Slow Sand Filtration
By Art Ludwig
Author of Water Storage (book)
Although slow sand filtration technology has been widely used in Europe since the early 1800s, its current use in North America has been primarily limited to smaller communities in New England. With the recent issuance of the Surface Water Treatment Rule by the U.S. Environmental Protection Agency and new filter requirements for all surface water systems to ensure removal of Giardia cysts, there is renewed interest in slow sand technology. Although extensive research has been conducted over the past ten years documenting the efficacy of this technology for removing cysts and other particles, no publication has provided the appropriate level of engineering information to design, construct, and operate a slow sand filter to meet the level of treatment performance specified in the new regulations.
The Manual of Design for Slow Sand Filtration provides state-of-the-art information on the feasibility of using this technology and on design, construction, and operations to best achieve desired production and performance. Written for civil engineers who are considering slow sand filtration, the manual provides the necessary guidance to design a slow sand facility. Site-specific considerations are also discussed.
Slow sand filtration technology is especially appropriate for small communities that are required to use filtration to comply with new regulations. This manual of design is intended to serve their special needs.
Safe Water From Every Tap Improving Water Service to Small Communities Committee on Small Water Supply Systems Water Science and Technology Board Commission on Geosciences, Environment, and Resources National Research Council
Evaluating Technologies For Small Systems
Before looking to technological answers to water quality problems, small water supply systems should exhaust other available alternatives for improving water quality. One option is to find a higher-quality source water, such as by switching from surface water to ground water or relocating a well to a cleaner aquifer. In general, ground water sources are a better choice for small water systems than surface water sources because they are less turbid and have lower concentrations of microbiological contaminants than surface water. A second, nontechnical option for improving small system water quality is to purchase treated water from a nearby utility. Such options are often more cost effective than attempting to remove contaminants from a poor-quality source water.
When other options are not available and small systems must turn to water treatment processes in order to provide water that meets the requirements of the Safe Drinking Water Act, they may have difficulty raising revenue for capital improvements. One option available for reducing the costs of water treatment for these communities is the use of pre-engineered "package plants." Package plants are off-the-shelf units that group elements of the treatment process, such as chemical feeders, mixers, flocculators, sedimentation basins, and filters, in a compact assembly. Package plants do not eliminate the need for an engineer to design the specifics of the on-site application of water treatment equipment. Nevertheless, because package systems use standard designs and factory-built treatment units that are sized, assembled, and delivered to the customer instead of being custom built on site, such systems have the potential to significantly reduce the engineering and construction costs associated with a new water treatment system.
Site-specific pilot testing requirements can significantly increase the costs of package water treatment plants, partially offsetting the cost savings these systems offer. State regulators often require pilot tests of all new treatment systems other than chlorinators. Often package plants must be evaluated over and over again for source waters having similar quality but located in different communities. Pilot tests can last anywhere from several weeks to 1 year or more. Extensive pilot testing reduces the savings achieved by having the package plants designed and assembled at a central facility. Manufacturers have reported that pilot testing can increase the costs of their equipment by more than 30 percent. For example, according to one manufacturer, a 6-month pilot test can add $16,000 to the cost of a $45,000 package filtration system.
Certification of package plant performance by an independent third party would reduce package plant costs by reducing, although not eliminating, the need for site-specific testing. Currently, no national program exists for certifying drinking water treatment systems other than point-of-use (POU) and point-of-entry (POE) devices used in individual homes. The National Sanitation Foundation (NSF) International, which certifies in-home water treatment equipment, is currently cooperating with the EPA to develop a verification program for package plants. This program, launched in late 1995, is in its beginning phases and is currently funded for a 3-year period. Support for the program should continue, because it could reduce the costs of drinking water treatment technologies for small communities. Once the program is established, testing fees provided by equipment manufacturers will sustain most of its costs.
A key component of a national pilot testing and verification program for package plants is standard protocols for equipment testing. Currently, such protocols do not exist. Water treatment system designers generally conduct bench and pilot studies using their own individual methods and parameters for documenting water quality. As a consequence, it is difficult to compare data sets developed by different investigators. Establishment of standard protocols that measure the parameters covered in Safe Drinking Water Act regulations would allow data collected in one location to be applied elsewhere.
