Revisiting Design Criteria for Stormwater Treatment Systems, Part 2
Fine-Media Filters
This is the second of a four-part series examining design criteria for stormwater treatment systems. In the first of this series (Stormwater, November/December 2004) basins were the focus. Now we shift our focus to fine-media filters, which are filters that use fine-grained media similar to those used for potable water treatment. The concept of using fine media for stormwater treatment was first adopted by the City of Austin, TX, in the late 1970s. The most commonly used medium is sand, although other media are discussed in this article. Not considered in this article are manufactured stormwater filters, which commonly use coarser media.
Since communities in the United States first began requiring post-development treatment of stormwater from new developments 25 years ago, local, regional, and state governments have published many manuals and handbooks identifying acceptable treatment systems (structural best management practices) and design criteria. As noted in part 1 of this series, few of the criteria were initially supported by laboratory or field research, and engineers relied on their best professional judgment in choosing design criteria. In light of the many studies completed in the past 20 years, it is time to reexamine some of these criteria.
In addition, treatment strategy is evolving from a focus on the general removal of pollutants using total suspended solids (TSS) as the surrogate for all pollutants to a more recent trend to consider specific pollutants. Some states now emphasize removal of total phosphorus, total nitrogen, and dissolved metals in certain situations. A particular design criterion can vary depending on which pollutant is targeted.
Table 1 provides design criteria for fine-media filters from representative manuals and handbooks. Sand filters are a commonly used treatment system in Texas and the Chesapeake Bay area, and are particularly attractive for the removal of sediment (TSS or suspended sediment concentration) and attached pollutants. They can be expected to give consistent performance irrespective of the size of the storm and flow rate, unlike many other systems such as swales and vortex separators. They also may provide relatively consistent effluent quality with respect to TSS and attached pollutants, irrespective of the variation in influent concentration (Caltrans 2004). Despite these advantages, sand filters are rarely used outside the two areas mentioned above. This likely occurs for two reasons: First, sand filters can be fairly large, particularly if the available head is modest given site topographic constraints. The second concern is maintenance. This article discusses how the selection of design criteria directly affects surface area and maintenance frequency. Alternative criteria are suggested that may result in wider use of fine-media filters.
Pretreatment
It is common practice to pretreat stormwater before it enters the filter bed. The objective is to reduce solids loading on the filter bed, thereby extending the time between necessary maintenance events. Subsurface filters commonly use a wet vault chamber separated from the filter bed by a common wall. The most common pretreatment method for surface filters is an extended detention basin. Wetponds are used where in some cases the filter is actually the outlet berm of the pond and not a separate unit. Flow-through grass swales have also been employed. However, wetponds and grass swales may not be the most appropriate choice for pretreatment given the potential for clogging of the filter bed—from algae in the case of ponds and eroded sediment in the case of swales. However, no problems have been published in this regard.
Media Type and Size Distribution
As previously mentioned, sand is the most commonly used filter medium. Mixtures of sand and a second medium are used, although infrequently. The objective is the removal of dissolved pollutants. Added media in full-scale facilities have included peat (for removal of metals), activated carbon (organics), calcite (phosphorus), and iron filings (phosphorus). There has been laboratory- or pilot-scale experimentation with dolomite (phosphorus) and soybean hulls (metals). An alternative is to coat the sand with an oxide of iron, manganese, or aluminum to remove dissolved metals.
The size distribution commonly specified today is ASTM (American Society of Testing and Materials) C33. This is the specification for fine aggregate for use in concrete. Its ready availability reduces the cost. It is, coincidently, fairly similar to the specification of sand for water treatment. Early manuals had a simple specification of "0.20 to 0.40 and fines are acceptable." It is believed that this specification has caused premature clogging of sand beds, sometimes through the entire depth of the bed.
Media Depth
Initially, the City of Austin specified a depth of 36 inches, mimicking the depth commonly used in potable water treatment. It was soon changed to 18 inches, likely because of the recognition that almost all of the removal of TSS and attached pollutants occurs on or near the surface of the bed. Substantial media depths are used in potable water treatment given the need to consistently and reliably meet a very strict potable water standard. Such a standard, less than one unit of turbidity, is not required of stormwater treatment.
The relevance of media depth relates to the area required of the filter to pass the design event. The greater the depth of the media, the greater the surface area of filter required to meet the specified drawdown time as shown in Table 2. Eighteen inches is taken as the standard with a unit area of one. Reducing the media depth from 18 to 6 inches reduces the required area of the filter by about 50%. Conversely, a filter depth of 36 inches increases the filter area by about 30%. How the relationships in Table 1 were produced is presented later in this article.
