A power plant that does not need a supply of expensive potable water could soon be a reality, thanks to a membrane system that recovers clean water from flue gas. As well as providing water, the system could cut an average plant’s annual energy bill significantly. KEMA’s F G de Vos, R Heijboer and H M van Deelen-Bremer discuss their experiments and findings.

F G de Vos, R Heijboer and H M van Deelen-Bremer, KEMA, the Netherlands

Most people who see white clouds billowing from the stack of a power plant or waste incinerator think that what they are looking at is pollution. Few people realize that these clouds contain harmless water. So, for PR reasons alone, the removal of water from flue gas is a worthwhile goal. However, the drivers for introducing water recovery are primarily economic.

In the late 1990s, it was anticipated that the cost of water for power plants would increase enormously. The decrease of potable water for industrial use, based on legislations from the Dutch government, lead to the assumption that the price of water would double.

One alternative source could be surface water, however, a lot of effort would be needed to produce water of a high enough quality and the total organic carbon (TOC) is difficult to remove1. In addition to surface water, KEMA felt other alternative sources needed to be investigated.

For the production of power a lot of water is required, in the form of steam, flue gas cleaning, etc. An average 400 MW power plant uses 30 m3/h of demineralized water for steam production. On the other hand, a lot of water escapes from the power plant through the stack. The same 400 MW power plant with a flue gas desulphurization (FGD) plant will emit 150 metric tonnes of water every hour through the stack2. If 20 per cent of this water could be recovered this plant would become self-supporting. If even more water was recovered, the power plant would become a water producer instead of a consumer.

To recover this water one could condense the flue gas, but this would require an enormous cooling capacity. Another disadvantage is that acidic flue gas compounds would dissolve in the condensed water, making it highly corrosive and therefore requiring an extra cleaning step.

This is why KEMA came up with an alternative solution – recovery of water vapour through membranes3. By using gas/gas separation membranes the water could be recovered, and with good selectivity the water cleaned in the same step. This technology, which is covered by a worldwide patent, allows one to recover water from flue gas.

Recovering water using membranes

The principle of this membrane technology is based on the breathable raincoat. This allows water vapour to be transported from the inside to the outside, while keeping the rain out. Fabrics like SympaTex and GORE-TEX are known for these properties.

The driving force in this process is the partial water vapour pressure. Close to the body the partial vapour pressure is higher than on the outside of the clothing. A drawback of these membranes, however, is that their use at flue gas temperatures is not suitable. Too many non-condensables, such as nitrogen, will pass through the membrane at elevated temperatures. To overcome this problem, the Netherlands’ University of Twente developed a polymer-based membrane material that is highly selective for water vapour at flue gas temperatures4. To obtain the difference in partial vapour pressure between the feed and permeation side of the membrane, a vacuum is applied on the permeation side. The vacuum not only creates a difference in partial vapour pressure, but also transports the water vapour from the permeation side to the condenser where condensation takes place.

Ideally, the membranes would be placed behind the FGD plant where the flue gas is saturated with water. The water recovered by the membranes is transported to the condenser where it is added to the water steam cycle as additional water to compensate for the steam/water loses. The condenser will also provide the vacuum because it normally has sufficient capacity. In Figure 1, this ideal situation is depicted.

Figure 1. Ideal method of water recovery
Click here to enlarge image

Besides the recovery of water an important energy saving can be made. Because there is less water in the flue gas, less heat is required for the flue gas to overcome condensation in the stack, or to reduce or even remove the plume.

In Practice

Before this theory could become a reality, experimentation had to prove the principle. The experiments were carried out with hollow fibre membranes, which have a high surface to volume ratio. The module was placed in a cylindrical membrane chamber. Flue gas from the stack of a coal fired power plant was then fed to this cylindrical membrane chamber and passed over the hollow fibre membranes.

Water vapour was permeated through the membranes and transported to a condenser by applying a vacuum. Any non-condensables were expelled by the vacuum system.

After carrying out the experiments, which were successful, the thermal and chemical stability of the membrane material was assessed. After 5300 hours under flue gas conditions, the main part of each hollow fibre membranes was defect-free and they complied with the ‘no measurable nitrogen flux’ specification. As nitrogen is the main compound in flue gas, this was chosen as the target compound for this assessment.

The water flux is also very important because it will determine the surface area required for the recovery of the water. In the first test, the water flux varied between 0.2 l/m2/h and 0.6 l/m2/h. This was too low because implementing such a high surface area into an existing power plant would present quite a challenge.

Water recovery experiment with curtain-shaped modules on top of the FDG
Click here to enlarge image

Firstly, one of the prerequisites was to keep the pressure drop over the membrane modules in the flue gas duct below 10 mbar, and secondly, a large surface area would be more costly, which would have a negative economical effect.