Another key component of a package plant testing and certification program is a national data base for reporting test results. Currently, no such data base exists. Considerable "reinvention of the wheel" occurs as new tests are required to verify technologies at each new location even if identical tests were performed elsewhere on water of a similar quality. Such a data base could be created by expanding the Registry of Equipment Suppliers of Treatment Technologies for Small Systems (RESULTS) data base at the National Drinking Water Clearinghouse in West Virginia. The expanded data base should cover all of the available technologies, use standard formats for reporting data, and include complete information about raw water quality, finished water quality, and operation and maintenance costs for each technology.
While development of standard protocols for testing drinking water treatment technologies is a desirable goal, it is essential to recognize that the degree to which pilot testing can be centralized in order to reduce site-specific testing varies considerably depending on the type of technology, the nature of the water to be treated, and the availability of data documenting the performance of the technology on waters of various qualities. For many technologies, some aspects of process performance can be tested in a central facility, while others need to be evaluated for each source water treated. Following are some general principles that apply to pilot testing of various classes of water treatment processes (see Chapter 4 for details):
Site-specific pilot testing of aeration systems is not necessary; performance can be predicted with design equations.
For membrane systems, much of the detailed evaluation can be based on pilot tests or full-scale applications elsewhere. However, systems using ground water will need to evaluate the potential for chemical scaling of the membranes. Surface water systems will need to test the potential for the source water to foul the membranes and determine whether pretreatment is required to remove particulate matter ahead of the membranes.
For granular activated carbon adsorption systems, some degree of source water-specific testing is necessary because the ability of the carbon to adsorb a target contaminant varies significantly with the chemical composition of the raw water. In cases where the raw water has a low concentration of organic matter, such as in ground water, inexpensive bench-scale columns can adequately predict performance; for surface water systems, pilot tests will be necessary.
Powdered activated carbon adsorption systems need to be evaluated in bench-scale tests, at a minimum, to determine the effectiveness of the powdered activated carbon on the particular raw water and with the mixing characteristics present in the system.
Ion exchange and activated alumina systems require some degree of source water-specific bench- or pilot-scale evaluation to determine the potential for competitive adsorption of ions other than the target contaminants, which can affect the life of materials used in treatment.
Because of the complexity of the chemical processes involved, coagulation/filtration systems require site-specific testing unless an identical coagulation/filtration system is already being used successfully on the same source water. The degree of testing required depends in part on the design of the system and in part on the characteristics of the raw water. In some cases, bench-scale tests using jars to determine appropriate coagulant doses will be adequate.
Diatomaceous earth filtration systems require a few weeks of pilot testing to establish the effectiveness of different grades of diatomaceous earth and to estimate the length of filter runs that might be expected with a full-scale plant.
For slow sand filtration systems, site-specific pilot testing is necessary, unless a slow sand filter is already treating the same source water at another location, because understanding of these systems is insufficient to allow engineers to predict what filtered water turbidity an operating slow sand filter will attain. Piloting of these systems need not be expensive. Pilot test units can be constructed from manhole segments and other prefabricated cylindrical products.
Bag filters and cartridge filters need not be pilot tested at each site. Performance of these filters depends on careful manufacture of the equipment and its use on waters of appropriate quality rather than on manipulation of the water or equipment during treatment.
Lime softening systems need not be pilot tested for small systems using ground water sources; jar testing to determine appropriate process pH and chemical doses is sufficient. Lime softening systems do need to be pilot tested if used on surface water sources with variable quality.
Disinfection systems using free chlorine, chloramine, chlorine dioxide, or ozone need not be tested at each individual site. The effectiveness of these systems is predicted based on laboratory test results, which regulators consider to be applicable to all systems.
Current regulations allow small systems to base corrosion control strategies on desk-top reviews of water quality, rather than on pilot tests.
