or through a surface treatment; polypropylene (PP), polytetrafluoroethylene (PTFE) and polyvinylidene fluoride (PVDF) are the most used.

Hydrophobicity is the basic requirement: the material must repel water itself, or its surface must be modified to do so. Of the three common polymers, PTFE is the one mostly used in pilot-scale and commercial MD. Membranes are made as flat sheets or hollow fibers, and flat sheets can be wound into spiral-wound modules.

Petro Sep's AQUA-SEP uses polymeric hollow-fiber modules, with membranes made of plastic (PE, PP, PVDF and others). Its brochure lists a "high velocity hollow fibre construction" among AQUA-SEP's advantages and states that the hollow-fiber modules are used to eliminate scaling or fouling at the membrane surface.

Where is membrane distillation used?

Mainly to make clean water from salty water, desalting seawater and brackish water, and increasingly to recover water from brines such as reverse osmosis (RO) reject; it also concentrates liquids and removes volatile organics.

Desalination. Petro Sep's AQUA-SEP brochure names desalting seawater and producing clean water from brackish water as the most common VMD applications, because non-volatile ionic particles are almost completely retained. Reviews add brine treatment.
RO brine. Recovering water from high-salinity solutions such as RO brine is seen by most academic and industry experts as the most promising future application of MD. Paired with RO, it forms a complementary system: RO recovers water first, and MD concentrates the brine to near saturation.
Crystallization and zero liquid discharge. In membrane distillation crystallization (MDC), MD produces pure water and the concentrated brine goes to a crystallizer that recovers a valuable crystal product. Petro Sep integrates VMD with zero liquid discharge systems "up to the point of crystallization"; see What is zero liquid discharge (ZLD)? and hybrid ZLD.
Concentrating liquids. MD concentrates dairy streams in the food industry, and the AQUA-SEP brochure lists fruit juices and milk, cooling liquids (glycols), non-volatile acids and oil-in-water emulsions, and streams that high-temperature processing would degrade. Petro Sep's VMD presentation adds removing water from acids (succinic, adipic, propanoic and sulfuric acid) and bases (sodium hydroxide).
Industrial water. The brochure lists water purification in the pharmaceutical, chemical and textile industries, and textile wastewater contaminated with dyes.
Volatile organics. SGMD and VMD are typically used to remove a volatile organic or dissolved gas from an aqueous solution.

How does membrane distillation compare with reverse osmosis and evaporators?

On feeds within its salinity limit, RO uses less energy, because it involves no phase change; MD evaporates water, as an evaporator does, which lets it treat much saltier feeds, and it does so below the boiling point, at low pressure, and can run on low-grade heat.

Table 2. Reverse osmosis and membrane distillation compared. The salinity and recovery figures are the maximums given in one 2025 review of membrane technologies for zero and minimal liquid discharge.
	Reverse osmosis (RO)	Membrane distillation (MD)
How water crosses	Forced through a semi-permeable membrane by a high-pressure pump	Evaporates and crosses a porous, hydrophobic membrane as vapor
Pressure	High, and rising with salinity, to overcome the osmotic pressure	At or below atmospheric; no feed pressure needed
Energy	Lower specific energy consumption, as no phase change is involved	A phase-change process that needs plenty of energy; low-grade or waste heat can supply it
Maximum feed salinity	70 g/L	350 g/L, reached in experiments near saturation under controlled conditions; lower concentrations are typically preferred for sustained operation
Maximum freshwater recovery	Up to 50%	Up to 90%
Fouling	Pretreatment is critical to prevent fouling and scaling	Less sensitive to concentration polarization and fouling than pressure-driven processes, and biofouling is less likely; scaling and wetting remain
Maturity	A strong commercial presence, widely used at full scale	An emerging technology, still early in its commercialization

Conventional thermal processes, such as evaporators, keep two advantages over RO that MD shares (Table 2): they are less susceptible to fouling and less sensitive to feed salinity. Petro Sep's own comparison of its VMD system with an evaporator notes that both recover water from brines, both can concentrate acid or base mixtures and both use mechanical vapor compression; the evaporator has to be built of corrosion-resistant metals and loses efficiency to scale on its heat-transfer surfaces, while the VMD membranes are plastic, though its condenser also has to be corrosion resistant. Petro Sep's AQUA-SEP brochure adds that MD could use low-grade and inexpensive heat sources, with a smaller plant footprint and lower capital costs than conventional distillation processes. The figures for Petro Sep's system, against RO and an evaporator and crystallizer, are on the VMD systems page.

How much energy does membrane distillation use?

It depends on the configuration and on how much heat is recovered: MD has to evaporate the water it produces, and published figures range from under 50 to several thousand kilowatt-hours per cubic meter.