To keep the recovery process feasible a higher flux was necessary, and based on the modelling and experiments, a curtain-shaped module was built based on hollow fibre membranes with an increased internal diameter. This internal diameter was increased because the pressure drop on the inside of the fibre influences the transportation of the water vapour, and as a consequence, the overall water flux.

The module, shown in the photo above was placed in a side stream of the flue gas directly after the FGD demisters. During the experiments a flux improvement was achieved – by a factor of seven.

Aside from the flux improvement, the quality of the recovered water improved from an average of 500 µS/cm to 20 µS/cm, implying a factor of 25. If one keeps in mind that this conductivity is only caused by acidic compounds one could compare this with a fresh water quality of 6 µS/cm to 7 µS/cm. The quality of the recovered water is shown in Figure 2.

Figure 2. Quality of the recovered water
Click here to enlarge image

From Figure 2 it can be concluded that the main compound in the recovered water is sulphate originating from sulphur dioxide.

Finally, experiments were carried out in the flue gas duct of a waste incinerator. The waste incinerator differs from a coal fired power plant in the following ways: the composition of the flue gas has more halogens present; the temperature after the wet scrubber (FGD) is higher; and more importantly, the flue gas contains twice as much water as from a coal fired power plant. A higher temperature implies a higher water vapour pressure. Before this experiment began the membrane modules had been in contact with flue gas for almost two months.

During the test period of over a period of a year, the water flux decreased. It started at 4 l/m2/h. During inspection it was observed that gypsum crystals were present on the membranes. This gypsum could easily be removed with water, so a nozzle was placed in front of the module, giving a ten second burst every hour.

The cleaning in place system (CIP) made sure that the flux could be kept between 3.5 l/m2/h and 4 l/m2/h. As expected, based on the composition of the flue gas, the quality of the water is slightly less than the water behind the FGD of the coal fired power plant – 40 µS/cm versus 20 µS/cm.

Making Economic Sense

The experiments presented above show that this technique to recover water from flue gas is very promising. The principle works, the water flux is good and the quality of the recovered water is high. The question now is, will the price of one cubic metre of water recovered from the flue gas be competitive with the price of one cubic metre of water from a conventional demineralization plant? It has to be kept in mind that the price of demineralized water is partly determined by the raw water price. This raw water price differs across the world. In dry areas, prices will be higher than in water-rich countries like the Netherlands.

From experience it is known that the average price of demineralized water is around €2/m3 ($2.5) in the Netherlands. Based on calculations, with a membrane lifetime of three years and a water flux of 2 l/m2/h, the price of the recovered demineralized water will be between €1.24/m3 and €1.38/m3 – dependent on the configuration of the power plant5. These prices are based on existing power plants and in new power plants this price could be even lower.

The calculated price, however, does not include energy savings for reheating, but could be as high as €1.2m per annum for a 400 MW coal fired power plant5.

Future Plans

For subsequent experiments a pilot installation will be built that should recover 1m3/h. To increase it to a larger scale it was decided to build the membranes in a modular structure. This modular structure would not only allow for up-scaling, but it would improve the ease of replacing membrane modules. Another option is to condense as close as possible to the membranes and to keep the diameter of the piping small, therefore reducing the investment costs.

From modelling and performing practical experiments it has been shown that the theory of recovering water from flue gas by using gas/gas separation membranes could be turned into practice. Based on the calculations, the price for the production of demineralized water from flue gas is competitive. An advantage is that no chemicals are needed and TOC is not a problem because the latter is not present in the flue gas.

Based on these principles KEMA is now leading an European Union project with 25 participants, called ‘NanoGloWa’ or ‘Nanotechnology against Global Warming’, to recover carbon dioxide from flue gas.


1 Heijboer R, van Deelen-Bremer MH, Butter L & Zeijseink AGL (2006). The behaviour of organics in a makeup water plant. Power Plant Chemistry, 8 (4), pp.197-202

2 Heijboer R, van Deelen-Bremer MH & Muller EF (2004). Eerste jaarrapportage project EETK01021 ‘Water- en energiewinning uit rookgassen’ (First annual report project EETK01021 ‘Water and energy recovery from flue gases’). KEMA, report number 50180318-TOS/MEC 04-7036

3 Zeijseink AGL, Beerlage M, Heijboer R and van Deelen-Bremer HM (2002). A Novel Process for Water Recovery from Flue Gases. 14th International Conference on the Properties of Water and Steam (ICPWS), Kyoto, Japan.

4 Sijbesma H, Nymeijer K, van Marwijk R, Heijboer R & Wessling M (2008). Flue gas dehydration using polymer membranes. Journal of Membrane Science, (313), pp.263-276

5 Heijboer R, van Deelen-Bremer MH & de Vos FG (2007). Derde jaarrapportage project EETK01021 ‘Water – en energiewinning uit rookgassen’ (Third annual report project EETK01021 ‘Water and energy recovery from flue gases’). KEMA, report number 50180318-TOS/MEC 07-9002.