Sand filters with 18 inches of media appear to perform more effectively than other stormwater treatment systems, in particular extended detention basins and swales. Therefore, some may consider it acceptable to reduce the depth of the media in a sand filter even if performance degrades somewhat. One study (Amini 1996) found that increasing the media depth from 6 to 12 inches, the maximum depth evaluated in this study, increased the removal efficiency of TSS by only 5%. A study of bacteria removal found little improvement with media depths above 12 inches (Bellamy et al. 1985). In a recently completed study (Caltrans 2004), several sand filters with 18 inches of media performed substantially better than a single filter that had only 12 inches, with mean effluent concentrations of about 6 and 12 mg/L, respectively. However, 12 mg/L is acceptable. While the results of these studies are inconsistent, they do indicate that shallower bed depths are acceptable in most situations, certainly 12 inches and possibly 6 inches.
A depth of 6 to 12 inches is likely impractical with large filters where mechanical methods are used to maintain the filter bed. Heavy equipment could damage the underdrain system. But it may be acceptable for small filters, less than a few hundred square feet, where cleaning by hand occurs. Also, more frequent cleaning may be feasible with small filters. Alternatively, where 18 inches is specified, the depth of the design storm could be reduced, commensurate with the higher efficiency of sand filters over other treatment technologies (Barrett 1999, King County 1998).
A shallow filter bed is likely acceptable where the objective is the removal of TSS and attached pollutants. It may not be appropriate if the objective is the removal of dissolved species. Removal of dissolved pollutants is discussed in the last of this series of articles.
Hydraulic Conductivity
Sometimes referred to as the coefficient of permeability or the filter factor, hydraulic conductivity describes the ability of water to pass through media. Its value is inherent to the particular medium, its size distribution, and sediment accumulation. Hydraulic conductivity is not to be confused with the infiltration rate, which is the actual rate of flow through the filter. Confusion occurs as both have the same units (feet or inches per hour or per day). The filtration rate increases with the depth, or head, of water above the filter bed, whereas the hydraulic conductivity is unaffected by head. Hydraulic conductivity, however, decreases with the accumulation of sediment.
Specified design values have ranged from 1.8 to 3.5 feet per day. The effect of its selection is shown in Table 3. The more conservative value of 1.8 feet per day increases the filter area by about 90%, a significant increase.
What does the selected hydraulic conductivity represent? The hydraulic conductivity of clean sand is on the order of 2 to 5 feet per hour. The values in Table 3 are 5%–10% of clean sand. These values therefore represent a filter almost totally clogged: the time to clean the filter. If the filter is not cleaned promptly, the management goal of treating, for example, 90% of the stormwater over time, is not met. The greater the selected hydraulic conductivity the smaller the filter area, but the more frequent the maintenance. The value chosen is arbitrary within a reasonable range, and represents the tradeoff between filter area and the frequency of maintenance. In effect, its selection represents a tradeoff between initial construction costs and long-term maintenance costs.
Elevation Drop
A major limitation in the use of fine-media filters is the available elevation drop—that is, the difference in the elevations of the development and of the public drainage system to which the development discharges. Table 4 illustrates the effect of the available maximum water depth over the top of the filter, called the head. As the head increases, the surface area of the filter decreases. Hence, the feasibility of fine-media filters increases with increasing head. However, the decrease in filter surface area corresponds with the required frequency of cleaning. It should also be noted that reducing the bed depth also reduces the elevation drop required of the site, and therefore the potential feasibility of a fine-media filter.
Drawdown Time
The design drawdown time affects the size of the filter. Therefore, its specification is important. Table 1 presents a significant range of 24 to 48 hours. Drawdown time affects filter area by its definition of Q, the average filtration rate during the design storm. The greater the allowed drawdown time, the lower the filtration rate and therefore the smaller the filter area.
Avoiding anaerobiosis in the filter bed is the common reason given as to why a maximum drawdown time must be specified. Bacteria attached to the filter media consume dissolved oxygen as they use organic matter and ammonia in the incoming stormwater. The concern is that under anaerobic conditions, dissolved phosphorus and metals previously removed may be released. An additional benefit from specifying a drawdown time may be to inhibit the excessive growth of bacteria that might clog the filter. Drying of the filter bed between storms inhibits growth.
