A Distinguished History and a Promising Future

(March 2006, as appeared in Journal AWWA)

As dedicated providers of safe, reliable, high-quality water, the water supply community has embraced new technologies and procedures as they have been developed over the past 200 years. In this overview, advances in the filtration of public drinking water supplies from the late 1800s to today are chronicled and thoughts on the filtration's future are expressed. This review reflects in part information published in The Quest for Pure Water, volumes I (Baker, 1948) and II (Taras, 1981).

Granular media's evolution over the years Slow sand filtration was the first. Successful filtration of public drinking water in the United States began in 1872 with the implementation of the English and European practice of slow sand filtration (SSF) at Poughkeepsie, N.Y. Other communities in New York and New England, including Ilion and Hudson (N.Y.), St. Johnsbury (Vt.), and Lawrence (Mass.), adopted SSF before 1900. SSF removed microorganisms effectively, but a typical filtration rate was 3 million gallons per acre per day (mgad) or 0.12 m/h. Thus, the early filtration works used by large cities required a considerable land area as well as a substantial labor force to scrape a thin layer of sand off the top of the filter bed once terminal head loss was attained. Slow sand filters proved ineffective for filtration of turbid surface waters that constituted the water supply in many parts of the country; with the development of effective clarification and rapid sand filtration, the filters gradually fell out of favor and were seldom installed in the early and mid-20th century.

Renewed interest in SSF was sparked by outbreaks of waterborne giardiasis in communities that had been using supposedly pristine, low-turbidity waters with disinfection as the only treatment. Because such waters seemed to be good candidates for SSF and the outbreaks were happening in small- to medium-sized communities, the US Environmental Protection Agency (USEPA) initiated research into SSF in the early 1980s. Subsequently SSF was implemented in numerous communities in the northeastern states, the Rocky Mountain region, and the Pacific Northwest. The perceived drawbacks ? the need for a large land area and an equally large labor force for filter cleaning ? proved to be no problem for small rural communities with readily available vacant land and labor for filter operation and maintenance.

Rapid sand filtration has early roots. Around the time AWWA was formed in 1881, the first attempts were made to coagulate and filter US surface waters. The process resembled what today might be called inline filtration, i.e., addition of coagulant chemicals ahead of a sand filter with little time allowed for coagulation and no flocculation or sedimentation. Such attempts failed to treat muddy surface waters, such as the Mississippi River near New Orleans, La. In the 1890s, the Louisville (Ky.) Water Company sponsored trials with filtration preceded by coagulation and sedimentation. George Fuller reported successful treatment of Ohio River water by coagulation, clarification, and filtration at a rate of 125 mgad (2 gpm/sq ft or 5 m/h). This filtration rate became the norm in the United States for many years beyond Fuller's time, and the treatment train of chemical coagulation, mixing, flocculation, and sedimentation followed by filtration became known as conventional treatment.

About the same time as Fuller's work at Louisville, Alan Hazen was evaluating filtration at Pittsburgh, Pa. Although turbidity measurement was still crude, bacteriology had developed to the stage at which measurements of plate count bacteria could be made. Hazen presented data showing a 97-99% reduction of the density of bacteria by coagulation and rapid sand filtration (Hazen, 1913).

Rapid rate filtration comes into use. With the widespread adoption of conventional treatment, water utilities and their engineers began to observe a need for process improvements. Over time, many of these have been implemented.

In the 1920s, the use of auxiliary scour was evaluated by John Baylis, who found that water sprayed laterally from a fixed grid of pipes enhanced the cleaning action during backwashing. In the late 1930s, rotary sweeps were developed for surface washing. The European practice of auxiliary scour tended toward the use of air scour, which in the 1980s also began to be used by US filtration plant designers. This change likely was influenced by wastewater filtration research conducted in the 1970s by John Cleasby and E.R. Baumann (1977) at Iowa State University in Ames and sponsored by the USEPA. This research showed that the combination of air scour and water wash at a low rise rate followed by water wash alone could clean filter media more effectively than air alone followed by water wash alone or by surface wash and backwashing. The homegrown research paid off; regulators and design engineers both tend to be more willing to accept new findings when they are demonstrated closer to home rather than on other continents.

