Introduction
The history of slow sand filtration in the United States has been one of reluctant
acceptance. Many European cities choose slow sand filtration as a water
treatment method because of its simplicity, reliability, and economy
(Collins). Slow sand filters in the United States are found primarily in smaller
communities with fewer than 10,000 people, 45% of which serve fewer than
1,000 people (Sims). This is primarily due to the associated low cost of
slow sand treatment facilities compared with alternative water treatment
technologies (Sims). The number of operating slow sand filters by state
in the United States as of 1991 can be seen in Figure 1.
Figure 1. Number of slow sand filters operating in each state as of 1991. (Sims)
Physical CharacteristicsSlow sand filtration is a water purification process in which water is passed
through a porous bed of filter medium. Slow sand filters are typically characterized
by certain design components: the supernatant (water above the filter sand that
provides hydraulic head for the process), filter sand varying in depth, the
underdrain medium (usually consisting of graded gravel), and a set of control
devices (Sims). In a mature sand bed, a thin upper sand layer called a
Schmutzedecke forms. The Schmutzedecke consists of biologically active
microorganisms that break down organic matter while suspended inorganic
matter is removed by straining (Van Duk). Slow sand filters are distinguished
from rapid sand filters by the biologically active sand medium
(including the Schmutzedecke), and slow detention times. Rapid sand filters
utilize primarily a physical removal process, are periodically backwashed for
cleaning, and operate with long detention times. Slow sand filters are
cleaned by periodically scraping the existing Schmutzedecke (Van Duk).
Figure 2 is a schematic of a common cross section of a slow sand filter.
Figure 2. Typical cross section of a slow sand filter.
Drawing by Mary Rust
The supernatant serves two distinct purposes. First, it provides a head of water sufficient to pass the raw water through the filter bed. Second, the supernatant creates a detention time of several hours for the treatment of the raw water. The supernatant should not be considered as a reservoir for sedimentation. If the raw water has a high content of suspended mater, then pretreatment should be considered to prevent rapid clogging of the filter bed. The supernatant depth is typically a meter (Van Duk).
The physical characteristics of a sand bed are important in maintaining
the slow sand filter’s efficiency. The effective size is the size opening that
will pass ten percent by weight of the filter material (Haarhoff). Effective
sizes in the range of 0.15 mm to 0.35 mm are used (Van Duk). The
uniformity coefficient is the ratio of the size openings that pass sixty
percent of filter material to the size openings that pass ten percent of filter
material, e.g. the effective size (Haarhoff). Uniformity coefficients range
between two and five; most facilities maintain uniformity coefficients
less than three (Haarhoff). The filter medium itself should consist of
inert and durable grains; sand should be washed so that it is free of
clays, loams, and organic matter. Depth of a filter bed ranges
between 1.0 and 1.4 meters (Van Duk). The clean filter medium
from a slow sand filter in a treatment plant near Paris, France can be
seen in Figure 3.
Photo by Dr. Robert Hoehn
Figure 3. Clean slow sand filter without supernatant water layer.
The underdrain system serves two purposes. It provides unobstructed
passage for the collection of treated water and it supports the bed of filter
medium. It is important that the underdrain system provide a uniform
velocity over the entire filter area (Van Duk). The underdrain gravel is
placed so that the finest gravel is directly underneath the sand and
the coarsest gravel is surrounding the underdrain pipes or covering
the underdrain block (Pyper). This prevents the filter sand grains from
being carried into the treated water system. An example of support
media and underdrain system is shown in Figure 4.
Photo by Dr. Robert Hoehn
Figure 4. Underdrain and support media.
Biological and Physical MechanismsBiological activity in the sand bed is not well understood. Scientists have a vague idea of the processes involved, but specific interactions are still unknown. Suggested biological removal mechanisms are predation, scavenging, natural death and inactivation, and metabolic breakdown (Haarhoff). In the Schmutzedecke, algae, plankton, diatoms, and bacteria break down organic matter through biological activity. It has been hypothesized that as the raw water passes through the bed, it constantly changes direction. Thus, the sand grains develop a uniform sticky layer of organic material that absorbs to the particles by various attachment mechanisms. The sticky layer around the sand grains is biologically active (bacteria, protozoa, bacteriophages) and the organic impurities are biologically converted to water, carbon dioxide and harmless salts. According to a study by Collins, the bacterial concentrations in the Schmutzedecke were a function of the elapsed time and potential for cell growth rather than the filtration of free-living bacteria from the source water (Collins). The biologically active section of the entire filter bed extends 0.4-0.5 m downward from the surface of the Schmutzedecke (Van Duk).