The relationship between drawdown time and either anaerobiosis or excessive bacteria growth has not been established. Nor has the significance of pollutant release been defined. Whether release occurs depends on the mechanism of removal. A possible mechanism of dissolved phosphorus removal is sorption/precipitation to ferric oxide present on the surface of the sand. Under anaerobic conditions, ferric iron changes to ferrous, disassociating the complex and releasing the pollutants. One study found a constant anaerobic condition in the bottom of the filter (Shapiro 1999). The conditions of the study are likely the most extreme that will be faced. The filter had a media depth of 36 inches and a typical drawdown time of 72 hours. Furthermore, it was located in a region with long storms and short interevent times, conducive to creating anaerobic conditions. Finally, iron filings were added to the sand to promote the removal of dissolved phosphorus. The transformation of the filings under anaerobic conditions appears to have caused sand particles to bind. The filter was found to be constantly anaerobic in the bottom 24 inches. However, despite this condition the filter removed on the order of 50% of the dissolved phosphorus as well as dissolved metals. All of the removal was likely occurring in the upper 12 inches of media that remained aerobic.
If nitrogen removal is the objective, a temporary anaerobic condition is desirable. Bacteria change ammonia to nitrate in the presence of dissolved oxygen, but the net reduction of nitrogen is zero. Other bacteria under anaerobic conditions change the nitrate to nitrogen gas. The ideal operation is for the upper area of the filter to be aerobic and the lower area to be anaerobic.
Specification of the drawdown time is therefore a function of the pollutant-removal objective, the interevent time between storms, and the time needed for the filter to dry. If the objective includes dissolved phosphorus and/or metals, a low drawdown time, perhaps on the order of 24 hours, is appropriate. If not, 72 hours may be satisfactory. A long drawdown time may also be desirable if nitrogen removal is the management objective.
An alternative to a "tight" drawdown time may be to incorporate aluminum into the sand. Aluminum oxides may be an effective remover of dissolved phosphorus and/or metals (Minton 2002). Unlike ferric oxide complexes, aluminum oxide/phosphorus/metal complexes will not disassociate under anaerobic conditions.
Determining the Filter Bed Area and Live Storage Volume
Manuals employ the same equation (Equation 1) for sizing the filter bed area, although the specific form of the equation varies with the units used.
The average filtration rate is equal to the runoff volume of the design event divided by the drawdown time at brimful. For example, assume a design runoff depth of 1 inch. The volume of runoff is the multiple of 1 inch, the drainage area, and the runoff coefficient. If a 48-hour drawdown time is selected, the average filtration rate, Q, is half that if the drawdown time is 24 hours. Hence, Equation 1 states that a drawdown time of 48 hours requires a filter area that is half that for a 24-hour drawdown time.
Previous observations in this article about the relationship between bed depth, hydraulic conductivity, drawdown time, available water depth, and filter surface area are understood with the examination of Equation 1. Hence, the selection of the values for the key design criteria must be considered as a whole, paying particular attention to the effect of these decisions on maintenance frequency.
It is important to recognize that the available live storage volume above the filter itself is usually less than the volume of the design event to be captured and treated. Commonly, the pretreatment unit provides the additional live storage volume. In effect, while a longer specified drawdown time deceases the filter area and the volume above it, it does not decrease the total storage volume that is required.
Furthermore, Equation 1 fails to consider the combined effect of the interevent time and the choice of drawdown time on the potential for some stormwater to be present in the treatment system when the next storm arrives. This aspect was discussed in the first article of this series with respect to extended detention basins. Similarly, the interrelated effects of interevent time and drawdown time should be considered when specifying sizing criteria for the total live volume of the combined pretreatment-filter system. One manual recognizes this point (CASQA 2003). A fine-media filter can be viewed as an extended detention basin with a system of underdrains rather than orifices. Counterbalancing this relationship is the recognition that the actual hydraulic conductivity is greater than the design hydraulic conductivity through much of the maintenance cycle. The only proper way to define the filter area and the live storage volume is through continuous simulation with a program that recognizes the gradual change in hydraulic conductivity as solids accumulate in the filter.
A complementary methodology sizes the area of the filter based on solids accumulation, thereby relating filter area to the desired maintenance cycle (Lenhart 2000, Minton 1995 and 2002, Urbonas 1999). There are two key elements to this approach: The first is the loading rate of sediment (TSS) to the filter, as pounds or kilograms per year. The second element is the amount of sediment that accumulates on the filter before the design hydraulic conductivity is reached, as pounds per square foot or kilograms per square meter.