Treatment pioneers focused on maximizing the medium. One of the chief limitations of rapid sand filtration as developed by Fuller was that typical sand filters with a depth of 24-30 in. (61-76 cm) and an effective size (es) of about 0.5 mm often could not be operated successfully at rates much higher than 2 gpm/sq ft (5 m/h). When filter materials of the same specific gravity are backwashed and then restratified, the finer grains settle at or near the top of the filter bed and the larger grains deeper in the bed. This results in floc being removed near the top of the filters, making little use of the storage capacity deeper in the bed and decreasing filter run length. To counteract this tendency, engineers developed dual-media filters consisting of a layer of larger anthracite over a layer of finer sand. Some of the first large-scale dual-media installations were at Hanford, Wash., and in Oak Ridge, Tenn., as a part of the Manhattan Project during World War II. This work was carried out first by Roberts Filter Company and Walter Conley of General Electric and later by Neptune Microfloc. The dual-media filters functioned well at 4 and 5 gpm/sq ft (10 and 12 m/h) and at higher rates on an intermittent basis.

As a follow-up to dual-media filters, Conley developed mixed media consisting of a layer of fine garnet overlaid by sand and anthracite. The fine garnet was able to provide added protection against turbidity breakthrough but carried with it the burden of additional head loss.

In the 1980s, the need for filters that could operate at higher rates than dual-media or mixed-media beds led to investigations into deep-bed monomedium filters. Coarser filtering materials induce lower head loss at higher filtration rates than the more traditional 0.5-mm-es sand and 1.0-mm-es anthracite, but their filtering efficacy for a given bed depth is not as good. The latter aspect is compensated for by deeper beds. The Los Angeles (Calif.) Aqueduct Filtration Plant, which was brought on line in the late 1980s by the Department of Water and Power, uses direct filtration (rapid mixing, flocculation, and filtration) and was designed for rates as high as 13 gpm/sq ft (32 m/h) on the basis of pilot-test results. The filter medium consists of 6 ft (1.8 m) of 1.5-mm-es anthracite. Since then other filtration plants have been built using deep-bed monomedium filters.

Because deeper filter beds with coarser filtering materials achieve effective filtration at higher rates than do sand filters and some dual-media filters, some existing filter plants are being retrofitted with deeper filters. In some existing plants, deeper filter beds can be installed by using new underdrain configurations that incorporate on the top of a filter tile a layer of porous material designed to prevent penetration by filter media.

Some water utilities faced with unpleasant tastes and odors caused by organic chemicals in their source water replaced the anthracite layer in their dual-media filters with granular activated carbon (GAC) or substituted GAC for some or all of the sand in their rapid sand filters, In the 1960s, utilities in Virginia and West Virginia were among the first to use GAC filter-adsorbers. When GAC filter-adsorbers are used for removing certain taste and odor compounds, replace­ment of GAC is typically performed at an interval much longer than that needed for removal of a broader range of organic compounds.

The Cincinnati (Ohio) Water Works installed GAC postfiltration adsorbers at its Ohio River plant in 1992 as a means of protecting against potential organic chemical contamination in the industrialized Ohio River watershed. This facility is intended to maximize the adsorption function of CAC for removing organics. It includes onsite regeneration, has a design capacity of 175 mgd (660 ML/d) at a 15-minute empty bed contact time (EBCT), and is the largest of its type in the United States.