Physical processes are also inherent to slow sand filter mechanisms. As the biological activity of the filter bed decreases, the physical processes of adsorption and chemical oxidation are the primary mechanisms (Van Duk). Adsorption accounts for removals that were traditionally thought to be purely biological. For example, the removal of chlorinated organics and the distribution of viruses are thought to follow adsorption isotherms (Haarhoff). Furthermore, suspended inorganic matter may be removed by the physical process of straining (Van Duk).
Organic Carbon RemovalAdsorption and biodegradation are considered to be the primary natural organic matter removal mechanisms (Collins). Literature cited by Collins suggests that large hydrophobic-humic organic molecules are removed by adsorption, and smaller organic molecules are removed by both adsorption and biodegradation. The smaller hydrophilic material (carbohydrates, aldehydes, and simple organic acids) are considered to be primarily removed by biodegradation. A common oxidant for the treatment of water in the United States is chlorine; the hydrophobic-humic organics (considered to be the more trihalomethane reactive) were removed in greater than 80% of all comparisons of organic parameters cited in the Collins study.
Another interesting aspect of the Collins paper was the fact that natural organic matter and organic precursor material were a function of filter media biomass: the greater the biomass, the greater the organic carbon removals. Three US sand filters were compared; the West Hartford filters use a unique Schmutzedecke cleaning procedure called the filter-harrowing cleaning technique (discussed in more detail under Filter Scraping). This procedure allows for the minimization of biomass removal from mature sand filters resulting in increased removals of biodegradation and bioadsorption (Collins).
Removal of Giardia and CryptosporidiaIn the past decade the protozoan parasite Cryptosporidium parvum has been recognized as a significant threat to public water supplies. The resistant stage of Cryptosporidia is called an oocyst; this stage is relatively untouched by a chlorination disinfection process. Slow sand filtration has been looked at in numerous studies to determine the viability of this treatment process for the removal of Cyrptosporidia. A study in England by Timms found reductions of oocysts greater than 99.97%; the oocysts were found in the filter media above 2.5 cm. Another study in British Columbia by Fogel contradicts the aforementioned study. Fogel found removal efficiencies of 48%; this figure is significantly different than the 100% removals Fogel cites from previous literature. However, a point to note concerning the British Columbia filters is that they were operating well out of the range of the recommended design limits for the uniformity coefficient at 3.5 (Fogel). Furthermore, temperature can adversely affect the performance of a slow sand filter; the British Columbia filters were operating at extremely low temperatures of less than 1°C (Fogel). Overall, the literature supports data that strongly suggests slow sand filtration is a viable alternative for Cryptosporidia removals.
Slow sand filters have also been proven highly efficient in removing Giardia lamblia, a frequently identified pathogenic intestinal protozoa. The same study by Fogel found that despite the uniformity coefficient parameter and the low temperatures, Giardia removals were complete. This data was further supported by literature cited by Fogel. Furthermore, fecal and total coliform counts were below the detection limit, and the removal rates were similar to Giardia removals (Fogel).
Schmutzedecke Scraping OperationsScraping typically involves the removal of the Schmutzedecke and the operation
is site specific. Frequency of scraping depends on the available head, the media grain-size
distribution, the influent water quality, and the water temperature (Letterman). Higher
frequencies of scraping are associated with increased water temperature, high solids
concentrations in the influent, low head, and small media pore size. A typical
operation involves draining the supernatant (usually by continuing filtration with no
influent) to 20 cm below the sand surface, skimming off one inch of the Schmutzedecke
and associated sand, and then filling the filter from the bottom of the bed using filtered
water to prevent air entrapment. The bed should be refilled until depth is sufficient to
continue normal operations (Letterman). A filter in the process of having the
Schmutzdecke scraped is shown in Figure 5.
Photo by Dr. Robert Hoehn
Figure 5. Removal of Schmutzdecke.
The study by Collins at West Hartford Treatment Plant suggests a
unique way to scrape the Schmutzedecke that minimizes the amount of
biomass removed. The supernatant is drained to a height roughly 30 cm
above the bed. A rubber-tired tractor equipped with a comb-tooth harrow
is place on top of the filter to rake the sand; simultaneously the filter drains
are opened, causing a steady discharge of overlying water. As the
Schmutzedecke is loosened, the colloidal debris is caught by the moving
water and discharged at the filter surface drain. The process is repeated as
necessary by backflushing until the entire filter surface has been harrowed (Collins).
Summary comments
by Dr. Robert Hoehn on the use of slow sand filtration in the United States.
If you are interested in the latest concerns about the operation of slow
sand filters in the United States you may want to take a look at the
AWWA Slow Sand Filtration Forum.
References
 Water Treatment Table of Contents |
 Previous Topic |
 Next Topic |