Equations 2 and 3 express the methodology. Equation 2 considers concentration. The analysis can also be based on annual unit loading—that is, pounds per acre or kilograms per hectare of drainage area per year (Minton 2002).
The solids-accumulation method explicitly considers the desired maintenance frequency. The area of the filter is calculated with both the flow and solids-accumulation methods. The larger area is selected. Alternatively, the selection of the filter area is based on Equation 1. Equations 2 and 3 are then used to specify the maintenance frequency for the particular development. A regulatory agency could assume the same sediment concentration or unit loading for each type of land use. It could then specify maintenance frequency as a function of the maximum water depth over the filter. This relationship recognizes that as the available water depth increases, the filter area decreases, and in turn the maintenance frequency increases. Finally, the two methods can be contrasted to determine the appropriate values for the drawdown time, hydraulic conductivity, and media depth for a particular climatic region.
Published data suggest that design hydraulic conductivity is reached at approximately 0.25 to 0.50 pounds of sediment per square foot of filter area (1.2 to 2.4 kg/m2) (Clark and Pitt 1999, Keblin et al. 1996, King County 1996, Shapiro 1999, Urbonas 1999). More field studies are needed.
Final Observations
Maintenance costs may be reduced, while performance is enhanced, by placing a suitable geofabric on the sand surface. The geofabric may extend the length of the maintenance cycle (Graham et al. 1994). Its replacement after clogging with most of the removed sediment may ease maintenance time and therefore costs. This approach is suggested for small filters, particularly if the media depth is reduced to between 6 and 12 inches.
Field inspection during construction is particularly important with fine-media filters. Care must be taken in the selection and protection of fine media when located onsite, and before placement in the filter bed. The hydraulic condition of the filter must be checked if the filter was used during construction of the development.
It is likely that fine-media filters are more attractive in regions with low annual rainfall. There are two reasons: The first is the unattractiveness of treatment systems that rely on water to sustain their performance, such as wet basins and flow-through swales. Secondly, in drier climatic regions the amount of sediment reaching the filter is less over the typical year than in wet climatic regions. Hence, the maintenance frequency will likely be less than in wetter climates.
Anecdotal information suggests that vigorous surface vegetation extends the maintenance cycle. The author knows of one sand filter in the Pacific Northwest in which a thick growth of grass was intentionally promoted. Although over five years old, the filter surface has never had to be cleaned. Other filters in the same area have required semi-annual maintenance. Sediment likely accumulates in the grass rather than the surface of the sand. Vegetation is known to enhance the infiltration capacities; more studies are needed on this question. Promotion of surface vegetation may be particularly attractive in wetter climates where filters experience a higher annual solids loading and longevity of agricultural soils (Minton 2002).
Some manuals suggest that fine-media filters not be used for drainage areas in excess of 10 acres (Table 1). However, sand filters are serving much larger areas without issue. The author is familiar with a dual-filter system that serves a residential area of 250 acres.
Washout of accumulated fine particles that are toxic has been observed under laboratory conditions (Clark and Pitt 1999). Whether this occurs under field conditions is not known.
Advertisement
Summary
The effect of the choices of key design criteria, their interrelationship, and their combined effect on filter area and maintenance frequency must be considered. There is no one appropriate value for the hydraulic conductivity. The value that is selected should be based on the desired frequency of maintenance, with consideration to the effect on filter surface area. Where the objective is the removal of sediment and attached pollutants, a media depth of 12 inches and possibly as little as 6 inches is likely sufficient in most situations. While performance will be less with the shallower bed depth, the effluent concentrations will be similar to that produced by other treatment systems such as basins and swales. A shallower bed depth allows for a smaller filter surface area. A shallower bed depth also reduces the elevation drop required of the design. However, a smaller filter area, whether from a reduction in the bed thickness or by the greater available elevation drop, may result in unacceptable frequency of maintenance, particularly in wetter climates. The selection of the drawdown time may differ depending on the pollutant-removal objective, but should take into consideration interevent time particularly in wet climates with frequent storms. It would be prudent to use both the flow and solids-accumulation methods to determine the filter area. More studies are needed on the relationship between solids accumulation and hydraulic conductivity, and the effect of surface vegetation on the maintenance cycle.
