In the last decade, the growth in membrane applications for municipal drinking water treatment has been explosive – in the UK, around 100 plants of all sizes have been built in the last ten years, with overall capacity close to 1,000Ml/d. The largest plant at Clay Lane in north London, with encased membranes, treats a flow of more than 150Ml/d, and an immersed microfiltration (MF) with a capacity close to 100Ml/d was started up last year in Farlington, Portsmouth.

In the US, there were only three MF plants in municipal applications in 1994. Ten years later, there are more than 100 plants larger than 4Ml/d, with a combined capacity of more than 2,000Ml/d. The largest plant about to be started up in Minneapolis is very similar to Clay Lane, using encased MF membranes for a flow close to 300Ml/d.

In Asia, about 350MF and ultrafiltration (UF) membrane plants are operating in Japan, treating a combined flow around 250Ml/d. The largest membrane unit operating in the world to date is the immersed membrane filtration system at the Chestnut Avenue water works (CAWW) in Singapore, where a 273Ml/d enhanced coagulation/ultrafiltration system to produce potable water was commissioned in November 2003.

The commercial success of membrane application was mostly driven by the need to positively control the removal of particles in a size smaller than parasite cysts, such as Giardia and Cryptosporidium, which are between three and 12µm in size. In addition, membranes need less treatment chemicals, leading to cost savings in dosage, control and residuals. Nevertheless, the operation of many membrane plants has been a trial and error experience, as site specific techniques to control membrane fouling and breakage had to be developed.

What was originally imagined as a maintenance-free, fully automatic operation, leading to water treatment without chemicals became sometimes a sophisticated operator-dependent technology, relying on creativity with chemical cocktails to keep the membranes clean. In the early days, ceramic membranes appeared attractive because of their mechanical stability, but organic membranes soon dominated the field as the technological developments and larger applications reduced their production cost by a factor ten.

One of the main cost factors is the membrane flux, which is heavily dependent on the pre-treatment and membrane type. Typically, with only minimal pre-treatment, such as 400µm screening and application of a coagulant, the membrane flux tends to be between 50-100 l/m2/h, whereas with a more complete pre-treatment, including a conventional coagulation and settling, the flux can be twice as high. Today, the cost of membrane plants is remains very site-specific and dependent on plant size, as the amount can double for smaller installations.

membrane performance

Recent examples in Minneapolis, Minnesota and Bakersfield in California, show turn-key cost close to £130/m3/d. Large UK plants had somewhat lower costs, with Clay Lane and Farlington lower than £65/m3/d. While not directly comparable regarding applications and treated water quality, conventional filtration for large plants at overflow rates of 10m3/m2/h would have a turnkey cost around £20/m3d. In Singapore, the membrane fouling was reduced significantly after the addition of 20mg/l of alum before the membranes, with a residence time in the flocculation basins of 5min. Stable membrane performance was achieved at a flux of 80l/m2/h and 95% recovery.

Figure 1 shows the membrane tanks of the Chestnut plant, with all civil work provided to enable the capacity to be increased to 480,000m3/d in the future. The layout consisted of a central process building with 14 membrane tanks on each side, but membranes and process equipment were only provided for eight trains per bank. The total land area required for including the feed channels, outdoor membrane tanks and the process equipment building is 53m x 47m or approximately 2,500m2, equating to approximately 190m3/d of capacity per m2 of plant surface area. With conventional gravity filtration, using filtration rate of 10m3/m2/h and counting an area of 50% of the filter surface for auxiliary structures, the land required would be comparable at 160m3/d/m2.

The fact a small coagulant dose with a short flocculation time increases membrane performance for both quality and flux has led to some recent innovations. On ceramic membranes, a pure water flux (at temperature of 25°C and pressure of 100kPa) of 1800 l/m2/h can be demonstrated, which is more than three times the standard flux of the most common polymeric membranes used for MF and UF, made of Polyethylene (PE), Polypropylene (PP) or Polyvinylidene Difluoride (PVDF).

Recent research examined the influence of coagulation time, dosage of Poly-Aluminium-Chloride (PAC) and pore size on ceramic membrane efficiency for virus removal, using bacteriophages as model virus at a count of 3-7 x 10 6pfu/ml. Coagulation time could compensate for very low dosages – with a short in-line mixing of 2.4sec, at a PAC dosage of 0.5mg Al/l, the average virus removal was 2.8log but the log removal could be more than doubled to 6.1 log with the conventional approach of 30min long mixing in coagulation tanks.

Conventional coagulation was not necessary for dosages of 1mg Al/l and higher – an in-line contact time of only 2.4sec achieved similar virus removals at 6.4log than a 30min long mixing. Another 0.5mg Al/l gained another log removal, independent of coagulation time. When coagulation is used on membranes, MF pore size does not have an important influence – with a coagulation time of 2.4sec and a dose of 1mg Al/l, both pore size of 0.5 and 1 micrometer achieved virus removal above 5log. Reducing the pore size to 0.1, the standard MF objective, only increased virus removal by 1 log.

Thus much higher membrane fluxes can be achieved by compensating the larger pore size with a small amount of coagulant and short in-line mixing. The combination of coagulation and ceramic membrane filtration can also remove organic matters with molecular weight >3,500, whereas typically polymeric membrane filtration (UF) can not remove matters of molecular weight <7,000. Since 1996, ceramic elements have gained a 10% share of the Japanese membrane market because of their mechanical resistance and chemical stability, which excludes the elution of any organic matter.


Membrane life is assumed in excess of 15 years and the material is easily recycled. After having installed close to 20 plants, with a total capacity of 7,500m3/d, a new membrane element design, much larger than the previous small tube of 1m length, with diameter of 30mm, has made the economics of ceramic membranes much more attractive.

The new ceramic membrane elements have a diameter of 180mm, with a length of 1,500mm, each element containing 2,000 channels of 2.5mm diameter. This represents a total membrane surface of 25m2 per element. As shown in Figure 2, the membrane elements are placed in a vertical position in the large-scale configuration, making the footprint comparable to other MF membrane configurations, with a specific treatment capacity per skid surface area around 260m3/m2/d. On full-scale demonstrations, with a pretreated surface water and a dosage of PAC of 1mg/l as Al, design fluxes between 150 l/m2/h to 200 l/m2/h are obtained. A 6h backwash interval translates into a water recovery above 99.0%. On groundwater, fluxes of up to 300 l/m2/h were observed and with backwash intervals spaced twice as long, just using a chloride dosage between 0.5-1mg Cl/l.

Backwash is performed by pressurised backflushes of air and water for a few seconds, yielding a sludge with solids content above 10% that can be dewatered to more than 50% solids. Energy consumption on ceramic membrane at 0.1kwh/m3 is claimed to be half that of conventional organic membranes.
Recent tests in Germany confirmed it is possible to treat river water directly on the ceramic membrane with coagulation, achieving up to 4log particle removal, with residual turbidity of 0.002NTU, as well as organics removal between 20-30% TOC. In the preliminary tests, a backwash regime to gain stable operation at a flux of 80 l/m2/h was found, and currently the membranes are operated at a flux of 160 l/m2/h.

After full-scale experience with the larger ceramic membranes in 15 smaller plants and two installations in the 3,500m3/d range in Japan, a new plant more than ten times larger for 39,800m3/d will be started in 2007 in Fukui. This
plant on the Japan’s west coast just accross Tokyo has an ultimate future capacity of 52,000m3/d.

Action inspires action. Stay ahead of the curve with sustainability and energy newsletters from edie