Black & Veatch's Frank Rogalla looks at multiple treatment goals with membranesTen years ago membrane filtration was relatively new and confined to a few small WTWs. When it became operational in 1995 the 12Ml/d membrane filtration plant in Fort Lupton, Colorado, was considered large. Today facilities like the 160Ml/d Clay Lane works are more than ten-times larger, and worldwide there are more than 100 municipal membrane plants of significant size. The engineering challenge is integrating membranes with other processes to achieve multiple treatment goals. One example is in Minneapolis. Granular media filters are being replaced with membranes downstream of lime softening and coagulation-sedimentation processes to treat 600Ml/d of water and yield a combination of softening, control of dissolved organic materials, filtration and disinfection. MF/UF membranes are formed into hollow fibres, slightly larger than a human hair, in which submicron-sized pores filter impurities. Membrane filters are made of plastics such as polyvinylidenefluoride (PVDF), often with proprietary chemistry.
The technology makes it easier for a WTW operator to consistently ensure removal of small particles and pathogens. "Membranes provide an excellent barrier to microbial contaminants. Generally, physical removal is better than chemical inactivation as it avoids possible future concerns over newly discovered byproducts," says Adam Kramer, director of the Minneapolis works. Putting up the barriers
In the 1990s, concern about certain pathogens, including Crypto. oocysts, in source water increased. Because these pathogens are resistant to traditional disinfectants, such as chlorine, it is advantageous to remove them ahead of chemical disinfection. The effectiveness of removal through conventional granular media filters depends on proper operation of the coagulation-flocculation process upstream. Coagulant chemicals, such as aluminum sulfate or ferric chloride, join small particles like oocysts into masses large enough to be settled or adsorbed onto the media, preventing their passage into the filtered water.
If coagulation, flocculation, sedimentation and granular media filtration processes are correctly combined and applied, protozoan removal credits of 2-2.5 log (which means 99-99.68 % removal) can be achieved. But to attain this level of performance with granular media filters, the coagulant dose must be continuously adjusted to obtain a filtered water turbidity close to 0.1 nephelometric turbidity units (NTU). The operating mechanism of a membrane is completely different to that of granular filter media. The pores in the membrane fibre wall are much smaller than the protozoan cysts, which have nominal sizes of 2-7µm for Crypto., easily removing them by size exclusion, independently of pre-treatment coagulation conditions. The automatic result is a very high level of removal of small particles and finished water turbidities that can reach 0.01NTU. Based on the results of an extensive microbial challenge test programme, the California Department of Health Services has approved MF/UF systems with 4-log (99.99%) removal credit for Crypto. Significantly higher values, up to 8-log, are routinely observed during tests.
Membranes offer a means of verifying the level of pathogen removal at full-scale facilities. WTWs using granular media filters monitor data from turbidimeters and particle counters to document the removal of cyst-size material. Membrane plants apply these same methods as well as automated integrity verification to check for leaks. This is similar to filling the inner tube of a bicycle tire with air while submerged in a bucket of water, helping to prove there are no significant defects or locate any leaks that may occur.
Utilities report fewer than one broken fibre a month per 40Mld of capacity, well under any significant level of concern. Even with a few breaks, large facilities maintain a high log removal. During the filtration cycle, material collects on the dirty side of the membrane surface. This filter cake is removed by a periodic automated backwash, similar to the operation of a granular media filter. To improve removal effectiveness, some systems apply an air scour during the backwash, and/or a chemically-enhanced backwash with a low concentration of citric acid or hypochlorite solution. After weeks or months of operation, the pressure drop across the membrane may increase because of permanent deposits. These are removed by chemically cleaning the membranes while in place (CIP) with acid, caustic sodium hydroxide and sometimes hypochlorite solutions.
In early applications of MF/UF for surface water treatment, in the mid-1990s, membranes were used primarily in a simple filtration mode. Without any additional process steps, the small pores in the membrane wall can produce well-filtered finished water, if removal of particulate material is the only goal. UK MF plants such as Homesford and North Mymms, which were set up as a result of Crypto. regulations, are examples of such sites.