Author's Bio: Gary R. Minton, Ph.D., P.E., is an independent consultant on stormwater treatment with Resource Planning Associates. He is the author of the book Stormwater Treatment: Biological, Chemical, and Engineering Principles.
January-February 2005
Revisiting Design Criteria for Stormwater Treatment Systems, Part 2
Fine-Media Filters
This is the second of a four-part series examining design criteria for stormwater treatment systems. In the first of this series (Stormwater, November/December 2004) basins were the focus. Now we shift our focus to fine-media filters, which are filters that use fine-grained media similar to those used for potable water treatment. The concept of using fine media for stormwater treatment was first adopted by the City of Austin, TX, in the late 1970s. The most commonly used medium is sand, although other media are discussed in this article. Not considered in this article are manufactured stormwater filters, which commonly use coarser media.
Since communities in the United States first began requiring post-development treatment of stormwater from new developments 25 years ago, local, regional, and state governments have published many manuals and handbooks identifying acceptable treatment systems (structural best management practices) and design criteria. As noted in part 1 of this series, few of the criteria were initially supported by laboratory or field research, and engineers relied on their best professional judgment in choosing design criteria. In light of the many studies completed in the past 20 years, it is time to reexamine some of these criteria.
In addition, treatment strategy is evolving from a focus on the general removal of pollutants using total suspended solids (TSS) as the surrogate for all pollutants to a more recent trend to consider specific pollutants. Some states now emphasize removal of total phosphorus, total nitrogen, and dissolved metals in certain situations. A particular design criterion can vary depending on which pollutant is targeted.
Table 1 provides design criteria for fine-media filters from representative manuals and handbooks. Sand filters are a commonly used treatment system in Texas and the Chesapeake Bay area, and are particularly attractive for the removal of sediment (TSS or suspended sediment concentration) and attached pollutants. They can be expected to give consistent performance irrespective of the size of the storm and flow rate, unlike many other systems such as swales and vortex separators. They also may provide relatively consistent effluent quality with respect to TSS and attached pollutants, irrespective of the variation in influent concentration (Caltrans 2004). Despite these advantages, sand filters are rarely used outside the two areas mentioned above. This likely occurs for two reasons: First, sand filters can be fairly large, particularly if the available head is modest given site topographic constraints. The second concern is maintenance. This article discusses how the selection of design criteria directly affects surface area and maintenance frequency. Alternative criteria are suggested that may result in wider use of fine-media filters.
Pretreatment
It is common practice to pretreat stormwater before it enters the filter bed. The objective is to reduce solids loading on the filter bed, thereby extending the time between necessary maintenance events. Subsurface filters commonly use a wet vault chamber separated from the filter bed by a common wall. The most common pretreatment method for surface filters is an extended detention basin. Wetponds are used where in some cases the filter is actually the outlet berm of the pond and not a separate unit. Flow-through grass swales have also been employed. However, wetponds and grass swales may not be the most appropriate choice for pretreatment given the potential for clogging of the filter bed—from algae in the case of ponds and eroded sediment in the case of swales. However, no problems have been published in this regard.
Media Type and Size Distribution
As previously mentioned, sand is the most commonly used filter medium. Mixtures of sand and a second medium are used, although infrequently. The objective is the removal of dissolved pollutants. Added media in full-scale facilities have included peat (for removal of metals), activated carbon (organics), calcite (phosphorus), and iron filings (phosphorus). There has been laboratory- or pilot-scale experimentation with dolomite (phosphorus) and soybean hulls (metals). An alternative is to coat the sand with an oxide of iron, manganese, or aluminum to remove dissolved metals.
The size distribution commonly specified today is ASTM (American Society of Testing and Materials) C33. This is the specification for fine aggregate for use in concrete. Its ready availability reduces the cost. It is, coincidently, fairly similar to the specification of sand for water treatment. Early manuals had a simple specification of "0.20 to 0.40 and fines are acceptable." It is believed that this specification has caused premature clogging of sand beds, sometimes through the entire depth of the bed.
Media Depth
Initially, the City of Austin specified a depth of 36 inches, mimicking the depth commonly used in potable water treatment. It was soon changed to 18 inches, likely because of the recognition that almost all of the removal of TSS and attached pollutants occurs on or near the surface of the bed. Substantial media depths are used in potable water treatment given the need to consistently and reliably meet a very strict potable water standard. Such a standard, less than one unit of turbidity, is not required of stormwater treatment.