For the smallest of water systems, in particular those serving a few dozen homes or less, POE or POU water treatment systems may provide a low-cost alternative to centralized water treatment. In POE systems, rather than treating all water at a central facility, treatment units are installed at the entry point to individual households or buildings. POU systems treat only the water at an individual tap. If a source water has acceptable quality for drinking except for exceeding the nitrate or fluoride standards, for example, using a POU system to treat the small number of liters per day needed for drinking and cooking might be less costly than installing a central treatment system that could remove the nitrate or fluoride from all water used by the community. Similarly, POE systems can save the cost of installing expensive new equipment in a central water treatment facility. POU and POE systems can also save the considerable costs of installing and maintaining water distribution mains when they are used in communities where homeowners have individual wells.
Regulators often have significant objections to using POE and POU devices. Concerns include the potential health risk posed by not treating all the water in the household (a problem for POU systems), the difficulty and cost of overseeing system operation and maintenance when treatment is not centralized, and liability associated with entering customers' homes. These objections have merit, particularly as system size increases and the complexity of monitoring and servicing the devices increases. Using centralized water treatment should be the preferred option for very small systems, and POE or POU treatment should be considered only if centralized treatment is not possible.
(then search on "sand filter")
Slow sand filtration was first developed to purify surface water sources for drinking purposes and became a legal requirement in Britain to assist in containing cholera epidemics. Since the mid-19th century, SSF has been widely employed in treating community water supplies in developing countries against water-borne disease and recently there has been a resurgence in its use in small municipal water treatment districts in the United States and England.
How SSF works
Slow sand filtration relies on both physical and biological activity in controlling plant pathogens.
In a slow sand filter, the filter bed is constructed of a medium with high surface area which can be colonized by suppressive micro-organisms. This fine media also presents a physical barrier to the passage of spores of plant pathogens. Bacteria, such as representatives of the genus Pseudomonas and Trichoderma have been demonstrated as biological control agents effectively controlling plant pathogens in hydroponics systems. In a SSF, plant pathogens recirculating in the irrigation water are captured in the filter media, and at slow rates of water filtration (100-200 l/hr/m2 surface area of filter), are acted upon by the antagonistic micro organisms that colonized the filter bed.
The efficiency of SSF depends on the particle size distribution of the sand, the ratio of surface area of the filter to depth and the flow rate of water through the filter. The finest grade sand fractions and granulated rockwool have been shown to be most efficient in controlling diseases such as Phytophthora, Pythium and Fusarium oxysporum, the most widespread nursery diseases.
Advantages of SSF
There are several advantages of slow sand filtration over other methods of water disinfestation:
It is a low energy consuming process
It has great adaptability in components and applications maintenance
Systems can be built and installed by laymen
Costs of building and running significantly lower than other disinfestation methods
In Australia, chlorination/bromination processes are most widely used for water disinfestation in nurseries. The effectiveness of such chemical treatments in controlling plant pathogen depends on correct dosages and treatment times, control of suspended particulate matter in the recycled water, and foolproof monitoring systems. Chemical treatments have proven effective when used properly, however they are relatively expensive and present safety issues to the handlers and the environment.
Other water treatment processes such as UV and Ozone are being adopted by some nursery crop producers. The perceived expense, difficulty and/or questionable effectiveness of some of these water treatment methods has not encouraged wide use.
Establishment of SSF in a nursery is similar to the installation of holding tanks and pumps to allow for batch treatments with chemical disinfectants. SSF requires no purchase of chemicals, there is no technical dosing equipment to service or replace and there is no chance of crop damage if dosing equipment or the operator miscalculates.
SSF used as a replacement to other forms of chemical, ozone or UV treatments would lead to significant savings in equipment upkeep and purchase of chemicals and avoid potential crop phytotoxicity.
Construction of a Slow Sand Filter
A slow sand filter consists basically of the following components:
Filters can be constructed in tanks with non-reactive surfaces such as plastic or fiberglass lined galvanized tanks, poly or concrete tanks of various sizes from 44gal drums (205 litres) up to 100,000 litre tanks. It may be advantageous to construct 2 smaller SSF units rather than one large unit so one can be shut down periodically for cleaning or repairs. Table 1 presents the volume of water filtered in a 24 hour period by filters of varying size. The capacity of the filter is determined by the surface area of the filter top and not the overall volume of the filter. Consideration must be made of the flow rate to be used when determining tank size (see Managing the SSF below).