New filters incorporate biomass. The term biologically active filtration (BAF) is applied to filtration through granular media filters having an active biomass on the media grains. Biologically active GAC filters are sometimes referred to as biological activated carbon. The use of BAF was prevalent in Europe in the 1970s and 1980s, with particle removal and biological removal generally accomplished in separate filtration steps. In the United States, BAF and particle removal usually occur in the same filter. Deliberate use of BAF commonly coincides with the use of ozonation, because ozone can increase the assimilable organic carbon (AOC), a measurement of the biodegradable fraction of dissolved organic carbon (DOC). Any granular media filter that does not have a disinfectant residual in the influent probably has some form of biological activity in the filter, so BAF may be occurring (perhaps unintentionally) in plants that have moved their primary chlorination point past the filters. Advantages of BAF include removal of DOC, AOC, disinfection by-product (DBP) precursor compounds, and taste and odor compounds; improved biostability of finished water and reduced potential for regrowth in the distribution system; and lower chlorine demand of the finished water. Disadvantages of BAF include increased heterotrophic plate counts in the effluent, the possibility of increased head loss, and the possible requirement for nonchlorinated backwash water.

The role that traditional granular media filters can play in the control of manganese is often unrecognized. In the latter 1980s, researchers at the Virginia Polytechnic Institute and State University in Blacksburg showed how the development of a coating of manganese dioxide on filter media in the presence of free chlorine provides a reactive surface that can remove soluble manganese that could otherwise pass through the plant into finished water. This manganese-removal mechanism has helped many plants avoid customer complaints, but at some plants the process may not be recognized until oxidation practice changes or filter media are replaced. If manganese removal has been occurring at a granular media filtration plant that is modified to use BAF or GAC filter-adsorbers or if granular media filtration is replaced with microfiltration (MF) or ultrafiltration (UF), another approach will be needed for manganese control. In this way, the presence of a contaminant not included in the National Primary Drinking Water Regulations may influence choices of processes used for controlling regulated contaminants.

Technological advances allow continuous monitoring of filtered water quality. Monitoring filter performance requires knowledge of the filtration rate and loss of head in a granular media filter. Over the decades, these measuring techniques have been refined to the point that rate of flow and head loss data are now detected by sensors and transmitted electronically to the operator. Measurement of filtered water quality has evolved in somewhat divergent directions. Early measurement of filtered water turbidity, performed using a Jackson candle turbidimeter, was imprecise for very clear filtered water. As late as 1971, the third edition of AWWA's Water Quality & Treatment explained that rapid measurement of turbidity in a filtration plant could be performed using an illuminated sightwell with a submerged light or a continuous-flow light-scattering nephelometer (AWWA, 1971). Modern turbidimeters date from about the 1960s when Hach Chemical Company introduced nephelometers for laboratory use and flow-through instruments that could produce a continuous readout of water turbidity. Continuous measurement of filtered water turbidity has been recommended in the water industry for at least two decades and is now a regulatory requirement.

Nephelometers of the type used for continuous turbidity monitoring have tended to lose precision at turbidity values of 0.1 ntu or lower, making it difficult for operators to assess the effect of subtle changes in pretreatment or filter operation when filtered water turbidity is near 0.1 ntu. This shortcoming has been remedied with the recent development of a laser nephelometer that appears to be capable of ultralow-level turbidity measurement.

Some utilities have found that online particle counters are effective in detecting changes in filtered water quality when dealing with filtrate turbidities of 0.1 ntu and lower. In 1981, the goal of treating 1-ntu Lake Mead water to produce a filtered water with very low turbidity led to the use of online particle counters at the Southern Nevada Water Authority's Alfred Merritt Smith Water Treatment Facility near Boulder City. Although particle counting has become more widely used in recent years, a drawback associated with the use of this method is the lack of a standardized instrument design. This issue remains unresolved.

In the early 1970s, monitoring of the turbidity of water within a filter bed was advocated by a filter media supplier (The Taulman Company). The concept involves continuously extracting an appropriate flow of water from within the filter bed and continuously measuring its turbidity. The original concept was applied to dual-media filters, with the sampling device (a screened orifice) located at the interface of the anthracite and the sand. A comparison of the turbidity of filter influent, the interface sample, and filter effluent provides information on where within the filter bed the particulate matter is being removed and can warn of impending turbidity breakthrough, which aids in scheduling filter backwashing. The concept of monitoring interface turbidity has been carried further at the Modesto (Calif.) Irrigation District's treatment plant with the installation of within-bed sampling devices at 1-ft (0.3-m) intervals in one of the 6-ft- (1.8-m-) deep monomedium anthracite filter beds. The monitoring data help operators interpret and understand the behavior of deep monomedium filter beds.