The relevance of media depth relates to the area required of the filter to pass the design event. The greater the depth of the media, the greater the surface area of filter required to meet the specified drawdown time as shown in Table 2. Eighteen inches is taken as the standard with a unit area of one. Reducing the media depth from 18 to 6 inches reduces the required area of the filter by about 50%. Conversely, a filter depth of 36 inches increases the filter area by about 30%. How the relationships in Table 1 were produced is presented later in this article.
Sand filters with 18 inches of media appear to perform more effectively than other stormwater treatment systems, in particular extended detention basins and swales. Therefore, some may consider it acceptable to reduce the depth of the media in a sand filter even if performance degrades somewhat. One study (Amini 1996) found that increasing the media depth from 6 to 12 inches, the maximum depth evaluated in this study, increased the removal efficiency of TSS by only 5%. A study of bacteria removal found little improvement with media depths above 12 inches (Bellamy et al. 1985). In a recently completed study (Caltrans 2004), several sand filters with 18 inches of media performed substantially better than a single filter that had only 12 inches, with mean effluent concentrations of about 6 and 12 mg/L, respectively. However, 12 mg/L is acceptable. While the results of these studies are inconsistent, they do indicate that shallower bed depths are acceptable in most situations, certainly 12 inches and possibly 6 inches.
A depth of 6 to 12 inches is likely impractical with large filters where mechanical methods are used to maintain the filter bed. Heavy equipment could damage the underdrain system. But it may be acceptable for small filters, less than a few hundred square feet, where cleaning by hand occurs. Also, more frequent cleaning may be feasible with small filters. Alternatively, where 18 inches is specified, the depth of the design storm could be reduced, commensurate with the higher efficiency of sand filters over other treatment technologies (Barrett 1999, King County 1998).
A shallow filter bed is likely acceptable where the objective is the removal of TSS and attached pollutants. It may not be appropriate if the objective is the removal of dissolved species. Removal of dissolved pollutants is discussed in the last of this series of articles.
Hydraulic Conductivity
Sometimes referred to as the coefficient of permeability or the filter factor, hydraulic conductivity describes the ability of water to pass through media. Its value is inherent to the particular medium, its size distribution, and sediment accumulation. Hydraulic conductivity is not to be confused with the infiltration rate, which is the actual rate of flow through the filter. Confusion occurs as both have the same units (feet or inches per hour or per day). The filtration rate increases with the depth, or head, of water above the filter bed, whereas the hydraulic conductivity is unaffected by head. Hydraulic conductivity, however, decreases with the accumulation of sediment.
Specified design values have ranged from 1.8 to 3.5 feet per day. The effect of its selection is shown in Table 3. The more conservative value of 1.8 feet per day increases the filter area by about 90%, a significant increase.
What does the selected hydraulic conductivity represent? The hydraulic conductivity of clean sand is on the order of 2 to 5 feet per hour. The values in Table 3 are 5%–10% of clean sand. These values therefore represent a filter almost totally clogged: the time to clean the filter. If the filter is not cleaned promptly, the management goal of treating, for example, 90% of the stormwater over time, is not met. The greater the selected hydraulic conductivity the smaller the filter area, but the more frequent the maintenance. The value chosen is arbitrary within a reasonable range, and represents the tradeoff between filter area and the frequency of maintenance. In effect, its selection represents a tradeoff between initial construction costs and long-term maintenance costs.
Elevation Drop
A major limitation in the use of fine-media filters is the available elevation drop—that is, the difference in the elevations of the development and of the public drainage system to which the development discharges. Table 4 illustrates the effect of the available maximum water depth over the top of the filter, called the head. As the head increases, the surface area of the filter decreases. Hence, the feasibility of fine-media filters increases with increasing head. However, the decrease in filter surface area corresponds with the required frequency of cleaning. It should also be noted that reducing the bed depth also reduces the elevation drop required of the site, and therefore the potential feasibility of a fine-media filter.
Drawdown Time
The design drawdown time affects the size of the filter. Therefore, its specification is important. Table 1 presents a significant range of 24 to 48 hours. Drawdown time affects filter area by its definition of Q, the average filtration rate during the design storm. The greater the allowed drawdown time, the lower the filtration rate and therefore the smaller the filter area.
Avoiding anaerobiosis in the filter bed is the common reason given as to why a maximum drawdown time must be specified. Bacteria attached to the filter media consume dissolved oxygen as they use organic matter and ammonia in the incoming stormwater. The concern is that under anaerobic conditions, dissolved phosphorus and metals previously removed may be released. An additional benefit from specifying a drawdown time may be to inhibit the excessive growth of bacteria that might clog the filter. Drying of the filter bed between storms inhibits growth.