The water layer above the filter bed provides the head to push water through the filter bed. It is convenient as a water storage zone and provides an effective temperature buffer to stabilize the filter and protect the biological activity occurring in the top layers of filter bed. A minimum depth of .5 m up to 1.5 m is most commonly used. Research in Australia suggests that this water layer should be maintained at a constant depth by the use of a small pump from an overflow tank or from the filtered water reservoir. This is particularly important if nutrient solution or recycled water is introduced into the filter in pulses and not continuously. The filter should never stand with an exposed dry top layer, or stagnant water which would accumulate only if the filter outlet was closed at the bottom. Continuous filtering assists in the development and maintenance of a healthy filter.
The filter bed consists of a uniform fine particle sand mixture as specified (see Sand Specifications). The most critical design feature of the SSF is using a correct sand or alternative media. The filter bed is built to a depth of 1-1.5 m (or more) with a minimum of .8m on smaller filters. This depth of sand will allow for losses which will occur if the top portion of the sand is removed when particulate matter and algae is cleaned from the top of the sand.
Sand is characterized by the diameter of the individual sand grains (eg 0.15-.35 mm) and the effective size of the composite sand, the ES or d10. d10 is defined as the sieve size in mm that permits passage of 10% by weight of the sand. The uniformity coefficient (UC) of a sand is defined as d60/d10.
Sand needs to be of a fine grade ( 0.15-.35 mm is recommended), uniform (the UC should always be less than 3 and preferably less than 2), and be washed free of loam, clay, and organic matter. Fine particles will quickly clog the filters and frequent cleaning will be required. A sand that is not uniform will also settle in volume, reducing the porosity and slowing the passage of water. Sand manufacturers should be able to supply or blend sand to these specifications.
A gravel drainage system is provided at the bottom of the filter to prevent movement of the fine sand into the filter outlet. In European specifications this has consisted of 3 graded layers, 2-8mm,8-16mm,16-32mm. The use of a geo-textile fabric may be considered to support the sand as an alternative to some gravel layers. The bottom layer of gravel supports perforated drainage pipes which may simply bisect the filter or in a large filter form a network of connecting pipes across the base. The use of granulated rockwool as an alternative media to sand can reduce the requirement for the gravel drainage system and thus reduce the depth of the filter. A fine screen over the outlet would still be recommended to prevent rockwool granules from passing into the outflow.
A regulating tap should be connected to the filter outlet to control the flow rate. On large filters a flow meter is sometimes installed for use in monitoring the flow rate. The flow rate is specified in terms of litres/hour per unit area of the surface of the filter ( m2). The flow rate through the filter is less than gravitational fall. Keeping the water level above the filter bed constant assists in maintaining an even flow. The flow rate will drop off with the build-up of material on the surface of the filter bed. An open clear pipe (poly tube) fixed to the exterior of the filter can be used to monitor head loss.
Schematic Diagram of SSF Integrated Into a Nursery Recycling System
Nutrient solution or recycled water drains from the crop and is collected in
a catchment tank or dam (1). Water is pumped to the SSF via a holding tank (3)
or through a fast particle filter (2) to remove suspended organic matter, algae
etc. Overflow outlet (4) back to holding tank maintains constant depth to water
The SSF consists of: a water storage layer (5), a sand or media filter bed (6) and gravel layers (7-9) to support filter bed. The outflow (10) may be fitted with a flow regulator valve and an open tube (11) to measure filter head loss. A small collection tank (12) lower than the SSF collects filtered water for distribution by pump (13). An optional holding tank (14) is used for the filtered water.
Managing the SSF
Water flow rates are relevant in relation to the disease load of the irrigation water. A slow flow rate of 100 l/hr per m2 of surface area have been found to be preferable in high risk situations such as the control of Fusarium or viruses in tomato crops. In German research, the rates of 200 or 300 l/hr per m2 are recommended for control of Pythium and Phytophthora in general nursery use. These rates are very important to consider when the SSF is being designed and a disease risk assessment should be conducted to determine the size of filters that will be required.
As a guideline, where high volumes of water are being treated and the disease pressure is low ( eg container nursery), then the higher flow rates may be most appropriate. There is no reported demonstration of filter effectiveness against plant diseases above the rate of 300 l/hr/m2. If there is a high disease presence in the crop or the disease of most concern is a disease with small spores such as Fusarium, low rates of flow would be recommended.