The measure of heat reuse is the gained output ratio (GOR): the heat recovered as latent heat of evaporation over the total heat supplied. The more of that latent heat a design reuses, the higher its GOR; in multi-effect MD, adding effects usually raises the GOR, though too many reduce the permeate flux.

Table 3. Examples of MD energy figures cited in a 2022 review of MD modules. The first two come from one pilot study, read in the original.
Study	Energy	Conditions
Vacuum-assisted AGMD, spiral-wound pilot modules	49 kWh/m³ of heat, GOR 13.5	Sodium chloride solution at seawater strength (35.1 g/L), 80 °C at the inlet
The same pilot	Below 515 kWh/m³ of heat, GOR 1.2	Sodium chloride solution at 292.2 g/L
AGMD pilot, 10 m³ per day, 50 days	182 to 208 kWh/m³ of heat (tap water); 199 to 231 (seawater); 294 to 373 (concentrated brine)	Various operating conditions
DCMD	100 kWh/m³ of heat, plus 1.5 to 3.65 kWh/m³ of electricity	60 to 80% recovery
Vacuum-enhanced DCMD	3,176.4 and 4,189.5 kWh/m³	Oil-field produced water at 10,000 and 230,000 ppm

In the AGMD pilot, the concentrated brine took the most heat per cubic meter, and in the vacuum-enhanced DCMD study the saltier water took more energy. Because MD runs at low temperature, the heat can come from low-grade waste heat or solar energy, and Petro Sep's General Presentation names unused waste or low-grade energy from power plants and refineries as a way to reduce energy further.

Petro Sep's AQUA-SEP uses mechanical vapor compression: the vapor drawn from the module is compressed and condensed in a heat exchanger, and its latent heat is captured and reused in the system, for example to keep the circulating water at temperature. On Petro Sep's energy data sheet for a 100 GPM unit, AQUA-SEP uses 26 kWh/m³ of electricity and 129 kWh/m³ of utilities, against 43 and 300 kWh/m³ for an evaporator and crystallizer. The sheet does not break the utilities figure down, so it is not directly comparable with the thermal figures in Table 3.

What are the limits of membrane distillation?

Wetting of the pores, fouling and scaling, heat lost through the membrane, and a lower flux and higher energy demand than RO; reviews place the technology early in its commercialization.

Wetting. When liquid feed penetrates the pores, the distillate quality falls and the membrane can foul. It can happen when the pressure difference across the membrane exceeds the liquid entry pressure, or when reactive components of the feed make the membrane surface hydrophilic during operation. Under vacuum, the membrane is more prone to wetting.
Fouling and scaling. Fouling is inorganic, organic or biological. Inorganic fouling, called scaling, is the deposit of minerals that precipitate from a supersaturated feed. Biofouling is less likely than in RO, because of MD's high salinity and temperature. Proper pretreatment of the feed, and suitable membranes and operating conditions, are needed to mitigate fouling; Petro Sep's AQUA-SEP brochure lists pretreatment as required.
Heat loss. 20 to 40% of the energy the hot feed supplies is lost by conduction through the membrane. The membrane surfaces also run at temperatures different from the bulk streams, forming thermal boundary layers; this temperature polarization limits mass transport.
Flux and energy. The permeate flux of MD is still lower than that of conventional RO, and reviews name wetting, fouling, low flux and high energy consumption as what has held MD back.
Maturity. Commercialization of MD is still at an early stage, and reviews class it as an emerging technology.

How does Petro Sep use membrane distillation?

As vacuum membrane distillation only: Petro Sep's system is AQUA-SEP®, a hollow-fiber VMD process for brines, RO reject and high-salinity wastewater.

AQUA-SEP pumps a heated brine along hydrophobic, microporous hollow fibers while a vacuum is held on the other side of the membrane; the vapor is compressed and condensed, and the brine is concentrated up to the point of crystallization. Petro Sep's documents give it the ability to process feeds of 25% salt and more, product water below 10 ppm from a 40,000 ppm feed in the brochure's example, and 85 to 90%+ total feedwater recovery; Petro Sep's RO presentation adds that routing an RO plant's reject to AQUA-SEP could raise water recovery up to about 95% overall. Its project records start with a modular skid for seawater desalination in Saudi Arabia in 2002 (Saudi Arabia desalination pilot), followed by full-scale membrane skids for RO reject treatment and brine concentration in China in 2008 (China brine concentration) and a mobile hazardous water treatment vacuum skid in Rhode Island in 2014 (Rhode Island hazardous water). Sizes, specifications and energy figures are on the VMD systems page and the AQUA-SEP page.