The relationship between drawdown time and either anaerobiosis or excessive bacteria growth has not been established. Nor has the significance of pollutant release been defined. Whether release occurs depends on the mechanism of removal. A possible mechanism of dissolved phosphorus removal is sorption/precipitation to ferric oxide present on the surface of the sand. Under anaerobic conditions, ferric iron changes to ferrous, disassociating the complex and releasing the pollutants. One study found a constant anaerobic condition in the bottom of the filter (Shapiro 1999). The conditions of the study are likely the most extreme that will be faced. The filter had a media depth of 36 inches and a typical drawdown time of 72 hours. Furthermore, it was located in a region with long storms and short interevent times, conducive to creating anaerobic conditions. Finally, iron filings were added to the sand to promote the removal of dissolved phosphorus. The transformation of the filings under anaerobic conditions appears to have caused sand particles to bind. The filter was found to be constantly anaerobic in the bottom 24 inches. However, despite this condition the filter removed on the order of 50% of the dissolved phosphorus as well as dissolved metals. All of the removal was likely occurring in the upper 12 inches of media that remained aerobic.
If nitrogen removal is the objective, a temporary anaerobic condition is desirable. Bacteria change ammonia to nitrate in the presence of dissolved oxygen, but the net reduction of nitrogen is zero. Other bacteria under anaerobic conditions change the nitrate to nitrogen gas. The ideal operation is for the upper area of the filter to be aerobic and the lower area to be anaerobic.
Specification of the drawdown time is therefore a function of the pollutant-removal objective, the interevent time between storms, and the time needed for the filter to dry. If the objective includes dissolved phosphorus and/or metals, a low drawdown time, perhaps on the order of 24 hours, is appropriate. If not, 72 hours may be satisfactory. A long drawdown time may also be desirable if nitrogen removal is the management objective.
An alternative to a "tight" drawdown time may be to incorporate aluminum into the sand. Aluminum oxides may be an effective remover of dissolved phosphorus and/or metals (Minton 2002). Unlike ferric oxide complexes, aluminum oxide/phosphorus/metal complexes will not disassociate under anaerobic conditions.
Determining the Filter Bed Area and Live Storage Volume
Manuals employ the same equation (Equation 1) for sizing the filter bed area, although the specific form of the equation varies with the units used.
The average filtration rate is equal to the runoff volume of the design event divided by the drawdown time at brimful. For example, assume a design runoff depth of 1 inch. The volume of runoff is the multiple of 1 inch, the drainage area, and the runoff coefficient. If a 48-hour drawdown time is selected, the average filtration rate, Q, is half that if the drawdown time is 24 hours. Hence, Equation 1 states that a drawdown time of 48 hours requires a filter area that is half that for a 24-hour drawdown time.
Previous observations in this article about the relationship between bed depth, hydraulic conductivity, drawdown time, available water depth, and filter surface area are understood with the examination of Equation 1. Hence, the selection of the values for the key design criteria must be considered as a whole, paying particular attention to the effect of these decisions on maintenance frequency.
It is important to recognize that the available live storage volume above the filter itself is usually less than the volume of the design event to be captured and treated. Commonly, the pretreatment unit provides the additional live storage volume. In effect, while a longer specified drawdown time deceases the filter area and the volume above it, it does not decrease the total storage volume that is required.
Furthermore, Equation 1 fails to consider the combined effect of the interevent time and the choice of drawdown time on the potential for some stormwater to be present in the treatment system when the next storm arrives. This aspect was discussed in the first article of this series with respect to extended detention basins. Similarly, the interrelated effects of interevent time and drawdown time should be considered when specifying sizing criteria for the total live volume of the combined pretreatment-filter system. One manual recognizes this point (CASQA 2003). A fine-media filter can be viewed as an extended detention basin with a system of underdrains rather than orifices. Counterbalancing this relationship is the recognition that the actual hydraulic conductivity is greater than the design hydraulic conductivity through much of the maintenance cycle. The only proper way to define the filter area and the live storage volume is through continuous simulation with a program that recognizes the gradual change in hydraulic conductivity as solids accumulate in the filter.