Table 1. Volume of water filtered in a 24 hour period by filters of varying size surface area
M2 surface area
Flow rate (litres/hour)
All forms of water disinfestation may need to use traditional sand filters to pre-clean recycled water if many particulates are present. Similarly, irrigation water from dams with a high algal or silt content will need pre-filtering prior to SSF. Figure presents a large SSF in California which recycles all the greenhouse nutrient solutions into an outside holding dam. The fast sand filter ( which can be easily backflushed) pre-filters all this dam water before it reaches the SSF.
When the flow rate through the filter slows (as indicated by head loss) the water layer above the sand bed can be drained and the top layer of sand scraped off. This layer should contain most of the suspended organic matter and silt that was slowing the water flow. In a granulated rockwool filter, some settling may occur over time and cleaning may also involve topping up with more rockwool at this stage.
Microbial activity builds up very quickly in greenhouse nutrient solutions and slow sand filters will become active in biological suppression without any special inoculation. Some researchers suggest a four week period to fully establish the activity of filters, however, the filters are physically active immediately and water needs to circulate through them to establish a microflora. In the future, the potential exists to inoculate the filters with specific biologically suppressive preparations as more of the complex microbial interactions occurring in the filters are understood.
Australian research has demonstrated that the majority of biological activity is occurring in the top 20 cm of the sand filter, when a modifying layer of water (nutrient solution) is passed continually over the filter head.
Safe drinking water is a high priority for people living without mains facilities. This tip-sheet outlines the necessary testing procedures and tells you how to construct a reliable cleansing filter for small volumes of water.
Slow gravity sand filters remove bacteria and other small particles from drinking water, making it safe to drink. This tipsheet provides a basic introduction to the subject and is for people who wish to maintain a constant supply of clean, running water using simple technology. The sand filter described below is designed for domestic use only.
It is important to start off with a fairly clean source of water, so you will need to get a water sample tested at a laboratory (details of which usually appear in the Yellow Pages). Sand filters cannot cope with heavy metals or other excessive pollutants. Their prime purpose is to remove bacteria and particles. It is not appropriate to use the technology to clean up water contaminated by chemicals. If your chosen water source does have a high level of contamination, ideally you should locate a new one. If this isn't possible other methods of filtration may be used, depending on the level of contamination. For example, even spring water or very clean river water should be checked for undesirable contaminants. If the water contains sediment, it should be passed through an initial settling tank before it gets to the sand filter.
How the filter works
The water passes through the sand from top to bottom. Any larger suspended particles are left behind in the top layers of sand. Smaller particles of organic sediment left in the sand filter are eaten by microscopic organisms including bacteria and protozoans which 'stick' in the layers of slime that form around the sand particles and the clean water which passes through the filter is safe to drink. Provided that the grain size is around 0.1mm in diameter, a sand filter can remove all fecal coliforms (bacteria that originate from feces) and virtually all viruses.
Slow Sand Filters
A typical slow sand filter configuration appears on the accompanying sketch. The raw water flows into the upper tank region in such a manner as to avoid disturbing the scmutzdecke; flows near that surface must be very gentle. The water in this compartment must have sufficient depth to drive through the schmutzdecke, the filter bed and into the support gravel - and initially should be about 2-3 meters, or 7-10 feet. The lower limit of the depth is somewhat controversial but 1.5 meters, or about 4 feet, should be a reasonable value. This figure, however, will be related to the sand properties and the porosity of the schmutzdecke.
The maximum water level may be automated by using a float and a control valve or by periodically adjusting the valve manually to maintain the water lever near the overflow line. The depth of the filter bed has a strong influence on the effectiveness of the filtration and should be at least 0.75 - 1.0 meters (30" - 40"). In the discussion of filter media the properties ES and UC are defined, and for the SSF, ES typically is 12 to 40 mm (0.5" to about 1.6")and UC should be less than about 2.5. Typical water processing rates in SSFs are about 2.5 m3/[m2 of filter cross-section area - day] = about 0.1 m/hour or about 0.33 ft/hour.