Pretreatment comprises coagulation, mixing, and clarification Coagulation increases efficacy. Although slow sand filters could be used without pretreatment on low-turbidity source waters, engineers quickly realized that rapid sand filters required the use of coagulant chemicals to be effective. Even before rapid-rate filtration came into use, coagulants had been used to aid sedimentation. The most commonly used coagulant was alum, but ferric coagulants were used at some utilities.

Proper application of coagulant chemicals proved essential for attaining best performance from plants using coagulation and filtration. Both ineffective coagulation and failure to add coagulant in the operation of rapid-rate filters led to giardiasis outbreaks in the 1970s and 1980s. In fact, a survey of filtration facilities across the United States by Cleasby and colleagues in the late 1980s (Cleasby et al, 1989) indicated that for meeting filtration goals for particle removal, the coagulation process was more important than the physical attributes of plants.

After World War II, floc strength became an issue with filters designed to operate at 4-5 gpm/sq ft (10-12 m/h) and higher. The use of synthetic organic polyelectrolytes as coagulant aids was introduced in the late 1950s. Some of the polyelectrolytes were used as filter aids to strengthen the floc so it would resist shear forces in the filter bed. Cationic polymers came into use as coagulants or as coagulant aids, replacing a portion of the inorganic coagulant chemical. Today many filtration plants depend on the use of polymers to attain their goal of very low filtered water turbidity. The search for effective coagulant chemicals has led to the development of new polymerized aluminum coagulants, such as polyaluminum chloride, as well as blends of metal salts and organic polymers.

Mixing and clarification aid coagulant performance. Early designers of conventional filtration plants did not place as much emphasis on effective mixing as is found in current practice. At a pre-World War I Pennsylvania plant, coagulant chemical was allowed to simply flow onto upwelling raw water in a large mixing tank, without mechanical mixing. This plant also employed baffled flocculation. Volume II of The Quest for Pure Water indicates that by the 1940s, plants designed for turbidity removal typically provided 30-45 min of flocculation and 4 h of sedimentation (Taras, 1981).

By the 1950s, the importance of rapidly mixing coagulant chemicals was recognized sufficiently to be included in water treatment textbooks. Recommended Standards for Water Works stipulated a detention time for rapid mixing not to exceed 30 seconds (Great Lakes-Upper Mississippi Board of State & Provincial Public Health & Environmental Managers,1982). More recently, the rapid dispersion of coagulant chemicals is attained with static inline mixers, motorized inline mixers, and pumped-jet mixers, with the last dispersing coagulant chemical at high velocity upstream into the raw water flowing in a pipe or conduit. Emphasis on dispersion as opposed to rapid-mix contact time recently has been recognized as the most essential consideration for effective use of coagulant chemicals.

The development of improved and different types of flocculation equipment, coupled with a better understanding of the process, has led to improvements in flocculation. With the advent of high-rate sedimentation processes, the importance of the uniformity of floc size became apparent: floc with a more uniform settling velocity is more effectively removed by equipment such as plate settlers. Thus the concept of well-baffled, multi-chambered flocculation basins that more closely approximate plug flow in flocculation grew in acceptance. Today flocculation times vary considerably, depending on the process used. For direct filtration and dissolved-air flotation (DAF) clarifiers, the goal is to provide a smaller floc size that can readily accommodate in-depth filtration or readily float. In both cases, floc of excess size is a liability. Flocculation times can be as short as 10 minutes or less for these processes, and the use of floc detention times that are 30-45 minutes long may be counterproductive.