A complementary methodology sizes the area of the filter based on solids accumulation, thereby relating filter area to the desired maintenance cycle (Lenhart 2000, Minton 1995 and 2002, Urbonas 1999). There are two key elements to this approach: The first is the loading rate of sediment (TSS) to the filter, as pounds or kilograms per year. The second element is the amount of sediment that accumulates on the filter before the design hydraulic conductivity is reached, as pounds per square foot or kilograms per square meter.
Equations 2 and 3 express the methodology. Equation 2 considers concentration. The analysis can also be based on annual unit loading—that is, pounds per acre or kilograms per hectare of drainage area per year (Minton 2002).
The solids-accumulation method explicitly considers the desired maintenance frequency. The area of the filter is calculated with both the flow and solids-accumulation methods. The larger area is selected. Alternatively, the selection of the filter area is based on Equation 1. Equations 2 and 3 are then used to specify the maintenance frequency for the particular development. A regulatory agency could assume the same sediment concentration or unit loading for each type of land use. It could then specify maintenance frequency as a function of the maximum water depth over the filter. This relationship recognizes that as the available water depth increases, the filter area decreases, and in turn the maintenance frequency increases. Finally, the two methods can be contrasted to determine the appropriate values for the drawdown time, hydraulic conductivity, and media depth for a particular climatic region.
Published data suggest that design hydraulic conductivity is reached at approximately 0.25 to 0.50 pounds of sediment per square foot of filter area (1.2 to 2.4 kg/m2) (Clark and Pitt 1999, Keblin et al. 1996, King County 1996, Shapiro 1999, Urbonas 1999). More field studies are needed.
Final Observations
Maintenance costs may be reduced, while performance is enhanced, by placing a suitable geofabric on the sand surface. The geofabric may extend the length of the maintenance cycle (Graham et al. 1994). Its replacement after clogging with most of the removed sediment may ease maintenance time and therefore costs. This approach is suggested for small filters, particularly if the media depth is reduced to between 6 and 12 inches.
Field inspection during construction is particularly important with fine-media filters. Care must be taken in the selection and protection of fine media when located onsite, and before placement in the filter bed. The hydraulic condition of the filter must be checked if the filter was used during construction of the development.
It is likely that fine-media filters are more attractive in regions with low annual rainfall. There are two reasons: The first is the unattractiveness of treatment systems that rely on water to sustain their performance, such as wet basins and flow-through swales. Secondly, in drier climatic regions the amount of sediment reaching the filter is less over the typical year than in wet climatic regions. Hence, the maintenance frequency will likely be less than in wetter climates.
Anecdotal information suggests that vigorous surface vegetation extends the maintenance cycle. The author knows of one sand filter in the Pacific Northwest in which a thick growth of grass was intentionally promoted. Although over five years old, the filter surface has never had to be cleaned. Other filters in the same area have required semi-annual maintenance. Sediment likely accumulates in the grass rather than the surface of the sand. Vegetation is known to enhance the infiltration capacities; more studies are needed on this question. Promotion of surface vegetation may be particularly attractive in wetter climates where filters experience a higher annual solids loading and longevity of agricultural soils (Minton 2002).
Some manuals suggest that fine-media filters not be used for drainage areas in excess of 10 acres (Table 1). However, sand filters are serving much larger areas without issue. The author is familiar with a dual-filter system that serves a residential area of 250 acres.
Washout of accumulated fine particles that are toxic has been observed under laboratory conditions (Clark and Pitt 1999). Whether this occurs under field conditions is not known.
Summary
The effect of the choices of key design criteria, their interrelationship, and their combined effect on filter area and maintenance frequency must be considered. There is no one appropriate value for the hydraulic conductivity. The value that is selected should be based on the desired frequency of maintenance, with consideration to the effect on filter surface area. Where the objective is the removal of sediment and attached pollutants, a media depth of 12 inches and possibly as little as 6 inches is likely sufficient in most situations. While performance will be less with the shallower bed depth, the effluent concentrations will be similar to that produced by other treatment systems such as basins and swales. A shallower bed depth allows for a smaller filter surface area. A shallower bed depth also reduces the elevation drop required of the design. However, a smaller filter area, whether from a reduction in the bed thickness or by the greater available elevation drop, may result in unacceptable frequency of maintenance, particularly in wetter climates. The selection of the drawdown time may differ depending on the pollutant-removal objective, but should take into consideration interevent time particularly in wet climates with frequent storms. It would be prudent to use both the flow and solids-accumulation methods to determine the filter area. More studies are needed on the relationship between solids accumulation and hydraulic conductivity, and the effect of surface vegetation on the maintenance cycle.