For a fuller treatment of the process, see Fakhir U. Baig's chapter "Pervaporation and Hybrid Vacuum Membrane Distillation Technology and Applications", in Membrane Separation Principles and Applications (Elsevier, 2019).

Membrane distillation vs pervaporation: what is the difference?

Both take a liquid feed and evaporate part of it at a membrane, but MD uses a porous, water-repellent membrane whose pores the vapor crosses, while pervaporation uses a dense, non-porous membrane and separates by affinity for it.

Table 4. Membrane distillation and pervaporation.
	Membrane distillation (MD)	Pervaporation (PV)
Membrane	Porous, hydrophobic	Dense, non-porous
What sets the separation	Volatility: what evaporates crosses the pores, and non-volatile solutes stay behind	Mainly the components' affinity for the membrane (in Petro Sep's AZEO-SEP® brochure, their polarity) rather than their volatility
Typical job	Water out of salty or contaminated water	Water out of organic solvents, or organics out of water
At Petro Sep	AQUA-SEP, a VMD system	AZEO-SEP (hydrophilic) and VOC-SEP® (hydrophobic)

Frequently asked questions

When was membrane distillation invented?

The first patent related to MD was filed in 1963, by Bodell, and the first MD paper was published in 1967, by Findley.

What temperature does membrane distillation run at?

Below the feed's boiling point: reviews describe MD run mostly below 90 °C, with saline feeds heated to 30 to 80 °C. Petro Sep's AQUA-SEP operates at about 70 °C.

Does membrane distillation need high pressure?

No. MD runs at or below atmospheric pressure and needs no feed pressure to overcome the osmotic pressure, as RO does. In vacuum MD the permeate side is under vacuum, and the pressure difference across the membrane has to stay below its liquid entry pressure, or the pores can wet.

Can membrane distillation run on waste heat?

Yes. Because it runs at low temperature, MD can use low-grade waste heat or solar energy. Petro Sep's General Presentation names unused waste or low-grade energy from power plants and refineries as a way to reduce energy further.

Can membrane distillation treat RO reject?

Yes. Recovering water from RO brine is seen by most academic and industry experts as MD's most promising future application. Petro Sep's RO material says that routing RO reject to an AQUA-SEP unit could raise water recovery up to about 95%.

Which configuration suits high-salinity feeds?

Petro Sep's AQUA-SEP brochure considers VMD a better option than direct contact MD for desalting feeds with high salt concentrations, because fouling in VMD is usually low and can be controlled through chemical or water wash. Reviews add that VMD has the highest flux and the lowest conductive heat loss of the four configurations, and that its membrane is more prone to wetting.

Does membrane distillation remove volatile compounds?

Not by itself. MD holds back non-volatile solutes, but volatile compounds can evaporate and cross with the water vapor; SGMD and VMD are in fact used to strip volatile organics and dissolved gases from water.

Is membrane distillation the same as pervaporation?

No. MD uses a porous, water-repellent membrane, and the vapor crosses its pores; pervaporation uses a dense, non-porous membrane and separates by the components' affinity for it.

Sources

Petro Sep, AQUA-SEP® brochure.
Petro Sep, AQUA-SEP® vacuum membrane distillation presentation.
Petro Sep, AQUA-SEP® energy consumption comparison data sheet (100 GPM unit).
Petro Sep, AQUA-SEP® and reverse osmosis presentation.
Petro Sep, General Presentation (January 2024).
Petro Sep, Company Profile (January 2024).
Petro Sep, AZEO-SEP® brochure.
Petro Sep, VOC-SEP® brochure.
Parani and Oluwafemi, Membrane Distillation: Recent Configurations, Membrane Surface Engineering, and Applications (Membranes, volume 11, issue 12, article 934, 2021, open access).
Francis, Ahmed and Hilal, Advances in Membrane Distillation Module Configurations (Membranes, volume 12, issue 1, article 81, 2022, open access).
Zhang and Xian, Review of Hybrid Membrane Distillation Systems (Membranes, volume 14, issue 1, article 25, 2024, open access).
Panagopoulos and Michailidis, Membrane Technologies for Sustainable Wastewater Treatment: Advances, Challenges, and Applications in Zero Liquid Discharge (ZLD) and Minimal Liquid Discharge (MLD) Systems (Membranes, volume 15, issue 2, article 64, 2025, open access).
Elsevier, Pervaporation and Hybrid Vacuum Membrane Distillation Technology and Applications, in Membrane Separation Principles and Applications (Fakhir U. Baig, pages 233 to 250, 2019 (further reading)).
Andrés-Mañas, Ruiz-Aguirre, Acién and Zaragoza, Performance increase of membrane distillation pilot scale modules operating in vacuum-enhanced air-gap configuration (Desalination, volume 475, article 114202, 2020).

What is membrane distillation (MD)?