Flocculation equipment has evolved from the paddle-wheel flocculators used in the early to mid-20th century to vertical turbine flocculators mounted above the water surface (with only the shaft and impeller submerged) to today's hydrofoil blade (somewhat resembling an aircraft propeller) that provides gentle flocculation without the risk of floc breakage. For paddlewheel flocculators, manufacturers now produce bearings of noncorrosive materials such as nylon and drive chains from fiberglass reinforced plastic. Clarification improvements made over the decades since the 1960s generally have been aimed at operating the process at ever-higher rates.

Advances shrink sedimentation basins, reduce detention times. Conventional sedi­mentation basins with overflow rates of approximately 0.5 gpm/sq ft (1.2 m/h) and detention times of around 4 hours were commonly used in large treatment facilities in the 1940s. However, these occupied a considerable footprint, with commensurate capital cost, so processes to hasten sedimentation were sought throughout much of the 20th century.

An early approach to eliminating large sedimentation basins and long detention times was to use solids-contact clarifiers. This process equipment combines a central mixing zone and an outer sedimentation zone in one basin, separated by a baffle wall. Massive recirculation of water within the mixing zone, mixed with incoming water and floc drawn from the bottom of the basin, promotes chemical reactions and floc formation. The units can achieve high solids in the mixing zone and may be operated with a sludge blanket in the settling zone. These systems have been operated at overflow rates of ~1.8 gpm/sq ft (4.4m/h) for softening and ~1.0 gpm/sq ft (2.4m/h) for chemical coagulation applications. Pulsed sludge-blanket clarifiers were also developed, with later designs using plate settlers to improve performance; these units have been operated with loading rates of 2.5 gpm/sq ft (6.1 m/h) and higher.

To shorten the detention times for sedimentation, Neptune Microfloc developed tube settlers in the 1960s. In this concept, a bundle of inclined tubes was placed in a settling basin, positioned so that flocculated water passed upward through the tubes. The vertical distance from the "ceiling" to the "floor" of the inclined tube was measured in inches instead of feet. A bundle of tubes provided many "false floors" onto which floc particles could settle, agglomerate, and slough down to the bottom end of the tube to be discharged and settle to the bottom of the settling basin. Tube settlers were initially employed in small preengineered "package plants" with the objective of maximizing treatment capacity in a small volume. They also have been used to uprate conventional sedimentation basins, with a typical gross overflow rate of 2 gpm/sq ft (5 m/h).

Somewhat analogous to the inclined tube settlers is the European concept of plate settlers, which operate with a gross overflow rate of ~3 gpm/sq ft (7 m/h). Treatment is enhanced by the formation of uniform-sized floc. Plate settlers have been used in large facilities and some package plants.

Contact adsorption clarifiers (CAC), or roughing filters, have been used in some package plants, with multimedia filters placed in series after the CAC. The system consists of a bed of coarse media through which coagulated water passes, making many twists and turns, which aids in flocculation and removal of floc, preparing the water for filtration in a dual- or mixed-media bed. Large CACs can serve as stand-alone pretreatment facilities ahead of conventionally constructed filters. Overflow (or filtration) rates for CACs and roughing filters can be as high as 10 gpm/sq ft (24 m/h), with application generally limited to low-turbidity sources with low coagulant demands. CAC systems are primarily an American technology.

A clarification technology adopted from Europe is the ballasted flocculation clarifier, which involves coagulation of raw water and the addition of very fine sand followed by polymer to cause the coagulated particles to stick to the sand. After a period of stirring to build up the floc, a short detention time is provided to attain effective sedimentation of the relatively dense floc. This process can achieve overflow rates of 16 gpm/sq ft (39 m/h) or higher.

Treatment approaches differ in other countries. For treatment of colored or low-turbidity waters and those containing algae, European practice has differed from the American practice of coagulation, flocculation, and sedimentation. The Scandinavian countries and the United Kingdom use DAF for treatment of such waters. Its effectiveness with respect to turbidity removal varies somewhat depending on the nature of particles; the presence of heavier particles tends to reduce efficacy of DAF.