The filtration rate may be determined by a flow meter in one of the lines or by a weir in the outlet tank. Unless the raw water is particularly well treated to about 20 turbidity units or less, this figure should be maintained, unless a very reliable post-ssf disinfection is in place. High turbidity levels in the raw water will prematurely block the ssf, leading to a much shortened time span between cleanings and an overall deterioration of the water quality. High turbidity in the raw water may shorten the filter life from several months to a matter of days. The Horizontal Roughing Filter is a very effective means of pre-treating the raw water to reduce the turbidity to acceptable levels, from an average of about 200 units, with occasional short-term peaks to around 1000, down to about 20. If the river water turbidity is around 20 units or less, except at certain periods of the year, the HRF could be by-passed most of the year and brought on-line during these periods. Other means of turbidity reduction include holding ponds and sedimentation tanks.
The processes that occur in the schmutzdecke are enormously complex and varied, but the principal one is the mechanical straining of most of the suspended matter in a thin dense layer in which the pores may be very much less than a micron. The thickness of this layer increases with time from the initial installation to the point where the flow rates become unacceptably small, when it is usually about 25 mm (1"). The greatest benefit of the SSF lies in its ability to trap bacteria and viruses in this schmutzdecke. Bacterial and biological activity maximizes in there but will continue at a decreasing level down into the sand of the filter bed. A certain minimum level of dissolved oxygen should be present to support the aerobic actions that occur in the bed. After the initial installation of the SSF, the formation of the schmutzdecke and bacterial/biological activity in the bed may take days or weeks depending strongly on the ambient temperature. During this period the processed water is unsafe for human consumption and should be discarded, or used as raw water for another filter, or for some other non-critical purpose.
Water quality tests should be done at a regular intervals until the required standard is reached. When the filtration rate drops to an unacceptable level, the SSF must be cleaned by carefully removing about 25 mm (1") of the top layer, which includes most of the existing schmutzdecke, after dropping the water level to slightly below the latter. Such work may be done by a mechanical scraper or manually and very carefully with broad flat-bottomed shovels. In strong hot sunlight this work should be carried out as quickly as possible to avoid excessive drying out and damage to the biological matter in the new top layer, which will be the base of the new schmutzdecke.
When the water flow is restarted, the processed water must again be diverted until the required level of quality is again reached. This usually requires several days. In time the depth of the original bed of sand will have been reduced by the cleaning processes to about 0.75 meters (30"), when the original depth must restored. As the new sand will be almost devoid of biological activity, placing it on top of the existing sand would require an excessively long time to develop a new schmutzdecke. To accelerate this process, the schmutzdecke is first removed and discarded, and then most of the existing bed is removed and set aside for re-use. The new sand is placed in position and is then covered with the original bed. In this way the biological activity and new schmutzdecke are more rapidly restored. The sand from the existing bed should not be allowed to dry out and should be set in place as quickly as possible.
A disadvantage of the SSF is the large amount of land required, which is a consequence of the slow rates of water filtration that are possible - typically only 10 percent or less of the rates that are possible in rapid filtration. Furthermore, these slow rates mandate water storage to accommodate peaks in the demand cycle. Great care must be exercised in the operation and maintenance, in particular in relation to the schmutzdecke which requires some time to form. The processed water cannot be considered safe until it has and the water that is passed through the system while the schmutzdecke is forming must be cycled back or discarded.
Please report errors on this page All contents copyright ©1996, 1997 waterinfo.com. All rights reserved Revised: URL:
More useful information for water systems
Links and info
- http://www.photometer.com/en/abc/abc_095.htm Procedure for measuring turbidity
- The Davnor BioSand Water Filter
- Manual of Design for Slow Sand Filtration, David Hendricks, Amer Water Works Assn (June 1991)
- Slow Sand Filtration, Huisman L. and Wood, W.E., WHO 1974
Prepared by Colorado State University, CH2M Hill Engineers, Denver (CO) Water Department, Colorado Health Department, Black & Veatch, Dayton & Knight, and RBD Consulting Engineers, June 1991. AWWA Research Foundation Order Number: 90578 Out of Print
www.oasisdesign.net Copyright © Art Ludwig 1997 -2015