However, many situations require removal of dissolved contaminants such as naturally-occurring organic material that reacts with chlorine to form disinfection byproducts (DBPs), trihalomethanes (THMs) and haloacetic acids (HAAs), or may cause taste and odour problems. Other processes can be integrated with membranes to achieve multiple treatment goals, as illustrated in the following examples. Many treatment plants use lime softening to produce softer, aesthetically pleasing water. Minneapolis is applying a new approach, integrating lime softening with membrane filtration. The city was confronted with the need to upgrade aging filtration equipment in treatment plants on the upper Mississippi River.
The replacement of granular media filters with membranes is under design at the 280Ml/d Columbia Heights works which, upon completion in 2004, will be the world's largest lime softening-membrane facility and the largest UF plant in North America. A similar design at the 360Ml/d Fridley plant is scheduled to follow two years later.
Pilot trials for two membrane systems were conducted onsite to verify operating conditions and microbial removal. After taking into account the initial capital cost as well as the present worth of major operating expenses, Norit's X-flow system was selected, which was also used at Clay Lane and North Mymms. Unlike the purchase prices for most types of equipment, membrane costs decrease each year. Minneapolis proved no exception, with an investment cost for the membrane system of £10M yielding a unit cost of less than £0.036 per Ml/d.
Many traditional water treatment plants supply coagulated-flocculated water directly to granular media filters. In Arizona, Scottsdale's new 120Ml/d Chaparral WTW will apply direct filtration to membranes. Raw water from the Arizona Canal will be dosed with approximately 15mg/l of ferric sulfate coagulant, which adsorbs dissolved contaminants into particles and allows membranes to filter them from the water.
A unique aspect of this plant is the use of coagulation-flocculation-MF/UF to remove dissolved arsenic, a pollutant of growing concern for many water suppliers. Scottsdale chose membranes for a variety of reasons, including space limitations. Four types of membranes were compared in on-site pilot studies and the full-scale equipment will be selected in an evaluated competitive bidding process. A post-membrane granular activated carbon (GAC) step will remove traces of taste and odour-causing compounds and DBP precursors - the plant will serve as an example of combining direct filtration and post-treatment methods.
In some cases, a sedimentation step between coagulation-flocculation and filtration improves the overall cost-effectiveness of treatment. In the Kern River, the source of raw water for the new 80Ml/d plant for Bakersfield in California, turbidity can increase from 15 to more than 1,000NTU during spring rains. Sedimentation will dampen these peaks and augment removal of dissolved colour and DBP precursors.
Before proceeding with detailed design, membrane equipment was pre-selected through a competitive bidding process based on life-cycle cost analysis. The cost of the entire project, including membranes, is £20M, with start-up planned in mid-2003. A similar 140Ml/d facility is being designed close by for the South San Joaquin Irrigation District.
In the past, each MF/UF membrane filtration train has been served by an individual pump. To simplify operation and beneficially use the hydraulic profile available at most WTWs, an innovative natural-siphon approach was developed for the 360Ml/d Choa Chu Kang works in Singapore, which will be the largest membrane water plant in the world. As part of a major plant upgrade, submerged membranes will be installed in the existing filter boxes to improve the already high quality of the finished water. After a siphon has been established, an 8m elevation change between the filter boxes and the filtrate storage tank will draw water through the membrane. Onsite pilot tests verified the operating conditions and the membrane pre-treatment of alum-based coagulation-flocculation-sedimentation.
The use of membrane filters for treatment of surface water has expanded rapidly, in terms of number of installations as well as geographic range. MF/UF membrane plants are now operating in more than half of the US states and in ten other countries. MF/UF yields high-log removals of Crypto. oocysts and produces low filtered water turbidity, independent of upsets in pre-treatment coagulation. The physical barrier provided by membranes, coupled with integrity verification, makes it relatively easy to produce well-filtered and safe potable water. New approaches are being developed to integrate membranes with other treatment steps to achieve multiple treatment goals including softening, removal of arsenic and elimination of compounds that can cause taste and odour problems or disinfection byproducts