In this process, a portion of clarified water is subjected to high pressure in an aerator so that supersaturated water can be returned to the DAF clarifier at the point of introduction of the coagulated and flocculated water. DAF clarifiers can be operated at overflow rates of 4 gpm/sq ft (10 m/h) and higher. More recently, a high-rate approach has been developed that applies a plenum for greater control over the short-circuiting associated with outflow conditions. This approach may allow higher loading rates. Another approach is a stacked DAF concept in which DAF and filtration are combined in a single unit. Although the United States has lagged behind European countries in the use of DAF clarification, a number of US plants using DAF are operating now. The largest to date is the 75-mgd (280-ML/d) plant in Greenville, S.C.

Utilities now choose from an array of filters and membranes Diatomaceous earth filters, cartridge filters, and bag filters are developed. Diatomaceous earth (DE) filters were developed during World War II to treat water for American troops on the Pacific front. After the war, DE filtration was studied by university researchers, including E.R. Baumann, and several manufacturers began to produce equipment for municipal use. DE filtration is simpler to operate than coagulation and filtration because typically it does not involve the use of coagulants and the only product fed into the raw water to accomplish filtration is DE. A limitation of DE filtration is that without pretreatment, dissolved constituents pass through the filter even though it is effective for the removal of particulate contaminants, including protozoan cysts and oocysts.

Bag filters and cartridge filters, which have been used in some small systems, remove pathogens and other particles by a straining mechanism but allow dissolved constituents such as color to pass through them. Bag filters and cartridge filters are listed as treatment techniques for the removal of protozoa in the recently promulgated Long Term 2 Enhanced Surface Water Treatment Rule (USEPA, 2006).

Membrane filtration first used for desalination. The initial use of membranes for municipal water treatment was for desalination of brackish waters or seawaters by high-pressure reverse osmosis (RO) membranes. The first American desalination facility, placed in operation in 1973 at Rotunda West, Fla., to treat 8,000-mg/L brackish well water, had a capacity of 0.5 mgd (1.9 ML/d) (DuPont de Nemours & Co., 1982). During the 1980s, the number of RO facilities increased along with process improvements, including the development of nanofiltration (NF), a subclass of RO. By 1990, the larger facilities in Florida had capacities in the range of 12-14 mgd (45-53 ML/d). RO/NF membranes are effective in removing multiple dissolved contaminants in a single step, which enables utilities to use water sources previously considered unsuitable. In addition to desalination, RO/NF controls organics such as color, pesticides, and DBP precursors and inorganics such as nitrate, fluoride, arsenic, hardness, and others. In the United States, the largest RO plant for treatment of seawater is the Tampa Bay (Fla.) Water facility, which is currently under renovation to increase its capacity to 25 mgd (95 ML/d).

Low-pressure membrane filtration using MF or UF radically changed the drinking water industry in the 1990s. With the success of the first major municipal facility, the San Jose Water Company's 15-mgd (57-ML/d) plant in Saratoga, Calif., in 1994, and with the acceptance of regulators across the country, utilities increasingly are applying MF/UF. Today, many new or refurbished surface water filtration plants use MF/UF. In 1996, 12 MF/UF drinking water facilities across North America had capacities of 1 mgd (3.8 ML/d) or larger. The number of plants increased to more than 60 by the year 2000 and exceeded more than 100 by 2003. Today more than 200 municipal facilities are in operation worldwide.

A major reason for the rapid increase in the number of MF/UF plants is concern over microbial contaminants, Giardia cysts, and disinfectant-tolerant Cryptosporidium oocysts. Four-log microbial removal integrity of full-scale MF/UF plants in the United States can be verified daily with a simple automated test. Challenge tests with protozoa and bacteria have shown 6-log removals or higher. Additionally, MF/UF plants produce filtered water with very low turbidity, routinely <0.05 ntu. Where they can be used in the absence of pretreatment, these plants can require less onsite labor than coagulation and filtration plants, have a compact footprint, and use modular equipment that can facilitate deferment of capital expenditure until more capacity is needed.

Membrane filters can be either encased or submerged. Encased MF units are constructed of hollow-fiber membranes mounted in pressure vessels and operate by positive pressure on the feed side driving filtered water through the membrane wall. In submerged MF units, the membranes are mounted in a tank at atmospheric pressure, and suction on the filtrate draws water through the membrane. Periodic backwashing, chemically enhanced backwashing, and chemical cleaning remove accumulated material from the membrane to maintain acceptable head loss. Most installations include pumping systems to provide the transmembrane pressure, but some more recent facilities rely on siphonic action as a cost-saving feature. Examples include the Chestnut Avenue Waterworks in Singapore, and a plant at Pendleton, Ore.

In the mid-1990s, MF/UF membranes were fabricated of cellulosic or polypropylene materials, but these more recently have been replaced by polyvinylidene fluoride, polysulfone, or polyethersulfone, which have better fouling resistance and chlorine tolerance.

Although MF/UF achieves excellent removal of particulate material (even of submicron particles), it removes few dissolved contaminants. With the use of pretreatment or post-treatment, the range of water quality that can be treated with MF/UF has been expanded. At the Chaparral Water Treatment Plant in Scottsdale, Ariz., inline coagulation with ferric coagulant is provided to control arsenic, and post-treatment with GAC removes taste and odor compounds. At Bakersfield, Calif., preoxidation, flocculation, and plate sedimentation are provided to control DBP precursors and manganese and to reduce membrane fouling. The South San Joaquin Irrigation District in California pretreats with DAF to prevent fouling by algae, and in Clovis, Calif., ballasted flocculation is used to remove dissolved organic material. The largest UF plant in North America combines lime softening and membrane filtration to treat 70 mgd (265 ML/d) of upper Mississippi River water for Minneapolis, Minn.

Filtration's future promises more advances Filtration of public water supplies began in the late 1800s and early 1900s to provide a supply of clear water and to prevent waterborne disease outbreaks. To some extent, the importance of filtration increased as microbiologists learned that protozoan microbes were considerably more resistant than bacteria to chemical disinfectants (especially chlorine).

Laws grow more stringent to protect public supplies. The need to provide an effective barrier against the passage of pathogens into public drinking water systems has led to progressively stricter regulatory requirements for water filtration, a trend that seems likely to continue. In 1942, the US Public Health Service recommended that turbidity should not exceed 10 Jackson turbidity units. In 1962, this was lowered to 5 turbidity units. In 1975, the USEPA established a maximum contaminant level (MCL) of 1 turbidity unit, which was followed by the 1989 Surface Water Treatment Rule setting a treatment requirement of 0.5 ntu in 95% of monthly combined filter effluent samples and the 1998 Interim Enhanced Surface Water Treatment Rule setting a treatment requirement of 0.3 ntu in 95% of monthly combined filter effluent samples for plants using coagulation and granular media filtration (Hess, 2005). In addition to this regulatory trend, the Partnership for Safe Water has recommended a goal of 0.1 ntu that encourages even higher levels of performance. The drive to attain lower and lower filtered water turbidity and control pathogenic protozoa is likely to favor membrane filtration in instances in which source waters do not present multiple contaminant removal challenges that would require substantial treatment efforts before or after MF/UF.

Another important trend in the regulation of public drinking water systems is the steadily increasing number of regulated contaminants for which treatment techniques or MCLs apply. As water utilities face the proliferating regulatory requirements, some may find that the most cost-effective approach is to continue using their existing treatment facilities, perhaps with modifications to operating procedures or infrastructure. For example, utilities that have to take action to meet an MCL for total organic carbon (TOC) may modify their coagulation practice.

With more regulated contaminants, filter technologies make sense. For new filtration plants, the cost savings attained by using monomedium filters or coarse dual-media filters and high filtration rates may need to be weighed against the need to lower the concentration of TOC, DBP precursors, or assimilable organic carbon at facilities using coagulation and filtration. A filter-adsorber operates with an EBCT of 5 minutes in a filter bed with a depth of 32 in. (0.81 m) and a filtration rate of 4 gpm/sq ft (10 m/h), whereas at 8 gpm/sq ft (20 m/h) a depth of 64 in. (1.6 nm) would be required to attain the 5-minute EBCT.

BAF is expected to play an important role in the future because it removes organics (DBP precursors) and produces a biostable effluent, requiring lower chlorine residuals to be maintained in the distribution system, both of which contribute to a reduction in DBPs. BAF may also be considered more often in the future because it provides removal of taste and odor compounds, especially when preceded by ozonation.

Endocrine disrupting compounds (EDCs) are becoming contaminants of interest. If some of these compounds are found to have adverse human health effects, filtration plants that treat surface waters receiving wastewater discharges appear particularly likely to need additional processes for controlling EDCs. Ultimately filtration by straining without any pre- or post-treatment other than disinfection may prove to be applicable only to pristine source waters not having high concentrations of TOC or color or other dissolved contaminants that would be covered by regulations.

In the current and anticipated regulatory environments, the comprehensive goals of water treatment at a given site need to be considered when a filtration process is selected. The need to produce finished water that meets multiple and sometimes apparently conflicting regulatory requirements is increasing the complexity of water treatment for many utilities. As the number of regulated contaminants rises, additional treatment processes or multifunctional processes (such as biological filtration) will be increasingly needed. The popularity of membrane filtra­tion has been growing rapidly, and this trend is expected to continue if pre- and post treatment processes plus MF/UF are competitive with granular media filters for meeting multiple water treatment goals at favorable capital and operating costs.

Acknowledgment

The authors thank Alan Hess for providing information on the history of turbidity regulations.

Gary S. Logsdon (to whom correspondence should be addressed) is an independent senior process consultant, 20 Springbok Dr., Fairfield, OH 45014-6616; e-mail garylogsdonpe@earthlink.net. He has more than 30 years of experience in water filtration and has been involved in the investigation of waterborne disease outbreaks. Logsdon earned a bachelor's degree in civil engineering and a master's degree in sanitary engineering from the University of Missouri at Columbia and a doctorate of science in environmental engineering from Washington University in St. Louis, Mo.

Michael B. Horsley is a project manager, Scott D.N. Freeman is a senior membrane process engineer, and Jeff F. Neemann is a process engineer at Black & Veatch in Kansas City, Mo. George C. Budd is a senior process specialist at Black & Veatch in Harborton, Va.

References

AWWA, 1971 (3rd ed.). Water Quality and Treatment (M.E. Flentje and R J. Faust, editors). John Wiley & Sons, New York.

Baker, M.N., 1948. The Quest for Pure Water, vol. 1. AWWA, Denver.

Cleasby, J.L. & Baumann, E.R., 1977. Backwash of Granular Filters Used in Wastewater Filtration EPA-600/2-77-016, Cincinnati.

Cleasby, J.L. et al, 1989. Design and Operation Guidelines for Optimization of the High-Rate Filtration Process: Plant Survey Results. AwwaRF, Denver.

DuPont de Nemours & Co., 1982. Permasep Products Engineering Manual. E.I. du Pont de Nemours & Co., Wilmington, Del.

Great Lakes-Upper Mississippi Board of State & Provincial Public Health & Environmental Managers, 1982. Recommended Standards or Water Works. Health Research Inc., Albany, NY.

Hazen, A., 1913. The Filtration of Public Water Sup­plies. John Wiley & Sons, New York.

Hess, A., 2005. Regulatory Background on Filtration. Techniques for Improving Filtration Workshop, Penn State-Harrisburg Environmental Training Center, Harrisburg, Pa.

Taras, M.J.,1981. The Quest for Pure Water, vol. 2. AWWA, Denver.

USEPA (US Environmental Protection Agency), 2006. Long Term 2 Enhanced Surface Water Treatment Rule. National Primary Drinking Water Regulations. Fed. Reg., 71:3:654 (Jan. 5).

Return to headlines